Professional Documents
Culture Documents
9 Hma Key
9 Hma Key
ANSYS, ANSYS Workbench, AUTODYN, CFX, FLUENT and any and all ANSYS, Inc. brand, product, service and feature
names, logos and slogans are registered trademarks or trademarks of ANSYS, Inc. or its subsidiaries located in the
United States or other countries. ICEM CFD is a trademark used by ANSYS, Inc. under license. CFX is a trademark
of Sony Corporation in Japan. All other brand, product, service and feature names or trademarks are the property
of their respective owners. FLEXlm and FLEXnet are trademarks of Flexera Software LLC.
Disclaimer Notice
THIS ANSYS SOFTWARE PRODUCT AND PROGRAM DOCUMENTATION INCLUDE TRADE SECRETS AND ARE CONFID-
ENTIAL AND PROPRIETARY PRODUCTS OF ANSYS, INC., ITS SUBSIDIARIES, OR LICENSORS. The software products
and documentation are furnished by ANSYS, Inc., its subsidiaries, or affiliates under a software license agreement
that contains provisions concerning non-disclosure, copying, length and nature of use, compliance with exporting
laws, warranties, disclaimers, limitations of liability, and remedies, and other provisions. The software products
and documentation may be used, disclosed, transferred, or copied only in accordance with the terms and conditions
of that software license agreement.
ANSYS, Inc. and ANSYS Europe, Ltd. are UL registered ISO 9001: 2015 companies.
For U.S. Government users, except as specifically granted by the ANSYS, Inc. software license agreement, the use,
duplication, or disclosure by the United States Government is subject to restrictions stated in the ANSYS, Inc.
software license agreement and FAR 12.212 (for non-DOD licenses).
Third-Party Software
See the legal information in the product help files for the complete Legal Notice for ANSYS proprietary software
and third-party software. If you are unable to access the Legal Notice, contact ANSYS, Inc.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. iii
Technology Demonstration Guide
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
iv of ANSYS, Inc. and its subsidiaries and affiliates.
Technology Demonstration Guide
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. v
Technology Demonstration Guide
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
vi of ANSYS, Inc. and its subsidiaries and affiliates.
Technology Demonstration Guide
13.6.5. Nonlinear Prestressed Full-Harmonic Cyclic Symmetry Analysis with Linear Perturbation ..... 185
13.6.6. Nonlinear Prestressed Mode-Superposition Harmonic Cyclic Symmetry Analysis with Linear
Perturbation ................................................................................................................................. 186
13.7. Results and Discussion ................................................................................................................ 187
13.7.1. Performance Benefits of Cyclic Symmetry Analysis .............................................................. 198
13.7.2. Performance Benefits of the Variational Technology (VT) Solver .......................................... 201
13.8. Recommendations ...................................................................................................................... 202
13.9. References .................................................................................................................................. 203
13.10. Input Files ................................................................................................................................. 203
14. Rotordynamics of a Shaft Assembly Based on a Representative Model of Nelson-Vaugh Rotor ..... 205
14.1. Introduction ............................................................................................................................... 205
14.2. Problem Description ................................................................................................................... 206
14.3. Modeling .................................................................................................................................... 207
14.3.1. 3-D Modeling of Flexible Rotor Component ........................................................................ 207
14.3.2. Axisymmetric Modeling of the Flexible Rotor Component from 3-D Geometry ..................... 208
14.3.3. Disk and Bearing Modeling ................................................................................................. 210
14.3.3.1. Disk Modeling ........................................................................................................... 211
14.3.3.2. Modeling Bearings .................................................................................................... 213
14.4. Material Properties ...................................................................................................................... 214
14.5. Boundary Conditions and Loading .............................................................................................. 215
14.6. Analysis and Solution Controls .................................................................................................... 216
14.6.1. Modal Analysis ................................................................................................................... 216
14.6.1.1. Modal Analysis without Gyroscopic Effects ................................................................. 216
14.6.1.2. Modal Analysis with Gyroscopic Effects ...................................................................... 217
14.6.2. Campbell Diagram Analysis ................................................................................................ 217
14.6.3. Unbalance Response Analysis ............................................................................................. 218
14.7. Results and Discussion ................................................................................................................ 219
14.7.1. Performance Benefits of the 2-D Axisymmetric Model ......................................................... 223
14.8. Recommendations ...................................................................................................................... 224
14.9. References .................................................................................................................................. 225
14.10. Input Files ................................................................................................................................. 225
15. Calibrating and Validating a Hyperelastic Constitutive Model ........................................................ 227
15.1. Introduction ............................................................................................................................... 227
15.2. Problem Description ................................................................................................................... 227
15.3. Material Properties ...................................................................................................................... 227
15.3.1. Calibration Experiments ..................................................................................................... 228
15.3.2. Validation Experiment ........................................................................................................ 229
15.4. Analysis and Solution Controls .................................................................................................... 229
15.4.1. Calibrating Parameters ....................................................................................................... 229
15.4.2. Validating Parameters ......................................................................................................... 230
15.5. Results and Discussion ................................................................................................................ 231
15.5.1. Calibration Results .............................................................................................................. 231
15.5.2. Validation Results ............................................................................................................... 235
15.6. Recommendations ...................................................................................................................... 236
15.7. References .................................................................................................................................. 236
15.8. Input Files ................................................................................................................................... 236
16. Evaluation of Mixed-Mode Stress Intensity Factors and T-stress for 3-D Surface Flaws .................. 237
16.1. Introduction ............................................................................................................................... 237
16.2. Problem Description ................................................................................................................... 238
16.2.1. Rectangular Block with a Semicircular Surface Flaw ............................................................. 238
16.2.2. X-Joint Pipe with Warped Surface Flaw ................................................................................ 239
16.3. Modeling .................................................................................................................................... 240
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. vii
Technology Demonstration Guide
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
viii of ANSYS, Inc. and its subsidiaries and affiliates.
Technology Demonstration Guide
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. ix
Technology Demonstration Guide
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
x of ANSYS, Inc. and its subsidiaries and affiliates.
Technology Demonstration Guide
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. xi
Technology Demonstration Guide
27.6.3. Solution Items Mapped from Original Mesh to New Mesh ................................................... 415
27.6.4. Analysis Resumes Using the New Mesh ............................................................................... 415
27.7. Results and Discussion ................................................................................................................ 416
27.7.1. Deformation Animations .................................................................................................... 420
27.8. Recommendations ...................................................................................................................... 421
27.9. Input Files ................................................................................................................................... 422
28. Friction Stir Welding (FSW) Simulation ............................................................................................. 423
28.1. Introduction ............................................................................................................................... 423
28.2. Problem Description ................................................................................................................... 424
28.3. Modeling .................................................................................................................................... 425
28.3.1. Workpiece and Tool Modeling ............................................................................................. 425
28.3.2. Contact Modeling ............................................................................................................... 426
28.3.2.1. Contact Pair Between the Plates ................................................................................. 426
28.3.2.2. Contact Pair Between Tool and Workpiece .................................................................. 427
28.3.2.3. Rigid Surface Constraint ............................................................................................ 428
28.4. Material Properties ...................................................................................................................... 429
28.5. Boundary Conditions and Loading .............................................................................................. 430
28.5.1. Thermal Boundary Conditions ............................................................................................ 430
28.5.2. Mechanical Boundary Conditions ....................................................................................... 431
28.5.3. Loading .............................................................................................................................. 431
28.6. Analysis and Solution Controls .................................................................................................... 432
28.7. Results and Discussion ................................................................................................................ 433
28.7.1. Deformation and Stresses ................................................................................................... 434
28.7.2. Temperature Results .......................................................................................................... 436
28.7.3. Welding Results .................................................................................................................. 440
28.7.4. Heat Generation ................................................................................................................. 441
28.8. Recommendations ...................................................................................................................... 444
28.9. References .................................................................................................................................. 444
28.10. Input Files ................................................................................................................................. 445
29. Rocket Nozzle Extension Simulation: Operation .............................................................................. 447
29.1. Introduction ............................................................................................................................... 447
29.2. Problem Description ................................................................................................................... 449
29.3. Modeling .................................................................................................................................... 449
29.4. Material Properties ...................................................................................................................... 452
29.5. Boundary Conditions and Loading .............................................................................................. 453
29.6. Analysis and Solution Controls .................................................................................................... 454
29.7. Results and Discussion ................................................................................................................ 454
29.8. Recommendations ...................................................................................................................... 460
29.9. Input Files ................................................................................................................................... 461
30. Acoustic Analysis of a Small Speaker System ................................................................................... 463
30.1. Introduction ............................................................................................................................... 463
30.2. Problem Description ................................................................................................................... 464
30.3. Modeling .................................................................................................................................... 465
30.4. Material Properties ...................................................................................................................... 466
30.5. Boundary Conditions and Loading .............................................................................................. 466
30.5.1. Structural Loads and Boundary Conditions .......................................................................... 466
30.5.2. Acoustic Loads and Boundary Conditions ........................................................................... 466
30.6. Analysis and Solution Controls .................................................................................................... 467
30.7. Results and Discussion ................................................................................................................ 467
30.7.1. Structural-Only Modes ........................................................................................................ 467
30.7.2. Plotting Sound Pressure Levels ........................................................................................... 467
30.7.3. Plotting Velocities ............................................................................................................... 471
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
xii of ANSYS, Inc. and its subsidiaries and affiliates.
Technology Demonstration Guide
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. xiii
Technology Demonstration Guide
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
xiv of ANSYS, Inc. and its subsidiaries and affiliates.
Technology Demonstration Guide
37.3.1.1. Model the True Threaded Bolt with Cover Plate and Base Plate .................................... 574
37.3.1.2. Model the Simplified Bolt with Cover Plate and Base Plate .......................................... 576
37.3.2. Model the Pretension Section ............................................................................................. 578
37.3.3. Model the Contact Pairs ...................................................................................................... 578
37.3.3.1. Contact in the Thread Region ..................................................................................... 579
37.3.3.1.1. Contact in the Thread Region for the True Thread Simulation Method ................. 579
37.3.3.1.2. Contact in the Thread Region for Bolt Section Simulation Method ...................... 579
37.3.3.1.3. Contact in the Thread Region for the MPC Simulation Method ........................... 581
37.3.3.2. Contact Between the Bolt Head and the Cover Plate ................................................... 581
37.3.3.3. Contact Between the Cover Plate and the Base Plate ................................................... 581
37.4. Material Properties ...................................................................................................................... 582
37.5. Boundary Conditions and Loading .............................................................................................. 583
37.6. Analysis and Solution Controls .................................................................................................... 583
37.7. Results and Discussion ................................................................................................................ 583
37.7.1. Results for 2-D Models ........................................................................................................ 584
37.7.2. Results for 3-D Models ........................................................................................................ 586
37.7.3. Comparison of Computation Time ...................................................................................... 589
37.8. Recommendations ...................................................................................................................... 590
37.9. Input Files ................................................................................................................................... 590
38. Large-Deformation Neo-Hookean Analysis (via UserMat Subroutine) ............................................ 591
38.1. Introduction ............................................................................................................................... 591
38.2. Problem Description ................................................................................................................... 591
38.3. Modeling .................................................................................................................................... 592
38.3.1. Neo-Hookean Model .......................................................................................................... 592
38.3.1.1. Stress ........................................................................................................................ 593
38.3.1.2. Tangent Stiffness ....................................................................................................... 593
38.3.2. Co-rotational Frame ............................................................................................................ 594
38.3.2.1. Polar Decomposition ................................................................................................. 594
38.3.2.2. Stress ........................................................................................................................ 595
38.3.2.3. Tangent Stiffness ....................................................................................................... 595
38.3.3. Voigt Notation .................................................................................................................... 595
38.3.3.1. Stress ........................................................................................................................ 596
38.3.3.2. Tangent Stiffness ....................................................................................................... 597
38.3.3.3. Change of Basis ......................................................................................................... 598
38.4. Material Properties ...................................................................................................................... 599
38.5. Boundary Conditions and Loading .............................................................................................. 599
38.6. Analysis and Solution Controls .................................................................................................... 599
38.7. Results and Discussion ................................................................................................................ 600
38.8. Recommendations ...................................................................................................................... 601
38.9. References .................................................................................................................................. 601
38.10. Input Files ................................................................................................................................. 602
39. Wire Bonding Ultrasonic Transducer ................................................................................................. 603
39.1. Introduction ............................................................................................................................... 603
39.2. Problem Description ................................................................................................................... 603
39.3. Modeling .................................................................................................................................... 605
39.4. Material Properties ...................................................................................................................... 607
39.5. Boundary Conditions and Loading .............................................................................................. 608
39.5.1. Structural Boundary Condition ........................................................................................... 608
39.5.2. Voltage Coupling and Boundary Condition ......................................................................... 608
39.6. Analysis and Solution Controls .................................................................................................... 608
39.6.1. Prestressed Modal Analysis with Linear Perturbation Method .............................................. 609
39.6.2. Prestressed Full Harmonic Response Analysis ...................................................................... 609
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. xv
Technology Demonstration Guide
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
xvi of ANSYS, Inc. and its subsidiaries and affiliates.
Technology Demonstration Guide
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. xvii
Technology Demonstration Guide
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
xviii of ANSYS, Inc. and its subsidiaries and affiliates.
Technology Demonstration Guide
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. xix
Technology Demonstration Guide
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
xx of ANSYS, Inc. and its subsidiaries and affiliates.
List of Figures
1.1. Brake Disc-Pad Assembly ........................................................................................................................ 2
1.2. Contact Pair Definition ........................................................................................................................... 3
1.3. Meshed Brake Disc-Pad Assembly ........................................................................................................... 4
1.4. Boundary Conditions (Displacement Constraints and Pressure Loading) .................................................. 5
1.5. Mode Shape for Unstable Mode (Mode 21) Obtained from the Linear Non-prestressed Modal Solution .... 9
1.6. Mode Shape for Unstable Mode (Mode 22) Obtained from the Linear Non-prestressed Modal Solution ... 10
1.7. Effect of Friction Coefficient on Unstable Modes .................................................................................... 12
2.1. Schematic of a Static Elastomeric Seal Assembly .................................................................................... 16
2.2. Schematic of Finite-Element Static Elastomeric Seal Assembly with Dimensions ..................................... 16
2.3. Progression of Seal Deformation ........................................................................................................... 17
2.4. Deformed Mesh Plot of Seal After Third Load Step Without Rezoning .................................................... 23
2.5. Deformation Profile at Load Step 1 -- Substep 20 (First Rezoning) ........................................................... 27
2.6. Deformation Profile at Load Step 1 -- Substep 50 (Second Rezoning) ...................................................... 28
2.7. Deformation Profile at Load Step 2 -- Substep 10 (Third Rezoning) ......................................................... 29
2.8. Final Mesh (After Third Load Step) ......................................................................................................... 31
2.9. Contact Pressure-Distribution Trends ..................................................................................................... 32
2.10. Seal Hydrostatic Pressures ................................................................................................................... 35
2.11. Seal σYY Profile .................................................................................................................................... 37
2.12. Seal σYY Plot Animation ....................................................................................................................... 39
2.13. Extrusion σYY Profile ............................................................................................................................ 40
2.14. Extrusion σXY Profile ............................................................................................................................ 40
2.15. Mesh Gradation Created with Nested Element Splitting During Vertical Rezoning ................................. 42
3.1. 3-D View of Sealing System ................................................................................................................... 46
3.2. Sealing System and Finite Element Model .............................................................................................. 47
3.3. Sealing System Contact Pairs ................................................................................................................. 49
3.4. Material Model Test Data ....................................................................................................................... 50
3.5. Sealing System with Applied Fluid Pressure to Contact Pairs .................................................................. 52
3.6. Sealing System Contact Pressure Following Housing Compression (First Load Step) ............................... 53
3.7. Sealing System fluid Penetration Pressure Distributions (During Second Load Step) ............................... 54
3.8. Animation of Fluid Penetration Pressure Distribution ............................................................................. 56
3.9. Sealing System Contact Pressure Distributions (During Second Load Step) ............................................. 57
3.10. Animation of Contact Pressure Distribution ......................................................................................... 59
3.11. Sealing System Von Mises Stress Contour (After Second Load Step) ...................................................... 60
3.12. Sealing System Equivalent Plastic Strain Contour (After Second Load Step) ........................................... 60
3.13. Sealing System Total Strain Energy and Stabilization Energy Time History ............................................. 61
4.1. Original Finite Element Model with Boundary Conditions and Loading .................................................. 65
4.2. Effective Plastic Strain and Deformed Mesh at Time of First Rezoning ..................................................... 68
4.3. Effective Plastic Strains and Deformed Mesh at Time of Second Rezoning ............................................... 69
4.4. New Mesh Read in During the First Rezoning ......................................................................................... 70
4.5. New Mesh Read in During the Second Rezoning .................................................................................... 71
4.6. Effective Plastic Strain Distribution After First Mapping .......................................................................... 72
4.7. Effective Plastic Strain After Second Mapping ........................................................................................ 73
4.8. Effective Strain Distribution and Deformed Shape at the Final Stage (Complete Die Fill) .......................... 75
4.9. Final Stage Animation ........................................................................................................................... 76
4.10. Von Mises Stress Distribution at the Final Stage .................................................................................... 77
4.11. Half Expanded Plot of Von Mises Stress at the Final Stage of Ring-Gear Forging .................................... 78
5.1. Geometry ............................................................................................................................................. 82
5.2. Symmetry Section ................................................................................................................................ 82
5.3. Meshed Geometry ................................................................................................................................ 83
5.4. Two Contact Pair Definitions (Initial Crack and CZM Area) ....................................................................... 83
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. xxi
Technology Demonstration Guide
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
xxii of ANSYS, Inc. and its subsidiaries and affiliates.
Technology Demonstration Guide
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. xxiii
Technology Demonstration Guide
14.10. Disk Modeled Using MASS21 Element (2-D Axisymmetric Model) ..................................................... 212
14.11. Bearings Modeled with COMBI214 Elements (3-D solid model) ......................................................... 213
14.12. Bonded Contact Pairs at Bearing Locations (3-D Solid Model) ........................................................... 214
14.13. Bearings Modeled with COMBI214 Elements (2-D Axisymmetric Model) ........................................... 214
14.14. Boundary Conditions (3-D Solid Model) ........................................................................................... 215
14.15. Campbell Diagram .......................................................................................................................... 219
14.16. Unbalance Response ....................................................................................................................... 221
14.17. Orbits Plot of 2-D Axisymmetric Model ............................................................................................ 223
14.18. CPU Time of 2-D Axisymmetric and 3-D Solid Models (Unbalance Response Analysis) ....................... 224
15.1. Hyperelastic Test Suite: Test Specimens ............................................................................................. 228
15.2. Hyperelastic Test Suite: Experimental Data ........................................................................................ 228
15.3. Tension-Torsion Test Specimen .......................................................................................................... 229
15.4. Tension-Torsion Experimental Data ................................................................................................... 229
15.5. Tension-Torsion Test Specimen Mesh ................................................................................................. 230
15.6. Comparison of the Data and Fits Over the Entire Range of Data .......................................................... 232
15.7. Parameters Fit to Experimental Data to About 100 Percent Strain ....................................................... 233
15.8. Comparison of the Data and Fits Showing Predictions Outside the Range of Fitted Data ..................... 234
15.9. Strain-Energy Density Contours of the Tension-Torsion Test ................................................................ 235
15.10. Comparison of Tension-Torsion Experiment to the Five-Parameter Mooney-Rivlin Model .................. 235
16.1. Rectangular Block Geometry with Semicircular Surface Flaw .............................................................. 238
16.2. Semicircular Crack and Torus Along the Crack Front .......................................................................... 239
16.3. X-Joint Pipe Full Model with Warped Surface Flaw at Welded Joint ..................................................... 239
16.4. Semi-elliptical Surface Crack Dimensions ........................................................................................... 240
16.5.Two-plane Symmetry of X-joint Pipe with Warped Surface Flaw Geometry .......................................... 240
16.6. Rectangular Block Model with Boundary Conditions and Loading Applied ......................................... 241
16.7. Sweep Mesh with SOLID186 Around Crack Front ................................................................................ 241
16.8. Two-plane Symmetry of X-joint Pipe with Warped Surface Flaw at Welded Joint ................................. 242
16.9. X-joint Pipe with Warped Surface Flaw at Welded Joint in Thickness Direction. .................................... 242
16.10. Sweep Mesh with SOLID186 Around the Crack Front ........................................................................ 243
16.11. Crack Tip Nodal Component: Rectangular Block Model ..................................................................... 244
16.12. Crack Tip Nodal Component: X-joint Pipe Model .............................................................................. 244
16.13. USUM Results (Rectangular Block) ................................................................................................... 245
16.14. Von Mises Stress Plot (Rectangular Block) ......................................................................................... 246
16.15. Normalized KI Results ..................................................................................................................... 246
16.16. Normalized T-Stress Results: Semicircular Surface Flaw .................................................................... 247
16.17. USUM Results (X-joint Pipe) ............................................................................................................. 247
16.18. Von Mises Stress Plot (X-joint Pipe) .................................................................................................. 248
16.19. Normalized SIF Results: Comparison with Chong Rhee[4] ................................................................. 248
16.20. Normalized T-Stress Results: X-Joint Pipe with Warped Flaw ............................................................. 249
17.1. Geometry and FE Model of a Metal Bar Impacting a Rigid Wall ........................................................... 252
17.2. Rigid Impact: Newmark Method with No Damping ............................................................................. 257
17.3. Rigid Impact: Newmark Method with Damping .................................................................................. 258
17.4. Rigid Impact: HHT Method with Damping ......................................................................................... 259
17.5. Elastic Impact: Newmark Method with No Damping ........................................................................... 260
17.6. Elastic Impact: Newmark Method with Damping ................................................................................ 261
17.7. Elastic Impact: HHT Method with Damping ........................................................................................ 262
17.8. Elastoplastic Impact: Newmark Method with No Damping ................................................................. 263
17.9. Elastoplastic Impact: Newmark Method with Damping ....................................................................... 264
17.10. Elastoplastic Impact: HHT Method with Damping ............................................................................. 264
17.11. Animation: Elastoplastic Impact Using the HHT Method with Damping ............................................. 265
18.1. FPD Viscoelastic (Glass) Veneer Geometry .......................................................................................... 270
18.2. FPD Ceramic Core Geometry (Inside the Veneer) ................................................................................ 271
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
xxiv of ANSYS, Inc. and its subsidiaries and affiliates.
Technology Demonstration Guide
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. xxv
Technology Demonstration Guide
22.9. Non-Master Nodes for Response Calculation in the Expansion Pass .................................................... 345
22.10. Nodal Responses Due to Harmonic Displacement Excitation at the Bottom of the Wheels ................ 349
22.11. Mode Shapes at the Undamped Natural Frequency of 187.22 Hz ...................................................... 349
22.12. Mode Shapes at the Undamped Natural Frequency of 223.63 Hz ...................................................... 350
22.13. Mode Shapes at the Undamped Natural Frequency of 237.48 Hz ...................................................... 350
23.1. Wind Turbine Blade Geometry and Structural Components ............................................................... 354
23.2. Representative Blade Cross-Section Areas at Various Locations ........................................................... 355
23.3. Line Diagram of Wind Turbine Blade Model ....................................................................................... 356
23.4. BEAM188 Blade Model with Fine Mesh and Linear Interpolation ........................................................ 357
23.5. BEAM188 Blade Model with Coarse Mesh and Cubic Interpolation ...................................................... 357
23.6. SHELL281 Reference Blade Model ...................................................................................................... 358
23.7. Boundary Conditions on the BEAM188 Blade Model .......................................................................... 359
23.8. Boundary Conditions on the Reference SHELL281 Blade Model .......................................................... 359
23.9. Comparison of Mode Shapes Predicted by BEAM188 and SHELL281 Models ....................................... 361
23.10. Averaged Bending-Stress Distribution (Corresponding to Third Mode Shape at Location x = 15m)
................................................................................................................................................................. 362
23.11. Nonaveraged Bending-Stress Distribution (Corresponding to Third Mode Shape at Location x = 15m)
................................................................................................................................................................. 362
24.1. 3-D Model of an Inflated Tire and a Road Surface ............................................................................... 366
24.2. 3-D Tire Model Using a Solid Element ................................................................................................ 367
24.3. Hydrostatic Fluid Element with Positive Volume ................................................................................. 367
24.4. Hydrostatic Fluid Element with Negative Volume ............................................................................... 368
24.5. Tire Reinforcing: Road Contact Area ................................................................................................... 369
24.6. Tire Reinforcing: Side Wall .................................................................................................................. 369
24.7. Tire Reinforcing: Sectional View ......................................................................................................... 370
24.8. Rim Model with Tire .......................................................................................................................... 371
24.9. Model of Tire Contact with Road Surface ............................................................................................ 372
24.10. Fluid Pressure Variation ................................................................................................................... 375
24.11. Fluid Volume Variation ..................................................................................................................... 375
24.12. Fluid Density Variation ..................................................................................................................... 376
24.13. Fluid Mass Variation ......................................................................................................................... 376
24.14. Tire Deformation (All Load Steps) ..................................................................................................... 376
25.1. Effect of Stent Placement in Increasing Blood Flow ............................................................................ 379
25.2. Cross-sectional View of Unloaded Artery and Stent ........................................................................... 380
25.3. Stent Model 3-D Expanded Solid Display ........................................................................................... 381
25.4. Simplified Atherosclerotic Artery Model ............................................................................................ 382
25.5. Standard Line-to-Surface Contact Between Stent and Inner Plaque Wall ............................................. 383
25.6. Artery Boundary Conditions .............................................................................................................. 384
25.7. Stent Boundary Conditions ............................................................................................................... 384
25.8. Uniform Pressure Loading on the Inner Plaque Wall (a) and Load History (b) ....................................... 385
25.9. Cross-Sectional View of Artery and Stent After Balloon Angioplasty (Load Step 1) .............................. 386
25.10. Arterial Wall Deformation During Balloon Angioplasty (a) and After Stent Placement (b) ................... 387
25.11. Arterial Wall Displacement and Tissue Prolapse Results .................................................................... 387
25.12. Arterial Wall von Mises Stress Results ............................................................................................... 388
25.13. Stent Retraction Resulting from Arterial Compression ...................................................................... 389
26.1. 3-D Model of Rubber Boot Seal and Rigid Shaft .................................................................................. 392
26.2. Half Symmetry Model of the 3-D Rubber Boot Seal ............................................................................. 393
26.3. Rigid-Flexible Contact Pair ................................................................................................................. 394
26.4. Self Contact Pairs .............................................................................................................................. 395
26.5. Boundary Conditions at a Symmetry Plane ........................................................................................ 396
26.6. Boundary Conditions at the Bottom of the Rubber Boot (in the Axial Direction) .................................. 396
26.7. Boundary Conditions at the Bottom of the Rubber Boot (in the Radial Direction) ................................ 397
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
xxvi of ANSYS, Inc. and its subsidiaries and affiliates.
Technology Demonstration Guide
26.8. Comparison of Cumulative Iterations for Different Contact Detection Methods .................................. 398
26.9. Displacement Vector Sum at the Maximum Shaft Angle ..................................................................... 399
26.10. Von Mises Stress at the Maximum Shaft Angle .................................................................................. 400
26.11.Total Strain at First Contact between Outer Plaits of Rubber Boot ...................................................... 400
26.12. Total Strain at Second Contact between Inner Plaits of Rubber Boot ................................................. 401
26.13.Total Strain at Third Contact between Inner Plaits of Rubber Boot and Shaft ...................................... 401
26.14. Animated Results ............................................................................................................................ 402
27.1. Hot-Rolling Model ............................................................................................................................. 405
27.2. Symmetric Hot-Rolling Model ............................................................................................................ 406
27.3. Hot-Rolling Quarter Model Geometry ................................................................................................ 407
27.4. Meshed Model with Dimensions ........................................................................................................ 407
27.5. Contact Between Block and Top Roller ............................................................................................... 408
27.6. Contact Between Block and Side Roller .............................................................................................. 409
27.7. Symmetric Boundary Conditions on the Block ................................................................................... 411
27.8. Loadings ........................................................................................................................................... 411
27.9. Deformation in First Load Step, 40th Substep ..................................................................................... 413
27.10. Deformed Mesh of the Block at 30th Substep of First Load Step ........................................................ 414
27.11. Original Deformed Mesh and Imported New Mesh ........................................................................... 414
27.12. Comparison of Contact Pressure Results Before and After Rezoning .................................................. 415
27.13. USUM Plots: Initial Run at Last Converged Substep ........................................................................... 416
27.14. USUM Plots: After Building Up the Rolling Process ............................................................................ 417
27.15. USUM Plots: End of Analysis (After Second Load Step) ...................................................................... 417
27.16. USUM Plot of Full Model after Symmetry Expansion ........................................................................ 418
27.17. Variation of the Moment (Mx) of the Top Roller ................................................................................. 418
27.18. Variation of the Reaction Forces of the Top Roller ............................................................................. 419
27.19. Equivalent Plastic Strain Plot in the Full Expanded Model ................................................................. 419
27.20. Deformation (USUM) in the Initial Run ............................................................................................. 420
27.21. Deformation (USUM) After Rezoning ............................................................................................... 420
27.22. Deformation (USUM) After Rezoning in the Expanded Model .......................................................... 421
28.1. 3-D Model of Workpiece and Tool ...................................................................................................... 425
28.2. 3-D Meshed Model of Workpiece and Tool ......................................................................................... 426
28.3. Contact Pair Between Plates .............................................................................................................. 426
28.4. Contact Pair Between Tool and Workpiece ......................................................................................... 427
28.5. Rigid Surface Constrained ................................................................................................................. 428
28.6. Thermal Boundary Conditions ........................................................................................................... 430
28.7. Mechanical Boundary Conditions ...................................................................................................... 431
28.8. Friction Stir Welding Animation ......................................................................................................... 434
28.9. Deflection at Workpiece After Load Step 1 ......................................................................................... 434
28.10. von Mises Stress After Load Step 1 ................................................................................................... 435
28.11. Temperature After Load Step 1 ........................................................................................................ 435
28.12. Frictional Stress After Load Step 1 .................................................................................................... 436
28.13. Frictional Stress After Load Step 2 .................................................................................................... 436
28.14. Temperature After Load Step 2 ........................................................................................................ 437
28.15. Temperature After Load Step 3 ........................................................................................................ 437
28.16. Maximum Temperature (on Workpiece Beneath the Tool) Variation with Time ................................... 438
28.17. Temperature Distribution on the Top Surface of Workpiece at Various Locations .............................. 438
28.18. Various Locations on the Workpiece ................................................................................................ 439
28.19. Temperature Distribution in Thickness Direction at Location 1 .......................................................... 439
28.20. Temperature Variation with Time on Various Joint Locations ............................................................. 440
28.21. Contact Status at Interface with Bonding Temperature 1000 °C ........................................................ 440
28.22. Contact Status at Interface with Bonding Temperature 900 °C .......................................................... 441
28.23. Total Frictional Heat Rate Variation with Time ................................................................................... 441
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. xxvii
Technology Demonstration Guide
28.24. Total Plastic Heat Rate Variation with Time ....................................................................................... 443
29.1. Element Coordinate System .............................................................................................................. 448
29.2. 3-D Solid Element Model ................................................................................................................... 448
29.3. Thermal-Structural Analysis ............................................................................................................... 449
29.4. Nozzle Extension Geometry ............................................................................................................... 450
29.5. Nozzle Extension 1° Base Sector Geometry ........................................................................................ 450
29.6. Ring Element Plot ............................................................................................................................. 451
29.7. Expanded Layer Representation of the Solid Mesh ............................................................................. 451
29.8. Thermal Boundary Conditions ........................................................................................................... 453
29.9. Structural Boundary Conditions ......................................................................................................... 454
29.10. Layered Temperature Results ........................................................................................................... 455
29.11. Layered TF Results ........................................................................................................................... 455
29.12. Layered Temperature Gradient Results ............................................................................................. 456
29.13. Layer 3 Temperature Results ............................................................................................................ 457
29.14. Homogeneous Thermal - Homogeneous Structural Results .............................................................. 457
29.15. Homogeneous Thermal - Layered Structural Results ......................................................................... 458
29.16. Layered Thermal - Homogeneous Structural Results ......................................................................... 458
29.17. Layered Thermal - Layered Structural Results ................................................................................... 459
29.18. Layer 1 Equivalent Stresses .............................................................................................................. 460
29.19. Single Element Coordinate System .................................................................................................. 461
29.20. Multiple Element Coordinate Systems .............................................................................................. 461
30.1. 3-D Model of the Speaker Assembly .................................................................................................. 464
30.2. Acoustic Fluid Elements .................................................................................................................... 465
30.3. Modal Analysis at 807 Hz ................................................................................................................... 467
30.4. Displacement Frequency Response Plot ............................................................................................ 468
30.5. SPL at 50 mm .................................................................................................................................... 469
30.6. SPL at 800 Hz .................................................................................................................................... 470
30.7. SPL at 860 Hz .................................................................................................................................... 471
30.8. Velocity Magnitude Contour Plot ....................................................................................................... 472
30.9. SPL Polar Plot at 800 Hz ..................................................................................................................... 473
31.1. Stable Hysteresis Strain-Controlled Data ............................................................................................ 477
31.2. Plot Comparing Curve-Fitting Tool Results ......................................................................................... 479
31.3. Loading History for Strain-Controlled Experiment ............................................................................. 480
31.4. Schematic for Single-Element Uniaxial Test ........................................................................................ 481
31.5. Plot Comparing Strain-Controlled Experimental Data with Simulation Data ........................................ 482
31.6. Plot Comparing Stress-Controlled Experimental Data with Simulation Data ........................................ 483
31.7. Plot Comparing Stabilized Hysteresis Data to Fit of Multiple Data Sets ................................................ 484
31.8. Plot Comparing Stress Data to Fit of Multiple Data Sets ...................................................................... 485
31.9. Plot Comparing Ratcheting Strain Data to Prediction from Fitted Chaboche Model ............................. 486
32.1. Human Knee Components ................................................................................................................ 490
32.2. Finite Element Model of an ACL ......................................................................................................... 491
32.3. Von Mises Stress (Anisotropic Visco-Hyperelastic and Neo-Hookean Viscoelastic) ................................ 494
32.4. ACL Stress-Strain Plot ....................................................................................................................... 495
32.5. Stress-Strain at Differing Strain Rates ................................................................................................. 496
32.6. Stress Relaxation ............................................................................................................................... 496
32.7. Von Mises Stresses Under 45° Knee Flexion ........................................................................................ 497
32.8. Von Mises Stresses Under 11.25° Knee Rotation .................................................................................. 498
33.1. Class IV Flextensional Underwater Acoustic Transducer Shell .............................................................. 502
33.2. Flextensional Transducer Components .............................................................................................. 503
33.3. Transducer Surrounded by Water ....................................................................................................... 504
33.4. 2-D Transducer Model Mesh .............................................................................................................. 505
33.5. 3-D Transducer Model Mesh .............................................................................................................. 506
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
xxviii of ANSYS, Inc. and its subsidiaries and affiliates.
Technology Demonstration Guide
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. xxix
Technology Demonstration Guide
37.4. 3-D Half Model of a Threaded Bolt Joint with Mesh Details ................................................................. 575
37.5. 2-D Axisymmetric Model of a True Threaded Bolt Joint with Mesh Details ........................................... 575
37.6. Simplified 3-D Model of an M120 Bolt ................................................................................................ 576
37.7. 2-D Axisymmetric Model of a Simplified Bolt Joint with Mesh Details ................................................. 577
37.8. 3-D Half Model of a Simplified Bolt Joint with Mesh Details ................................................................ 577
37.9. Pretension Section in the Bolt Model at y = 260 mm ........................................................................... 578
37.10. Contact and Target Surfaces at the Bolt Thread and the Base Plate Thread ........................................ 579
37.11. Contact and Target Surfaces at the Bolt Surface and Base Plate ......................................................... 580
37.12. Contact and Target Surfaces at the Bolt Head and Cover Plate .......................................................... 581
37.13. Contact and Target Surfaces at the Cover Plate and Base Plate .......................................................... 582
37.14. Boundary Conditins and Loading on the 3-D Model ......................................................................... 583
37.15. UY Displacement in the 2-D Axisymmetric Model for all Three Methods ............................................ 584
37.16. von Mises Stress in the 2-D Axisymmetric Model for all Three Methods ............................................. 585
37.17. Linearized Stress in the Bolt Shank Along a Path at y = 280 for all Three Methods. ............................. 586
37.18. Contact Status in the Thread Region for all Three Methods ............................................................... 587
37.19. UY Displacement in the 3-D Half Model for all Three Methods .......................................................... 587
37.20. von Mises Stress in the 3-D Half Model for all Three Methods ............................................................ 588
37.21. von Mises Stress in the Bolt for all Three Methods ............................................................................. 588
37.22. von Mises Stress in the Bolt Shank at Section y = 280 for all Three Methods ....................................... 589
38.1. Stress-vs.-Strain Plot .......................................................................................................................... 601
39.1. Ultrasonic Transducer Model ............................................................................................................. 604
39.2. Approximate Location for Positioning the Holder ............................................................................... 605
39.3. Meshed Ultrasonic Transducer Model ................................................................................................ 606
39.4. Top and Bottom Plate Connection ..................................................................................................... 607
39.5. First Longitudinal Mode of Interest (16) at 26.6 kHz ............................................................................ 610
39.6. Second Longitudinal Mode of Interest (30) at 58.9 kHz ....................................................................... 611
39.7. Third Longitudinal Mode of Interest (39) at 87.3 kHz ........................................................................... 612
39.8. Electrical Impedance ......................................................................................................................... 613
39.9. Tip Displacement .............................................................................................................................. 613
40.1. Spinal Spacer Implant ....................................................................................................................... 619
40.2. Loading and Recovery of a Spinal Spacer Implant .............................................................................. 620
40.3. Spinal Spacer 1/4 Model .................................................................................................................... 620
40.4. Displacements of Central Node A of the Spinal Spacer ....................................................................... 622
40.5. Spinal Spacer Deformation and Stress ................................................................................................ 623
40.6. Motion of a Vertical Helical Spring ..................................................................................................... 624
40.7. Finite Element Models of a Spring Actuator ....................................................................................... 625
40.8. Temperature and Force Load Steps .................................................................................................... 626
40.9. Spring Actuator Deformation at Step 1 .............................................................................................. 627
40.10. Spring Actuator Deformation at Step 2 ............................................................................................ 627
40.11. Spring Actuator Deformation at Step 3 ............................................................................................ 628
40.12. Displacement History: Bottom of a Spring Actuator with Temperature .............................................. 629
41.1. 3-D Model of a Resonator Panel ......................................................................................................... 632
41.2. 3-D Model of a Quarter-Wave Tubes Resonator .................................................................................. 632
41.3. Resonator Cross-Section .................................................................................................................... 633
41.4. 3-D Mesh of Resonator Panel ............................................................................................................ 634
41.5. Absorption Coefficient of the Resonator: BLI Model ............................................................................ 636
41.6. Absorption Coefficient of the Resonator: LRF Model ........................................................................... 637
41.7. BLI and LRF Results Comparison ........................................................................................................ 637
42.1. 3-D Crimp Joint Model ...................................................................................................................... 640
42.2. 3-D Meshed Model of Grip with Seven Stranded Wires ...................................................................... 641
42.3. Rigid Punch and Rigid Base ............................................................................................................... 642
42.4. Surface Contact Elements (a) and Edge Contact Elements (b) ............................................................ 643
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
xxx of ANSYS, Inc. and its subsidiaries and affiliates.
Technology Demonstration Guide
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. xxxi
Technology Demonstration Guide
45.20. CYCSPEC Histogram Response Plot at Frequency 522.76 in the Tuned Analysis ................................. 709
45.21. Aerodynamic Stiffness and Damping Coefficients vs. Nodal Diameter ............................................... 709
45.22. Modal Damping Ratio vs. Modal Frequency for Tuned and Mistuned Models .................................... 710
45.23. Total Deformation Pattern (Real Only) at Frequency of 522.76 Hz in the Mistuned Analysis ................ 711
45.24. Equivalent Stress Pattern (Real Only) at Frequency of 522.76 Hz in the Mistuned Analysis .................. 711
45.25. Equivalent Strain Pattern (Real Only) at Frequency of 522.76 Hz in the Mistuned Analysis .................. 712
45.26. Nodal Solution Plots with Respect to the Frequency of Excitation in the Mistuned Analysis ............... 712
45.27. Directional Deformation at the Blade Tip Along the Z-Direction in the Mistuned Analysis .................. 713
45.28. CYCSPEC Histogram Response Plot at Frequency 522.76 Hz in the Mistuned Analysis ........................ 714
46.1. 2-D Schematic of a Disc-Shaped Reservoir at a Depth (D) in a Subsurface ........................................... 718
46.2. Typical Mesh of Reservoir in a Subsurface .......................................................................................... 718
46.3. Surface Subsidence for Values of η(=D/R) ........................................................................................... 721
46.4. Surface Subsidence for Values of Reservoir Height (H) ........................................................................ 722
46.5. Maximum Subsidence as a Quadratic Function of η(=D/R) and a Linear Function of Reservoir Height
(H) ............................................................................................................................................................ 723
46.6. Surface Subsidence ........................................................................................................................... 724
46.7. Reservoir Compaction ....................................................................................................................... 724
47.1. Solder Ball Model .............................................................................................................................. 728
47.2. Solder Ball Model Dimensions ........................................................................................................... 728
47.3. Meshed Solder Joint .......................................................................................................................... 729
47.4. Hydrostatic Stress ............................................................................................................................. 733
47.5. Final Temperature ............................................................................................................................. 734
47.6. Current Density ................................................................................................................................. 734
47.7. Normalized Concentration After Two Years ........................................................................................ 735
47.8. Minimum Concentration vs. Time ...................................................................................................... 736
47.9. Maximum Concentration vs. Time ...................................................................................................... 736
48.1. Soil Geometry and Active/Passive Movement of Left Retaining Wall ................................................... 740
48.2. Vertical Displacement for Gravity Load Step (Initial Stresses Not Applied) ........................................... 741
48.3. Vertical Displacement for Gravity Load Step (Initial Stresses Applied) .................................................. 742
48.4. Horizontal-to-Vertical Stress Ratios Along the Retaining Wall .............................................................. 744
48.5. Horizontal Displacement for Active Pressure Load Step (Negative Displacement of Retaining Wall in x-
Direction) ................................................................................................................................................. 745
48.6. Equivalent Plastic Strains for Active Pressure Load Step ...................................................................... 746
48.7. Horizontal Displacement for Passive Pressure Load Step (Positive Displacement of Retaining Wall in x-
Direction) ................................................................................................................................................. 746
48.8. Equivalent Plastic Strains for Passive Pressure Load Step ..................................................................... 747
49.1. Quarter Model: Concrete Slab and Depiction of Dirichlet Boundary Conditions ................................... 750
49.2. Modeling Steel Reinforcing Using Mesh-Only Elements (MESH200) .................................................... 751
49.3. 3 Generated Reinforcing Elements (REINF265) ................................................................................... 752
49.4. Reaction Force/Displacement Curve: Load-Limit Analysis ................................................................... 754
49.5. Vertical Displacements: Concrete Slab After Appplying Dead Load and Load Limit (Top View) ............. 755
49.6. Horizontal Stress Component SX: Concrete Slab at Ultimate Limit Load (Top and Bottom View) ........... 756
49.7. Horizontal Stress Component SZ: Concrete Slab at Ultimate Limit Load (Top and Bottom View) ........... 757
49.8. Stress Components SX: Steel Reinforcing at Ultimate Load Limit (Top View) ........................................ 758
49.9. Equivalent Plastic Strains: Concrete Slab at Ultimate Limit Load (Bottom View) .................................... 758
49.10. Reaction Force/Displacement Curve: Load-Limit Analysis Using the Arc-length Method .................... 759
50.1. Schematic of Brain Axial Section with Boundary Conditions ............................................................... 762
50.2. Temperature Distribution Contour of Brain Cross-section at Time 0.02 (a) and Time 1 (b) ..................... 764
50.3. Pore-pressure Distribution From the Last Time Step ........................................................................... 765
50.4. Stress Contour of the Last Time Step .................................................................................................. 766
50.5. Deformation Contour of the Last Time Step (a) and Deformed / Undeformed Shapes (b) ..................... 767
51.1. 3-D Model of a Five-Filar Metal Coil Inside a Polymer Tube .................................................................. 770
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
xxxii of ANSYS, Inc. and its subsidiaries and affiliates.
Technology Demonstration Guide
51.2. Meshed Model for Case 1, Case 2, and Case 3 ..................................................................................... 771
51.3. Models: (a) Solid Elements, (b) Beam Elements ................................................................................... 772
51.4. Tube Models: (a) Solid Elements, (b) Line Elements ............................................................................. 773
51.5. Contact Pair Between Coil and Tube for Case 1, Case 2, and Case 3 ...................................................... 774
51.6. Self Contact Between Filars: (a) Case 1, (b) Case 2 and Case 3 .............................................................. 775
51.7. MPC-Based Rigid Constraints (a) and Boundary Conditions at Pilot Nodes (b) ..................................... 776
51.8. USUM Displacement ......................................................................................................................... 777
51.9. von Mises Stress ................................................................................................................................ 777
52.1. Schematic of an Embankment on a Clay Layer ................................................................................... 780
52.2. Finite Element Model of the Problem ................................................................................................. 781
52.3. Settlement and Excess Pore Pressure ................................................................................................. 784
52.4. Evolution of Pore Pressure During 280 Days of Sequential Embankment Construction ........................ 785
52.5. Settlement Process of the Model During 280 Days of Sequential Embankment Construction .............. 786
52.6. Excess Pore-Pressure Evolution During 280 Days of Sequential Embankment Construction ................. 787
53.1. 3-D Model of the Condenser MEMS Microphone with One Symmetry Plane ....................................... 790
53.2. Silicon Microphone Structure ............................................................................................................ 790
53.3. Mesh of the Condenser MEMS Microphone ........................................................................................ 791
53.4. Air Gap Capacitance as a Function of Voltage ..................................................................................... 793
53.5. Microphone Sensitivity as a Function of Frequency ............................................................................ 794
53.6. Membrane Deformation at 20,000 Hz ................................................................................................ 794
53.7. Housing Sound Pressure Level at 20,000 Hz ....................................................................................... 795
54.1. RC Joint Geometry and Reinforcing ................................................................................................... 798
54.2. Finite Element Model of the RC Joint ................................................................................................. 799
54.3. RC Joint Boundary Conditions and Loading ....................................................................................... 801
54.4. Force Displacement Curve Results ..................................................................................................... 802
54.5. Evolution of the Equivalent Plastic Strain (u = Applied Displacement) ................................................. 802
54.6. Tension and Compression Damage Profiles at the End of Simulation .................................................. 803
54.7. Plasticity and Damage Contours at the End of Simulation ................................................................... 803
54.8. Deformed Shape and Plastic Strain in the Steel Reinforcing ................................................................ 804
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. xxxiii
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
xxxiv of ANSYS, Inc. and its subsidiaries and affiliates.
List of Tables
1.1. Solution Output .................................................................................................................................... 10
12.1. X-Direction Input .............................................................................................................................. 168
12.2. X-Direction Input (with Missing-Mass Effect) ...................................................................................... 169
12.3. X-Direction Input (with Missing-Mass and Rigid-Response Effects) ..................................................... 170
12.4. X-Direction Input Motion .................................................................................................................. 171
12.5. X-Direction Input (with Missing-Mass and Rigid-Response Effects) ..................................................... 172
12.6. 3-Direction Input (with Missing-Mass and Rigid-Response Effects) ...................................................... 173
12.7. 3-Direction Input Motion ................................................................................................................... 174
22.1. Comparison of Eigenfrequencies for Full and CMS Models .................................................................. 346
22.2. Comparison of CPU and Elapsed Times for Modal Analysis ................................................................. 347
22.3. Comparison of Response Amplitudes for Full and CMS Models ........................................................... 347
22.4. Comparison of CPU and Elapsed Times for Harmonic Analysis ............................................................ 348
26.1. Simulation Statistics for Different Contact Detection Methods ............................................................ 399
28.1. Workpiece Material Properties ........................................................................................................... 429
28.2. Material Properties of the PCBN Tool .................................................................................................. 429
28.3. Load Steps ........................................................................................................................................ 432
28.4. Solution Settings ............................................................................................................................... 432
28.5. Locations on Weld Line ...................................................................................................................... 439
31.1. Initial and Fitted Parameters for the Strain-Controlled Experiment ..................................................... 480
31.2. Comparison of Strain-Controlled Experimental Data with Simulated Data ........................................... 481
31.3. Initial and Fitted Parameters for the Stress-Controlled Experiment ..................................................... 482
31.4. Initial and Fitted Parameters for Fit to Both Data Sets ......................................................................... 484
34.1. Coherency Function Coefficients ....................................................................................................... 517
34.2. Comparison of Wall Time ................................................................................................................... 530
34.3. Comparison of Wall Time for Various Numbers of Cores (R14.5) ........................................................... 531
35.1. Implicit Creep Models ....................................................................................................................... 536
35.2. Generalized Garofalo Creep Constants from Curve Fitting and Reference ............................................ 541
35.3. Flip Chip Material Properties .............................................................................................................. 542
35.4. Temperature-Dependent Properties .................................................................................................. 542
37.1. Material Properties of Bolt ................................................................................................................. 582
37.2. Material Properties of Cover Plate and Base Plate ............................................................................... 582
37.3. Simulation Time and Cumulative Iterations for 2-D Models ................................................................. 589
37.4. Simulation Time and Cumulative Iterations for 3-D Models ................................................................. 589
42.1. Material Properties of Grip and Wire .................................................................................................. 644
45.1. Aliased Engine Order (Excited Harmonic Index) .................................................................................. 698
45.2. Comparison of Modal Frequencies ..................................................................................................... 704
45.3. Comparison of Prestressed Modal Frequencies .................................................................................. 704
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. xxxv
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
xxxvi of ANSYS, Inc. and its subsidiaries and affiliates.
What You Need to Know
The purpose of these technical demonstration problems is to encourage you to take advantage of the
extraordinarily broad simulation capabilities of ANSYS Mechanical APDL. The real-world problems
showcase the features and effectiveness of Mechanical APDL by presenting a series of analyses from a
variety of engineering disciplines.
The problems are more substantive and complex than examples found in the standard documentation
set. The documentation thoroughly examines the physics involved with each problem and the consid-
erations necessary for translating problems into numerical models. Approximation issues, accuracy
considerations, and recommended practices are discussed.
A comprehensive results and discussion section carefully examines analysis results (often comparing
them to baseline or known good results using more traditional analysis methods), and illustrates why
specific strategies and methods were chosen.
Each problem concludes with valuable hints and recommendations for performing a similar type of
analysis. In many cases, references are provided for additional background information. You can therefore
use any given example problem as a template for setting up a similar type of simulation.
• use the link provided under Input Files in the documentation for each problem, or
• use the "Download files" link for each problem in the summary table below.
You can also download all td-nn file sets in a single zip file.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. xxxvii
What You Need to Know
1. Open the file and extract or copy the td-nn input files to a convenient location on your computer.
Some problems involve more than one analysis. In such cases, a separate.dat file is provided for each
analysis.
Problem Summary
The following technology demonstration problems are available:
td-1 Brake Squeal Analysis (p. 1) Solves a brake squeal problem. Three analysis
methods (p. 5) are highlighted: linear non-prestressed
Download modal, partial nonlinear prestressed modal, and full
files nonlinear prestressed modal. The problem demonstrates
sliding frictional contact and uses complex eigensolvers
to predict unstable modes.
td-2 Nonlinear Analysis of a 2-D A nonlinear analysis of a 2-D hyperelastic seal assembly
Hyperelastic Seal Using using manual rezoning with remeshing via the
Download Rezoning (p. 15) element-splitting method. The problem shows how
files multiple vertical rezoning steps can be used to ensure
convergence and completion of an analysis.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
xxxviii of ANSYS, Inc. and its subsidiaries and affiliates.
td-4 Ring-Gear Forging Demonstrates the efficiency and usefulness of rezoning
Simulation with in a 2-D simulation of metal-forming processes.
Download Rezoning (p. 63) Rezoning facilitates the convergence of a nonlinear
files finite element simulation in which excessive element
distortion occurs.
td-6 Thermal Stress Analysis of a Shows how to easily set up and perform a thermal-stress
Cooled Turbine Blade (p. 93) analysis of a cooled turbine blade. The problem uses
Download surface-effect capabilities to simulate convection loading
files on solid regions, and one-dimensional fluid-flow
capabilities to obtain a highly accurate thermal solution
for convection loading.
td-7 Nonlinear Transient Analysis Shows how to easily set up and perform an analysis
of a Camshaft involving both axisymmetric and nonaxisymmetric
Download Assembly (p. 105) components. The problem demonstrates how modeling
files with general axisymmetric element technology can
reduce computational resources significantly while
maintaining the same degree of accuracy as a
simulation using a full 3-D model.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. xxxix
What You Need to Know
td-11 Rocket Nozzle Extension Simulates thermal stresses during the manufacturing
Simulation: of a rocket nozzle. Uses current structural-shell element
Download Fabrication (p. 149) technology to accurately model the orthotropic thermal
files expansion in curved-shell structures. Section offsets are
applied when connecting shell-to-shell or to
shell-to-other element types.
td-12 Dynamic Simulation of a Accounts for the missing-mass effect and rigid-response
Nuclear Piping System Using effects in a spectrum analysis, and demonstrates how
Download RSA Methods (p. 161) including those effects improves results accuracy (as
files compared to full-transient analysis results). Uses a
piping system model from an actual nuclear power
plant.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
xl of ANSYS, Inc. and its subsidiaries and affiliates.
td-13 Centrifugal Impeller Analysis A centrifugal impeller blade analysis using cyclic
Using Cyclic Symmetry and modeling methods and linear-perturbation solution
Download Linear Perturbation (p. 177) approaches. The problem includes a modal analysis, a
files full-harmonic analysis, a prestressed modal analysis
using linear perturbation, and a prestressed
full-harmonic analysis using linear perturbation.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. xli
What You Need to Know
td-17 Impact of a Metal Bar on a An impact simulation using a model of a 3-D metal
Rigid Wall (p. 251) bar hitting a rigid wall. The problem shows the
Download advantages of using impact constraints for
files modeling contact in a nonlinear transient dynamic
analysis. Several combinations of available
time-integration methods and contact algorithms
are also investigated, using different material
models to show how various choices affect the
performance and accuracy of the finite-element
solution of impact problems.
td-20 Dynamic Simulation of a Uses residual vectors to improve the solution accuracy
Printed Circuit Board in modal subspace based analysis methods, such as
Download Assembly Using Modal modal superposition and power spectral density (PSD)
files Analysis Methods (p. 301) analyses. The problem includes a study of the
computational efficiency of the results-expansion
procedure used to obtain the full-model solution.
td-21 Buckling and Post-Buckling A nonlinear buckling and post-buckling analysis using
Analysis of a Ring-Stiffened nonlinear stabilization. The problem uses a stiffened
Download Cylinder Using Nonlinear cylinder subjected to uniform external pressure to
files Stabilization (p. 313) demonstrate how to find the nonlinear buckling loads,
achieve convergence at the post-buckling stage, and
interpret the results.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
xlii of ANSYS, Inc. and its subsidiaries and affiliates.
td-22 Modal and Harmonic Modal and full-harmonic analyses of an automotive
Frequency Analyses of an suspension assembly exhibiting harmonic displacement
Download Automotive Suspension excitation at the wheel bottoms. Component mode
files Assembly Using CMS (p. 335) synthesis (CMS) is used to generate dynamic
superelements for use in downstream linear dynamics
analyses. Analysis results from the full (non-substructure)
and the CMS generated models are compared to
highlight the benefits of CMS technology.
td-23 Modal Analysis of a Wind Simulates a wind turbine blade, a slender composite
Turbine Blade Using Beam structure, using a current-technology beam element. A
Download Elements (p. 353) simplified 1-D beam-based model of the typically
files complex blade geometry is especially useful in the early
design stage, when small design variations can lead to
partial or even complete reconstruction of a 3-D model,
generally an impractical solution given the difficulty of
building the model.
td-24 Hydrostatic Fluid Analysis of Models a fluid that is fully enclosed within a solid
an Inflating and Rolling (container). Shows how loading and deformation on
Download Tire (p. 365) the container affects the contained fluid pressure,
files volume, density and mass.
td-25 Cardiovascular Stent Simulates stent-artery interaction during and after stent
Simulation (p. 379) placement. The analysis uses advanced modeling
Download techniques including contact, element birth and death,
files mixed u-P formulation, and nonlinear stabilization
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. xliii
What You Need to Know
td-26 Nonlinear Analysis of a A nonlinear analysis of a rubber boot seal that includes
Rubber Boot Seal (p. 391) geometric nonlinearities (large strain and large
Download deformation), nonlinear material behavior
files (incompressible hyperelastic material model), and
changing status nonlinearities (contact). The analysis
demonstrates the advantages of the
surface-projection-based contact contact method.
td-28 Friction Stir Welding (FSW) Simulates the friction stir welding (FSW) process. Several
Simulation (p. 423) characteristics of FSW are presented, including plastic
Download deformation, tool-workpiece surface interaction, and
files heat generation due to friction and plastic deformation.
Thermal and mechanical behaviors are mutually
dependent and coupled together during the FSW
process. A nonlinear direct coupled-field analysis is
performed.
td-29 Rocket Nozzle Extension Simulates the thermal stresses induced during the
Simulation: operation of a rocket nozzle. Hot gases flow through
Download Operation (p. 447) the nozzle, subjecting the inside and outside of the
files nozzle body to convection heat loading. Solid thermal
and structural elements accurately simulate the
multiphysics of the problem.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
xliv of ANSYS, Inc. and its subsidiaries and affiliates.
td-30 Acoustic Analysis of a Small Uses acoustic elements coupled with structural elements
Speaker System (p. 463) to analyze the performance of a speaker assembly.
Download Features structural-acoustic coupling using
files fluid-structure interaction (FSI) in 3-D, a symmetric FSI
algorithm, perfectly matched layers (PML) to absorb
outgoing acoustic waves, postprocessing sound pressure
level (SPL) and velocity, far-field postprocessing of
acoustic field, and user-defined symmetric expansion
options.
td-32 Anterior Cruciate Ligament Simulates the response of an anterior cruciate ligament
(ACL) Simulation (p. 489) (ACL) of a human knee subjected to tension, flexion,
Download and rotation. The problem uses an anisotropic
files hyperelastic material model with viscoelasticity.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. xlv
What You Need to Know
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
xlvi of ANSYS, Inc. and its subsidiaries and affiliates.
user-programmable feature for developing a custom,
constitutive model to define the stress-strain behavior
of a material.
td-40 Shape Memory Alloy (SMA) Two shape memory alloy (SMA) simulations: a spinal
with Thermal Effect (p. 617) spacer implant and a spring actuator. An SMA is a
Download material that, after being subjected to mechanical
files loading/unloading cycles, is able to undergo large
deformations without showing residual strains
(pseudoelasticity) or that can recover from large
deformations via temperature change (shape memory
effect). The simulations demonstrate the utility of the
SMA material model using martensite and austenite
(nitinol), and the behavior of the SMA with thermal
loading.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. xlvii
What You Need to Know
td-42 Wire Crimping Modeled with Demonstrates the ease of contact modeling using the
General Contact (p. 639) general contact method. Thes method offers automated
Download contact creation and requires minimal user input. The
files general contact procedure is especially useful when a
large number of contacting surfaces are involved in the
model and the geometry makes it difficult to determine
contact pairs.
td-43 Contact Surface Wear Demonstrates how to model contact surface wear using
Simulation (p. 649) the Archard wear model or a user-defined wear model
Download (userwear subroutine). The wear model calculates how
files much, and in what direction, a contact node is to be
moved to simulate wear based on the contact results.
Mesh nonlinear adaptivity based on a wear criterion is
used to improve the mesh during the wear process.
td-44 C*-integral Evaluation for Evaluates the C*-integral for cracks in structural
3-D Surface Flaws (p. 665) components. Analyses of a simple semicircular surface
Download flaw in a rectangular block and a warped flaw along a
files tubular joint are presented.
td-45 Forced Response Analysis of Demonstrates the free vibration and forced response
a Mistuned Bladed Disk analysis of a tuned and mistuned NASA Rotor 67 fan
Download Using Cyclic Symmetry and using cyclic modeling methods and linear perturbation
files Linear Perturbation (p. 685) solution approaches. Includes a modal analysis, a
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
xlviii of ANSYS, Inc. and its subsidiaries and affiliates.
prestressed modal analysis using linear perturbation,
and a prestressed mode-superposition harmonic analysis
using linear perturbation.
td-46 Surface Subsidence Caused Predicts solid deformation caused by the coupling of
by Reservoir pore-pressure diffusion and solid-matrix deformation.
Download Depletion (p. 717) Surface subsidence of a disc-shaped compartmentalized
files reservoir in a homogenous poro-elastic continuum is
used as a demonstration case. The problem highlights
geomechanics (soil analysis), coupled pore-pressure
mechanical solid elements, and the modified Cam-clay
soil material model.
td-48 Active and Passive Lateral Simulates soil behavior under active and passive
Earth-Pressure earth-pressure loadings. Shows how the nonlinear
Download Analysis (p. 739) plastic behavior of soil can be modeled with a
files Mohr-Coulomb material.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. xlix
What You Need to Know
td-51 Multi-Filar Coil Inside a Tube Three methods for modeling a multi-filar coil inside of
Modeled with a tube. Each model uses a different contact scenario:
Download Beam-to-Beam surface-to-surface, beam-to-surface, or beam-to-beam.
files Contact (p. 769) A comparison shows that the beam model using
beam-to-beam contact offers the best advantage in
terms of simplified modeling and reduced computation
time.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
l of ANSYS, Inc. and its subsidiaries and affiliates.
td-53 Acoustic Analysis of a MEMS Analyzes the response of a silicon micro-machined
Microphone (p. 789) microphone using acoustic elements and
Download electrostatic-structural coupled-field elements.
files
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. li
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
lii of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 1: Brake-Squeal Analysis
This example analysis shows how to solve a brake-squeal problem. Three analysis methods (p. 5) are
highlighted: linear non-prestressed modal, partial nonlinear perturbed modal, and full nonlinear perturbed
modal. The problem demonstrates sliding frictional contact and uses complex eigensolvers to predict
unstable modes.
1.1. Introduction
Eliminating brake noise is a classic challenge in the automotive industry. Brake discs develop large and
sustained friction-induced oscillations, referred to simply as brake squeal.
• Stick-Slip Theory -- The self-excited vibration of a brake system occurs when the static friction coefficient is
greater than the sliding friction coefficient. Variable friction forces introduce energy into the system which
is not properly dissipated during the squealing event, resulting in large vibration.
• Mode-Coupling Theory -- When two similar characteristic modes couple with each other, instability is intro-
duced to the braking system. This instability is caused primarily by improperly selected geometric parameters.
Both theories attribute brake squeal to variable friction forces at the disc-pad interface.
• Low-frequency noise -- An example of a low-frequency noise is the “groaning” noise which occurs in the
frequency range between 100 and 1000 Hz. Any noise having a frequency above 1000 Hz is considered a
squeal.
• Low-frequency squeal -- A result of mode coupling occurring between the out-of-plane modes of the rotor
and the bending modes of the brake pad.
• High-frequency squeal -- A result of mode coupling occurring between the in-plane modes of the rotor.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 1
Brake-Squeal Analysis
Low- and high-frequency squealing can be determined via complex eigensolvers. The presence of unstable
modes suggests that the geometry parameters and material properties of the braking system should
be modified.
For more information, see Brake-Squeal (Prestressed Modal) Analysis in the Structural Analysis Guide.
1.3. Modeling
The following modeling topics are available:
1.3.1. Understanding the Advantages of Contact Element Technology
1.3.2. Modeling Contact Pairs
1.3.3. Generating Internal Sliding Motion
1.3.4. Meshing the Brake Disc-Pad Model
3-D contact elements (CONTA17x) offer a more efficient alternative by modeling surface-to-surface
contact at the pad-disc interface. With contact surface-to-surface contact elements, a matched mesh is
unnecessary at the contact-target surface, and there is no need to calculate the unsymmetric terms.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
2 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
Contact surface-to-surface elements offer many controls for defining contact pairs, such as the type of
contact surface, algorithm, contact stiffness, and gap/initial penetration effect.
The augmented Lagrange algorithm is used for the frictional contact pairs, as the pressure and frictional
stresses are augmented during equilibrium iterations in such a way that the penetration is reduced
gradually. The augmented Lagrange algorithm also requires fewer computational resources than the
standard Lagrange multiplier algorithm, which normally requires additional iterations to stabilize the
contact conditions. The augmented Lagrange is well suited for modeling general frictional contact, such
as the contact between the brake pad and disc defined in this example.
An internal multipoint constraint (MPC) contact algorithm is used for bonded contact because it ties
contact and target surface together efficiently for solid-solid assembly. The MPC algorithm builds
equations internally based on the contact kinematics and does not require the degrees of freedom of
the contact surface nodes, reducing the wave front size of the equation solver. A contact detection
point is made on the Gauss point for frictional contact pairs, and on the nodal point (normal-to-target
surface) for MPC bonded contact pairs.
Frictional contact between pad and disc (left) and bonded contact between other braking
components (right)
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 3
Brake-Squeal Analysis
on the disc surface and the contact elements are defined on the pad surface. The target elements at-
tached to the disc surface are grouped to form a component named E_ROTOR which is then later spe-
cified on the CMROTATE command to generate a sliding frictional force.
Material Properties
Young's Modulus (Nm-2) 2.0 E+11 Pa
Density 7800 Kg/m3
Poisson's Ratio 0.3
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
4 of ANSYS, Inc. and its subsidiaries and affiliates.
Analysis and Solution Controls
Following is the process for solving a brake-squeal problem using this method:
When using the QRDAMP solver, you can reuse the symmetric eigensolution from the previous load
steps (QRDOPT), effective when performing a friction- sensitive/parametric analysis, as it saves time
by not recalculating the real symmetric modes after the first solve operation.
For this analysis, the UNSYM solver is selected to solve the problem. (Guidelines for selecting the
eigensolver for brake-squeal problems appear in Recommendations (p. 12).)
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 5
Brake-Squeal Analysis
The frequencies obtained from the modal solution have real and imaginary parts due the presence of
an unsymmetric stiffness matrix. The imaginary frequency reflects the damped frequency, and the real
frequency indicates whether the mode is stable or not. A real eigenfrequency with a positive value in-
dicates an unstable mode.
The following input shows the solution steps involved in this method:
Modal Solution
/SOLU
ANTYPE, MODAL ! Perform modal solve
NROPT, UNSYM ! To generate non symmetric
CMSEL, S, C1_R ! Select the target elements of the disc
CMSEL, A, C2_R
CM, E_ROTOR, ELEM ! Form a component named E_ROTOR with the selected target elements
ALLSEL, ALL
CMROTATE, E_ROTOR, , , 2 ! Rotate the selected element along global Z using CMROTATE command
Following is the process for solving a brake-squeal problem using this method:
Use the unsymmetric Newton-Raphson method (NROPT,UNSYM). Specify the restart control points
needed for the linear perturbation analysis (RESCONTROL)
The static solution with external loading establishes the initial contact condition and generates a
prestressed matrix.
2. Restart the previous static solution from the desired load step and substep, and perform the first phase of
the perturbation analysis while preserving the .ldhi, .rnnn and .rst files (ANTYPE,STATIC,RESTART,,,PER-
TURB).
Generate forced frictional sliding contact (CMROTATE), specifying the component names created
in the previous step.
The contact stiffness matrix is based only on the contact status at the restart point.
Regenerate the element stiffness matrix at the end of the first phase of the linear perturbation
solution (SOLVE,ELFORM).
3. Obtain the linear perturbation modal solution using the QRDAMP or UNSYM eigensolver (MODOPT).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
6 of ANSYS, Inc. and its subsidiaries and affiliates.
Analysis and Solution Controls
When using the QRDAMP solver, you can reuse the symmetric eigensolution from the previous load
steps (QRDOPT), effective when performing a friction-sensitive/parametric analysis, as it saves time
by not recalculating the real symmetric modes after the first solve operation.
4. Expand the modes and postprocess the results (from the Jobname.RSTP file).
The following inputs show the solution steps involved with this method:
Static Solution
ANTYPE, STATIC ! Perform static solve
OUTRES, ALL, ALL ! Write all element and nodal solution results for each sub steps
NROPT, UNSYM ! Specify unsymmetric Newton-Raphson option to solve the problem
RESCONTROL,DEFINE,ALL,1 ! Control restart files
NLGEOM, ON ! Activate large deflection
AUTOTS, ON ! Auto time stepping turned on
TIME, 1.0 ! End time = 1.0 sec
CMSEL,S,C1_R ! Select target elements of the disc
CMSEL,A,C2_R
CM,E_ROTOR,ELEM ! Form a component named E_ROTOR
SOLVE ! Solve with prestress
Following is the process for solving a brake-squeal problem using this method:
Specify the restart control points needed for the linear perturbation analysis (RESCONTROL).
3. After obtaining the second static solution, postprocess the contact results.
Determine the status (that is, whether the elements are sliding, and the sliding distance, if any).
4. Restart the previous static solution from the desired load step and substep, and perform the first phase of
the perturbation analysis while preserving the .ldhi, .rnnn and .rst files (ANTYPE,STATIC,RESTART,,,PER-
TURB).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 7
Brake-Squeal Analysis
Regenerate the element stiffness matrix at the end of the first phase of the linear perturbation
solution (SOLVE,ELFORM).
5. Obtain the linear perturbation modal solution using the QRDAMP or UNSYM eigensolver (MODOPT).
6. Expand the modes and postprocess the results (from the Jobname.RSTP file).
The following inputs show the solution steps involved with this method:
The mode shape plots for the unstable modes suggest that the bending mode of the pads and disc
have similar characteristics. These bending modes couple due to friction, and produce a squealing noise.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
8 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Figure 1.5: Mode Shape for Unstable Mode (Mode 21) Obtained from the Linear Non-prestressed
Modal Solution (p. 5)
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 9
Brake-Squeal Analysis
Figure 1.6: Mode Shape for Unstable Mode (Mode 22) Obtained from the Linear Non-prestressed
Modal Solution (p. 5)
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
10 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
An examination of the results obtained from the modal analysis of a free disc and pad shows that the
second bending mode of the pad and ninth bending mode of the disc can couple to create dynamic
instability in the system. These pad and disc bending modes can couple to produce an intermediate
lock, resulting in a squeal noise at a frequency close to 6470 Hz.
The following plot suggests that modes with similar characteristics approach each other and couple as
the coefficient of friction increases:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 11
Brake-Squeal Analysis
1.8. Recommendations
The following table provides guidelines for selecting the optimal analysis method to use for a brake-
squeal problem:
The following table provides guidelines for selecting the optimal eigensolver (MODOPT) for obtaining
the brake-squeal solution:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
12 of ANSYS, Inc. and its subsidiaries and affiliates.
Input Files
For further information, see Brake-Squeal (Prestressed Modal) Analysis in the Structural Analysis Guide.
1.9. References
The following references are cited in this example problem:
1. Triches M., Jr, S.N.Y. Gerges, R. Jordon.“Reduction of Squeal Noise from Disc Brake Systems Using
Constrained Layer Damping.” Journal of the Brazilian Society of Mechanical Sciences and Engineering.
Volume XXVI: 340–343. July-September 2004.
2. Allgaier, R. et al.“Mode Lock-In and Friction Modeling.” Computational Methods in Contact Mechanics
IV. 35-47. WIT Press: Southampton (1999).
3. Schroth, R., N. Hoffmann, R. Swift. Mechanism of Brake Squeal--From Theory to Experimentally Measured
Mode Coupling. Robert Bosch, Corporate Research and Development, Germany.
• disc_pad_model.cdb -- Common database file used for the linear non-prestressed modal analysis, the
partial prestressed modal analysis, and the full nonlinear prestressed modal analysis (called by the lin-
ear_non_prestressed.dat, partial_prestressed.dat, full_non_linear.dat and
linear_non_prestressed_par.dat files, respectively).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 13
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
14 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 2: Nonlinear Analysis of a 2-D Hyperelastic Seal Using
Rezoning
This example problem is a nonlinear analysis of a 2-D hyperelastic seal assembly using rezoning via the
element-splitting method. The problem shows how multiple vertical rezoning steps can be used to
ensure convergence and completion of an analysis.
2.1. Introduction
Seals create a mating region between two connecting parts so that the complete assembly can transfer
forces and motion without separation. For typical seals, the seal material must:
• Conform to all irregularities of the mating surface and prevent any fluid passage or leakage
• Resist extrusion (caused by shear forces) due to pressure differential between the pressured and nonpressured
sides of the seal.
The most common sealing applications are rotating-shaft sealing, elastomeric compressive sealing, and
flange sealing. Sealing applications can be classified into two types: static and dynamic sealing. This
problem focuses on a static elastomeric sealing application modeled as a 2-D nonlinear finite-element
problem.
Static elastomeric seals are characterized by a fixed mating surface, a moving mating surface, and the
bead (seal material). The bead is typically compressed between the mating surfaces, providing a fluid-
tight interface between them, as shown:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 15
Nonlinear Analysis of a 2-D Hyperelastic Seal Using Rezoning
Figure 2.2: Schematic of Finite-Element Static Elastomeric Seal Assembly with Dimensions
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
16 of ANSYS, Inc. and its subsidiaries and affiliates.
Problem Description
• The coefficient of friction (between the bead and the mating surface) is not critical for static seals.
The pressure improves seal integrity, but can also create extrusions of the bead. The extrusions are
due to shear stresses (caused by differential axial pressure created by the fluid, as an extrusion gap
is provided).
• As fluid pressure increases, sealing capacity improves but bead extrusion can occur.
The fluid pressure on the seal surfaces tends to compress the seal axially, forcing the deformed bead
further into the gland therefore improving contact. As the fluid pressure increases, the seal's internal
shear stresses also increase slowly, driving the seal material into the extrusion gap.
The degree and severity of the extrusion depends on the operating pressures, operating temperatures,
geometry of the clearance gap, and seal shape and material.
This type of extrusion typically causes nibbling damage [1] to the seal, caused by large tensile stresses
close to the free surface of the extruded material.
The following figure illustrates the progression of seal deformation, showing the seal gland being filled
and the corresponding bead extrusion as fluid pressure is applied:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 17
Nonlinear Analysis of a 2-D Hyperelastic Seal Using Rezoning
If one or more regions are selected for rezoning at the same time during the analysis, the process is
called horizontal rezoning.
If one or more rezoning operations are performed at different times during the analysis, the process is
called vertical rezoning.
1. Within a given load step, select the substep at which rezoning should be initiated.
3. Remesh the selected region(s), replacing the old (distorted) mesh with a new mesh.
After remeshing has occurred, boundary conditions and loading from the old mesh are mapped to
the new mesh.
4. Map state variables from the old mesh to the new mesh.
This step establishes equilibrium in the new mesh by balancing residual forces.
Remeshing can occur by reading in a generated mesh or a generic third-party mesh (CDB file), or by
using element-splitting refinement of selected regions of the distorted mesh.
In this example, remeshing via the element-splitting method is used because splitting increases the
number of degrees of freedom in a given region without changing mesh topology. This remeshing
method is especially useful for modeling material flow through confined cavities (the primary physical
behavior of hyperelastic seals).
With element splitting, the mesh topology does not change in the interior of the selected regions where
elements are simply split evenly. In the transition region between the refined (new mesh) and the old
mesh, however, layers of transition of elements are generated, changing the mesh topology in these
regions. The transitions can be all quadrilateral or can be composed of degenerate quadrilaterals.
Typically, an all-quadrilateral transition requires more than one element layer to create a compatible
mesh between the split and unsplit regions. The degenerate elements can do the same in just one
element layer; however, a degenerate element transition is more prone to locking than a quadrilateral
element transition zone.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
18 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
2.3. Modeling
This example is modeled as a plane strain problem.
The moving and static mating parts are considered to be rigid. The seal bead is modeled as a hypere-
lastic material.
The seal is chemically bonded (zero applied displacement) to the static rigid mating part, as shown in
Figure 2.2: Schematic of Finite-Element Static Elastomeric Seal Assembly with Dimensions (p. 16).
The pressurization gap is the location for the fluid insertion. The moving rigid mating part moves
downward, reducing the 0.02-inch gap to a 0.002-inch gap, at which an extrusion may form.
The notch indicates the region where self-contact in the seal bead is expected to occur.
1. The top of the seal is compressed first by moving the rigid part downward by 0.018 inches, leaving a small
gap of 0.002 inches at the top (TIME = 0 - 1 s). This load step simulates the first phase of the seal assembly
where the rigid mating parts come together to form the seal gland which the seal bead must eventually
fill.
2. The temperature of the seal is increased from the current room temperature (72° F) to 302° F (TIME = 1 - 2
s).
This load step simulates the thermal loading phase, where the seal is allowed to expand laterally,
filling most of the clearance gaps.
3. A uniform pressure of 4000 PSI is applied in the pressurization gap (to all open status contact elements),
effectively pressurizing the seal (TIME = 2 - 3 s). This is the pressure exerted by the sealing fluid in the
pressurization gap.
This load step simulates the injection of the fluid, which pressurizes the seal laterally. The seal fills
all clearance gaps and possibly creates some material extrusion.
The nonlinear sparse solver is used for the solution. A static analysis is required when geometric nonlin-
earity is present (NLGEOM,ON).
The contact elements are modeled with CONTA172 with augmented Lagrangian formulation (KEYOPT(2)
= 0). The stiffness updates of the contact elements are done at each iteration, based on the mean stress
of the underlying solid element (KEYOPT(10) = 0).
The fixed mating part is chemically bonded to the seal bead, as shown in Figure 2.2: Schematic of Finite-
Element Static Elastomeric Seal Assembly with Dimensions (p. 16). The bonding is modeled by constrain-
ing the displacements on the seal bead in the chemically bonded boundaries (D).
The seal bead material is modeled using an Ogden hyperelastic material, as shown in this input fragment:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 19
Nonlinear Analysis of a 2-D Hyperelastic Seal Using Rezoning
TB,HYPE, 1, 1, 2, OGDEN
TBTEMP,0.000000
TBDATA, 1, 2.80000e+00, 7.90000e+00, -1.86000e+02,
TBDATA,4,-1.85000e+00, 1.00000e-05, 0.00000e+00,
For more information about the seal bead material, see Material Properties (p. 21).
In load step 3 (p. 19), the fluid-penetration loads are applied as a uniform element pressure in the
pressurization gap (shown in Figure 2.2: Schematic of Finite-Element Static Elastomeric Seal Assembly
with Dimensions (p. 16)). The fluid-penetration loads are applied via the following command sequences:
• Specific contact elements (on which the fluid penetration loads are to be applied) are selected as follows:
esel,s,,,2166,2186
esel,a,,,2248
esel,a,,,2250,2261
esel,a,,,2556
The third argument is set to 2 to allow the SFE command to specify the starting points for the fluid penet-
ration. The argument -1 specifies that none of the selected element nodes can be considered a starting
point for fluid penetration.
The command specifies a pressure magnitude of 4000 PSI on face 1 (the default face for 2-D contact elements)
of the selected elements.
The input specifies that the nodes of the contact elements 2186, 2248, and 2556 are starting points, as they
are initially exposed to the fluid. Also, depending on the contact status (open or closed), the node can either
be a fluid penetrating point (for “open” contact) or can no longer be a starting point (if contact closes).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
20 of ANSYS, Inc. and its subsidiaries and affiliates.
Analysis and Solution Controls
finish
In this case, the load step selected to start rezoning is 2. Element-splitting refinement is used to increase
the number of degrees of freedom in the selected region to enhance flexibility. All-quadrilateral transition
elements are generated to connect the new refined mesh to the old unrefined mesh.
Vertical rezoning is used for this problem. Three nested rezonings occur at:
Material properties are provided for one temperature data point. The material properties are assumed
not to change with temperature.
For details about this material model's strain-energy function, see Ogden Hyperelastic Material Constants
in the Material Reference.
Loads are generally applied as pressure/displacement on the moving mating surface, as temperature
on the seal bead, and sometimes as fluid pressure in the seal cavity. The loads cause progressive de-
formation, as seen in Figure 2.1: Schematic of a Static Elastomeric Seal Assembly (p. 16).
Although a moving mating part exists, the analysis is considered to be quasi-static; that is, velocity of
the moving mating part is very small, constant, and does not change direction.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 21
Nonlinear Analysis of a 2-D Hyperelastic Seal Using Rezoning
• Load Step 1:
time,1.0 ! analysis up to 1.0s
nsubst,50,2000,50 ! initial, maximum and minimum number of
! substeps
The corresponding time step sizes are 0.02s, 5.0E-04s, and 0.02s, respectively. To ensure solution accuracy,
the smallest time step is given.
• Load Step 2:
time,2.0 ! analysis upto 2.0s
deltim, 0.01, 1.e-05, 0.1 ! initial, minimum and maximum time step
• Load Step 3:
time,3.0 ! analysis upto 3.0s
deltim, 0.001, 1.e-05, 0.01 ! initial, minimum and maximum timestep
To minimize the possibility of convergence issues, the initial time step for load step 3 is the smallest.
Convergence with fluid penetration loads is more difficult to achieve because, depending on the contact
status, loaded contact surfaces may become unloaded and reloaded during solution.
Remeshing is usually necessary when the mesh becomes too distorted (when element angles approach
180 degrees). If old elements are too distorted at the rezoning step, however, mapping variables from
the old mesh to a new mesh can be difficult. At extreme distortions, it is not possible to map new nodes
at the corners of old elements accurately, affecting boundary condition mapping. This concern is critical
in element splitting because old nodes are shared in new elements generated by splitting.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
22 of ANSYS, Inc. and its subsidiaries and affiliates.
Analysis and Solution Controls
Element splitting does not change the mesh topology in the interior of the rezoned domain. The element
angles before and after splitting remain the same, and the element aspect ratios remain the same;
therefore, a badly shaped element cannot be improved by splitting alone. For best results, element
splitting should occur a few substeps before the elements angles approach 180 degrees.
Mapping on state variables and residual equilibration can be difficult if the remeshing is done in the
presence of very high stress/strain gradients. Remeshing should therefore be attempted at substeps
where the gradients are not so high.
In general, the earlier remeshing is done in a load step, the more difficult it becomes for the problem
to converge during restart. (It is also difficult to determine the correct rezoning region in the beginning
of a loadstep, especially if it is the first loadstep.) Conversely, if remeshing occurs too late, the mapping
may not succeed because the base mesh is already too distorted or large stress/strain gradients have
developed.
For best results, attempt the first remeshing at a substep (S1) toward the later stages of a load step if
that substep is not very close to divergence and where rezonable regions in the problem are clear. If
successful, the solution can proceed; otherwise, attempt another remeshing at a different substep (S0)
prior to S1, which reduces the element sizes by the time the remeshing at S1 can occur. This size reduc-
tion smooths out possible gradients and creates less element distortion by the time the solution reaches
S1.
In this problem, the first rezoning is attempted first at load step 1, substep 20 (TIME = 0.4 s), which
corresponds to substep S0. The next rezoning is done at load step 1, substep 50 (TIME = 0.8996 s) which
corresponds to substep S1. If rezoning at S1 is attempted on its own, convergence does not improve
(because at 0.8996 seconds, the seal bead already deforms to a large degree and creates mapping dif-
ficulties).
The focus of rezoning should be to perform the minimum number of remeshing steps possible while
ensuring that the simulation results and the model mirror the actual physical problem behavior as
closely as possible. It is possible that a nonlinear problem may converge without rezoning; however,
the results may not be indicative of true physical behavior. For example, this seal problem was run
without rezoning, producing the following results:
Figure 2.4: Deformed Mesh Plot of Seal After Third Load Step Without Rezoning
Because an extrusion gap is present and the problem is classified as a medium-to-low pressure seal,
the extrusion is expected to occur. Without rezoning, a successful convergence in this case fails to
provide a physically meaningful solution, as the mesh size is greater than the extrusion gap and therefore
extrusion does not occur.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 23
Nonlinear Analysis of a 2-D Hyperelastic Seal Using Rezoning
Using rezoning in the third load step (where fluid pressure is applied) causes convergence difficulties
during remapping. Large shear-strain gradients develop near the extrusion due to the pressure differ-
ential caused by the fluid pressure. It is therefore necessary that remeshing occur before the fluid pressure
is applied so that mapping does not occur when large strain gradients are present. To enable material
flow in the extrusion gap, however, element sizes near the gap should be sufficiently small by the time
the third load step is applied.
For this problem, optimal rezoning substeps occur in load steps 1 and 2. The magnitude of deformation
in the seal bead is much larger in load step 1 (where the mating parts move) as opposed to load step
2 (a uniform temperature load). Because most of the deformation occurs in load step 1, remeshing is
more critical there than in load step 2 (and two vertical rezonings in load step 1 are justified).
Load step 2 is a temperature load which expands the seal axially, thereby filling up most of the clearance
gaps. The thermal expansion also contributes to the seal bead material being pushed through the ex-
trusion gap. At least one local rezoning is required (around the extrusion) at load step 2 so that the
incipient extrusion behavior can be modeled, and so the third and final rezoning is done in load step
2, substep 10 (TIME = 1.145s).
• Manually (ESEL)
Only solid elements are candidates for splitting. Contact and target elements are ignored.
If no elements are selected prior to mesh splitting (REMESH,SPLIT), all solid elements in the model are
split and no transition elements are created.
Existing contact and target elements attached to base solid elements which have been split, and those
which have been replaced by smaller transition elements, are automatically deleted.
or
REMESH, SPLIT,,,,TRAN,QUAD
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
24 of ANSYS, Inc. and its subsidiaries and affiliates.
Analysis and Solution Controls
An all-quad transition is usually more than a single element layer deep. It is topologically difficult to
resolve nodal incompatibilities from the split to the unsplit elements in a single-element layer, especially
in an unstructured quad mesh.
Using degenerate transitions between the split and unsplit elements usually resolves nodal incompat-
ibilities in a single element layer. Triangular elements tend to lock, however.
Provided that the transition elements are not too distorted and do not overlay other transition elements,
it is preferable to use an all-quad transition for nonlinear problems. The degenerate transition is useful
in cases where the quad transitions become too distorted or where the problem is not bending- or
compression-dominated. In this problem, an all-quad transition is used to prevent volumetric locking.
With nested splitting schemes, it is possible to prevent the all-quad transition regions from intersecting.
The splitting operation subdivides a quad element into four quad elements, a triangular element into
four triangular elements, and a degenerate element into three quadratic elements. After one cycle of
splitting a quad base element, therefore, the split elements are 1/4 the area of the base element.
If N such splittings occur, the final split elements have an area of (1/4)N of that of the base element. If
multiple horizontal rezonings overlap or multiple vertical rezonings are done on the same region, it
becomes increasingly difficult to generate proper transitions due to the increasing size disparity of the
split and unsplit elements.
One method for overcoming the size disparity between split and unsplit elements is to design a nesting
scheme for splitting during vertical rezoning. The nesting should allow the Nth remeshing region to be
sufficiently larger than the (N+1)th remeshing so that the transitions between the two regions do not
overlap.
Using such nesting schemes, it is possible to design mesh grading from coarse to fine, improving
solution performance. In this problem, the nesting scheme allows the remeshed regions in the first,
second, and third rezonings to be successively reduced while overlapping, therefore creating a good
gradation of the mesh.
Nesting schemes can also alleviate the effect of large tensile strains on element splitting strategies,
where aspect ratios of split elements can be adversely affected.
2.6.3.3. Step 3. Map Quantities from Old to New Mesh and Rebalance Residuals
After remeshing (REMESH,FINI), corresponding new contact and target elements which respect the to-
pology of the split solid elements are generated automatically for the meshed region. Isolated rigid
target elements (as done in the modeling of the moving mating part) cannot be remeshed and persist
throughout the life cycle of the analysis.
At this stage, it is necessary to map boundary conditions, loads, and temperatures from the old mesh
to the new mesh.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 25
Nonlinear Analysis of a 2-D Hyperelastic Seal Using Rezoning
The MAPSOLVE command maps the state variables from the old mesh to the new mesh and balances
the residuals by modifying the displaced state of the new mesh. The presence of large stress or strain
gradients in the old mesh may cause nonconvergence of the residuals in the new mesh. View the result
plot to determine if the mesh substep can be split depending on the presence of stress/strain gradients.
Convergence problems during mapping can also occur if a very large size difference exists between
the elements of the old and new meshes. Regions are allowed to overlap during horizontal splitting
operations, so it is possible to generate an arbitrarily small element size if many such regions overlap
(given that the new element area after N such overlapped horizontal splitting operations is (1/4)N of
the old element size). Nesting (p. 25) the remeshed regions can alleviate this problem.
In most cases, convergence difficulties during mapping also indicate convergence difficulties during
the analysis restart. It is therefore good practice to remesh a few substeps earlier (if possible) if MAP-
SOLVE requires a large number of bisections to converge.
For contact-element generation and remapping, the program attempts to detect a new node (of a
split element) in an old element. Because element splitting retains the nodes of the old element in
the new elements, the algorithm often involves detecting a new node at the corner of an old element.
Element splitting also retains the element angular values, so for badly skewed elements, it is difficult
to map a corner node of a new element to the corresponding parent element. This mapping difficulty
can cause the creation of new contact elements to fail, and boundary conditions may not be properly
generated for the new mesh.
It is therefore important to perform element splitting before the element becomes too badly skewed.
The guidelines used to select the substep (p. 22) and region (p. 24) for splitting also apply when
considering how base element skewness may affect contact- and boundary-condition generation.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
26 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
(a) Deformation profile at load step 1, substep 20, time = 0.4 sec, at which the first
rezoning occurs
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 27
Nonlinear Analysis of a 2-D Hyperelastic Seal Using Rezoning
(c) Close-up view of the refined region in (b), showing the element edge pivot at the
corner around which the mesh must “rotate”
At the first rezoning, shown by (a) in the figure, the mesh is split around the region where the extrusion
is expected to take place. The mesh topology is adjusted so that there is an edge pivot at the corner,
around which the material is expected to flow into the extrusion gap (c).
A slightly larger area in (a) is selected for refinement because subsequent rezoning refinements are in-
tended (using nesting). The larger area ensures better self contact in the notch (p. 16) in later stages,
as shown:
(a) Deformation profile at step 1, substep 50, at which the second rezoning occurs
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
28 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
(c) Close-up view of the refined region in (b) showing the start of material flow into the
extrusion gap about the pivot
In the absence of the edge pivot at the corner of the extrusion gap (shown by (c) in Figure 2.5: Deform-
ation Profile at Load Step 1 -- Substep 20 (First Rezoning) (p. 27) and Figure 2.6: Deformation Profile at
Load Step 1 -- Substep 50 (Second Rezoning) (p. 28) the element at that location would collapse from
compression as the seal material begins to flow into the extrusion gap.
The nested element-splitting scheme (p. 25) gradually increases the number of degrees of freedom in
the extrusion region to model the material flow in that region with increasing loads, as shown:
(a) Deformation profile at load step 2, substep 10, at which the third rezoning occurs
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 29
Nonlinear Analysis of a 2-D Hyperelastic Seal Using Rezoning
(c) Close-up view of the refined region in (b), showing the material flow into the extrusion
gap about the pivot
As shown in (b) and (c), the nested refinement by element splitting can gradually increase the number
of degrees of freedom in the region of the extrusion gap and correctly predict the material flow. The
nested refinement creates a thin layer of transition elements for each block of split elements and places
them far from each other, preventing the formation of badly shaped elements close to this high strain
zone (which can lead to premature convergence).
After the three rezoning operations have been performed, the problem converges correctly at the end
of the third load step. The following figure shows the final meshes:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
30 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
(b) Close-up view of the refined region in (a), showing the extent of extrusion of the
seal bead through the extrusion gap
As shown by (a) and (b), the primary objectives of the seal assembly simulation have been met:
• The seal bead model is adequate to enable it to establish proper contact and fill the seal gland.
Subsequent results (shown below) quantify the extent of fluid sealing capacity and the uniformity
of the contact pressures.
Depending on the seal pressure, the extrusion may or may not damage the material. The addition
of extra degrees of freedom (via multiple vertical nested rezonings) allows the material enough
flexibility to flow through the extrusion gap.
To determine if the seal assembly analysis predicts the extent of the seal's integrity (that is, the capacity
to successfully fill the gland and develop sufficient pressure to prevent fluid leakage), the contact
pressure distribution must be evaluated at the three load steps. Because the seal is molded in the
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 31
Nonlinear Analysis of a 2-D Hyperelastic Seal Using Rezoning
metal groove in the bottom and sides (p. 16), it is necessary to consider only the contact conditions
on the top part of the seal.
The following figure shows the contact pressure-distribution trends at the load steps:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
32 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
As shown by (a), at the end of load step 1 (where the moving rigid mating part descends by 0.018
inches), the maximum contact pressure develops as expected at the region where the maximum geometry
change occurs. The incomplete and uneven distribution of the contact pressure indicates that the seal
assembly forces are not yet activated. The notch (p. 16) is not completely closed at this stage, providing
incomplete sealing capacity.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 33
Nonlinear Analysis of a 2-D Hyperelastic Seal Using Rezoning
As shown by (b), at the end of load step 2 (where the seal temperature is raised from 720 F to 3020 F),
the maximum contact pressure now develops at a self-contact zone bounding the pressurization gap.
The reason for this behavior is that load step 2 is a temperature load, where the temperature of the
seal bead is raised uniformly by 2300 F. The uniform temperature increase expands the seal into the
pressurization gap (and, to a smaller extent, the extrusion gap). As the seal expansion occurs, some
pressure releases from the crown region, and contact pressure increases in the self-contact zone, as the
material tries to close the pressurization gap.
Compared to load step 1 (shown by a), the contact pressures are better distributed at the rigid flexible
interface between the mobile mating part and the seal bead. The notch is closed better than in load
step 1, and the contact pressure of the self-contact developed in the closed notch is higher. In order
to maintain seal integrity in case the rigid mating parts move during operation, it is essential that the
self-contact pressures in the notch are high.
Parts (c) and (d) of the figure show the final contact pressure distribution and the corresponding fluid
pressure distribution (in the pressurization gap) at the end of the third load step (fluid pressure in the
pressurization gap). The action of the applied fluid pressure is to expand the pressurization gap, which
in turn creates an axial (along the X axis) pressure on the seal. The axial pressure causes the maximum
contact pressure to develop in the notch, thereby creating better seal integrity. The application of fluid
pressure also causes a pressure differential, leading to the extrusion. Some contact pressure is created
at the neck of the extrusion (which by design should not rupture the seal material). The interface pressure
(as shown in (c)) indicates the contact pressure at the interface of the penetrated fluid and the solid
material. The interface pressures range from 4300-4500 PSI; the pressures are greater than the applied
fluid pressure of 4000 PSI, implying that the seal will not leak under the present set of loads. Downstream
of the interface pressure regions, the contact pressure increases, guaranteeing the fluid-sealing capacity.
The overall contact pressure profile (at the seal-mating surface boundary) after load step 3 is more
uniform than that in load steps 1 or 2 (as shown in (a) and (b)), so the seal integrity is established at
this point. Local high-contact pressure peaks can cause surface damage to the seal. The evolution of
the contact pressures show that local high-contact pressures have not developed except in the extrusion
region.
To ensure seal bead life, the assembly and functional loads on the seal must not create large pressure
differentials in the seal bead. Pressure differentials cause large shear stresses which can rupture the
material, so be sure to check the hydrostatic pressure development in the seal (p. 35). Unusually high
hydrostatic pressures at finite deformation may also indicate volumetric locking in the element.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
34 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 35
Nonlinear Analysis of a 2-D Hyperelastic Seal Using Rezoning
As shown by (a) and (b), pressure differentials exist (especially around the notch and the pressurization
gaps), indicating incomplete seal assembly. At the end of load step 3 (c), the addition of fluid pressure
causes a nearly uniform pressure in the seal, except for a peak in the vicinity of the extrusion. The peak
pressure near the extrusion and the free surface of the seal bead at the extrusion gap actually drives
the extrusion and causes shearing strains.
The largest assembly forces in the seal are applied vertically via the rigid mating surfaces. For optimal
seal stresses, the σYY profile inside the seal must be as uniform as possible when all assembly loads
have been applied. The following figure shows the σYY profile of the seal bead under the three load
steps, respectively:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
36 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 37
Nonlinear Analysis of a 2-D Hyperelastic Seal Using Rezoning
As shown by (a) in the figure, the maximum tensile stress occurs at the extrusion, and the maximum
compression occurs at the crown region where maximum geometry change has occurred. The tensile
stress occurs because the extrusion gap causes a reentrant corner to be formed at the seal (where stress
concentrations may occur).
At the stage shown by (b), the temperature increase tries to expand the seal through the pressurization
and the extrusion gaps, thereby moving the maximum compressive stress to a location close to the
extrusion.
At (c) in the figure, fluid pressure is applied and the overall stress distribution is nearly uniform, creating
near-optimal sealing conditions. Maximum tensile and compressive stresses are localized near the ex-
trusion.
At the end of the third load step, the rigid-flexible contact pressures are near-uniform, preventing fluid
leakage. The overall pressure and the σYY stresses are near-uniform in the seal bead, indicating optimal
seal assembly loading.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
38 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Evidence of shearing-contact pressure spikes due to pressure differentials exists and is localized near
the extrusion. This condition has been modeled adequately by the nested step-wise rezoning with
element-splitting refinement used in this problem.
The following figure shows the σYY profile at the extrusion after the third load step:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 39
Nonlinear Analysis of a 2-D Hyperelastic Seal Using Rezoning
The maximum tensile and compressive stress peaks at the end of the third load step are due to the
singularity near the reentrant corner. Thus, with reference to (c) in Figure 2.10: Seal Hydrostatic Pres-
sures (p. 35), the σYY profile is virtually uniform in the seal bead.
The following figure shows the σXY profile at the extrusion after the third load step:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
40 of ANSYS, Inc. and its subsidiaries and affiliates.
Recommendations
The maximum and minimum shearing stresses occurring at the location shown are caused by the
pressure differential between the free surface of the seal bead and the seal bead itself. The shearing
stress occurs as a result of the shearing strain created by the extrusion.
2.8. Recommendations
To perform a similar type of analysis, consider the following recommendations:
Rezoning with element splitting is essential for solving this type of problem properly. Extrusion
modeling is critical for simulating accurate physical behavior of the seal. A gradual reduction of element
size in the vicinity of the extrusion gap is necessary as the material flow into the gap increases over
time.
See the edge pivot shown by (c) in Figure 2.5: Deformation Profile at Load Step 1 -- Substep 20 (First
Rezoning) (p. 27). The pivot prevents the elements at the reentrant corner from deforming too severely
under compression and can, in general, improve convergence. Providing for a pivot in the starting
mesh allows it to remain until the end of the analysis, as element splitting does not change mesh
topology.
Large stress and strain gradients are not easily equilibrated (MAPSOLVE) and can cause nonlinear
convergence issues. The problem may converge better if rezoning is done early enough when the
gradients have a lesser chance of occurring.
Performing rezoning too early in the analysis process, however, may be counterproductive as the
mesh may not have sufficiently deformed, possibly causing the rezoned mesh itself to deform later
in the analysis. It is also difficult to discern regions likely to benefit from rezoning too early in the
analysis.
When element-splitting refinement is used for rezoning, ensure that the transition regions of over-
lapped rezoned meshes do not intersect to a great degree. Excessive intersections can cause badly
shaped elements.
If using vertical (p. 18) rezoning with element splitting, nesting the refinements (p. 25) ensures that
transition regions do not overlap, resulting in better element quality. Maintain mesh gradation during
the vertical rezoning process to avoid mesh distortion in large strains and to better resolve large
stress/strain gradients.
The following figure shows the mesh gradation obtained from nesting the remeshing zones properly:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 41
Nonlinear Analysis of a 2-D Hyperelastic Seal Using Rezoning
Figure 2.15: Mesh Gradation Created with Nested Element Splitting During Vertical Rezoning
The mesh area reduces from h to h/64 in three rezoning steps. The nesting scheme ensures that the
transition zones do not overlap, and no badly shaped elements occur. Despite the presence of large
strains, a remeshing strategy using only element splitting refinement does not create elements with
bad aspect ratios, or skew.
Rezoning Hints:
Contact elements (rigid-flex contact, self-contact, and flex-flex contact) and target elements (for
self-contact and flex-flex contact) for the rezoned solid elements are generated automatically at
the end of the remeshing operation (REMESH,FINISH).
Excess rezoning around a reentrant corner does not improve convergence due to the effect of the
singularity.
2.9. References
The following references are used in this example problem:
1. Muller, H. K., B.S. Nau. Fluid Sealing Technology, Principles and Applications. Mechanical engineering: 117.
Marcel Dekker. 1998.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
42 of ANSYS, Inc. and its subsidiaries and affiliates.
Input Files
ANSYS, Inc. gratefully acknowledges Parker Hannifin Corporation for providing the initial mesh, material
description, loading information, and illustrations for the seal used in this problem.
• seal_assembly.cdb -- Common database file for the Parker seal assembly model (called by the
seal_assembly.dat file).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 43
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
44 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 3: Fluid-Pressure-Penetration Analysis of a Sealing System
This example problem demonstrates a method for examining fluid-pressure-penetration effects on a
sealing system. The use of seals is primarily to prevent the transfer of fluid (liquid, solid, or gas) between
two or more regions.
• Frictional contact
• Fluid-penetration loading
• Nonlinear stabilization
3.1. Introduction
The applications of mechanical seals are numerous. The most common sealing systems involve rotating-
shaft, elastomer, and flange sealing.
The sealing capability of an elastomeric seal depends upon the contact stresses that develop between
the seal and the surfaces with which it comes into contact. Leakage may occur when the pressure dif-
ferential across the seal exceeds the contact stress.
As a general nonlinear finite element tool, the ANSYS Mechanical APDL program can:
• Predict contact stress profiles after installation, in operation, and under various loading conditions
• Account for the effects of fluid pressure penetration between seals and other structural components
in the analysis.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 45
Fluid-Pressure-Penetration Analysis of a Sealing System
All of this information is crucial to understanding how mechanical seal systems are designed, how they
operate, and how best to apply them to prevent fluid leakage.
At the start of each iteration, the program finds all possible starting points which are exposed to the
fluid pressure. Among the starting points, the program then finds fluid-penetrating points where the
contact status is open or lost, or where the contact pressure is smaller than the user-defined pressure-
penetration criterion.
When a contact-detection point has a contact condition of "penetrating," both it and its nearest
neighboring nodes are considered to be starting points that are exposed to the fluid pressure.
The fluid pressure begins to penetrate into the interface between contact and target surfaces from the
starting points. The fluid penetration can be cut off when contact between the surfaces is reestablished
or when contact pressure is larger than the fluid-penetration criterion.
The components of the sealing system to be modeled are an elastomer o-ring and a rolled-shape plastic
cap. The figure below is a 3-D view of the model with a one-quarter cutout to show the o-ring config-
uration.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
46 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
3.3. Modeling
Following is an axisymmetric model of the seal components, an elastomer o-ring and a plastic cap. The
rigid surface on the right represents the groove, and the left rigid surface represents the shaft.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 47
Fluid-Pressure-Penetration Analysis of a Sealing System
Property Value
O-ring radius 0.89 mm
Cap width 0.623 mm
Cap height 2.123 mm
Two contact pairs are defined using CONTA172 and TARGE169 elements. One rigid-to-flexible contact
pair models the contact between the entire exterior surface of seals (o-ring and cap) and the rigid sur-
faces, as shown by (a) in the following figure:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
48 of ANSYS, Inc. and its subsidiaries and affiliates.
Material Properties
Another flexible-to-flexible contact pair models the contact between the o-ring and the cap, as shown
by figure (b).
A frictional interaction is defined for the contact pairs and the friction coefficient of 0.1 is applied (via
the MP command). The contact formulation used is the “Lagrange multiplier on contact-normal and
penalty-on-tangent” (KEYOPT(2) = 3 on CONTA172).
In the following figure, part (a) displays the original uniaxial test data and fitted curve with the Ogden
first-order hyperelastic form. The plastic cap is modeled by an elastic-plastic material which is softer
than the o-ring material. The isotropic hardening stress-strain curve (initialized via the TB,PLAS,,,,MISO
command) is shown in part (b).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 49
Fluid-Pressure-Penetration Analysis of a Sealing System
(a) Uniaxial Test Data and Fitted Curve with Ogden First
Order
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
50 of ANSYS, Inc. and its subsidiaries and affiliates.
Boundary Conditions and Loading
Material Properties
Coefficient of Friction
O-Ring [1] Cap
for Contact Interface
µ1 = 7.5605 MPa
E = 171.47 MPa
µ = 0.1 α1 = 0.54275
υ = 0.4
d1 = 1e-4
For more information about the Ogden form, see Ogden Hyperelasticity in the Material Reference.
For more information about the multilinear curve, see Multilinear Isotropic Hardening in the Material
Reference.
Load Step 1
The housing compression is analyzed in the first load step. The rigid shaft surface (left
side) moves 0.2794 mm in the x direction and the rigid groove surfaces (right side) move
0.05715 in the negative x direction, simulating the housing assembly.
Load Step 2
In the second load step, a fluid penetration pressure of 4.1 MPa is applied to the contact
pairs (via the SFE command with the load key value LKEY set to 1), as shown:
esel,s,real,,6 ! select rigid-flexible contact pair
esel,r,ename,,172 ! reselect contact elements only
esel,a,real,,8 ! select flexible-flexible contact pair
sfe,all,1,pres,,4.1 ! apply fluid pressure
allsel
The following input prevents fluid penetration loads from being applied multiple times
on overlapping contact elements from different pairs:
esel,s,real,,8 ! select flexible-flexible contact pair
nsle
esln,s,1
esel,r,real,,6 ! reselect rigid-flexible contact pair
sfedele,all,all,all ! remove overlapping fluid pressure loads
allsel
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 51
Fluid-Pressure-Penetration Analysis of a Sealing System
Figure 3.5: Sealing System with Applied Fluid Pressure to Contact Pairs
The fluid-penetration starting points are also specified in order to model the system fluid pressure ex-
posed from the bottom part (the opening between the shaft and the lower part of groove). For the rigid-
to-flexible contact pair, no default starting points exist because the contact surfaces are in a closed
loop, as shown in (a). To activate the fluid penetration pressure, two contact elements are chosen as
starting points initially exposed to the fluid (via the SFE command, with the load key value LKEY set
to 2 and the first starting point status value VAL1 set to 1), as follows:
SFE,3121,2,PRES,,1 ! starting point for o-ring
SFE,3308,2,PRES,,1 ! starting point for cap
One starting point is for the contact surface of the o-ring, and the other is for the contact surface of
the cap, respectively.
For the flexible-to-flexible contact pair, the ending points of the contact and target surfaces are the
default starting points. To force the fluid to penetrate from the bottom only, the two default starting
points on the top part of the seals are suppressed (via the SFE command, with the load key value LKEY
set to 2 and the first starting point status value VAL1 set to -1), as follows:
SFE,2529,2,PRES,,-1
SFE,2625,2,PRES,,-1
The gaps (between the o-ring and the shaft, the o-ring and the cap, and the cap and lower part of the
groove) are closed after the first load step, so the fluid does not initially penetrate into those contacting
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
52 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
regions. The fluid gradually opens the contact and penetrates into those regions with increasing pressure
loading. Since a default value of zero for the pressure penetration criterion (PPCN input via the real
constant command R) is used, the fluid pressure penetration occurs immediately as soon as the contact
is open or lost.
The entire system is very unstable during the second load step. The seals are pushed up due to the
fluid pressure. Rigid body motion may occur when the frictional stresses are not large enough to hold
the seals vertically. Convergence difficulty due to an unstable problem is usually the result of a large
displacement for smaller load increments. To overcome the instability, the nonlinear stabilization cap-
ability (STABILIZE command) is used in the second load step, as follows:
STABILIZE, CONSTANT,ENERGY,0.01,ANYTIME
The deformed configuration and the contour of the contact pressures on the seals at the end of the
first load step are shown in the following figure:
Figure 3.6: Sealing System Contact Pressure Following Housing Compression (First Load Step)
• The middle part of the contact interface between the o-ring and the shaft
• The middle part of the contact interface between the o-ring and the cap
• The top and bottom part of the contact interface between the cap and the groove.
The top and bottom part of contact interfaces between the seals and the shaft remains open.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 53
Fluid-Pressure-Penetration Analysis of a Sealing System
The fluid penetration pressure loads are applied during the second load step. The fluid pressure is active
immediately on elements near the bottom part of contact interfaces where the contact status is open.
The fluid pressure does not spread to the elements around the top part of the seals because gaps have
been closed near the middle part of the interfaces.
The propagation of the fluid penetration is captured in the following figures showing the deformed
shape, the fluid pressure contour, and the contact pressure contour corresponding to different stages
of the applied loading. The fluid pressures applied to the surfaces range from 0.041 MPa in the first
substep to 4.1 MPa in the last substep.
Figure 3.7: Sealing System fluid Penetration Pressure Distributions (During Second Load Step)
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
54 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 55
Fluid-Pressure-Penetration Analysis of a Sealing System
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
56 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Figure 3.9: Sealing System Contact Pressure Distributions (During Second Load Step)
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 57
Fluid-Pressure-Penetration Analysis of a Sealing System
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
58 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Increasing pressure penetration loads applied in the second load step force the o-ring and cap to be
pushed up against the shaft, eventually causing complete contact between the seals and upper part
of the shaft. Throughout the loading history, there is no evidence of complete fluid penetration as the
contact in certain regions remains closed.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 59
Fluid-Pressure-Penetration Analysis of a Sealing System
Figure 3.11: Sealing System Von Mises Stress Contour (After Second Load Step)
Figure 3.12: Sealing System Equivalent Plastic Strain Contour (After Second Load Step)
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
60 of ANSYS, Inc. and its subsidiaries and affiliates.
Recommendations
The analysis also shows a certain degree of extrusion of the plastic cap through the extrusion gap of
the groove near the end of the second load step. Because the plastic cap is softer than the o-ring, de-
formations and stresses in the cap are higher than those in the o-ring.
Figure 3.13: Sealing System Total Strain Energy and Stabilization Energy Time History
In the early stages of loading (TIME < 1.27), slight strain energy (SENE) is developed and the seals are
pushed up almost rigidly. The stabilization energy (STEN) is developed to prevent rigid body motion;
without stabilization, the solution diverges.
In the later stage of the loading (TIME > 1.27), the seals deform as the result of compression and strain
energy is developed while the stabilization energy remains constant. Eventually, the stabilization energy
is less than the strain energy by a factor of three. The stabilization feature helps to prevent potential
rigid body motion and improves the convergence while having little effect on the final results.
3.8. Recommendations
To perform a similar type of analysis, consider the following recommendations:
• Fluid pressure penetration is a path-dependent loading. Therefore, you should define the “starting points”
with great care. In many cases, you might need to remove invalid default starting points and properly define
the physical location of the fluid penetration loading. For more information, see Specifying Fluid Penetration
Starting Points in the Contact Technology Guide.
• You should be careful not to apply fluid pressures multiple times when contact pairs overlap. Duplicate
fluid penetration loads should be removed as demonstrated in Boundary Conditions and Loading (p. 51).
• Determining an appropriate stabilization constant (STABILIZE command) may require some experimentation.
The defined constant should be large enough to prevent rigid body motion, but also ensure that the resulting
stabilization energy is much smaller than the strain energy. Trial and error is often required to obtain a
suitable value.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 61
Fluid-Pressure-Penetration Analysis of a Sealing System
3.9. Acknowledgments
ANSYS, Inc. gratefully acknowledges Trelleborg Sealing Solutions for providing the geometry and ma-
terial properties used in this example.
• oring.cdb -- The common database file for the axisymmetric model of the o-ring (called by oring.dat).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
62 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 4: Ring-Gear Forging Simulation with Rezoning
This example problem demonstrates the efficiency and usefulness of rezoning in a simulation of metal-
forming processes. Rezoning facilitates the convergence of a nonlinear finite element simulation in
which excessive element distortion occurs.
4.1. Introduction
Finite element analysis plays a significant role in designing and analyzing complex metal-forming pro-
cesses that improve product quality. Successful simulation of these processes is often complicated due
to the nonlinear nature of the problems involved. Causes for nonlinearity in such problems include
large-deformation analysis, material plasticity, and the contact needed between the work piece and the
dies.
Rezoning is a powerful tool for overcoming convergence difficulties that can arise due to mesh distortion.
Through its capabilities, a distorted mesh can be repaired and the simulation can be continued to
completion. Rezoning is not intended for cases where the material or structure becomes unstable or
where the mesh deformation has become too severe to repair.
If necessary, several regions can be repaired at the same time (horizontal rezoning), and multiple
rezoning operations are allowed on a region at different times during the analysis (vertical rezoning).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 63
Ring-Gear Forging Simulation with Rezoning
The forging process is simulated using a 2-D axisymmetric model meshed with a 2-D four-node struc-
tural solid element (PLANE182). The model represents an elastoplastic cylindrical block (the work piece)
sitting in a rigid surface (static die).
The block is deformed by another rigid surface (moving die) that moves at an infinitesimal speed, such
that the final shape of the work piece becomes a ring gear with complete die fill, as shown in Fig-
ure 4.1: Original Finite Element Model with Boundary Conditions and Loading (p. 65).
The initial analysis diverges due to excessive mesh distortion when 96 percent of the total loading (TIME
= 0.96) is reached. The initial rezoning is applied at TIME = 0.716 (substep 26), and the entire work piece
is remeshed.
The analysis continues with the new mesh and converges to completion, but the mesh is severely dis-
torted.
A second rezoning is performed at TIME = 0.9 (substep 45), and the entire work piece is remeshed again
to ensure proper die fill and more accurate results at TIME = 1.
The results of effective plastic strains and total strains are compared with results of a similar problem.[1]
A single rezoning at a later stage of the analysis may have been sufficient to solve this problem, but
the solution described here was chosen instead so that the problem could also demonstrate how ver-
tical rezoning is performed.
4.3. Modeling
The cylindrical block is modeled using 396 plane elements (PLANE182) with axisymmetric behavior
(element key option KEYOPT(3) = 1). Options for full integration with the B-bar method (KEYOPT(1) =
0) and mixed u-P formulation (KEYOPT(6) = 1) are also specified.
et,1,182
keyopt,1,1,0 ! B-bar method (default)
keyopt,1,3,1 ! Axisymmetric
keyopt,1,6,1 ! Mixed u-P formulation
The B-bar method and the mixed u-P formulation prevent the volumetric locking that can be triggered
by large plastic deformation. Following is the initial finite element model with loading and boundary
conditions applied:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
64 of ANSYS, Inc. and its subsidiaries and affiliates.
Material Properties
Figure 4.1: Original Finite Element Model with Boundary Conditions and Loading
4.3.1. Contact
Two rigid-deformable contact pairs without friction are defined using target and contact elements
(TARGE169 and CONTA172, respectively) with the Augmented Lagrangian formulation (KEYOPT(2) = 0)
activated.
The first contact pair is defined between the top die (moving) and the top and right edges of the work
piece. The second contact pair is defined between the bottom die (static) and the bottom and right
edges of the work piece.
The material is approximated as an elastoplastic material with a nonlinear isotropic hardening behavior.
The power-hardening law for nonlinear isotropic behavior is represented as:
(4.1)
where:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 65
Ring-Gear Forging Simulation with Rezoning
tb,nliso,1,,power
tbdata,1, 0,n
No boundary conditions are applied to the rigid targets, as they are automatically restrained via the
TARGE169 target element's default behavior (KEYOPT(2) = 0).
The top rigid surface is displaced downwards by 138 mm in one load step. The downward displacement
compresses the cylindrical block (work piece), and the material flows to fill up the space between the
dies.
The end time is 1. The initial, minimum, and maximum time step sizes are 20, 2000, and 10, respectively;
they are applied with an initial time step size of 0.05 and a minimum time increment of 0.5e-3.
Because the program execution ends if the analysis fails to converge, the analysis is continued after
failure by performing rezoning and a multiframe restart. (Without the NCNV command, running the
full analysis in batch mode would not be possible due to analysis termination after the first divergence.)
All restart files must be available; otherwise, it is impossible to know which substeps to perform to ini-
tiate the rezoning. It is also a good idea to save the results; doing so makes it easier to investigate the
reason for solution termination and decide which substep to use to activate rezoning. If it is not possible
to save results (due to disk space limitations, for example), they can be generated for a specific substep
from the restart files (ANTYPE,,REST,Loadstep,Substep,RSTCREATE).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
66 of ANSYS, Inc. and its subsidiaries and affiliates.
Analysis and Solution Controls
Generally, the best way to select a substep is by visual inspection of results in the postprocessing
module (/POST1). It is helpful to use the most important element solutions (PLESOL) to create animation
files that capture the specific moment when the mesh becomes sufficiently distorted, but not yet
severely distorted (that is, when internal angles approach or exceed 180 degrees).
It is good practice to correlate the element solutions plots with corresponding steps in the monitor file
(.mntr) and the convergence pattern in general. To initiate rezoning, select a substep where the current
time increment is not close to the specified minimum and where the convergence pattern of the solution
has not slowed down significantly (before bisections have started to occur).
In this example, no more than two rezonings are needed because the initial divergence occurs close
to the end of analysis (TIME = 0.96). The initial rezoning cannot be performed very close to the end of
the last converged substep, because another rezoning is planned before the end of the solution. If the
initial rezoning is performed at the beginning of the analysis, large subsequent deformations distort
the refined mesh too early in the process, necessitating more vertical rezonings.
In the following figure, the element solution of effective plastic strains at this stage shows the deformed
mesh, where some of the elements exhibit a very high aspect ratio and internal angles close to 180
degrees:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 67
Ring-Gear Forging Simulation with Rezoning
Figure 4.2: Effective Plastic Strain and Deformed Mesh at Time of First Rezoning
Substep 47 (TIME = 0.9138) or thereabouts appears to be best suited for remeshing from the perspective
of limited element distortions. The final selection of substep 45 (TIME = 0.9s) as the optimal substep,
however, is guided by the convergence from the prior balancing of the residual forces (MAPSOLVE).
Element distortions guide the general range of substeps best suited for remeshing. The residual-force
balancing determines the precise substep.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
68 of ANSYS, Inc. and its subsidiaries and affiliates.
Analysis and Solution Controls
Figure 4.3: Effective Plastic Strains and Deformed Mesh at Time of Second Rezoning
Generally, if the model is too large or too complex or if the mesh distortion is localized, horizontal
(multiple) remeshing is a better option than remeshing the entire model.
If the model is not large or the mesh distortion is widely spread as in this problem, the entire model
can be selected for remeshing.
Horizontal rezoning may be necessary depending on the model size, computational costs, and the degree
of accuracy required.
A region selected for remeshing should contain all of the highly distorted elements and, if possible,
should be slightly larger than the area containing the distorted elements. The only condition is that all
regions must be of the same element type, material type, thickness (for plane stress), and nodal coordin-
ate system (except for boundary nodes).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 69
Ring-Gear Forging Simulation with Rezoning
in the new mesh can work, provided that the new element connections and topology reflect better
shape characteristics.
For this problem, a new mesh generated as a .cdb file is read in (REMESH,READ). The new mesh contains
only solid elements, as the contact elements are generated automatically when remeshing is complete
(REMESH,FINISH).
The rigid target elements remain the same throughout the analysis and cannot be remeshed. While it
is possible for the new mesh to contain contact and target information as well, it is faster and more
reliable to read in only the remeshed solid elements and allow the contact/target elements to be gen-
erated automatically.
The initial element size is approximately 4. The first rezoning decreases it by a factor of two, and the
second rezoning uses an element size of 1.7. The new meshes appear in Figure 4.4: New Mesh Read in
During the First Rezoning (p. 70) and Figure 4.5: New Mesh Read in During the Second Rezoning (p. 71)
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
70 of ANSYS, Inc. and its subsidiaries and affiliates.
Analysis and Solution Controls
The quality of the new mesh, and the mapping operation in general, can be sensitive to mesh size. If
the mesh quality is mediocre due to a less-than-optimal element size, mapping difficulties may occur,
leading to convergence failure. In such cases, a better mesh using a different element size is necessary.
Although the default is five substeps, it is generally better to use more substeps, especially when contact
is included in the problem. This problem uses 500 substeps as the maximum in the case of convergence
problems becoming an issue. Using a larger number of substeps does not affect computational efficiency
and is recommended for solving nonlinear problems characterized by large strains.
The results of the element solution of effective plastic strains after the first and second mapping are
shown in the following two figures. The results are compared (p. 74) with the same plots before mapping
(as shown in Figure 4.2: Effective Plastic Strain and Deformed Mesh at Time of First Rezoning (p. 68)
and Figure 4.3: Effective Plastic Strains and Deformed Mesh at Time of Second Rezoning (p. 69)).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 71
Ring-Gear Forging Simulation with Rezoning
If the old mesh is heavily distorted by rezoning, or if the differences in mesh density and topology
between the old and new mesh are especially large (specifically in the elements on boundaries), mapping
variables may become difficult, possibly resulting in errors or convergence failure.
If mapping leads to a successful convergence despite the distortion, the results may be slightly different
than those which existed before mapping. To observe this effect, compare Figure 4.2: Effective Plastic
Strain and Deformed Mesh at Time of First Rezoning (p. 68) to the following figure
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
72 of ANSYS, Inc. and its subsidiaries and affiliates.
Analysis and Solution Controls
Convergence issues after mapping often indicate that the restarted analysis will also have convergence
difficulties; therefore, if mapping uses many substeps to achieve equilibrium, try to rezone from an
earlier substep or improve the quality of the new mesh.
The following input shows the vertical rezoning process as described for this problem:
/clear,nostart
/file,ringforging
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 73
Ring-Gear Forging Simulation with Rezoning
rezone,manual,1,45
remesh,start
remesh,read,mesh2,cdb ! Read in the second mesh with esize=1.7
allsel,all
remesh,finish
mapsolve,500,pause
finish
/solu ! Second multiframe restart
antype, , rest, , , continue
solve
finish
The effective plastic strain distribution before and after the first rezoning is shown in Figure 4.2: Effective
Plastic Strain and Deformed Mesh at Time of First Rezoning (p. 68) and Figure 4.6: Effective Plastic Strain
Distribution After First Mapping (p. 72), respectively.
The same contours apply before and after the second rezoning, as shown in Figure 4.3: Effective Plastic
Strains and Deformed Mesh at Time of Second Rezoning (p. 69) and Figure 4.7: Effective Plastic Strain
After Second Mapping (p. 73), respectively. The effective plastic strain distribution and the similar contours
both indicate that the mapping of data was achieved with accuracy.
Because the mesh changes significantly during the first rezoning, it is expected that results before and
after first mapping are different (by about 0.1 percent in this case). The difference is due to the balancing
of residuals that are larger when compared to the situation when mesh changes are not significant (as
seen in the second rezoning).
Few new elements were added during the second rezoning; therefore the maximum effective strains
are identical before and after mapping. The maximum value of effective strain is 3.53, which is close to
the reference effective strain value.[1] The contours of effective strains are similar to those of the refer-
ence, with a region of very concentrated strains in the middle part where the material flows.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
74 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Figure 4.8: Effective Strain Distribution and Deformed Shape at the Final Stage (Complete Die
Fill)
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 75
Ring-Gear Forging Simulation with Rezoning
The von Mises stress distribution in the following figure shows that elastic deformation is extremely
small compared to plastic deformation:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
76 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The tiny elastic deformation occurs because the minimum stresses are much higher than the initial yield
stress of the material.
In this figure, a half 2-D axisymmetric expansion is performed on the von Mises stress, which offers a
better visualization of the plastic flow in this process:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 77
Ring-Gear Forging Simulation with Rezoning
Figure 4.11: Half Expanded Plot of Von Mises Stress at the Final Stage of Ring-Gear Forging
4.8. Recommendations
To perform a similar ring-gear forging simulation using rezoning, consider the following hints and re-
commendations:
Use rezoning only in problems experiencing mesh distortion as a result of large deformation.
Determine the optimal substep after carefully examining the element solution, the deformed mesh
(in /POST1, and the convergence pattern.
The best substep to rezone is the one where the mesh is distorted (though not too severely distorted)
and the time increment is not close to (or at) its minimum.
If any of the following situations are encountered, initiate rezoning at an earlier substep or improve
the quality of the new mesh:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
78 of ANSYS, Inc. and its subsidiaries and affiliates.
Input Files
A new mesh can be program-generated (AREMESH and AMESH), created via mesh splitting
(REMESH,SPLIT), or read in (REMESH,READ).
If an external mesh (.cdb file) created by a third-party meshing application is used, two considerations
exist:
-- If a partial region is rezoned, the nodal locations on the interior boundaries must be retained
by the third-party meshing application when creating the new mesh. The locations are ne-
cessary to maintain nodal compatibility between the new and existing meshes.
-- If PLANE183 (quadratic) elements are used in the original mesh, the third-party meshing
software should retain the location of midside nodes by generating the new nodes using
element-based quadratic interpolation.
Multiple regions can be remeshed at the same substep via horizontal rezoning. Multiple regions where
the new mesh is read in (REMESH,READ) cannot intersect, though they may touch at a boundary.
Repeated rezonings can be performed at different locations in time via vertical rezoning. The only
vertical rezoning restriction is that a program-generated mesh (AREMESH and AMESH) cannot be
used after obtaining a new mesh either by reading it in (REMESH,READ) or by splitting the mesh
(REMESH,SPLIT). (The restriction is due to the loss of associativity of the mesh and geometry which
occurs in cases where the new mesh is read in or created from mesh splitting.)
If the new mesh size is drastically different from that of the old mesh, convergence issues may occur
during mapping. It may be necessary to resize the new mesh to more closely match the old one.
A larger maximum number of substeps may need to be specified to handle convergence difficulties
during mapping and in the restarted analysis.
4.9. References
The following reference was used in this example problem:
1. Kwak, Dae-Young, J.S. Cheon, Y.T. Im.“Remeshing for Metal Forming Simulations - Part I: Two-Dimensional
Quadrilateral Remeshing.” International Journal for Numerical Methods in Engineering . 53.11 (2002): 2463-
2500.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 79
Ring-Gear Forging Simulation with Rezoning
• ringforging.cdb -- Common database file containing the initial mesh information (called by ringfor-
ging.dat).
• mesh1.cdb -- Common database file for the new mesh, read in during the first rezoning (called by ring-
forging.dat).
• mesh2.cdb -- Common database file for the new mesh, read in during the second rezoning (called by
ringforging.dat).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
80 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 5: Delamination of a Stiffened Composite Panel Under a
Compressive Load
This example problem demonstrates the use of solid-shell element technology to model a layered-
composite structure. The problem simulates interface delamination through the debonding capability
of contact elements.
• Bonded contact with a cohesive zone model to simulate delamination (also known as debonding)
• Nonlinear stabilization
5.1. Introduction
Stiffened composite panels are ideal for aircraft fuselage construction because of their excellent durab-
ility and optimal strength-to-mass ratios. The program provides a variety of element types for modeling
layered composite structures. In this example, the eight-node solid-shell element SOLSH190 is chosen
for its general applicability to layered structures and its continuum element connectivity that greatly
simplifies the modeling process.
The unique properties of the SOLSH190 element used in this example greatly simplify the modeling of
contact between thin parts. For example, when using SOLSH190 instead of shells, you do not have to
worry about the section offset, contact surface orientation, or thickness change in large deflection.
The stiffened panel may undergo various local and global failure modes when subjected to a service
load. This example focuses primarily on the global buckling of the panel and the progressive failure of
the bonding material between different structural components. To simulate this highly nonlinear and
unstable phenomenon, the nonlinear stabilization method and bonded contact with a cohesive zone
model are used.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 81
Delamination of a Stiffened Composite Panel Under a Compressive Load
5.3. Modeling
Due to the symmetry of the problem, it is possible to model only one representative section of the
whole panel. The representative section shown below contains a 600 mm x 160 mm portion of the
panel skin and one stiffener assembly.
The section is meshed with SOLSH190 elements, as shown below. Note that the elements on the skin
and the flange do not need to match when the interfaces are modeled with contact elements.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
82 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
A secure bond is assumed between the web and the flange. This perfect bonding is simulated through
the use of common nodes for both components.
The interfaces between the skin and the flange are meshed with CONTA174 and TARGE170 elements.
Two distinct contact pairs are established, as shown in Figure 5.4: Two Contact Pair Definitions (Initial
Crack and CZM Area) (p. 83). Since debonding is permitted in the entire interfacing area between the
skin and flange, KEYOPT(12) = 6 is set for the CONTA174 elements to allow only an initially bonded
contact, and a cohesive zone material (CZM) is assigned to these elements for modeling any subsequent
debonding. An area of artificial imperfection is introduced in the skin-flange interface. In this area, the
bonding material is completely missing and the standard contact behavior (KEYOPT(12) = 0) is assigned
to the CONTA174 elements.
Figure 5.4: Two Contact Pair Definitions (Initial Crack and CZM Area)
The following two figures show the coordinate systems of the contact and target elements.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 83
Delamination of a Stiffened Composite Panel Under a Compressive Load
Refer to the Contact Technology Guide for more information on the extensive contact options and
modeling procedures.
The properties of the two orthotropic materials used in this problem are summarized in the tables below.
They are assumed to be linear elastic and temperature independent. These materials are used to define
the layers of the composite materials that make up the skin, stringer flange, and stringer web (refer to
the shell section definitions (p. 85) later in this section).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
84 of ANSYS, Inc. and its subsidiaries and affiliates.
Material Properties and Section Definitions
The properties of the bilinear cohesive zone material model with the CBDE option are summarized in
the table below. The TB and TBDATA commands are used to define this material model. See Cohesive
Zone Material for Contact Elements in the Material Reference for details on the definition of cohesive
zone models.
Shell sections are used for defining the layup of the composite materials. See the SECTYPE and SECDATA
commands for details on how to define the material, thickness, material orientation, and the number
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 85
Delamination of a Stiffened Composite Panel Under a Compressive Load
of integration points of each layer. The following tables summarize the section properties for the skin,
the flange, and the web.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
86 of ANSYS, Inc. and its subsidiaries and affiliates.
Boundary Conditions and Loading
The periodic symmetry requires that any node on one cut boundary of the representative model moves
in the same way as the corresponding node at the other cut boundary. The coupling condition (CP) is
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 87
Delamination of a Stiffened Composite Panel Under a Compressive Load
shown in (b) in the above figure. APDL scripting is adopted for locating coupled nodal pairs. (See Input
Files (p. 91) for more details.)
An in-plane compressive force of 76666 N in the negative X direction is applied at the pilot node to
induce buckling and debonding.
The energy dissipation ratio is carefully chosen to avoid excessive artificial stabilization energy. As shown
in Figure 5.8: Time History Plot of Strain and Stabilization Energies (p. 88), the level of stabilization energy
(STEN) is low compared to the total strain energy (SENE) throughout the simulation; therefore, the
validity of the simulation results can be guaranteed. The figure also shows that the solution becomes
increasingly unstable toward the end of simulation. If you encounter a convergence difficulty, you can
increase the energy dissipation ratio to an adequate level; however, validation of the new energy dis-
sipation ratio is required.
The buckling of the stiffener and the separation of the skin-flange connection are clearly indicated by
the final deformed shape, shown below.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
88 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
This highly nonlinear deformation leads to complex overall stress distributions and a number of signi-
ficant stress concentrations, as shown below.
The following two figures show the final contact status on the skin and the flange. As expected, debond-
ing of the skin-flange interface starts at the edges of the imperfection and propagates further into the
bonded areas as the load increases.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 89
Delamination of a Stiffened Composite Panel Under a Compressive Load
Figure 5.12: Contact Status for the Initial Bonded Contact Region
The figure above shows that the initial sticking status of the large skin-flange interface areas has changed
to open or near-contact status at the final converged stage. The structure may undergo catastrophic
failure and fail to withstand any further load as debonding progresses. In this case, a transient simulation
would be more suitable.
5.8. Recommendations
To perform a similar analysis, keep the following recommendations in mind:
• Consider the SOLSH190 element for modeling general layered composite structures, especially where contact,
variable thickness, 3-D constitutive relations, and/or thin to thick part transitions are present.
• Choose the proper technology (bonded contact or cohesive zone elements) for simulating interface debond-
ing. In this example, contact with a cohesive zone model simulates the interface debonding. Another option
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
90 of ANSYS, Inc. and its subsidiaries and affiliates.
Input Files
is available for debonding analysis via interface elements (INTER205 in this case). See Crack-Growth Simulation,
Interface Delamination, and Fatigue Crack Growth in the Fracture Analysis Guide; for more information about
both methods.
• To ensure a reliable solution, avoid an excessive energy dissipation ratio or mass damping factor in nonlinear
stabilization.
• delamination.cdb -- The common database file used in this analysis, called by the delamination.dat
file.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 91
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
92 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 6: Thermal-Stress Analysis of a Cooled Turbine Blade
This example problem shows how to easily set up and perform a thermal-stress analysis of a cooled
turbine blade.
• Use of one-dimensional fluid flow capabilities to obtain a highly accurate thermal solution for convection
loading.
6.1. Introduction
It is common practice in the turbine industry to cool turbine blades with a fluid flowing through cooling
holes. As a result of the thermal gradients setup in the blade, thermal stresses are induced which can
lead to failure of the blades.
In a typical thermal-stress analysis, temperatures are calculated and then applied as load conditions for
the stress analysis. While it is possible to solve for the temperature using a conjugate heat transfer
capability of a computational fluid dynamics (CFD) code, it can significant computational resources. A
reduced-order model for CFD, assuming a one-dimensional flow through the holes, can provide a relat-
ively inexpensive solution without much loss in accuracy. Typically, the mass flow rates are known
through the cooling holes, and certain empirical correlations are used to estimate the film coefficients
for heat transfer from the solid to the fluid.
Because of the proprietary nature of these types of analyses, it was not possible to obtain a realistic
model and loading conditions for this analysis from industry sources. A simplified model described in
a NASA report (p. 103) is used instead, and an understanding that turbine blade cooling passages and
loading conditions are much more complex in realistic models is assumed. For example, the following
figure illustrates that the cooling passages in an actual turbine blade (shown in yellow) can join or
branch with each other inside the blade, and the coolant is allowed to bleed through the cooling holes
and cool the external surface of the blade:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 93
Thermal-Stress Analysis of a Cooled Turbine Blade
In this problem, however, the cooling holes do not interact with each other and the coolant is confined
to the interior of the blade.
• SURF152 - 3-D Thermal Surface Effect and FLUID116 - Coupled Thermal-Fluid Pipe in the Mechanical
APDL Theory Reference.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
94 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
6.3. Modeling
The solid and fluid regions of a parasolid model are already meshed, and the model is read into the
database (CDREAD). The solid region is meshed with SOLID70 elements, as shown in this figure:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 95
Thermal-Stress Analysis of a Cooled Turbine Blade
The fluid region is meshed with FLUID116 line elements. The cross-sectional area of each fluid element
is set through a real constant. Appropriate components of nodes and elements have also been created
in order to facilitate the application of loads. The components can be viewed if necessary (CMLIST).
SURF152 elements are generated on the inside of the holes (ESURF); however, these surface-effect
elements are not coupled to the fluid elements. The following example input couples the SURF152 and
FLUID116 elements:
keyop,3,5,2 ! Allow 2 extra nodes for SURF152 elements for hole # 1
esel,s,type,,13 ! Select Fluid116 elements corresponding to hole # 1
cm,com116,elem ! Create a component using the fluid116 elements in each channel
esel,s,type,,3 ! Select SURF152 elements corresponding to hole# 1
cm,com152,elem ! Create a component using the surf152 elements in each channel
allsel ! Select all nodes & elements before using MSTOLE
mstole,1,'com152','com116' ! Map component 'com152' to 'com116' using projection method
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
96 of ANSYS, Inc. and its subsidiaries and affiliates.
Analysis and Solution Controls
Hole Film
Number Coefficient
1 295.43
2 296.29
3 300.76
4 314.16
5 314.95
6 301.99
7 302.47
8 443.43
9 285.27
10 895.86
Inflow
Hole Mass Flow Rate
Temperature
Number (kg/sec)
(°K)
1 348.83 0.228E-01
2 349.32 0.239E-01
3 339.49 0.228E-01
4 342.30 0.243E-01
5 333.99 0.239E-01
6 364.95 0.242E-01
7 343.37 0.232E-01
8 365.41 0.799E-02
9 408.78 0.499E-02
10 453.18 0.253E-02
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 97
Thermal-Stress Analysis of a Cooled Turbine Blade
For the stress analysis, SOLID70 elements are converted to SOLID185 elements, and the SURF152 and
FLUID116 elements are suppressed.
The following two figures show the fluid and solid surface temperatures, respectively. The fluid temper-
ature rises from inlet to outlet. The solid surface temperature shows a similar trend.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
98 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 99
Thermal-Stress Analysis of a Cooled Turbine Blade
The following two XY plots show the fluid and solid temperatures, respectively, along the fluid path of
hole number 1.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
100 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The next figure shows the von Mises stresses for the solid region. Maximum stresses occur inside hole
number 10.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 101
Thermal-Stress Analysis of a Cooled Turbine Blade
6.8. Recommendations
When performing a similar type of analysis, consider the following hints and recommendations:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
102 of ANSYS, Inc. and its subsidiaries and affiliates.
Input Files
• Create surface-effect elements that are much smaller than the fluid elements.
Doing so ensures that all fluid elements are coupled with the surface-effect elements.
• Mapping the surface-effect elements onto the fluid elements (MSTOLE) creates the two extra nodes for the
surface-effect element.
Three methods are available for mapping: Minimum Centroid Distance, Projection, and Hybrid.
As shown in this analysis, the MSTOLE command is applied to each hole. Using a single MSTOLE
command for multiple holes is inefficient and can lead to incorrect mappings.
If the passage has a serpentine path, several MSTOLE commands may be necessary to produce the
correct mapping for that hole. For example, if the passage is a U-tube, use the MSTOLE command
three times.
• When higher-order elements are used for the solid region, the midside nodes on the convecting elements
should be dropped.
The lower-order surface-effect element can then be used. Using higher-order surface-effect elements
can sometimes lead to an unrealistic temperature distribution.
6.9. References
This example problem was based on the following reports:
1. Hylton, L.D. et al.“Analytical and Experimental Evaluation of the Heat Transfer Distribution Over the Surfaces
of Turbine Vanes.” NASA CR 168015, May 1983.
2. Turner, E. R. et al.“Analytical and Experimental Evaluation of Surface Heat Transfer Distributions with
Leading Edge Showerhead Film Cooling.” NASA CR 174827, July 1985.
3. Hylton, L. D. et al.“The Effects of Leading Edge and Downstream Film Cooling on Turbine Vane Heat
Transfer.” NASA CR 182133, Nov. 1988.
• turbinebladestruc.dat -- Input file for the structural analysis portion of the problem.
• tblade.cdb -- The Parasolid file of the meshed turbine blade model (called by turbineblade.dat).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 103
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
104 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 7: Nonlinear Transient Analysis of a Camshaft Assembly
This example problem demonstrates the ease with which you can set up and perform an analysis in-
volving both axisymmetric and nonaxisymmetric components. The problem shows how modeling with
general axisymmetric element technology can reduce computational resources significantly while
maintaining the same degree of accuracy as a simulation using a full 3-D model.
• Use of general axisymmetric element technology and 3-D element technology in a model consisting of both
axisymmetric and nonaxisymmetric parts.
• Generating axisymmetric elements in an assembly having more than one axisymmetric part with multiple
axes of symmetry and subjected to nonaxisymmetric loading.
• Use of contact element technology to couple general axisymmetric elements with standard 3-D elements.
7.1. Introduction
Axisymmetric modeling greatly reduces modeling and analysis time when compared to equivalent 3-D
modeling. In some cases, however, certain components of the model may be nonaxisymmetric, or the
geometry may be axisymmetric but loading is nonaxisymmetric.
• Introduce the Fourier series into interpolation functions to describe the change of displacements in the
circumferential (θ) direction. (The elements can therefore apply to any analysis type, including geometric
nonlinear analyses, and can support any load and deformation mode.)
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 105
Nonlinear Transient Analysis of a Camshaft Assembly
• Do not require the input of peak loads and multiple load steps for each Fourier term. (Loads can therefore
be applied anywhere in 3-D space and only one solve operation is required to obtain the solution.)
With general axisymmetric elements, it is necessary only to define base elements (quadrilaterals or tri-
angles) on a master plane. (See General Axisymmetric Element Terminology.) The program generates a
3-D mesh (based on a 2-D mesh) on the master plane, after which boundary conditions and loading
can be applied at nodes in 3-D space.
• SOLID272 - General Axisymmetric Solid with 4 Base Nodes and SOLID273 - General Axisymmetric Solid with
8 Base Nodes in the Mechanical APDL Theory Reference.
One of the cams is in contact with a valve. As the shaft rotates, the motion of the valve is controlled
by the cam, which pushes the valve according to the cam profile.
A transient analysis of the assembly is performed by rotating the shaft for one full rotation.
7.3. Modeling
This section covers the following modeling topics:
7.3.1. Camshaft Modeling
7.3.2. Contact Modeling
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
106 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
The 2-D valve and 2-D shaft are modeled with general axisymmetric SOLID272 elements. Because the
valve and shaft should not have high localized deformation in the circumferential direction (θ), the
number of Fourier nodes is set to 3. (The SOLID272 element's KEYOPT(2) option controls the number
of Fourier nodes in the circumferential direction.)
The remaining parts of the assembly such as the valve ball and cams are modeled with standard 3-D
elements (SOLID187 in this case), as shown:
Figure 7.3: Meshed Camshaft Model Prior to Generating the 3-D Mesh
Before generating the 3-D mesh (based on the 2-D mesh) on the master plane, axes of symmetry for
the valve and shaft are defined separately (via SECTYPE and SECDATA commands), as follows:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 107
Nonlinear Transient Analysis of a Camshaft Assembly
After defining the axes of symmetry, the NAXIS command generates the 3-D mesh, as shown:
Figure 7.4: Meshed Camshaft Model After Generating the 3-D Mesh
All components of the assembly are modeled with standard 3-D elements (SOLID187 in this case), and
all boundary conditions, loadings, material properties, solution settings and hardware settings are
identical to the general axisymmetric model.
A comparison of the analysis results based on the general axisymmetric model and the full 3-D model
appears in Results and Discussion (p. 112).
Because the SOLID272 general axisymmetric element has nodes (and not elements) in 3-D space after
the NAXIS command has generated the 3-D mesh, the CONTA175 2-D / 3-D node-to-surface contact
element is a good choice for making contact pairs.
As shown in the following figure, bonded contact pairs with the MPC algorithm are defined to couple
the shaft with the cams and the valve ball with the valve. The bonded contact pairs connect the general
axisymmetric elements SOLID272 with the standard 3-D elements SOLID187.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
108 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
The following example input shows how to create the contact pairs between the shaft and cams:
ET,1,175 ! Define CONTACT175 element
ET,2,170 ! Define TARGET170 element
KEYOPT,1,12,5 ! Bonded Contact
KEYOPT,1,2,2 ! MPC style contact
CMSEL,S,SHAFT_NODES ! "SHAFT_NODES" is the user-defined nodal component
! which has all the nodes of the shaft that are in
! contact with cams
TYPE,1
REAL,1
NN=0
*GET,NUMN,NODE,,COUNT
*DO,I,1,NUMN
NN=NDNEXT(NN)
E,NN ! Create contact elements on nodes one by one.
*ENDDO
Similarly, the other bonded contact pair is defined between the valve and valve ball.
A spring is modeled using the COMBIN14 element (as shown in Figure 7.4: Meshed Camshaft Model
After Generating the 3-D Mesh (p. 108)). The spring ensures that the valve maintains contact with the
shaft during analysis. One end of the spring is attached to the valve via the bonded contact pair; the
other is fixed in all degrees of freedom.
In the following figure, a standard surface-to-surface contact pair using the augmented Lagrangian al-
gorithm is defined between the cam and valve ball. Also shown are two revolute joints (one at either
end of the shaft).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 109
Nonlinear Transient Analysis of a Camshaft Assembly
Shown are two revolute joints (one at each end of the shaft) and a surface-to-surface
contact pair between the cam and valve ball.
CONTA174 and TARGE170 elements are used in the contact pair between the cam and valve ball. The
status of this contact pair changes during analysis.
The following example input shows how to create the joint elements:
! Creating joint elements
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
110 of ANSYS, Inc. and its subsidiaries and affiliates.
Analysis and Solution Controls
NN=0
*GET,NUMN,NODE,,COUNT
*DO,I,1,NUMN
NN=NDNEXT(NN)
E,NN
*ENDDO
ALLSEL,ALL
TYPE,5
REAL,4
TSHAPE,PILOT
EN,999,99 ! Creates target element with pilot node
TSHAPE
CSYS,11
NROT,99
CSYS,0
ALLSEL,ALL
SECTYPE,3,JOINT,REVO, ! Creates section type for joint elements
SECJOINT,,11,11 ! Defines local coordinate systems at joint element nodes
REAL,3
TYPE,3
SECNUM,3
EN,888,,99 ! Defines body to ground joint (revolute joint)
After the joints are defined at both ends of the shaft, loading (360-degree rotation) is applied on the
revolute joint at one end of the shaft (via the DJ command), as follows:
ESEL,S,TYPE,,3 ! Selects joint elements
DJ,ALL,ROTZ,%LOAD% ! Applies load on joint elements
When the shaft rotates, the cams attached to the shaft also rotate. The cam in contact with the valve
ball (via the surface-to-surface contact pair) pushes the valve along its axial direction.
The valve moves along its axial direction only and, due to the spring force, it returns to its original po-
sition when the cam does not push it further.
To control the movement of the valve, all valve nodes are constrained so that it moves along its axial
direction only. To do so, a local coordinate system is defined at the valve such that the local z axis is
along the valve axis.
Large-deflection effects and transient effects are included in the analysis (via NLGEOM,ON and TIMINT
commands, respectively).
Automatic time stepping is used, with an initial time increment of 3.3 milliseconds.
Solution items are stored in the .rst results file for 20 equally spaced time steps.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 111
Nonlinear Transient Analysis of a Camshaft Assembly
The plots show that the valve always remains in touch with the cam throughout the analysis.
The following figure shows the maximum and minimum principal stress plots of the valve and shaft at
the end of the analysis:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
112 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The shaft and valve are subjected to nonaxisymmetric deformation. The results show that general
axisymmetric elements are capable of modeling nonlinear general 3-D deformation. The problem also
demonstrates how general axisymmetric elements can have any axis as the axisymmetric axis.
7.7.1. Results Comparison: General Axisymmetric Model and Full 3-D Model
To verify the accuracy and efficiency of the general axisymmetric element model, a transient analysis
was performed on the full 3-D model of the crankshaft assembly. Both the general axisymmetric model
and the full 3-D model simulations were run on the same computer. Solution settings for both models
were identical.
The following figure shows a direct comparison of results between the full 3-D model and the general
axisymmetric model:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 113
Nonlinear Transient Analysis of a Camshaft Assembly
Figure 7.9: Comparison of USUM Plots: General Axisymmetric Model and Full 3-D Model
In the full 3-D model, the number of degrees of freedom (DOFs) is approximately 300,000. In the gen-
eral axisymmetric model, the number of DOFs is only 18,000. The smaller number of DOFs in the gen-
eral axisymmetric model is possible due to the combination of axisymmetric and 3-D modeling.
Following is a comparison of the general axisymmetric model and the full 3-D model in terms of com-
putational efficiency:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
114 of ANSYS, Inc. and its subsidiaries and affiliates.
Input Files
1. In the general axisymmetric model, the valve and shaft components are meshed with
general axisymmetric SOLID272 elements. The remaining components consist of standard
3-D elements.
2. In the full 3-D model, all components are meshed with 3-D elements.
The analyses show that using general axisymmetric elements can reduce computational time significantly
with no loss of accuracy.
7.8. Recommendations
To perform a similar type of analysis, keep the following hints in mind:
• For every axisymmetric part of the assembly, define a separate axis of symmetry (SECDATA) before generating
nodes around the axis of an axisymmetric section (NAXIS).
• Choose an appropriate number of Fourier nodes in the circumferential direction to ensure better accuracy
and to minimize computational cost.
The KEYOPT (2) option for the general axisymmetric elements (SOLID272 and SOLID273)
controls the number of Fourier nodes in the circumferential direction.
If any part of your model which uses general axisymmetric elements does not have
high localized deformation in the circumferential direction, the number of Fourier
nodes can be set as low as 3.
In the case of high localized deformation in the circumferential direction, the number
of Fourier nodes can be set as high as 12, according to your requirements.
• Use a node-to-surface contact pair to connect general axisymmetric elements to standard 3-D elements.
• camshaft_assembly.inp -- Input file for the transient analysis (used for both the general axisymmetric
and the full 3-D models)
• full_3d_model.cdb -- The common database file for the full 3-D model (called by the camshaft_as-
sembly.inp file).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 115
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
116 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 8: Nuclear Piping System Under Seismic Loading
This example problem demonstrates the advantages of elbow element technology over traditional shell
and pipe element technology for modeling pipe bends in a typical nuclear piping system.
8.1. Introduction
Nuclear piping systems under seismic loading are typically analyzed in two stages:
1. A transient analysis is performed with a simplified model of the entire piping system to obtain the global
response of the system.
2. Using the global response as the imposed boundary condition, nonlinear static analyses are performed
separately on critical parts to investigate local stress and strain responses over time. The critical parts are
typically modeled with refined 3–D elements.
This conventional global-local analysis method requires the conversion of results between the global
and local models and is usually time-consuming and computationally expensive.
Compared to the straight-pipe segments, curved-pipe bends (elbows) have much greater flexibility and
can therefore develop more significant deformation and stresses. ANSYS, Inc. elbow element technology
offers a simple 1-D geometry (similar to a standard pipe element) and excellent accuracy (matching
that of a 3-D shell element). The elbow can be used directly in both global and local simulation to avoid
any cumbersome global-to-local model and boundary-condition conversion.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 117
Nuclear Piping System Under Seismic Loading
The system is supported by nozzles, an anchor, three two-directional supports, a horizontal support, a
vertical support and a spring hanger [1]. The following analyses are performed on the piping system
to study both global and local responses of the system when subjected to seismic loading:
• Nonlinear static analysis of one of the critical elbows of the piping system using elbow element technology
• Nonlinear static analysis of one of the critical elbows of the piping system using shell element technology
8.3. Modeling
Three models are prepared for this problem:
8.3.1. Global Nuclear Piping System Model
8.3.2. Local Elbow Model Meshed with ELBOW290 Elements
8.3.3. Local Elbow Model Meshed with SHELL281 Elements
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
118 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
Straight segments are meshed with PIPE289 elements, and elbows are meshed with ELBOW290 elements:
Figure 8.3: Entire Nuclear Piping System Meshed with Pipe and Elbow Elements
An ELBOW290 element key option controls the level of accuracy in cross-sectional deformation. In this
case, KEYOPT(2) = 4 allows general section deformation, including nonuniform radial expansion, ovaliz-
ation, and warping.
The average diameter and average thickness of the pipes are assigned to ELBOW290 and PIPE289 ele-
ments (via SECDATA and SECTYPE commands), as follows:
SECTYPE,1,PIPE
SECDATA,Do,T ! Where Do is the outer diameter and T is thickness of the pipe
! In this example, Do = 219.2 mm and T = 10.38 mm
SECNUM,1
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 119
Nuclear Piping System Under Seismic Loading
Subsequently, to create necessary transition zones from the elbows to straight-pipe segments, the ELBOW
command automatically converts a few PIPE289 elements adjacent to the elbows into ELBOW290 ele-
ments, as follows:
ELBOW,ON,,,
Elbow and pipe elements must have the same pipe section ID for the pipe-to-elbow
transition to occur.
A modal analysis is performed on the global model to obtain the fundamental natural frequency. The
result is compared to the experimental result given in [1].
The model has branches that are 645.2 mm long and an elbow with a radius of 304.8 mm (for a total
centerline length of 950 mm). The diameter of the pipe is 219.2 mm and the wall thickness is 10.38
mm. The elbow is meshed with ELBOW290 elements:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
120 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
Time-varying-displacement boundary conditions, extracted from a transient analysis of the entire piping
system model under seismic loading, are applied at one end of the model. A nonlinear static analysis
using the Chaboche material model is performed on this elbow model to obtain the stress and strain
response over time.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 121
Nuclear Piping System Under Seismic Loading
Material properties and loadings considered in this model are identical to those of the local ELBOW290
model. A conversion of boundary conditions from the global line mesh to the local 3-D shell mesh is
necessary, however. Time varying displacement boundary conditions are applied to the pilot node at
one end of the model. The pilot nodes are coupled with edge nodes at both ends via contact elements
as shown in Figure 8.7: Elbow Model Meshed with SHELL281 Elements (p. 122).
A Chaboche nonlinear kinematic hardening material model is used in the nonlinear static analysis of
the elbow model (for both the ELBOW290 and SHELL281 models).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
122 of ANSYS, Inc. and its subsidiaries and affiliates.
Boundary Conditions and Loading
The mass density of the pipe material for all models is 12388 kg / m3. All material properties shown are
taken from Nie (p. 128).
• To simulate the nozzles and anchor supports, the nodes at all three ends of the system are completely
constrained.
• To simulate the three two-directional supports, three nodes are constrained in either the Y and Z directions
or the X an Z directions.
• To simulate a horizontal support and a vertical support, two nodes are constrained respectively in the X and
Z directions.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 123
Nuclear Piping System Under Seismic Loading
Internal pressure and gravity load are applied in the model as follows:
SFE,ALL,1,PRES,,10.7 ! P (internal pressure) = 10.7 MPa
ACEL,,,9.8 ! G = 9.8 m/sec^2
Cross-section deformation is not allowed at both ends (D). The input file (p. 128) for this model shows
the time-varying-displacement data.
Internal pressure and gravity loads are identical to those applied in the global model (p. 123).
Pilot nodes are created both ends of the model, at the center of the cross-sections. The pilot nodes are
coupled with the edge nodes of the cross-sections via contact pairs. One pilot node is fixed in all degrees
of freedoms and time-varying-displacement data are applied at the other pilot node, as shown in Fig-
ure 8.7: Elbow Model Meshed with SHELL281 Elements (p. 122).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
124 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Initially, a large-deflection (NLGEOM) static analysis is performed, followed by the prestressed modal
analysis (PSOLVE). The Block Lanczos solver (MODOPT) extracts the first fundamental mode.
PSTRESS,ON
MODOPT,LANB,1
MXPAND,1, , ,0
PSOLVE,EIGLANB
To examine the effect of the prestressed analysis, a modal analysis without prestress effects is performed
on the same model, as follows:
MODOPT,LANB,1
MXPAND,1, , ,0
SOLVE
A constant time increment of 5 ms is specified to obtain the stress and strain response over a period
of 20 seconds.
Results are stored (in the RST file) for all substeps.
The close agreement between the simulation and the expected results validates the methods used in
this analysis:
• Using PIPE289 and ELBOW290 separately for the straight and curved segments, respectively.
• Defining transition zones (ELBOW) for building the global piping system model.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 125
Nuclear Piping System Under Seismic Loading
Figure 8.9: USUM Plot Comparison: Local ELBOW290 and SHELL281 Models
Von Mises stress and total mechanical strain for both models are shown respectively in the following
figures:
Figure 8.10: Von Mises Stress Comparison: Local ELBOW290 and SHELL281 Models
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
126 of ANSYS, Inc. and its subsidiaries and affiliates.
Recommendations
Figure 8.11: Von Mises Strain Comparison: Local ELBOW290 and SHELL281 Models
Again, the close agreement between the stress and strain results from the two models is apparent.
Following is a comparison of the local ELBOW290 and SHELL281 elbow models in terms of computational
efficiency:
While offering simplicity for line-based modeling, the ELBOW290 element is capable of producing an
accurate global response as well as detailed local solutions for a piping system. ELBOW290 is ideal for
modeling bends in the pipe systems or straight-pipe segments that may undergo significant cross-section
deformation.
8.8. Recommendations
To perform a similar type of analysis, consider the following hints and recommendations:
Computational costs also rise sharply as the number of Fourier terms increases.
The ELBOW290 element's KEYOPT(2) setting controls the number of Fourier terms.
See the element documentation for more information about specifying an appropriate
number of Fourier terms.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 127
Nuclear Piping System Under Seismic Loading
• To minimize computational costs, use the PIPE289 element to model straight-pipe segments.
• Transition zones (ELBOW) between the curved ELBOW290 and straight PIPE289 elements are generally de-
sirable for obtaining a smooth solution.
• Because the ELBOW290 element is based on shell theory, its applicability is limited to pipes with thin to
moderately thick walls.
8.9. References
The following references are used in this example problem:
1. Nie, J. et al.“Nonlinear Seismic Correlation Analysis of the JNES/NUPEC Large-Scale Piping System Tests.”
2008 ASME Pressure Vessels and Piping Division Conference. PVP2008 Proceedings. CD-ROM. ASME. 2008.
• nuclear_piping_system.dat -- Combined input file for performing the modal and static analyses of
the nuclear piping system.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
128 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 9: Reliability Study of a Composite Overwrapped Pressure
Vessel
This example problem is a reliability study of a composite overwrapped pressure vessel (COPV). The
model uses reinforcing fibers in a layered composite. A finite-element simulation of a COPV is performed
first to gain insight into its mechanical behaviors, then simulation results are processed using failure
analysis to determine the most vulnerable layer. The problem generates linearized stress output for
pressure-vessel design optimization and code compliance.
9.1. Introduction
Composite overwrapped pressure vessels (COPVs) are manufactured by filament winding, a process in-
volving a continuous application of high-stiffness fibers over a thin metal shell (liner). For their excellent
corrosion resistance and high strength-to-mass ratios, COPVs are commonly used for critical applications
(such as deep-sea and space missions).
A typical COPV usually contains numerous fiber-reinforced layers and exhibits a complex overall mater-
ial orthotropy. The material properties may undergo significant changes due to fiber realignment when
the COPV is under intense operational load. The material properties and their load-dependent variations
are difficult to obtain by experimental means; therefore, finite-element simulations are often conducted
by designers and researchers to gain insight into the mechanical behaviors of COPVs.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 129
Reliability Study of a Composite Overwrapped Pressure Vessel
A nonlinear static analysis is performed to obtain stress and strain results of all layers under large-de-
formation conditions.
Failure criteria (FC) analysis and stress linearization are conducted based on the simulation results.
9.3. Modeling
The pressure vessel consists of a cylindrical main tank, and an inlet and an outlet at each end of the
tank, as shown below:
The tank wall has a composite construction of one homogeneous aluminum alloy liner and four
T800/EPOXY fiber-reinforced layers. The inlet and outlet are made of aluminum alloy only. Due to the
symmetry, only one-eighth of the entire geometry is modeled (shown by (b) in the above figure).
For this problem, the composite wall is modeled as two distinct components:
• One isotropic matrix component containing the liner and all bonding material (resin) in the reinforced layers
The modeling approach described allows one to independently study the failure limits of the matrix
and fiber components based on their own strength characteristics.
The shell element SHELL281 is used for meshing the matrix component.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
130 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
The layup of the matrix material is defined using shell sections. Following are the section properties,
including material IDs, layer thickness, material orientation (irrelevant if the material is isotropic), and
the number of layer integration points:
For proper internal pressure application, the shell sections are offset to the bottom (SECOFFSET).
It is assumed that the fibers in each reinforced layer have a unique cross-section area, material property,
spacing, and orientation; therefore, a smeared reinforcing approach is used. With smeared reinforcing,
the fibers are represented by an equivalent homogeneous layer (instead of the fibers being modeled
individually). The smeared approach allows for better computational efficiency. The smeared reinforcing
element used is REINF265. The fiber properties (material, cross section area, spacing, orientation, and
location) are defined via the reinforcing section using the smear subtype (SECTYPE,REINF,SMEAR), as
follows:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 131
Reliability Study of a Composite Overwrapped Pressure Vessel
The expanded solid display (/ESHAPE) of both SHELL281 (shown in translucency) and the embedded
REINF265 elements is shown by (b) in Figure 9.2: COPV Mesh (p. 131).
The material properties of the bonding matrix and reinforcing fibers are as follows:
The fibers have a one-dimensional constitutive relation and are significantly stiffer than the bonding
material. For simplicity, it is assumed that all materials in this problem are linear-elastic and temperature-
independent.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
132 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
To compensate for the asymmetric section construction, cyclic symmetry conditions (CP) are applied
for coupling degrees of freedom (DOFs) at the AB edge with the corresponding DOFs at the CD edge
(shown by (b) in the above figure).
A uniform temperature of 1000° is applied to the inner surface. A 500° temperature gradient is allowed
through the wall thickness, resulting in a temperature of 500° at the outer surface.
The temperature at each base shell layer is specified via the BFE command. See the SHELL281 element
documentation for more information about defining layered temperature loading.
Temperatures on the reinforcing fibers are automatically determined according to their locations in the
shell elements.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 133
Reliability Study of a Composite Overwrapped Pressure Vessel
The deformation is uniform in the hoop direction; however, a significantly nonuniform pattern is apparent
in the meridian direction.
A detailed through-the-wall strain distribution in the matrix component is shown in this figure:
The improved shell formulation of SHELL281 incorporates the membrane strain effects into the curvature
calculation, leading to a more accurate through-the-wall strain distribution. If the matrix is simulated
with regular shell elements, strains would be largely uniform through the wall thickness.
Material-specific strength limits are defined (FC) for both matrix and fiber components. The failure-cri-
teria analysis automatically determines the layer with the maximum failure criteria value (or the most
vulnerable layer that is likely to fail first). The layer number and failure-criteria value/type of the most
vulnerable layer can be listed and plotted (via standard postprocessing listing and visualization commands
[PRESOL and PLESOL]).
The following figure shows that the liner would most likely fail at the inlet, while the bonding material
would likely fail at the fourth layer:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
134 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Figure 9.6: Layer with Maximum Failure Criteria in Matrix and Fibers
Similarly, according to (b) in the above figure, the fibers would fail first at the first reinforcing layer.
The following figure shows the type of maximum failure criteria on the matrix (a) and fiber (b) compon-
ents, respectively:
Based on the value of maximum failure criteria, the tank-to-inlet transition region (b) is easily identified
as the critical location for the reinforcing fibers (shown by (b) in the following figure):
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 135
Reliability Study of a Composite Overwrapped Pressure Vessel
Consistent with the failure criteria analysis results, the fiber strain distribution also shows a spike in this
transition region:
The results indicate that a design modification, either in the liner shape or fiber orientation, is necessary
to further strengthen the fibers in this area.
Strains, stresses, and other solution items at the layer of maximum failure criteria can be inspected
(LAYER,FCMAX, followed by postprocessing listing and visualization commands [PRESOL and PLESOL]).
The following figure shows the equivalent stresses at the layer of maximum failure criteria in the matrix
and fiber components, respectively:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
136 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Linearized stresses are commonly cited in boiler and pressure vessel codes and standards. The program
computes the linearized stresses, including membrane, bending, and peak stresses, and stores them as
the SMISC quantities for current-technology shell and solid-shell elements. These quantities can be re-
trieved during postprocessing (ETABLE or *GET) for further inspection and code-compliance verification.
The following two figures show the membrane and bending stress, respectively, of the matrix component
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 137
Reliability Study of a Composite Overwrapped Pressure Vessel
9.8. Recommendations
To perform a reliability study of a composite overwrapped pressure vessel, consider the following hints
and recommendations:
• Fibers in composite materials can be simulated separately using reinforcing elements. In addition to the
smeared reinforcing element REINF265 used in this study, a discrete reinforcing element REINF264 is also
available for modeling individual fibers with arbitrary orientations or nonuniform geometry or materials.
• Use the FCMX item (and accompanying components) available in the postprocessing listing and visualization
commands (PRESOL and PLESOL) to conveniently determine the location, type, and value of the highest
failure criteria.
• Select layers of maximum failure criteria for more standard postprocessing (LAYER,FCMAX).
• Take advantage of the linearized stress output to optimize the design and to check for code compliance.
9.9. References
The following reference was used as the basis of this example problem:
1. Kim, Cheol-Ung et al.“Optimal Design of Filament Wound Structures under Internal Pressure Based on the
Semi-geodesic Path Algorithm.” Composite Structures. 67 (2004): 443-452.
• pressure_vessel.dat -- Input file for the reliability study of a composite overwrapped pressure vessel.
• pressure_vessel.cdb -- The common database file used for this problem, called by pressure_ves-
sel.dat.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
138 of ANSYS, Inc. and its subsidiaries and affiliates.
Input Files
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 139
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
140 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 10: Simulation of a Lumbar Motion Segment
This example problem demonstrates the use of coupled pore-pressure-thermal mechanical solid element
technology to study the creep response of a lumbar motion segment under compression. The simulation
reveals the interaction between the solid phase and the fluid phase in soft tissues.
10.1. Introduction
Approximately 26 million Americans between the ages of 20 and 64 are affected by frequent lower
back pain [1], one of the major causes of worker disability. The problem costs the economy roughly
$50-100 billion each year [2].
Most lower back problems involve the intervertebral disc (IVD), shown in this illustration [4]:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 141
Simulation of a Lumbar Motion Segment
The IVD is the smallest segment of the spine that exhibits biomechanical characteristics similar to that
of the entire spine [3]. The IVD is composed of annulus bulk and nucleus pulposus, both soft tissues.
Finite element analysis (FEA) has been applied in orthopedic biomechanics and spine biomechanics
research since 1983. Typically, the IVD is modeled by solid elements. Many experimental observations,
however, suggest that it is more realistic to consider soft tissue in the IVD as a biphasic model. For ex-
ample, when the disc is compressed, fluid flows out from either the entire disc [5] or from disc slices;
therefore, one can consider the soft tissues in IVD to be a multiphase system.
Clinical studies shows that back pain is associated with the mechanical behavior of the intervertebral
disc. A significant finding is that the degenerated intervertebral disc has lower permeability than a
normal disc [3]; therefore, using coupled pore-pressure-thermalelements in a lumbar motion segment
simulation provides a realistic model for clinical study of degenerated intervertebral discs.
10.3. Modeling
Following is finite element model of a lumbar motion segment:
Because the IVD is composed of soft tissue, it is modeled using CPT217 coupled pore-pressure-thermal
elements. Bones in the segment are modeled using SOLID187 solid elements.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
142 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The permeating fluid and the solid phase are both assumed to be incompressible, so the Biot modulus
α is set to 1.
Creep is studied over a period of five days (about 4.26E4 seconds). Results show that small time steps
induce nonphysical pressure oscillations in the computation; therefore, the entire solution is completed
in 100 substeps with no oscillation present.
For IVD stresses, the annulus bulk takes more load than the nucleus since it has higher stiffness, as
shown in this figure:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 143
Simulation of a Lumbar Motion Segment
With the IVD modeled using coupled pore-pressure-thermal elements, vertical displacement gradually
increases with time, as shown in this animation:
Figure 10.5: Lumbar Motion Segment Animation of Vertical Displacement Increasing with Time
The animation also illustrates the difference between solid elements and coupled pore-pressure-thermal
elements. With the same loading and boundary conditions, the upper vertebral body modeled with
coupled pore-pressure-thermal elements drops gradually under compression, as compared to the constant
displacement of solid elements. For soft-tissue modeling, coupled pore-pressure-thermal elements
provide a more realistic simulation.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
144 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The creep response of the lumbar motion segment is due to fluid flowing out of the IVD, as shown:
When the fluid is exhausted, the final displacement is equal to that of an IVD modeled by solid elements
only:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 145
Simulation of a Lumbar Motion Segment
The creep response under compression demonstrates the diffusive interaction between the solid matrix
and the interstitial fluid in soft tissues.
10.8. References
The following references are used in this example problem:
2. Paremer, A., S. Fumer, D. P. Rice. Musculoskeletal Conditions in the United States. 1st ed. Park Ridge: American
Academy of Orthopaedic Surgeons, 1992.
3. Gilbertson, L. G. et al.“Finite Element Methods in Spine Biomechanics Research.” Critical Reviews in Biomed-
ical Engineering. 23.5-6 (1995): 411-473.
5. Charnley, J.“Imbibation of Fluid as a Cause of Herniation of the Nucleus Pulposus.” Lancet. 6699 (1952):
124–127.
6. Argoubi, A., A. Shirazi-Adl.“Poroelastic Creep Response Analysis of a Lumbar Motion Segment in Compres-
sion.” Journal of Biomechanics. 29.10 (1996): 1331-1339.
ANSYS, Inc. gratefully acknowledges Materialise NV for providing the vertebral body geometry for this
example problem.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
146 of ANSYS, Inc. and its subsidiaries and affiliates.
Input Files
• vertebral_body.cdb -- Common database file for the lumbar motion segment model (called by the
vertebral_body_input.dat file).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 147
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
148 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 11: Rocket Nozzle Extension Simulation: Fabrication
This is the first of two example problems that simulate a rocket nozzle:
• The example problem presented here simulates thermal stresses during the manufacturing (fabrication)
stage of the nozzle.
• The second example, Rocket Nozzle Extension Simulation: Operation (p. 447), simulates thermal stresses in-
duced during the operation of the nozzle.
While both examples are based on the same geometry, the assumptions used are different, therefore
leading to different meshes and different element types.
This example problem demonstrates how current structural-shell element technology can accurately
model the orthotropic thermal expansion in curved-shell structures. Section offsets are applied when
connecting shell elements to shell elements, or shell elements to other element types. An appropriate
results-storage option is selected for postprocessing the layered-element solution.
11.1. Introduction
Thermal loading can produce different thermal strains in the in-plane and through-the-thickness directions
in shell structures with orthotropic thermal expansion coefficients. If the structure is curved, this difference
causes shell curvature changes and additional stresses.
Classic shell elements do not account for thickness strains and therefore lack the ability to account for
the orthotropic thermal expansion effect [1]. Shell element technology that incorporates the thickness
strain in the shell curvature calculation is well suited for this type of simulation.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 149
Rocket Nozzle Extension Simulation: Fabrication
The extension materials exhibit different in-plane and through-the-thickness thermal expansion properties;
therefore, the cooling process may induce significant residual stresses, particularly in parts with large
curvatures.
A linear static analysis is performed to study the residual stresses on the nozzle extension under this
uniform thermal (cooling) load. Special attention is given to the reinforcing ring at the jet exhaust where
the highest stresses are expected.
11.3. Modeling
The nozzle extension consists of a main curved wall and a reinforcing ring close to the jet exhaust, as
shown in this figure:
Both the wall and the reinforcing ring are made of layered composite materials. To maintain a degree
of simplicity for this example, the model uses a single orthotropic material for all layers; however, the
material orientation varies from layer to layer.
Because the model exhibits symmetry, optimal computational efficiency is achieved by modeling only
a single 1° sector (the base sector) of the entire nozzle extension. The geometry model of the 1° sector
is shown in this figure, where (a) is the global geometry and (b) is a detail view of the reinforcing ring:
The reduced 1° geometry model is meshed with SHELL281 elements. The following figures show respect-
ively the mesh and the expanded solid representation of the shell mesh (/ESHAPE).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
150 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
Using the SHELL281 element's option for improved shell formulation (KEYOPT(2) = 1), the element in-
corporates initial curvature effects more accurately. With this option, both shell-membrane and thickness
strains are accounted for in the calculation of effective shell curvature change. This option is essential
for capturing the orthotropic thermal expansion effects in curved shell structures. To properly postprocess
a layered solution, layered data storage is set to store results for the tops and bottoms of all layers
(KEYOPT(8) = 1).
Shell sections define the layers of the composite material. (The SECTYPE command specifies the section
type, and the SECDATA command specifies the material, thickness, material orientation, and the number
of integration points of each layer.) Two sections are created, respectively, for the main wall and the
reinforcing ring. The following table summarizes the shell section properties:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 151
Rocket Nozzle Extension Simulation: Fabrication
Assume that the reinforcing ring is securely bonded to the outer surface of the main extension wall.
The secure bonding is easily simulated via the use of common nodes at the connection areas between
the wall and the reinforcing ring. (See Figure 11.3: Ring Element Plot (p. 151).) It is necessary to apply
shell offsets, however, so that the wall and the ring can be properly connected. This example offsets
the shell section for the main wall to its top, and the shell section for the ring to the bottom (SECOFFSET),
as shown in this figure:
Figure 11.5: Nozzle Wall and Ring Connection with Shell Offsets
Without the offsets, the shell sections are offset to the midplane by default, leading to an incorrectly
overlapped model, as shown in this figure:
Figure 11.6: Nozzle Wall and Ring Connection without Shell Offsets
To obtain an accurate reference solution, results are compared to a refined 3-D model with multiple
quadratic layered solid elements (SOLID186) through the wall thickness.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
152 of ANSYS, Inc. and its subsidiaries and affiliates.
Analysis and Solution Controls
The material is assumed to be linear elastic and temperature-independent. The material has different
thermal expansion coefficients in the in-plane (X and Y) and through-the-thickness (Z) directions.
The first set of conditions (a) constrains rigid body motions while allowing free radial wall expansion.
The second set of conditions (b) provides necessary symmetry constraints on this reduced sector model.
For the symmetry constraints to work properly, nodal degrees of freedom must be rotated to the cyl-
indrical coordinate system.
Only thermal loads are considered in this simulation. The uniform cooling of 1600K is specified via the
uniform temperature command (TUNIF).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 153
Rocket Nozzle Extension Simulation: Fabrication
The maximum displacement for the shell model (26.667) nearly matches the reference value (26.704),
with an error of only 0.14 percent.
As expected, the highest stresses caused by the thermal load appear in the reinforcing ring, which has
the largest curvature and curvature variation. Distributions of all three normal stress components (X -
radial, Y - hoop, and Z - axial) agree closely with the reference results, as shown in the following three
figures, where (a) refers to the reference 3-D model and (b) refers to the shell model:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
154 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The following figures compare the detailed stress distributions from the reference model and the shell
model, at both the bottom and top of the reinforcing ring:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 155
Rocket Nozzle Extension Simulation: Fabrication
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
156 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 157
Rocket Nozzle Extension Simulation: Fabrication
The generally accurate agreement between the reference and simulation stress results is clearly observ-
able.
Some small local differences exist in the stress results, particularly at the areas where the ring connects
with the main extension wall. The discrepancy is noticeable because the problem applies the SHELL281
element (which is based on the plane stress assumption) to a complex 3-D domain.
If the problem were to be simulated with standard shell elements, no residual stresses would be obtained.
The stress contour display on the SHELL281 model with the expanded solid shapes reveals one graphics
irregularity: result averaging may fail where the pseudo expanded interelement connectivity does not
match, as shown in the circled area of (b) in Figure 11.10: Nozzle Axial (Z) Stress (p. 155). To overcome
this issue, use the (default) simple area element display option (/ESHAPE,0).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
158 of ANSYS, Inc. and its subsidiaries and affiliates.
Input Files
11.8. Recommendations
To perform a similar type of analysis, consider the following hints and recommendations:
• Select a suitable shell element for simulating initially curved shell structures. Shell element technology that
incorporates the thickness strain in the shell curvature calculation (such as SHELL281) is well suited to this
type of simulation.
• Apply appropriate section offsets when connecting shell elements to shell elements, or shell elements to
other element types.
11.9. References
The following reference was used for this example problem:
1. Sach, E. D., G. Fleischmann, W. Ulrich.“Influence of Modeling for Orthotropic Material Properties.” MSC
1999 Aerospace User’s Conference Proceedings.
• nozzle.cdb -- The common database file used in the orthotropic thermal expansion analysis (called by
nozzle.dat).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 159
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
160 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 12: Dynamic Simulation of a Nuclear Piping System Using
RSA Methods
This example problem is a dynamic simulation of a piping network used in a nuclear power generation
system. The problem demonstrates the analysis methods available to account for the effect of higher
frequency modes of a structure in a response-spectrum analysis (RSA).
For verification, RSA results are compared to the full-transient analysis results performed on the same
model. (In an industrial application, however, a transient analysis approach is computationally intensive
and often impractical for analyzing a nuclear piping network.)
• Use of the missing-mass response to include the responses of the higher modes in the RSA.
• Use of rigid responses (Gupta or Lindley-Yow method) to more accurately combine the modal responses
(consisting of both periodic and rigid-response components).
12.1. Introduction
The response-spectrum analysis (RSA) is a widely used method for studying seismic responses of structures
such as nuclear power plants, boilers, and pressure vessels. The current practice is to calculate only the
responses of the modes having a frequency within the frequency range of the input spectrum.
In some cases, however, the responses of the higher modes may not be negligible. The missing-mass
method is a convenient, computationally efficient, and accurate method used to account for:
• The contribution of all modes with frequencies above fZPA, at which the response spectrum returns to
the Zero Period Acceleration (ZPA).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 161
Dynamic Simulation of a Nuclear Piping System Using RSA Methods
Terms Used
Zero Period Acceleration (ZPA) -- The acceleration value (peak) at zero period (or infinite frequency)
of an input spectrum. It corresponds to the response of a rigid system.
ZPA frequency (fZPA) -- The minimum frequency beyond which the input spectrum curves for
various damping ratios converge towards the same ZPA acceleration.
For frequencies occurring in the amplified region of the response spectrum (f < fZPA), the mode re-
sponses generally have both in-phase (periodic) and out-of phase (rigid-response) components.[2]
To separate and combine the response components, the methods proposed by Gupta and Lindley-Yow
are applied and the results are compared.[3] The in-phase modal response components and the missing-
mass contributions are combined algebraically to produce the total in-phase response component. For
the combination of the out-of-phase modal response components, either the Square Root of the Sum
of the Squares (SRSS) method or the Der Kiureghian CQC method is applied.
For more information, see the following resources (in the Mechanical APDL Theory Reference):
• Missing-Mass Response
• Rigid Responses
• Combination of Modes
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
162 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
To measure the influence of each effect and compare the response-component combination methods,
seven single-point response spectrum (SPRS) analyses are performed.
12.3. Modeling
The following modeling topics are available:
12.3.1. BM3 Piping System Modeling
12.3.2. Elastic Support Modeling
Figure 12.2: BM3 Model Meshed with PIPE289 and ELBOW290 Elements
The default settings for the PIPE289 and ELBOW290 elements apply.
To more accurately capture the curvature of a bend, it is divided into four parts, and each part is meshed
with ELBOW290.
The section properties (SECTYPE and SECDATA) of the pipe and elbow elements are defined via the
following input:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 163
Dynamic Simulation of a Nuclear Piping System Using RSA Methods
One-dimensional longitudinal and torsional springs are defined via element key options (KEYOPT(2) =
1 through 6).
Spring Properties
Spring Constant
Direction
(N/m)
UX 1.00E+05
UY 1.00E+08
UZ 1.00E+11
ROT (X,
1.00E+20
Y, Z)
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
164 of ANSYS, Inc. and its subsidiaries and affiliates.
Boundary Conditions and Loading
The input-acceleration response is applied in the X direction. It exhibits a sharp, highly amplified peak
at the fundamental frequency of the structure/soil system. Its characteristics are fZPA ≈ 16.50 Hz and
ZPA = 0.54 g.
Generally, piping systems experience this type of seismic input instead of ground motion associated
with broad-banded response spectra. For this study, the unbroadened spectrum is used to provide a
direct comparison to the time history analysis results.
The ground motion is consistent with the input-acceleration response spectrum shown in Figure 12.4: In-
put-Acceleration Response Spectra (1% X Direction) (p. 165).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 165
Dynamic Simulation of a Nuclear Piping System Using RSA Methods
For full-transient and response-spectrum analyses performed using all three-directional inputs, the same
input is considered for both horizontal X and Z directions. In the vertical Y direction, the input is con-
sidered to be 2/3 of the horizontal input.
Input Mode-Combination
Missing-Mass
Case Rigid-Response Effect
Spectra Method Effect
A1 SRSS No No
A2 No No
A3 No
Yes
A4
X-direction (RIGRESP,ON,LINDLEY,0.54)
input Yes
Yes
A5
CQC
(MMASS,ON, (RIGRESP,ON,GUPTA,2.80,11.90)2 (p. 166)
0.54)1 (p. 166) Yes
A6
(RIGRESP,ON,GUPTA,2.80,6)2 (p. 166)
X-, Y-, and Yes
A7 Z-direction
inputs (RIGRESP,ON,LINDLEY,0.54)
1. The input spectrum unit is g. To transform it into SI units, set FACT = 386.4 on the SVTYP
command.
2. The rigid-response effect initial frequency is 2.80 Hz (VAL1 on the RIGRESP command),
and the final frequency is 6 Hz or 11.90 Hz (VAL2).
Fourteen modes exist below fZPA. All 14 modes are used in all seven spectrum analyses.
The equation ξi = α/2ωi + βωi/2 is used for the damping coefficients, where ωi is the natural circular
frequency of mode i.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
166 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The coefficients are determined by specifying the target modal damping at the fundamental frequency
(2.91 Hz) and at an intermediate frequency between this frequency and fZPA. A frequency of 14.32 Hz
is chosen to achieve the best fit over the 2.91 Hz to 16.50 Hz range of interest.
The values obtained are α = 0.304 (ALPHAD) and β = 1.85e-4 (BETAD). Small variations of these coeffi-
cients have a negligible effect on the results.
Automatic time stepping ensures that the time step is small enough to accurately calculate the higher
mode responses. The duration of the analysis is 14.99 seconds.
When performing the time history analysis with three-directional input motions, the individual responses
for each input motion (X, Y, and Z directions) are combined using the SRSS method.
Mode Frequency
(Hz)
1 2.91
2 4.44
3 4.86
4 5.02
5 6.95
6 7.58
7 7.82
8 10.94
9 11.65
10 11.78
11 12.80
12 14.32
13 15.17
14 15.79
The ratio of the RSA results to the transient results is reported in separate columns. The mean and
standard deviation of these ratios are evaluated at the bottom of each table. This form of representation
facilitates easy recognition of over-prediction and under-prediction by the RSA method.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 167
Dynamic Simulation of a Nuclear Piping System Using RSA Methods
A ratio of 1.0 indicates exact agreement, a ratio of > 1.0 indicates RSA over-prediction, and a ratio of <
1.0 indicates RSA under-prediction.
The absolute acceleration solutions are also compared for the x-direction and 3-direction input motions.
Coupled Modes
Mode Frequency Coupling
Number (Hz) Mode Coefficient
Freq. (Hz)
No.
3 4.857 4 5.017 0.276
6 7.581 7 7.816 0.299
9 11.653 10 11.775 0.787
12 14.316 13 15.173 0.106
Because of the closely spaced modes, the CQC results are closer to the reference (transient-analysis
results). The mean and standard-deviation values of spectrum results using CQC are 1.57 and 1.11, re-
spectively.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
168 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
12.7.1.2. RSA A3
To improve accuracy, the missing-mass response is included in the analysis. The standard deviation
decreases to 1.05. The mean value of 1.72 still shows over-prediction in the results.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 169
Dynamic Simulation of a Nuclear Piping System Using RSA Methods
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
170 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The absolute acceleration values at support (node 4) and far from support (nodes 16 and 34) are com-
pared in the following table:
It is clearly shown that the absolute acceleration value at node 4 (at support) shows close comparison
with the full-transient solution after the addition of the missing-mass effect.
12.7.1.4. RSA A6
With the Gupta method, there is a limitation lying in the semi-empirical basis of the definition of the
rigid-response coefficient αi, as a function of fi.[2][3] The choice of key parameter f2 (RI-
GRESP,,,,VAL2), which defines the frequency above which modal responses are combined algebraically,
has a significant effect on the predicted response.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 171
Dynamic Simulation of a Nuclear Piping System Using RSA Methods
To show the effect of the f2 value, two different values of f2 are chosen: 6.0 Hz [2 Appendix H] and
11.90 Hz.[4] Both frequencies are within a range where the input acceleration is almost constant and
the acceleration value is very close to the ZPA. A value of f1 = 2.80 Hz is maintained for both analyses.
The respective results are shown in the following two tables:
For f2 = 6.0 Hz, the mean and standard-deviation values are 0.91 and 0.18, respectively, implying under-
prediction of the reaction forces. Conversely, for f2 = 11.90 Hz, the mean and standard-deviation values
are 1.10 and 0.17, respectively, implying over-prediction of reaction forces.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
172 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
12.7.1.5. RSA A7
This analysis considers the inputs in the X, Y and Z directions. The mean and standard-deviation values
obtained are 1.00 and 0.10, respectively, implying that the correlation between the spectrum-analysis
and transient-analysis results is better than for single-directional input. The better spectrum-analysis
correlation is a result of the reactions having directions orthogonal to the input, which are not significantly
improved by including the missing-mass and rigid-response effect; however, these reactions should
remain smaller than the primary reactions.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 173
Dynamic Simulation of a Nuclear Piping System Using RSA Methods
The absolute acceleration values at support (node 4) and far from support (nodes 16 and 34) are com-
pared in the table below:
It is clearly shown that the absolute acceleration value at node 4 (at support) shows close comparison
with the full-transient solution after the addition of the missing-mass effect.
12.8. Recommendations
When performing a spectrum analysis that takes into account missing-mass and rigid-response effects,
consider the following hints and recommendations:
• Define the ZPA frequency value (fZPA) corresponding to the input spectrum. (This value is the beginning
of the frequency range for which the acceleration remains constant and equal to the ZPA.)
• Include all modes having a frequency below fZPA in the spectrum analysis.
• Choose a CQC mode combination to correctly combine modes with closely spaced frequencies (if any).
• Always include the missing-mass effect in the first simulation to verify its significance. (The missing-mass
response is the result of a static analysis of the structure, so its computational cost is small.)
• Include the rigid-response effect. The Lindley-Yow method is easier to use, as only the ZPA value must be
input.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
174 of ANSYS, Inc. and its subsidiaries and affiliates.
Input Files
12.9. References
The following references are used in this example problem:
2. Morante, R., Y. Wang. "Reevaluation of Regulatory Guidance on Modal Response Combination Methods for
Seismic Response Spectrum Analysis." NUREG/CR-6645, BNL-NUREG-52576. December 1999.
3. U.S. Nuclear Regulatory Commission. "Combining Modal Responses and Spatial Components in Seismic
Response Analysis." Regulatory Guide 1.92. Rev. 2. July 2006.
4. Gupta, A.K. Response Spectrum Method in Seismic Analysis and Design of Structures. CRC, 1993.
• BM3model.dat -- Input file for modeling the BM3 nuclear piping system.
X-Direction Input
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 175
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
176 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 13: Centrifugal Impeller Analysis Using Cyclic Symmetry
and Linear Perturbation
This example problem demonstrates a centrifugal impeller blade analysis using cyclic modeling methods
and linear-perturbation solution approaches. The problem includes a modal analysis, a full-harmonic
analysis, a prestressed modal analysis using linear perturbation, a prestressed full-harmonic analysis
using linear perturbation, and a prestressed mode-superposition harmonic analysis using linear perturb-
ation.
The results of the cyclic symmetry analysis are verified against reference results obtained from an ana-
lysis of the full (360-degree) model.
13.1. Introduction
Cyclic symmetry modeling is a powerful tool for analyzing structures with a repetitive geometric pattern
360 degrees around an axis of symmetry. Cyclic symmetry is found in many civil engineering structures
such as domes, cooling towers, and industrial chimneys. It can also be found in mechanical equipment
such as milling cutters, turbine blade disks, gears, fans, and pump impellers.
Cyclically symmetric models can be solved using a single portion (called the basic sector) of the whole
structure, enforcing the continuity and compatibility boundary conditions between the cyclic substruc-
tures. A cyclic symmetry analysis vastly reduces model sizes and computational cost.
The following model shows a cyclically symmetric sector of a single centrifugal impeller blade:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 177
Centrifugal Impeller Analysis Using Cyclic Symmetry and Linear Perturbation
The model consists of a shroud and an impeller blade assembly with a sector angle of 27.692 degrees.
The full model is composed of 13 primary blades and splitters, as shown:
Modal, perturbed prestressed modal with linear and nonlinear base static solution, full-harmonic, per-
turbed prestressed full-harmonic with nonlinear base static solution, and perturbed prestressed mode-
superposition harmonic analyses with nonlinear base static solution are performed on the cyclic-sector
model.
The perturbed modal cyclic symmetry analysis includes the initial prestressed condition from both the
linear and nonlinear static analyses. The initial stress state with the linear static solution is produced by
the spinning impeller assembly as well as the pressure load applied on the impeller blade. The initial
stress state with the nonlinear static analysis is produced by the spinning impeller blade, a pressure
load applied on the impeller blade, and thermal loading applied on all nodes of the impeller blade as-
sembly model.
The perturbed full-harmonic and perturbed mode-superposition harmonic cyclic symmetry analyses
include the initial prestressed condition due to the nonlinear static analysis. The initial stress state is
produced by the spinning of the impeller assembly and thermal loading applied on all nodes of the
impeller blade assembly model.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
178 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
Modal, perturbed modal, full-harmonic, and perturbed full-harmonic analyses of the full (360-degree)
model are also performed to validate the accuracy of cyclic-sector model results.
13.3. Modeling
Following are the topics related to the cyclic symmetry modeling for this problem:
13.3.1. Impeller Blade Modeling
13.3.2. Contact Modeling
SURF154 elements are created on the surface of the impeller blade to apply a pressure load.
The following model shows the low- and high-edge components of the cyclic-sector model:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 179
Centrifugal Impeller Analysis Using Cyclic Symmetry and Linear Perturbation
A tetrahedral mesh is used to obtain the matched node pattern along the high and low edges. To ensure
the most accurate solution, it is best to have identical node and element face patterns on the low and
high edges of the cyclic sector.
If you issue the CYCLIC command before meshing the cyclic sector (AMESH or VMESH ), the mesh will
have identical node and element face patterns on the low and high edges when possible.
For more information about sector low and high edges, see Edge Component Pairs in the Cyclic Symmetry
Analysis Guide.
Figure 13.6: Bonded Contact Pair Between the Shroud and Impeller
The pure penalty contact algorithm is used for bonded contact. Because MPC-based bonded contact
can create overconstraint along the edge of the cyclic sectors (due to the internally generated constraint
equations), the pure penalty or augmented Lagrangian penalty methods are preferred for bonded
contact when used in conjunction with cyclic symmetry application.
The contact surface is meshed using CONTA174 elements. The target surface is meshed using TARGE170
elements.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
180 of ANSYS, Inc. and its subsidiaries and affiliates.
Boundary Conditions and Loading
Units
Unit System Metric (mm, kg, N, s,
mV, mA) Degrees
rad/s Celsius
Angle Degrees
Rotational Velocity rad/s
Pressure MPa
Figure 13.7: Static Fluid Pressure Load on Hub Wall, Main Blade, and Splitter
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 181
Centrifugal Impeller Analysis Using Cyclic Symmetry and Linear Perturbation
• Thermal loads caused by the difference in reference temperature and applied temperature and due to
thermal coefficient of expansion
The applied load is cyclic in nature. It is also possible to apply non-cyclic loading with different loading
values on each of the cyclic sectors (CYCOPT,LDSECT,SECTOR).
The rotational velocity (OMEGA,0,6000,0) is applied along the global Y axis. The impeller blade assembly
is modeled with a thermal coefficient expansion of 1.2e-005 °C-1. The reference temperature is held at
22° C and a body temperature of 50° C is applied on all nodes of the model to generate the thermal
load vector.
The thermal load vector generated from the base static solve can be ignored in the subsequent analysis
(THEXPAND).
For the full-harmonic, perturbed full-harmonic, and perturbed mode-superposition harmonic analyses,
the pressure load applied on the main blade, splitter, and hub wall is treated as a harmonically varying
load.
For information about linear perturbation, see Linear Perturbation Analysis in the Structural Analysis
Guide.
The CYCLIC command automatically detects cyclic symmetry model information such as edge compon-
ents, the number of sectors, the sector angles, and the corresponding cyclic coordinate system.
You can display the fully expanded model in the preprocessor (/CYCEXPAND,1,ON).
The Block Lanczos eigensolver is used to extract 10 modes. By default, all harmonic indices (0 to 6 in
this case) are calculated (based on (N - 1) / 2, where N is the number of sectors). You can ignore some
harmonic indices if desired (CYCOPT,HINDX).
/solu
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
182 of ANSYS, Inc. and its subsidiaries and affiliates.
Analysis and Solution Controls
/post1
set, list,,,,,,,order ! Sorts the harmonic index results in ascending order
! of eigen frequencies
finish
13.6.2. Linear Prestressed Modal Cyclic Symmetry Analysis with Linear Per-
turbation
The procedure to perform a linear prestressed modal cyclic symmetry analysis is essentially the same
as a standard modal analysis, except that you must first prestress the structure by performing a static
cyclic symmetry analysis.
1. Obtain a linear static solution. Enable file writing for performing a subsequent linear perturbation analysis
(RESCONTROL,LINEAR).
2. Perform a linear perturbed modal analysis where the linear prestress effects are included automatically.
3. Expand the modes and then postprocess the results from jobname.rstp.
/solu
antype,static,restart,,,perturb ! Perform a static restart with perturbation
! from the last load step and substep of
! the previous static solve
perturb,modal,,,allkeep ! Set the analysis options for perturbed modal
! analysis
solve,elform ! Reform element matrices
outres,all,all
modopt,lanb,10 ! Use Block Lanczos mode extraction procedure
mxpand,10 ! Expand the modes
solve
finish
/post1
file, ,rstp ! Specify the data file where results are to be found
/out,
finish
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 183
Centrifugal Impeller Analysis Using Cyclic Symmetry and Linear Perturbation
2. Enable the file writing option for multiframe restart from a static analysis (RESCONTROL).
4. Expand the modes and then postprocess the results from jobname.rstp.
To better demonstrate the nonlinear effects in the prestressed model, the rotational velocity is doubled
(OMEGA,0,6000,0) from the linear prestressed analysis, and thermal loading is added.
csys,0
antype,static ! Perform Static analysis
nlgeom,on ! Include large deformation effects
rescontrol,define,all,1 ! Enable the file writing in multiframe restart
nsub,10,10,10 ! Number of substeps = 10
allsel,all,all
nsle
bf,all,temp,50 ! Define temperature = 50
time,1.0
omega, 0, 6000, 0, ! Centrifugal load
kbc,0
solve
finish
/solu
antype,static,restart,,,perturb ! Perform a static restart with perturb
! from the last load step and substep of the previous
! static solution
perturb,modal,,,allkeep ! Set the analysis options for perturbed modal analysis
solve,elform ! Reform element matrices
outres,all,all
modopt,lanb,10 ! Use Block Lanczos mode extraction procedure
mxpand,10 ! Expand the modes
solve
finish
/post1
file, ,rstp ! Specify the data file where results are to be found
/out,
finish
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
184 of ANSYS, Inc. and its subsidiaries and affiliates.
Analysis and Solution Controls
The following input fragment shows the steps involved in this analysis:
/solu
solve
finish
Note
csys,0
antype,static ! Perform Static analysis
nlgeom,on ! Include large deformation effects
rescontrol,define,all,1 ! Enable the file writing in multiframe restart
nsub,10,10,10 ! Number of substeps = 10
allsel,all,all
nsle
bf,all,temp,50 ! Define temperature = 50
time,1.0
omega, 0, 6000, 0, ! Centrifugal load
kbc,0
solve
finish
/solu
antype,static,restart,,,perturb ! Perform a static restart with perturb
! from the last load step and substep of the previous
! static solution
perturb,harm,,,allkeep ! Set the analysis options for perturbed harmonic analysis
solve,elform
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 185
Centrifugal Impeller Analysis Using Cyclic Symmetry and Linear Perturbation
csys,0
antype,static ! Perform Static analysis
nlgeom,on ! Include large deformation effects
rescontrol,define,all,1 ! Enable the file writing in multiframe restart
nsub,10,10,10 ! Number of substeps = 10
allsel,all,all
nsle
bf,all,temp,50 ! Define temperature = 50
time,1.0
omega, 0, 6000, 0, ! Centrifugal load
kbc,0
solve
finish
/solu
antype,static,restart,,,perturb ! Perform a static restart with perturb
! from the last load step and substep of the previous
! static solution
perturb,modal,,,allkeep ! Set the analysis options for perturbed modal analysis
solve
outres,all,all
modopt,lanb,2
mxpand,2,,,yes
solve
finish
/solu
csys,0
antype,harm
hropt,MSUP ! Choose MSUP harmonic analysis option
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
186 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The linear (NLGEOM,OFF) prestressed modal frequencies of the cyclic sector obtained from the linear
perturbation analysis also show strong agreement with the full-model results, as shown in the following
table:
The nonlinear (NLGEOM,ON) prestressed modal frequencies obtained from the linear perturbation
analysis match within reasonable tolerance with the corresponding full-model results, as shown in the
following table:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 187
Centrifugal Impeller Analysis Using Cyclic Symmetry and Linear Perturbation
The full-harmonic analysis postprocessing (/POST1) results of the cyclic-sector model are compared to
the results of the full-model analysis:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
188 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The results show strong agreement between the total deformation plots of the cyclic-sector model and
the full model.
The nodal solution plots (NSOL) show the amplitude of a nodal degree-of-freedom (DOF) value with
respect to the frequency of excitation:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 189
Centrifugal Impeller Analysis Using Cyclic Symmetry and Linear Perturbation
Figure 13.10: Nodal Solution Plots with Respect to the Frequency of Excitation
(b) Comparison of Nodal Displacement Amplitude of Cyclic Sector Model and Full Model
The plots indicate the occurrence of a resonance condition at 2920 Hz. The displacement amplitude of
nodes at symmetric angular positions on sectors 1 through 10 is also plotted. The plots show strong
agreement between the cyclic and full-model results.
Part (b) of the figure shows the comparison of the nodal solution result (NSOL) of a node located at
an identical location on the cyclic and full models. The agreement of results is very good and plotted
curves are superimposed.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
190 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The following table compares the displacement amplitude (UY) of nodes placed symmetrically across
each sector at a 2920 Hz frequency excitation. The values show strong agreement between the cyclic-
sector and full-model results.
For the nonlinear prestressed perturbed full-harmonic analysis, the following figure compares the
postprocessing (/POST1) results of the cyclic-sector model to the results of the full-model analysis:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 191
Centrifugal Impeller Analysis Using Cyclic Symmetry and Linear Perturbation
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
192 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The results show strong agreement between the total deformation plots of the cyclic-sector model and
the full model.
The nodal solution plots (NSOL) show the amplitude of a nodal degree-of-freedom (DOF) value with
respect to the frequency of excitation:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 193
Centrifugal Impeller Analysis Using Cyclic Symmetry and Linear Perturbation
Figure 13.13: Nodal Solution Plots with Respect to the Frequency of Excitation
(b) Comparison of Nodal Displacement Amplitude of Cyclic Sector Model and Full Model
The plots indicate the occurrence of a resonance condition at 2920 Hz. The displacement amplitude of
nodes at symmetric angular positions on sectors 1 through 10 is also plotted. The plots show strong
agreement between the cyclic and full-model results.
Part (b) of the figure shows the comparison of the nodal solution result (NSOL) of a node located at
an identical location on the cyclic and full models.
The following table compares the displacement amplitude (UY) of nodes placed symmetrically across
each sector at 2920 Hz (frequency of excitation). The values show strong agreement between the cyclic-
sector and full-model results.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
194 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
For the nonlinear prestressed perturbed mode-superposition harmonic analysis, the following figures
show the postprocessing (/POST1) results of the cyclic-sector model:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 195
Centrifugal Impeller Analysis Using Cyclic Symmetry and Linear Perturbation
The results show close comparison between the total deformation plots obtained from nonlinear
prestressed perturbed mode-superposition harmonic and the nonlinear prestressed perturbed full-har-
monic analyses.
The nodal solution plots (NSOL) show the amplitude of a nodal degree-of-freedom (DOF) value with
respect to the frequency of excitation:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
196 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Figure 13.16: Nodal Solution Plots with Respect to the Frequency of Excitation
The plots indicate the occurrence of a resonance condition at 2920 Hz. The displacement amplitude of
nodes at symmetric angular positions on sectors 1 through 10 is also plotted. The plots show close
comparison with the nonlinear prestressed perturbed full-harmonic analysis results.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 197
Centrifugal Impeller Analysis Using Cyclic Symmetry and Linear Perturbation
Part (b) of the figure shows the comparison of the nodal solution result (NSOL) of a node located at
an identical location for the nonlinear prestressed perturbed mode-superposition and nonlinear
prestressed perturbed full-harmonic analyses.
The following table compares the displacement amplitude (UY) of nodes placed symmetrically across
each sector at 2920 Hz (frequency of excitation). The values show close comparison between the non-
linear prestressed perturbed mode-superposition and nonlinear prestressed perturbed full-harmonic
analyses results.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
198 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
1. The difference in the number of elements is due to the surface elements used to apply the pressure
load.
The following figure shows the gain in CPU time by solving the analysis via cyclic symmetry modeling:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 199
Centrifugal Impeller Analysis Using Cyclic Symmetry and Linear Perturbation
Figure 13.17: Comparison of CPU Time Between the Cyclic Sector Model and the Full Model for
Various Analyses
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
200 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The following figure shows that the VT solver is approximately four times faster than the sparse solver
for the full-harmonic analysis, and nearly twice as fast for the nonlinear prestressed perturbed full-har-
monic analysis:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 201
Centrifugal Impeller Analysis Using Cyclic Symmetry and Linear Perturbation
Figure 13.18: Comparison of CPU Time Between the VT Solver and the Sparse Solver for Harmonic
Analysis on the Cyclic Sector Model
13.8. Recommendations
To perform a similar type of analysis, consider the following hints and recommendations:
• Specify a cyclic symmetry analysis (CYCLIC) before meshing the cyclic sector.
The program can then automatically create matched node pairs at the low and high edges, providing
the most accurate results from a cyclic analysis (AMESH and/or VMESH).
For example, applied boundary conditions (D) and/or loading (such as SF) on the cut boundaries of
a cyclic sector model are not applicable.
• In a harmonic or static cyclic analysis with non-cyclic loading, all applicable harmonic index solutions are
constructed automatically.
• Use the linear perturbation procedure for linear or nonlinear prestressed modal analyses and harmonic
analyses.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
202 of ANSYS, Inc. and its subsidiaries and affiliates.
Input Files
13.9. References
The theory behind the cyclic symmetry formulation is available in the following reference:
1. Dickens, John M. Numerical Methods for Dynamic Substructure Analysis. Ph.D. Thesis from University of
California, Berkeley. 1980.
Cyclic-sector model:
• sector_model.cdb -- The common database file for the cyclic-sector model of the centrifugal impeller
blade
• surf154_gen.dat -- The common input file for the cyclic-sector model of the centrifugal impeller blade
to read SURF154 elements
• cyclic_modal_analysis.dat -- Input file for the cyclic-symmetry modal analysis of the impeller blade
• cyclic_modal_perturb.dat -- Input file for the cyclic-symmetry linear prestressed perturbed modal
analysis of the impeller blade
• cyclic_model_harmonic.dat -- Input file for the cyclic-symmetry full-harmonic analysis of the impeller
blade
• full_model.cdb -- The common database file for the full-model analysis of the centrifugal impeller
• surf154_gen_360_degree.dat -- The common input file for the full-model analysis of the centrifugal
impeller to read SURF154 elements
• full_modal_analysis.dat -- Input file for the modal analysis on the full impeller model
• full_modal_perturb.dat -- Input file for the linear prestressed perturbed modal analysis on the full
impeller model
• full_model_harmonic.dat -- Input file for the full-harmonic analysis on the full impeller model
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 203
Centrifugal Impeller Analysis Using Cyclic Symmetry and Linear Perturbation
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
204 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 14: Rotordynamics of a Shaft Assembly Based on a
Representative Model of Nelson-Vaugh Rotor
Rotordynamics plays a crucial role in identifying critical speeds, and to ultimately design rotating
structures that tolerate extremely high vibrations. This example illustrates the application of rotordy-
namics analysis procedures using the Nelson-Vaugh rotor model (p. 225).
A 2-D axisymmetric representation of the 3-D solid model is used to perform a rotordynamic analysis.
The results of the 2-D axisymmetric model analyses are compared to the full 3-D solid model results.
• Orbit plot
14.1. Introduction
Rotating machines such as steam or gas turbines, turbo-generators, internal combustion engines, motors,
and disk drives can develop inertia effects that can be analyzed to improve the design and decrease
the possibility of failure. Current trends in rotating equipment design focus on increased speeds, which
increase operational problems caused by vibration. At higher rotational speeds, the inertia effects of
rotating parts must be consistently represented to accurately predict rotor behavior.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 205
Rotordynamics of a Shaft Assembly Based on a Representative Model of Nelson-
Vaugh Rotor
Inertia effects in rotating structures are usually caused by gyroscopic moment introduced by the precise
motions of the vibrating rotor as it spins. As spin velocity increases, the gyroscopic moment acting on
the rotor becomes critical. Not accounting for inertia effects at the design level can lead to bearing and
support structure damage. It is also important to consider bearing stiffness, support structure flexibility,
and damping characteristics to understand the stability of a vibrating rotor.
In the sections that follow, modeling details and analysis procedures for a rotating structure are detailed.
Generally, a 3-D model directly available from the CAD can be used for the analysis; however, 3-D
models result in a large number of nodes and elements models. This example demonstrates how to
extract a plane 2-D model from the 3-D model, which can be analyzed using far fewer nodes and ele-
ments. The ease of use, accuracy, and performance of 2-D and 3-D model analyses are compared.
For more information about rotordynamics, refer to the Rotordynamic Analysis Guide and Rotating
Structure Analysis in the Advanced Analysis Guide.
Modal, Campbell diagram, and unbalance response analyses are performed on the 2-D axisymmetric
model represented in the figure below. The effect of gyroscopic moment on the rotating structure dy-
namics is observed in each analysis.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
206 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
To validate the accuracy of the results obtained with the 2-D axisymmetric model, the same analyses
were performed with the corresponding 3-D solid model. The results are then used as a benchmark for
the 2-D axisymmetric model results.
14.3. Modeling
The following modeling topics are discussed in this section:
14.3.1. 3-D Modeling of Flexible Rotor Component
14.3.2. Axisymmetric Modeling of the Flexible Rotor Component from 3-D Geometry
14.3.3. Disk and Bearing Modeling
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 207
Rotordynamics of a Shaft Assembly Based on a Representative Model of Nelson-
Vaugh Rotor
14.3.2. Axisymmetric Modeling of the Flexible Rotor Component from 3-D
Geometry
The following process uses the ANSYS Workbench product to extract the 2-D axisymmetric geometry
the 3-D geometry:
2. Use the Slice by Plane operation on the unfrozen portion of the model as shown below.
3-D Geometry
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
208 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
The extracted 2-D geometry is then meshed with the general axisymmetric SOLID272 elements as shown
in the figure below.
For modeling, three Fourier nodal planes along the circumferential direction are considered with KEY-
OPT(2) = 3 as pictured in the figure below.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 209
Rotordynamics of a Shaft Assembly Based on a Representative Model of Nelson-
Vaugh Rotor
Figure 14.7: 2-D Axisymmetric Model with Fourier Nodal Planes
The 3-D mesh is generated based on the 2-D mesh of master planes using the axis of symmetry of the
rotor model. It is defined using the SECTYPE and SECDATA commands as follows:
! Define the axis of symmetry
SECT, 1, axis
SECDATA, 1, x1,y1,z1, x2,y2,z2 ! Coordinates to define the axis of symmetry
Finally, using the axis of symmetry and the NAXIS command, the 3-D mesh is automatically generated
as shown in the figure below.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
210 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
Figure 14.9: Rigid Disk Modeled Using MASS21 Element (3-D Solid Model)
1. Create the contact pairs at the interface corresponding to the location of the rigid disk (the geometry is
sliced beforehand at this location).
2. Mesh the contact surfaces of the 3-D solid model with CONTA174 elements, and contact surfaces of the
2-D axisymmetric model with CONTA175 elements.
In the case of the 3-D solid model, bonded surface-to-surface contact pairs are used to define contact
between the cross sectional elements and the pilot node. The pilot node is located at the center of the
cross section. The cross sectional elements and/or contact surfaces are meshed using CONTA174 elements
(3-D 8-node surface-to-surface contact) as shown in the figure above.
Similarly, for the 2-D axisymmetric model, bonded node-to-surface contact pairs are used to define
contact between the cross sectional nodes and the pilot node. CONTA175 elements (2-D/3-D node-to-
surface contact) are used to mesh the cross sectional nodes and/or contact surfaces as shown in the
figure below.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 211
Rotordynamics of a Shaft Assembly Based on a Representative Model of Nelson-
Vaugh Rotor
Figure 14.10: Disk Modeled Using MASS21 Element (2-D Axisymmetric Model)
The Multipoint Constraints (MPC) algorithm is used for bonded contact through KEYOPT(2) = 2 when
modeling both contact pairs.
TARGE170 elements are used to mesh the pilot node in both models. All degrees of freedom (DOF) of
this pilot node are constrained through KEYOPT(4) = 111111.
The following input fragments show the steps involved in creating the contact pairs for the 2-D
axisymmetric model:
/prep7
type, 1
real, 1
mat, 1
nn = 0
*get, numn, node,, count
*do, i, 1, numn ! Loop to create contact elements on selected nodes
nn = ndnext(nn)
e, nn ! Create contact elements on node one by one
*enddo
allsel, all, all
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
212 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
KEYOPT(2) = 1 is used to define the elements in a plane parallel to the YZ plane. The DOFs of these
elements are then UY and UZ.
The X, Y, and Z axes should be added to the following figure (see Figure 14.4: 3-D Geometry (p. 208) for
the directions).
Figure 14.11: Bearings Modeled with COMBI214 Elements (3-D solid model)
1. Create the contact pairs at the bearing locations (the geometry is sliced beforehand at these locations).
2. Mesh the contact surfaces with CONTA174 elements for the 3-D solid model, and CONTA175 elements for
the 2-D axisymmetric model.
3. Create nodes at the center of the cross section of the rotor at each bearing location (at the pilot node loc-
ations). It will represent the second node of the bearing.
4. Create the COMBI214 elements using these nodes and the pilot nodes.
The contact pairs shown in the figure below must be created to model the bearings. The contact pairs
are modeled in a similar fashion to the disk modeled in the previous section.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 213
Rotordynamics of a Shaft Assembly Based on a Representative Model of Nelson-
Vaugh Rotor
Figure 14.12: Bonded Contact Pairs at Bearing Locations (3-D Solid Model)
To model an orthotropic bearing, an additional node is created at the center of the cross section of the
rotor at the bearing location. This node is then connected to pilot node using COMBI214 elements. To
visualize this element, offset the node along the Y-direction without altering the results as shown in
the figure below and in Figure 14.11: Bearings Modeled with COMBI214 Elements (3-D solid model) (p. 213).
Figure 14.13: Bearings Modeled with COMBI214 Elements (2-D Axisymmetric Model)
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
214 of ANSYS, Inc. and its subsidiaries and affiliates.
Boundary Conditions and Loading
Material properties of the 3-D/2-D Model of a Nelson Rotor are shown in the table below.
The properties of the point mass (rigid disk) are shown in the table below.
The stiffness properties of the two identical orthotropic bearings are shown in the table below.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 215
Rotordynamics of a Shaft Assembly Based on a Representative Model of Nelson-
Vaugh Rotor
The effect of the unbalanced mass on the disk is represented by forces acting in the two directions
perpendicular to the spinning axis. The forces are applied on a node located on the axis of rotation at
the same location as the point mass. The amplitude of the unbalance force is equal to the mass multiplied
by the distance of the mass unbalance to the spin axis.
The following input fragment shows the steps to apply the unbalance force:
/com, Force, f0 = mass * radius
f0 = 1.90e-3
nodeUnb = node(89.01, 0, 0)
The following input fragment shows the steps to extract these modes:
/solu
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
216 of ANSYS, Inc. and its subsidiaries and affiliates.
Analysis and Solution Controls
The modal analysis with gyroscopic effects is performed on the model using the complex QRDAMP ei-
gensolver.
The following input fragment shows the steps to perform the modal analysis with gyroscopic effects:
/SOLU
esel,,ename,,272
esel,a,ename,,21
cm,rot_part, elem
esel, all
pival = acos(-1)
spinRpm = 50000 ! Rotational velocity in rpm
spinRds = spinRpm*pival/30
cmomega, rot_part, spinRds ! Apply rotational velocity along rotating velocity (X)
The following input fragment shows the steps to perform the Campbell Diagram Analysis:
/SOLU
esel,,ename,,272
esel,a,ename,,21
cm, rot_part, elem
esel, all
pival = acos(-1)
spinRpm = 0 ! Rotational velocity in rpm
spinRds = spinRpm*pival/30
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 217
Rotordynamics of a Shaft Assembly Based on a Representative Model of Nelson-
Vaugh Rotor
cmomega, rot_part, spinRds
solve
/POST1
In this analysis, the unbalance is considered as loading. (see Boundary Conditions and Loading (p. 215)
for more details).
The frequency of excitation is specified as synchronous with the rotational velocity (SYNCHRO). The
rotational velocity (CMOMEGA) determines only the rotational velocity direction vector of the rotating
component. The spin of the rotor is automatically calculated (HARFRQ).
The following input fragment shows the steps to perform the unbalance response analysis:
/SOLU
finish
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
218 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Mode 2-D Axisymmetric Model (Hz) 3-D Solid Model (Hz) Error
# (%)
1 189.72 189.00 0.38
2 208.96 208.04 0.44
3 639.91 639.59 0.05
4 654.54 653.25 0.20
5 733.26 732.77 0.07
6 807 805.64 0.17
7 990.36 991.73 0.14
8 1780.5 1782.50 0.11
9 1781.1 1785.20 0.23
10 2016.9 2009.60 0.36
11 2092.6 2086.90 0.27
12 3291.7 3287.80 0.12
The natural frequencies of the 2-D axisymmetric model in rotation (50,000 rpm) also show good
agreement with the 3-D solid model results, as shown in the following table.
Mode 2-D Axisymmetric Model (Hz) 3-D Solid Model (Hz) Error
# (%)
1 169.07 168.25 0.49
2 232.54 231.75 0.34
3 627.97 627.19 0.12
4 652.32 651.39 0.14
5 752.14 751.59 0.07
6 808.51 807.25 0.16
7 990.36 991.73 0.14
8 1763.00 1766.00 0.17
9 1798.90 1802.00 0.17
10 1931.30 1923.50 0.41
11 2192.50 2187.60 0.22
12 3291.70 3287.80 0.12
The /POST1 results of the Campbell diagram analysis are shown in the figure that follows.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 219
Rotordynamics of a Shaft Assembly Based on a Representative Model of Nelson-
Vaugh Rotor
With the help of the Campbell diagram analysis, we can identify the forward (FW) and backward (BW)
whirls, as well as possible unstable frequencies (though none are present in this example). In the table
below, the whirls and natural frequencies of the 2-D axisymmetric model at maximum rotational speed
(100,000rpm) are compared with the 3-D solid model results.
Frequency (Hz)
Whirl 100000 RPM
Mode # 2-D Axisymmetric 3-D Solid Model 2-D Axisymmetric 3-D Solid Model
Model Model
1 BW BW 144.98 144.17
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
220 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Frequency (Hz)
Whirl 100000 RPM
Mode # 2-D Axisymmetric 3-D Solid Model 2-D Axisymmetric 3-D Solid Model
Model Model
2 FW FW 264.59 263.92
3 BW BW 602.73 601.48
4 BW BW 649.45 648.72
5 FW FW 792.94 791.73
6 FW FW 814.48 813.99
7 FW BW 990.36 991.73
8 BW BW 1745.08 1748.14
9 FW FW 1817.30 1820.05
The Campbell diagram analysis helps to determine the critical speeds of the rotating structure (PRCAMP
). Critical speeds are compared in the table below. For a synchronous excitation, the critical speeds
correspond to the intersection points between the frequency curves and the 1.0 slope line. The critical
speeds of the 2-D axisymmetric and 3-D solid models show strong agreement.
The results of the unbalance response analysis post processed in /POST26 are shown in the following
figure. The logarithmic plots show the variation of the displacement amplitudes of two selected nodes
with respect to the frequency of excitation. The first node is located near the rigid disk, and it corresponds
to the light blue curve. The second node is located near the bearing location, and it corresponds to the
purple curve.
The critical frequencies appear where the amplitudes are largest, and correspond to the critical speeds.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 221
Rotordynamics of a Shaft Assembly Based on a Representative Model of Nelson-
Vaugh Rotor
The orbits after a full harmonic analysis can be plotted in POST1 as shown in the figure below (PLORB
command). For the solid element as well as for the axisymmetric element model, we need to add
massless line elements using BEAM188 on the rotational velocity axis to plot these orbits. The orbits of
the 2-D axisymmetric model at a frequency of 1666.67 Hz are shown in the figure below. The rotor line
is in dark blue, while the orbits are in light blue.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
222 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The following input fragment shows the steps to produce an orbits plot at a given frequency:
/POST1
esel,r,ename,, 188 ! Select BEAM188 elements to produce orbits
set,1, 200 ! Visualize orbits at frequency 1666.67 Hz
/view,,1,1,1
plorb ! Displays the orbital motion of a rotating structure
The memory and CPU usage of the 3-D model is shown in the following table.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 223
Rotordynamics of a Shaft Assembly Based on a Representative Model of Nelson-
Vaugh Rotor
3-D Solid Model
Elements # Nodes # No. of Memory CPU Time
Equations required for (Sec)
in-core (MB)
Campbell
Diagram 186.141 45.56
Analysis
9239 15123 45341
Unbalance
Response 605.464 4645.95
Analysis
The CPU times for the unbalance response analysis are represented in the following bar graph.
Figure 14.18: CPU Time of 2-D Axisymmetric and 3-D Solid Models (Unbalance Response Analysis)
14.8. Recommendations
• In the CAD geometry, identify the axisymmetric and non-axisymmetric parts. A non axisymmetric part should
be considered as follow:
– If inertia is negligible, create a non-rotating component based on this part. The gyroscopic effect will not
be taken into account.
– If inertia is not negligible, delete the part and create an equivalent axisymmetric geometry so that its
gyroscopic effects are included. The simplest way to do this is to add a point mass on the rotational velocity
axis. The point mass characteristics are based on the part mass and inertias. The two rotary inertias per-
pendicular to the rotational velocity axis must be equal to guarantee the axisymmetry.
• For 3-D and 2-D axisymmetric modeling, the geometry is sliced at rigid disks and bearing locations so that
those components are easily created and connected to remote points attached to the interfaces.
• When meshing a 2-D axisymmetric model using SOLID272 or SOLID273, choose an appropriate number of
Fourier nodes in the circumferential direction to ensure good accuracy and minimize the computational
cost. For typical rotordynamics problems in linear dynamics, three Fourier nodes are sufficient.
• When performing a Campbell diagram of a structure, always check the eigenfrequencies at zero rotational
velocity first. If the supports (bearings or boundary conditions) are symmetric, bending frequencies should
appear in pairs. If that is not the case in a 3-D model, try refining the mesh.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
224 of ANSYS, Inc. and its subsidiaries and affiliates.
Input Files
• To perform the unbalance response analysis of 3-D and 2-D axisymmetric models, the unbalance response
is introduced using complex forces defined at a node on the rotational velocity axis. The unbalance response
may be defined using a point mass away from the rotational velocity axis only in the case of a nonlinear
transient analysis.
14.9. References
The following references are used in this example problem:
1. H.D. Nelson and J.M. Mc Vaugh. The Dynamics of Rotor-Bearing Systems Using Finite Elements. Journal of
Engineering For Industry. ASME. May 1976.
2. Beley, A., C. Rajakumar, P. Thieffry.“Computational Methods for Rotordynamics Simulation.” NAFEMS World
Congress, 2009.
• 2-D_axi_symm.cdb -- The common database file for a 2-D axisymmetric model of the Nelson-Vaugh rotor.
• 2-D_axi_symm_campbell.dat -- Input file for performing a Campbell diagram analysis of the 2-D
axisymmetric model.
• 2-D_axi_symm_harm.dat -- Input file for performing an unbalance response analysis of the 2-D
axisymmetric model.
• 3D_solid_model.cdb -- The common database file for a 3-D solid model of the Nelson-Vaugh rotor.
• 3D_solid_campbell.dat -- Input file for performing Campbell diagram analysis of the 3-D solid model.
• 3D_solid_harm.dat -- Input file for performing unbalance response analysis of the 3-D solid model.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 225
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
226 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 15: Calibrating and Validating a Hyperelastic Constitutive
Model
This example problem demonstrates the hyperelastic curve-fitting capabilities used to select constitutive
model parameters to fit experimental data. Several issues that influence the accuracy of the curve fit
are discussed. Validation of the resulting constitutive model is demonstrated by comparison with a
tension-torsion experiment.
15.1. Introduction
Several hyperelastic constitutive models can be used to model the large deformation behavior of
elastic materials; however, it is sometimes difficult to select parameters to adequately match the beha-
vior of the material. The curve-fitting process fits the hyperelastic constitutive model parameters to a
set of experimental data using a least-squares minimization.
Curve fitting is relatively simple, but certain conditions can affect the accuracy of the resulting constitutive
model. The constitutive model should therefore be compared with experimental data to ensure that it
adequately reproduces the material behavior over the actual range of deformation.
In this problem, the experimental data are obtained from a simulation of a hyperelastic test suite (uni-
axial, biaxial, and planar tension tests) using common experimental test specimens. Using this data,
parameters for a constitutive model are determined using hyperelastic fitting capabilities that focus on
use of the three-, five-, and nine-parameter Mooney-Rivlin hyperelastic models.
After demonstrating the fitting procedure and selecting a suitable constitutive model, a tension-torsion
experiment is simulated and compared to the experimental data to validate the predictions for the
model.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 227
Calibrating and Validating a Hyperelastic Constitutive Model
The uniaxial specimen is similar to ASTM D412-C (ASTM Standard D412, 2006).
The crosshead is displaced by 396 mm, giving a measured engineering strain in the gage section of
662 percent and a calculated engineering stress of 58.1 MPa.
The equibiaxial specimen is disc-shaped, with 16 equally spaced tabs about the circumference. The tabs
are stretched 127.3 mm, resulting in a measured engineering strain in the gage section of 336 percent
and a calculated engineering stress of 22.1 MPa.
For the planar specimen, the crosshead is displaced by 191.6 mm, giving a calculated engineering strain
of 639 percent and a calculated engineering stress of 54.7 MPa.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
228 of ANSYS, Inc. and its subsidiaries and affiliates.
Analysis and Solution Controls
The experiment consists of clamping each end of the specimen into the test apparatus, then stretching
the specimen by 50 percent of its original gage length and twisting one end of the specimen for four
complete revolutions. Following is the resulting moment-vs.-rotation data:
The command input shown here is for illustrative purposes only. While curve fitting can be done
using command input, ANSYS, Inc. recommends using the graphical user interface (GUI) to perform
the curve fitting, or at least visually validating the results using the GUI to ensure a sound fit.
Following is an input example of the commands used to fit a hyperelastic constitutive model to a set
of uniaxial stress-strain data:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 229
Calibrating and Validating a Hyperelastic Constitutive Model
/PREP7
TBFT,FADD,1,HYPER,MOONEY,3
TBFT,EADD,1,UNIA,UNIAX.LOG
TBFT,SOLVE,1,HYPER,MOONEY,3
TBFT,FSET,1,HYPER,MOONEY,3
The TBFT,FADD command initializes the curve-fitting procedure for a hyperelastic, three-parameter,
Mooney-Rivlin model assigned to material identification number 1.
TBFT,EADD reads the uniaxial experimental data in the uniax.log file as the fitting data for material
number 1. The experimental data in the file is a set of engineering-strain vs. engineering-stress input:
0.819139E-01 0.82788577E+00
0.166709E+00 0.15437247E+01
0.253960E+00 0.21686152E+01
0.343267E+00 0.27201819E+01
0.434257E+00 0.32129833E+01
0.526586E+00 0.36589498E+01
0.619941E+00 0.40677999E+01
0.714042E+00 0.44474142E+01
0.808640E+00 0.48041608E+01
0.903519E+00 0.51431720E+01
0.998495E+00 0.54685772E+01
0.109341E+01 0.57836943E+01
TBFT,SOLVE determines the three constitutive parameters for the Mooney-Rivlin model, minimizing the
difference between the model and the experimental data.
For this problem, the fitted parameters for the three-parameter Mooney-Rivlin model are:
C10 = 1.338856
C01 = 5.236214 x 10-1
C11 = - 1.648364 x 10-2
The mesh consists of 1,332 SOLID186 elements using the default formulation (a mixed-displacement
pressure formulation with reduced integration).
The attachment of the test specimen (p. 228) to the test apparatus is simulated by boundary conditions
applied to the specimen in the region of the clamps, as described here:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
230 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
• The back-right clamp region is attached to a rigid-contact surface and fixed in place.
• The front-left clamp region is attached to a rigid-contact surface and displaced in the z direction to
simulate a clamping displacement equal to 25 percent of the specimen thickness. The same is true for
the front-right clamp region.
The stretching to 50 percent engineering strain is simulated by displacing the rigid-contact surfaces
attached to the right clamp regions while holding left clamp regions fixed.
The torsion of the specimen is simulated by holding the left clamp region in place and twisting the
keypoints attached to the right contact surfaces about the longitudinal axis.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 231
Calibrating and Validating a Hyperelastic Constitutive Model
Figure 15.6: Comparison of the Data and Fits Over the Entire Range of Data
Thus far, it is obvious that none of the models provide a suitable fit to the entire range of experimental
data. The reason is that the least-squares fitting procedure is minimizing the error over the entire range
of data; therefore, it can be detrimental to include data that is not representative of the actual range
of use.
If the experimental data range is limited to about 100 percent strain, however, the fitted parameters
shown in the following table are obtained:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
232 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The following figure is a comparison of the models with the parameters fit to the modified experimental
data:
Figure 15.7: Parameters Fit to Experimental Data to About 100 Percent Strain
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 233
Calibrating and Validating a Hyperelastic Constitutive Model
For the equibiaxial and planar experiments, any of the three models might be acceptable; however, the
comparison with the uniaxial data might indicate that none of the three models are acceptable.
The behavior of the model outside the fitted range can significantly differ from the actual response of
the material. For example, the model parameters fit to the experimental data to 100 percent strain have
been used to simulate the hyperelastic test suite to strains of about 200 percent, as shown in the fol-
lowing comparisons:
Figure 15.8: Comparison of the Data and Fits Showing Predictions Outside the Range of Fitted
Data
Beyond 100 percent strain, it becomes apparent that some of the predictions quickly deteriorate. In all
three comparisons, the nine-parameter model quickly loses accuracy, and it appears that the three- and
nine-parameter Mooney-Rivlin models have lost stability for the biaxial deformation case.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
234 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The following figure shows a contour plot of the strain energy density at the end of simulation. The
plot offers a general idea of the overall deformation of the specimen.
With the exception of the clamp regions, the deformation shows a uniform pattern in the gage region
along the axis of twisting. Perpendicular to the axis of twisting is a large strain-energy density near the
outside edge of the specimen, decreasing toward the center.
The following figure shows a comparison of the model with the experimental moment vs. theta data:
After a seemingly anomalous first data point, the error between the simulation and experiment is in
the range of 2 to 4 percent. Throughout the entire simulation, the five-parameter Mooney-Rivlin model
predicts a higher moment for an equivalent twist, which is not entirely expected by the error plots for
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 235
Calibrating and Validating a Hyperelastic Constitutive Model
the hyperelastic test suite comparisons; nevertheless, a maximum four percent error appears to be a
reasonable margin of error for this simulation.
15.6. Recommendations
When performing a similar type of calibration and validation, consider the following recommendations:
• Obtain test data from at least two (and preferably all three) of the experiments in the hyperelastic test suite.
• Ensure that the test data covers the range of deformation over which the constitutive model will be used.
• If the error between the experimental data and the constitutive model is too great, try limiting the experi-
mental data to the range of deformation over which the constitutive model will be used.
• Use the constitutive model within the range of fitted data only.
• Use an independent experiment to validate that the constitutive model adequately matches the material
behavior.
15.7. References
The following references are used in this example problem:
1. ASTM Standard D1043. Standard Test Method for Stiffness Properties of Plastics as a Function of Temperature
by Means of a Torsion Test. ASTM International. West Conshohocken. 2006.
2. ASTM Standard D412. Standard Test Methods for Vulcanized Rubber and Thermoplastic Elastomers-Tension.
ASTM International. West Conshohocken. 2006.
• tension_torsion.cdb -- The common database file containing the model information for this problem
(called by tension_torsion.dat ).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
236 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 16: Evaluation of Mixed-Mode Stress Intensity Factors and
T-stress for 3-D Surface Flaws
This example problem shows how to evaluate mixed-mode stress-intensity factors, J-integrals, and T-
stresses for cracks in structural components. Analyses of a simple semicircular surface flaw in a rectan-
gular block and a warped flaw along a tubular joint are discussed.
• Evaluating mode-I stress-intensity factors and T-stresses for a semicircular surface flaw in a rectangular block.
• Evaluating mixed-mode stress-intensity factors and T-stresses for a warped semi-elliptical surface flaw in a
tubular joint.
16.1. Introduction
Fracture analysis is widely used to predict component failure caused by preexisting small cracks, allowing
one to take precautions to prevent further crack growth or to determine the remaining life of the
structure.
To assess the fracture damage, stress intensity factors (SIFs) must be evaluated accurately. Because it
is difficult to determine accurate SIFs using a closed-form analytical solution for cracks in complex
structures, finite-element analysis is used instead.
• Interaction Integral Method -- Performs the SIF calculation during the solution phase of the analysis
and stores the results for later postprocessing.
• Displacement Extrapolation Method -- Performs the SIF calculation during postprocessing. This
method is limited to problems involving linear elasticity with homogeneous, isotropic materials near
the crack region.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 237
Evaluation of Mixed-Mode Stress Intensity Factors and T-stress for 3-D Surface Flaws
The interaction integral method is used here as it allows for the calculations to be performed during
the solution phase. It is also more accurate, as the domain integral expressions are naturally aligned
with the FE formulation.
A torus is created around the crack front to control the mesh at the crack front. The objective is to find
KI along the crack front and validate the results with reference results.[1][2][3]
Following is the semicircular crack and torus along the crack front used to create a sweep mesh with
SOLID186:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
238 of ANSYS, Inc. and its subsidiaries and affiliates.
Problem Description
Figure 16.2: Semicircular Crack and Torus Along the Crack Front
Figure 16.3: X-Joint Pipe Full Model with Warped Surface Flaw at Welded Joint
The problem consists of two tubular members attached to each other by a welded joint. The tubular
members have outer diameters of 323.85 mm (D1), 219.08 mm (D2), and thicknesses of 15.88 mm (t1)
and 8.18 mm (t2), respectively. The semi-elliptical surface crack lies on a plane parallel to the radial
direction of the heavier running pipe.
A torus is created around the crack front to control the mesh at the crack front. The semi-elliptical surface
crack at the weld toe is warped along the welded joint and it is perpendicular to the outer surface of
the 323.85 mm diameter pipe in the thickness direction.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 239
Evaluation of Mixed-Mode Stress Intensity Factors and T-stress for 3-D Surface Flaws
The warped crack profile at the welded joint is created via the interaction of a rotated semi-ellipse on
the 323.85 mm diameter pipe and an extruded circle normal to the inner surface of the same pipe at
the welded joint, as shown:
Figure 16.5: Two-plane Symmetry of X-joint Pipe with Warped Surface Flaw Geometry
16.3. Modeling
For 3-D models, the recommended element type near the crack front is SOLID186, the 20-node brick
element.
The rectangular block and x-joint pipe models are meshed with SOLID187 elements (except for the regions
around the crack tip), as shown:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
240 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
Figure 16.6: Rectangular Block Model with Boundary Conditions and Loading Applied
Regions around the crack tips are meshed with SOLID186, as shown:
Due to the two-plane symmetry of the x-Joint problem, a quarter model is considered for analysis, as
shown:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 241
Evaluation of Mixed-Mode Stress Intensity Factors and T-stress for 3-D Surface Flaws
Figure 16.8: Two-plane Symmetry of X-joint Pipe with Warped Surface Flaw at Welded Joint
If the crack surface is not normal to any component in the global coordinate system, a local coordinate
system should be created in such a way that one component of this coordinate system is perpendicular
to the crack surface. For example, in the x-joint model, a local cylindrical coordinate system is created
such that one of its components is normal to the crack surface, as shown:
Figure 16.9: X-joint Pipe with Warped Surface Flaw at Welded Joint in Thickness Direction.
In the x-joint model, the warped crack in the thickness direction is perpendicular to the inner surface
of the 323.85 mm diameter pipe.
A torus is created around the crack front to obtain a good sweep mesh (VSWEEP). A common area
separates the torus and the remaining volume at the interface.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
242 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
In 3-D models, a fracture has two surfaces with a common edge at the crack front; the same is true for
the torus. The two torus surfaces are used to create a clean sweep mesh around the crack front. One
surface is defined as the source, the other as the target. In this procedure, the sweep mesh generates
brick elements with one layer of prism elements around the crack front. Following is the sweep mesh
with SOLID186 around the crack front in the x-joint pipe model, with the source and target areas for
the sweep mesh indicated in the deformed body:
Figure 16.10: Sweep Mesh with SOLID186 Around the Crack Front
The warped surface contains a sweep mesh and a set of nodes, along with source and target areas in
the deformed torus body. The source and target areas existed at the same location in the nondeformed
structure.
After meshing, crack parameters must be defined. The following example inputs show how to define
parameters associated with contour integral calculation:
! Initiate a new contour integral calculation and define type
CINT, NEW, 1 ! Initiate new calculation and assign ID
CINT,TYPE,SIFS ! Replace SIFS with JINT to find J-integral
! Define crack tip nodal component to find SIFs or J-integral for specified flaw
CINT,CTNC,component_name,,1
A line component of the crack front lines (CM) is created. This component is useful for selecting nodes
attached to the crack front (NSLL). The nodal component of these nodes is used to define crack tip
node components (CINT,CTNC), as shown in the following figures:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 243
Evaluation of Mixed-Mode Stress Intensity Factors and T-stress for 3-D Surface Flaws
Material Properties
Young's Modulus (Pa) 2E+11
Poisson's Ratio 0.3
Analysis of the x-joint model is performed with two-sided symmetry. Two-plane symmetric boundary
conditions are applied and one midside node at the horizontal plane is constrained in the opposite
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
244 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
direction to restrict rigid-body motion. A pressure load of -1000 MPa is applied on the top of the small-
diameter tube, as shown in Figure 16.8: Two-plane Symmetry of X-joint Pipe with Warped Surface Flaw
at Welded Joint (p. 242).
Output controlled (OUTRES,CINT) is used to write the SIFs, J-integral, and T-stress results to the results
(.rst) file and output files.
The following postprocessing commands can be used to list and/or plot and list the results:
/POST1
PLCINT,FRONT,1,,,K1 ! Plot KI results
PLCINT,FRONT,1,,,K2 ! Plot KII results
PLCINT,FRONT,1,,,K3 ! Plot KIII results
PLCINT,FRONT,2 ! Plot J-integral results
PLCINT,FRONT,3,,,TSTR ! Plot T-Stress results
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 245
Evaluation of Mixed-Mode Stress Intensity Factors and T-stress for 3-D Surface Flaws
The plot shows the maximum Von Mises stress occurring at the crack tip.
To verify the SIF accuracy obtained via the interaction integral method, the results are compared to
analytical results of Newman and Raju (p. 249) and finite-element result of Kamaya (p. 249). The results
agree closely with those obtained in the references.
Because the crack exists at the center of the rectangular block, the results for half of the crack are
therefore plotted in this figure, where normalized SIF = :
The T-stress results for the semicircular surface flaw are plotted in this figure, where normalized T-stress
= T-stress / σo :
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
246 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 247
Evaluation of Mixed-Mode Stress Intensity Factors and T-stress for 3-D Surface Flaws
The SIFs of the mixed-mode 3-D problem with the warped crack surface were compared with those of
Chong Rhee.[4] The comparison shows that the normalized KI, KII and KIII results agree well with the
reference results near the end points, as shown:
where T is the thickness of the pipe, and the normalized distance is the distance from
center of the crack along the crack front divided by half of the crack tip length.
A fine mesh (using meshing bias [LESIZE,,,,,SPACE]) at the end-point regions improves the results.
The T-stress results for the X-joint pipe with warped flaw are plotted in this figure, where normalized
T-stress = T-stress / σo :
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
248 of ANSYS, Inc. and its subsidiaries and affiliates.
References
Figure 16.20: Normalized T-Stress Results: X-Joint Pipe with Warped Flaw
16.8. Recommendations
When setting up a fracture analysis for 3-D structures, consider the following hints and recommendations:
• The recommended element type for 3-D fracture models along the crack tip is the 20-node brick element
SOLID186.
• A fine sweep mesh along the crack front yields more accurate results. Create a torus along the crack front
in the model geometry for this purpose.
• If any component in the global coordinate system is not orthogonal to the crack surface, create one local
coordinate system with one component normal to the crack surface. This action is necessary to define the
crack plane normal (CINT,NORM).
• For multiple cracks, use unique crack tip IDs and separate sets of CINT commands for each.
• To determine SIFS and J-integrals for the same crack, use two separate CINT command sets.
• The interaction integral method gives accurate results because the contour integral is evaluated at points
far away from the crack-tip. Disregard the first contour results, however, as they are less accurate than the
other results due to the nearness of the crack tip.
For more information, see Understanding Fracture Mechanics in the Fracture Analysis Guide.
16.9. References
The following references are used in this example problem:
1. Newman, J. C., Jr., and I. S. Raju.“Analysis of Surface Cracks in Finite Plates Under Tension or Bending Loads.”
NASA Technical Paper 1578 . National Aeronautics and Space Administration (1979).
2. Kamaya, M.“Stress Intensity Factor of Surface Crack with Undulated Front.” JSME International Journal49.4
(2006): 529-535.
3. Anderson, T.L. Fracture Mechanics - Fundamentals and Applications 3rd ed. Boca Raton: CRC, (2005).
4. Rhee H. C. and M. M. Salama.“Mixed-Mode Stress Intensity Factor Solutions of a Warped Surface Flaw by
Three-Dimensional Finite Element Analysis.” Engineering Fracture Mechanics 28-2 (1987): 203-209.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 249
Evaluation of Mixed-Mode Stress Intensity Factors and T-stress for 3-D Surface Flaws
• semicircular_surface_flaw.dat -- Input file for the analysis of the semicircular surface flaw in the
rectangular plate.
• xjoint_pipe_with_warped_flaw.dat -- Input file for the fracture analysis of the x-joint model.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
250 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 17: Impact of a Metal Bar on a Rigid Wall
This example problem is an impact simulation using a 3-D model of a metal bar hitting a rigid wall. The
problem shows the advantages of using impact constraints for modeling contact in a nonlinear transient
dynamic analysis.
Several combinations of time-integration methods and contact algorithms are also investigated, using
different material models to show how various choices affect the performance and accuracy of the finite-
element solution of impact problems.
17.1. Introduction
Simulating contact in a transient dynamic analysis can be challenging. The presence of inertial forces
can adversely affect convergence in contact problems. The simulated response must also be accurate
to ensure that it remains stable, and consistent with physical behavior, over a long time period.
The impact of a metal bar on a rigid wall is ideal for demonstrating various solution options, as this
problem has been extensively studied and documented (p. 266), and there are existing analytical and
numerical solutions available for comparison.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 251
Impact of a Metal Bar on a Rigid Wall
Figure 17.1: Geometry and FE Model of a Metal Bar Impacting a Rigid Wall
Several transient analyses are performed, taking into account the following criteria:
• Rigid, elastic, and elastoplastic (p. 253) material behavior for the bar
For rigid and elastic material behavior, the results of displacements, velocities, strain energies (SE), and
kinetic energies (KE) are compared to the analytical solution For the elastoplastic material behavior, the
results of mushroom radius, final length, equivalent plastic strain, and von Mises stress are compared
to a reference solution.
17.3. Modeling
The bar is modeled with a 3-D coarse mesh using 495 SOLID186 elements, as shown in Figure 17.1: Geo-
metry and FE Model of a Metal Bar Impacting a Rigid Wall (p. 252).
Frictionless contact between the rigid wall and the end of the bar is modeled using TARGE170 and
CONTA174 elements.
• Location of the contact-detection point on the nodal point-normal to target surface is activated (KEY-
OPT(4) = 2). This setting is required for the rigid-impact case, as geometric irregularities that may exist
on the contact surface could create a nonsymmetric contact-force distribution and affect solution
convergence adversely.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
252 of ANSYS, Inc. and its subsidiaries and affiliates.
Material Properties
• Contact predictions are changed (KEYOPT (7) = 3) -- Maintains the minimum time/load increment
whenever a contact status change occurs.
• Impact constraints are used KEYOPT (7) = 4) -- The time increment is adjusted automatically.
The latter two time-incrementation controls (KEYOPT (7) = 3 and KEYOPT (7) = 4) activate time-step size
control to capture all contact status changes.
The bar is modeled as a rigid body using only TARGE170 elements, with automatically constrained
boundary conditions for rigid target nodes (KEYOPT (2) = 1). The target elements are located on
the exterior surface of the bar which has been premeshed with SOLID186 elements.
The program builds internal multipoint constraints between the nodes on the exterior surface
of the rigid body and a pilot node located at the center of gravity. The pilot node is also shared
by a 3-D point mass with rotary inertia (modeled with a MASS21 element).
The location of the center of gravity, and the mass and moments of inertia properties for MASS21,
are estimated by performing a single load step solution with the option for precalculating masses
(IRLF,-1).
For more information, see Modeling Rigid Bodies in a Multibody Analysis in the Multibody Ana-
lysis Guide.
Before obtaining the solution, the underlying SOLID186 mesh is unselected (ESEL,U command).
The bar is modeled as a flexible body with linear elastic material properties.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 253
Impact of a Metal Bar on a Rigid Wall
An initial velocity of 227 m/s is applied to all nodes on the bar (including the rigid-body pilot node for
rigid impact).
The rigid wall, defined using TARGE170 elements, is internally constrained in all six directions.
For each of the three impact scenarios (p. 253), three transient analyses are performed using the following
settings and commands:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
254 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The total time (simulation time) represents a slightly larger time than the time needed for the bar to
impact the rigid wall; the bar moving at a velocity of 227 m/s requires about 0.4405e-5 seconds to
cover a 1 mm gap before impacting the rigid wall.
The minimum of 100 substeps ensures a smooth response. The maximum of 10000 substeps allows the
automatic time-stepping method to cut back the time increment to satisfy the impact constraints. Because
the goal in this case is to study the displacement and velocity response of the bar at some points of
interest over the total time period, the nodal displacement and velocity solution data is written at every
substep to the results file:
OUTRES,NSOL,ALL
OUTRES,V,ALL
The nodal-displacement solution data is written to the results file at every substep, and stresses and
plastic strains are written at the last substep only, as follows:
OUTRES,NSOL,ALL
OUTRES,STRS,LAST
OUTRES,EPPL,LAST
As shown in the following table and subsequent figures, a comparison of the rebound velocity and the
total energy at post-impact separation indicates that the rebound velocity is greater than the initial
velocity when using either:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 255
Impact of a Metal Bar on a Rigid Wall
• element-level time-incrementation control (based on contact-status changes (KEYOPT(7) = 3)) without nu-
merical damping.
The greater rebound velocity after impact is related to the absence of energy conservation, evidenced
by the increase in the total energy. With impact constraints (KEYOPT(7) = 4), energy conservation is
enforced, so there is no increase in the total energy or the rebound velocity.
Precise satisfaction of impact constraints is necessary to conserve energy and obtain the same rebound
velocity as the initial velocity, as evidenced by a comparison of the results for impact constraints (KEY-
OPT(7) = 4) with augmented Lagrangian contact (KEYOPT(2) = 0), and Lagrange multiplier contact
(KEYOPT(2) = 4).
Note
Lagrange multiplier contact for rigid bodies is generally not recommended due to the pos-
sibility of overconstraints. It is used here with CONTA175 at a single node (to avoid overcon-
straints) to demonstrate that energy conservation is dependent upon the exact satisfaction
of impact constraints.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
256 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
As shown in the following figure, adding numerical damping to the Newmark method helps to improve
the results when using either:
Numerical damping removes some of the energy from the system and therefore counterbalances the
increase in the total energy due to non-energy-conserving time integration, lowering the rebound ve-
locity by a small amount.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 257
Impact of a Metal Bar on a Rigid Wall
The HHT method removes energy only from the higher frequency modes (which are not dominant in
the rigid-impact case). When compared to the Newmark method with damping, therefore, the HHT
method does not yield as much improvement when KEYOPT(7) = 0 or 3, as shown in this figure:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
258 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Generally, impact constraints yield a close-to-expected or an expected solution, with or without damping.
Other types of element-level time-incrementation controls (KEYOPT(7) = 0 or 3) rely heavily on numer-
ical damping for a stable solution.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 259
Impact of a Metal Bar on a Rigid Wall
KEYOPT(7)
225.39 291.60 218.73 239.67 220.84 275.20
=3
KEYOPT(7)
222.67 236.31 216.46 221.80 221.10 230.30
=4
Finite-element discretization introduces additional nonphysical high-frequency modes into the structural
response of the flexible bar. When using no control (KEYOPT(7) = 0) and no numerical damping, the
displacement and velocity responses of the flexible bar become corrupted by high-frequency modes
after impact, as shown in this figure:
The higher modes are excited due to absence of enforcement of energy conservation, as evidenced by
the total energy output. The response becomes worse over time due to the continued increase in the
total energy.
Adding numerical damping stabilizes the response by dissipating some of the extra energy, as shown
in Figure 17.6: Elastic Impact: Newmark Method with Damping (p. 261) and Figure 17.7: Elastic Impact:
HHT Method with Damping (p. 262). Eliminating higher modes, however, requires a great deal of numer-
ical damping, which can yield a solution that is significantly different from the expected physical response.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
260 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 261
Impact of a Metal Bar on a Rigid Wall
Using impact constraints (KEYOPT(7) = 4) ensures energy conservation implicitly at the contact interface,
preventing excitation of higher modes. Exact energy conservation is possible only when the impact
constraints are satisfied exactly. When using the augmented Lagrangian or penalty-contact options
(KEYOPT(2) = 0 or 1), a slight violation of constraints exists, leading to a small loss of energy. The energy
loss is minimized via the program’s automatic time-increment adjustments; however, the adjustment
decreases the rebound velocity after separation by only a small amount, giving a stable and smooth
response much closer to the expected physical response and other published results for similar prob-
lems.[1]
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
262 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
In the table above, R and L are expressed in mm, and σeqv is expressed in MPa.
The numerical response is stable because the plastic dissipation of the total energy in the bar is much
more significant than the jump in energy due to either non-conservation of energy during time integ-
ration or the dissipation of energy because of added numerical damping.
Although the peak values for stresses and strains are comparable for different choices, the contour plots
for equivalent plastic strain and von Mises stress indicate that satisfaction of impact constraints (which
conserves energy for the contact interface) gives a better distribution of the stresses and strains, as
shown in the following figures:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 263
Impact of a Metal Bar on a Rigid Wall
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
264 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Figure 17.11: Animation: Elastoplastic Impact Using the HHT Method with Damping
In the following table, a comparison of computational times (in seconds) indicates that using impact
constraints with automatic time-increment adjustment (KEYOPT(7) = 4) requires the least amount of
time in all cases:
CONTA174 Key Option NMK with No Damping NMK with Damping HHT with Damping
Setting (p. 257) (p. 258) (p. 259)
Rigid Impact
KEYOPT(7) = 0 39 42 45
KEYOPT(7) = 3 179 321 190
KEYOPT(7) = 4 45 41 44
KEYOPT(7) = 4 with
38 41 41
Lagrange Multiplier
Elastic Impact
KEYOPT(7) = 0 783 638 753
KEYOPT(7) = 3 12186 11511 13091
KEYOPT(7) = 4 404 413 507
Elastoplastic Impact
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 265
Impact of a Metal Bar on a Rigid Wall
By using impact constraints with automatic time-increment adjustment, fewer substeps and equilibrium
iterations are necessary to obtain the transient response.
With no control (KEYOPT(7) = 0), more analysis time is required, as more substeps and equilibrium iter-
ations are required. The absence of energy conservation at the contact interface forces smaller time
increments.
Using element-level time-incrementation control based on contact status changes (KEYOPT(7) = 3) requires
the most time, as smaller time increments are necessary to avoid sudden changes in the contact status.
17.8. Recommendations
When performing a similar type of impact simulation, consider the following recommendations for ob-
taining better accuracy and faster performance:
• Activate impact constraints (KEYOPT(7) = 4 on the CONTA174 element) to enforce energy conservation at
the contact interface. This option helps to maintain accuracy of the nonlinear transient response over long
simulation times.
• For rigid- or elastic-impact scenarios, use the HHT time-integration method with small numerical damping
(0.1) to damp out high-frequency noise.
17.9. References
The following references are used in this example problem:
1. Carpenter, N.J., R.L. Taylor, and M.G. Katona. "Lagrange constraints for transient finite element surface
contact." International Journal for Numerical Methods in Engineering. 32 (1991): 103-128.
2. Hallquist, J.O. and D.J. Benson. "DYNA3D User's Manual." Report No. UCID-19592 (1987). Rev. 3. Lawrence
Livermore National Laboratory.
3. Kamoulakos, A. "A Simple 'Benchmark' for Impact." Bench Mark. (1990): 31-35.
4. Simo, J.C. "Algorithms for Static and Dynamic Multiplicative Plasticity that Preserve the Classical Return
Mapping Schemes of the Infinitesimal Theory." Computer Methods in Applied Mechanics and Engineering.
68 (1998): 1-31.
5. Wilkins, M.L. and M.W. Guinan. "Impact of Cylinders on a Rigid Boundary." Journal of Applied Physics. 44.3
(1973): 1200-1206.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
266 of ANSYS, Inc. and its subsidiaries and affiliates.
Input Files
• impact_rigid.cdb -- Common database file for the rigid-impact simulation (called by impact_ri-
gid.dat ).
• impact_elastic.cdb -- Common database file for the elastic-impact simulation (called by im-
pact_elastic.dat).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 267
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
268 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 18: Viscoelastic Analysis of an All-Ceramic Fixed Partial
Denture
This example problem demonstrates the fictive-temperature material model using the Tool-Naray-
anaswamy (TN) shift function to examine residual stresses in an all-ceramic fixed partial denture (FPD).
A coupled-field solution process, including transient thermal and nonlinear structural analyses, is used
in the problem simulation.
• Running a transient thermal analysis to determine the temperature profile at various time steps.
• Running a nonlinear structural analysis to determine residual stresses due to thermal loading.
18.1. Introduction
Materials are generally considered to be viscoelastic if they have an elastic and viscous behavior. The
elastic behavior is typically rate-independent and represents the recoverable deformation due to loading,
while the viscous behavior is typically rate-dependent and represents dissipative mechanisms within
the material.
A wide range of materials (such as polymers, glassy materials, soils, biological tissue, and textiles) exhibit
viscoelastic behavior. Viscoelastic materials exhibit viscous fluid behavior at high temperatures and
solid behavior at low temperatures.
For most viscoelastic materials, the effect on the material properties caused by changes in temperature
is similar to that of the effect caused by changes in the time scale. Such materials are considered to be
thermorheologically simple. A general material property called the shift function can reduce the con-
stitutive relation at a reference temperature and shifted time. The shift function can lessen the amount
of experimentation needed to determine the material parameters.
The following shift functions are available for representing thermorheologically simple materials:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 269
Viscoelastic Analysis of an All-Ceramic Fixed Partial Denture
• TN with fictive temperature -- Suitable for the melting and solidification process of viscoelastic materials
such as glass and stiff polymers.
The shift functions reproduce the behavior of a wide range of viscoelastic materials. For special require-
ments, user-defined shift functions can also be defined.
The fictive temperature is the temperature at which the current microstructure of glass is in an equilib-
rium state. For the TN shift function with fictive temperature model, the fictive temperature is used to
model materials containing an intrinsic equilibrium temperature that typically differs from the ambient
temperature of the material. The fictive temperature relaxes toward the ambient temperature similar
to the way that deviatoric and volumetric stiffness constants of the viscoelastic material relax toward
the long time-elastic constants.
With the shift function, the evolution of the fictive temperature for any thermal history can be calculated.
As the fictive temperature approaches the actual temperature, the viscoelastic material becomes more
relaxed. The fictive temperature model is often used to model the melting and solidification process of
viscoelastic materials such as glass and stiff polymers. This problem uses a fixed partial denture (FPD)
model to determine the residual stresses due to the solidification of a glass veneer on a ceramic core
material.
Metal-free ceramic materials are biocompatible, chemically durable, and aesthetically desirable. Such
materials are therefore ideal for FPDs. Thermal loading during the glass layer manufacturing process
causes residual stresses in FPDs. Higher residual stresses caused by thermal contraction incompatibility
between the veneer and core materials can lead to failure under occlusal loading in the oral cavity.[1]
The ability to determine residual stresses in an FPD subjected to thermal loading is useful for predicting
the life of the FPD.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
270 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
A 3-D model of the FPD is constructed from the digitized scanning data of a reference FPD.[1] Because
the original model is unavailable, a similar geometry of a three-unit FPD is created in ANSYS Design-
Modeler for use in this problem.
Sintering is commonly used for glass coatings. The high temperatures (~700° C) used during sintering
are reduced to room temperature (30° C) via free convection. Initially, some free thermal expansion
occurs in the core material due to the high temperature; however, this free expansion may not affect
stresses developed during free convection of the veneer and core together. For demonstration purposes,
a uniform initial temperature for both the veneer and the core is assumed, with no initial thermal ex-
pansion of the ceramic core.
The temperature distribution in the FPD is needed to determine the thermal residual stresses in the
veneer and core. A transient thermal analysis with a duration of 600 seconds is performed on the FPD.
The initial temperature is 700° C, and free convection cools the FPD to a room temperature of 30° C. A
convective heat transfer coefficient of 3.4E-5 W/mm2 °C is specified on all free surfaces, reducing sintering
temperature to the steady-state room temperature.
The temperature results are stored every 60 seconds. The results are used as input for a subsequent
nonlinear structural analysis to determine the residual stresses.
18.3. Modeling
Following are the modeling topics available for this problem:
18.3.1. Denture Modeling
18.3.2. Contact Modeling
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 271
Viscoelastic Analysis of an All-Ceramic Fixed Partial Denture
A 3-D surface-to-surface contact is used for creating the contact pair. The contact between veneer and
core is meshed with CONTA174 and TARGE170 elements, as shown in the following figure:
Figure 18.4: Contact Between Veneer and Core (CONTA174 and TARGE170)
The temperature profile (obtained in the thermal analysis) is used in a nonlinear structural analysis to
determine the residual stress in the veneer. The temperature results are read in as thermal loads at
different time steps (LDREAD).
The structural analysis requires the same mesh as that used for the thermal analysis. It is therefore ne-
cessary to change the thermal elements to structural elements (ETCHG). In this case, thermal element
SOLID87 is converted to structural solid element SOLID187. The contact CONTA174 and target TARGE170
elements are compatible with SOLID187, so those elements remain; contact element options are modified
as necessary (KEYOPT).
The FPD model contains 164,041 nodes. The model has 110,275 3-D 10-node tetrahedral structural
solid elements (56,811 veneer elements and 53,464 core elements), and 10,568 3-D 8-node surface-to-
surface contact elements.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
272 of ANSYS, Inc. and its subsidiaries and affiliates.
Material Properties
Core
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 273
Viscoelastic Analysis of an All-Ceramic Fixed Partial Denture
Coefficients of
Young’s Modulus
Poisson's Ratio Density (kg/mm3) Temperature (°C) Thermal
(GPa)
Expansion
30 1.019E-5
40 1.007E-5
50 9.955E-6
60 9.848E-6
100 9.499E-6
96 0.24 2.514E-6
200 9.190E-6
300 9.689E-6
400 1.099E-5
500 1.311E-5
700 1.975E-5
Veneer
Reference
Young’s Modulus
Poisson's Ratio Density (kg/mm3) Temperature Tref H/R (° K) *
(GPa)
(°C)
65 0.26 2.531E-6 700 46400
Following are the polynomial coefficients for glass and liquid thermal expansion:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
274 of ANSYS, Inc. and its subsidiaries and affiliates.
Boundary Conditions and Loading
The following example input defines the initial temperature and convection coefficient:
NSEL,ALL ! Select all nodes
IC,ALL,TEMP,700 ! Specifies initial temperature at nodes
The following figure shows the model with the initial temperature and convection on the FPD surface
applied:
For the contact pair, a thermal contact conductance of 4E-05 J/(sec-°C) is used for heat conduction
between the veneer and the core.
The following example input reads the temperature from the thermal analysis results:
LDREAD,TEMP,,,TIME,0,'thermal','rth' ! Read temperature from thermal analysis results file
The temperature is read at every load step (LDREAD). The TIME value represents the final solution time
of each load step.
The following figure shows the displacement boundary conditions and body force at TIME = 60 seconds.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 275
Viscoelastic Analysis of an All-Ceramic Fixed Partial Denture
The following example input obtains the solution for the thermal analysis:
ANTYPE,TRANS ! Perform a transient analysis
TRNOPT,FULL ! Specifies transient analysis options
TIME,600 ! Final time for solution
OUTRES,ALL,-10 ! Saved results at every 60 seconds
A linear static analysis (ANTYPE,STATIC) is performed. For the stress analysis, SOLID87 elements are
converted to SOLID187 elements.
The Tool-Narayanaswamy (TN) shift function with fictive temperature is used with the viscoelastic con-
stitutive model. As with other time-temperature superposition models, the shift function is accessible
via the TB,SHIFT command, as shown in the following input example:
TB,SHIFT,1,1,4,FICT ! Specify Tool-Narayanaswamy shift function
TBDATA, 1,700,46400,0.3 ! TN constants - Reference temperature, H/R and X
TBDATA, 4, 750, 0.25, 1.58E-5 ! Fictive T, weight, relaxation time
....
TBDATA, 16, 10.7510E-6, -2.4208E-8, 5.7267E-11 ! glass CTE coefficients
TB,PRONY,1,1,4,SHEAR
TBDATA,1, 0.48844 , 1.58E-5 ! Prony pairs
....
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
276 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
According to the reference results, volume relaxation for glass occurs much less rapidly than the shear
relaxation.[1] The Prony series input for volume decay is therefore not considered in this problem.
• A small increment is more computationally intensive, requiring a longer analysis time, but offers greater
accuracy.
Important
The structural portion of the analysis requires 10 solve operations. The first solution (time =
1 - 60) requires more time to converge than the remaining nine; however, it is not good
practice to increase the time increment for the first solution.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 277
Viscoelastic Analysis of an All-Ceramic Fixed Partial Denture
The following three figures show the von Mises stress distribution at different time steps in both the
FPD veneer and core:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
278 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Figure 18.11: Von Mises Stress in Veneer and Core at 300 Seconds
Figure 18.12: Von Mises Stress in Veneer and Core at 600 Seconds
The von Mises stresses in both veneer and core change with temperature due to natural convection
on the veneer surface. The stresses correspond to temperature loadings at 60, 300 and 600 seconds.
At high temperatures, the von Mises stresses are very high in the veneer as compared to the core.
As the veneer cools through the transition, the elastic moduli no longer relax on the time scale of the
cooling and--due to the different thermal contraction coefficients of the veneer and core--a jump in
the stress distribution occurs at the veneer-core interface, bringing the higher von Mises stresses in the
veneer close to the core stress at room temperature.
Stresses in the inner veneer surface at the interface are not significantly different from the core outer
surface stresses due to the stabilized fictive temperature (equal to the actual temperature) at lower
temperatures.
The following figure shows tension in the core and compression in the veneer at the room temperature
residual stress pattern:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 279
Viscoelastic Analysis of an All-Ceramic Fixed Partial Denture
Figure 18.13: Residual Principal Stress Distribution in Veneer and Core 600 Seconds
The tension is present because the surface cools most rapidly and becomes rigid, while the inner surface
of the veneer (at interface with the core) is still at a higher temperature. When the interface becomes
rigid and cools, it forces the surface under more compression. These residual stresses can be used for
analyzing FPDs under occlusal loading in the oral cavity and subsequently to predict the life of the FP-
Ds.[1]
The following figure shows the maximum residual principal stress distribution in the veneer and the
core from the reference results [1], which agree closely with the simulated results from this problem:
Figure 18.14: Reference Results: Residual Principal Stress Distribution in Veneer and Core
18.8. Recommendations
To perform a similar type of viscoelastic analysis, consider the following hints and recommendations:
• Save the thermal analysis results (OUTRES) at required time steps so that the temperatures can be read in
for the structural analysis.
• Specify the correct path for the thermal analysis results (.rth file) during the structural analysis (LDREAD).
• The sum of the fictive-temperature relaxation coefficients (specified in the table data activated via the
TB,SHIFT,,,,FICT command) should be 1.0.
For more information, see Viscoelasticity in the Material Reference and Shift Functions in the Mechanical
APDL Theory Reference.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
280 of ANSYS, Inc. and its subsidiaries and affiliates.
Input Files
18.9. References
This analysis was based on the following reference work:
1. DeHoff, Paul H. et al.“Viscoelastic Finite Element Analysis of an All-Ceramic Fixed Partial Denture.” Journal
of Biomechanics. 39 (2006): 40-48.
• denture_model.cdb -- Common database file for the fixed partial denture (FPD) finite-element model
and contact pair.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 281
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
282 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 19: Transient Dynamic Analysis of a Digger-Arm Assembly
This example problem uses a digger-arm assembly to demonstrate a transient dynamic analysis of a
multibody system.
The example also uses transient dynamic analyses for the following three cases:
• Some of the parts are flexible and the remainder are rigid.
19.1. Introduction
A multibody system is an assembly of bodies or parts in which some or all parts move relative to one
another. These assemblies may be simple or complex and may be composed of all rigid bodies or a
combination of rigid and flexible parts. The parts are constrained to each other by a set of kinematically
admissible constraints modeled as joints.
Common examples of multibody systems include land transportation systems, aviation systems, nautical
systems, and robotic systems.
The components of a multibody system may experience finite-strain effects and large displacements
and/or large rotations.
Dynamic analysis of a multibody system allows one to understand the interaction of the parts, evaluate
the stress and deformation fields in the parts, and calculate the fatigue life of critical components.
For further information about multibody systems, see the Multibody Analysis Guide.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 283
Transient Dynamic Analysis of a Digger-Arm Assembly
A mass of 500 kg (not shown) is attached to the bucket to simulate the load the bucket carries.
The various parts of the digger-arm assembly are connected to each other using joints. Actuation of
the two piston-cylinder arrangements causes the frame, arms, and connecting rods to move, thereby
causing the bucket to move as well. The entire system has essentially two free degrees of freedom.
The motion of the frame, arms, connecting rods, and the bucket are dependent on the two free degrees
of freedom. For this problem, the motion of the digger-arm is restricted to in-plane motion.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
284 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
• The second analysis assumes that the connecting rods are flexible and all other parts are rigid. The
flexible parts are modeled using 3-D finite elements.
• The third analysis is a variation of the second analysis. The flexible connecting rods are now modeled
as CMS superelements.
19.3. Modeling
The finite-element models for simple 2-D and 3-D problems are usually generated via the Mechanical
APDL command interface. For complicated assemblies, the ANSYS Workbench product is used, as it allows
one to define the geometry natively and to set up a project workflow that allows the entire analysis,
from model generation to results processing, to occur in a well-defined manner.
In this problem, the digger-arm assembly is modeled using ANSYS Workbench. The modeling description
identifies the relevant Mechanical APDL features that define the eventual finite-element model for
analysis.
The following modeling topics for the digger-arm assembly are available:
19.3.1. Modeling Rigid Parts
19.3.2. Modeling Joints
19.3.3. Modeling Flexible Parts
19.3.4. Modeling Flexible Parts with CMS Superelements
For details about the ANSYS Workbench program, see Workbench User's Guide.
1. A mass element, with mass equal to the mass of the connecting rod, is defined at the center of mass
(B) of the part.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 285
Transient Dynamic Analysis of a Digger-Arm Assembly
2. The node defining the mass element location (B) is identified as a pilot node using the TARGE170
element.
3. The locations at which the rod is connected to other parts are identified (A and B).
4. TARGE170 elements are defined between the mass element and the identified locations (A-B and B-
C).
5. Points A and B are used in the joint definitions when connecting to other parts.
For further information, see Modeling Rigid Bodies in a Multibody Analysis in the Multibody Analysis
Guide.
All rigid parts of the digger-arm assembly are essentially modeled in the same manner. The ANSYS
Workbench program automates the process for modeling the complicated geometries of the digger-
arm assembly. The following figure shows the entire rigid model for the assembly as generated by
ANSYS Workbench:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
286 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
Two nodes define a joint element. The relative motion between the two nodes is characterized by six
relative degrees of freedom. A joint element is defined based on the type of constraints imposed on
these relative degrees of freedom.
So that the constraints can be suitably imposed, local coordinate systems must be defined at the nodes
of the joint elements. In Mechanical APDL, the constraints are implemented via the Lagrange multiplier
method.
• Joint actuation
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 287
Transient Dynamic Analysis of a Digger-Arm Assembly
secnum,secid#
en,elem#,n1,n2 ! define revolute joint element
The parts of the assembly are connected to each other by various types of joints, depending on the
interaction between the parts. For example, the piston-cylinder arrangement requires a translational
joint; other interactions between parts seem to indicate that one part rotates relative to another about
an axis, requiring a revolute joint. Thus, as a first attempt, the joints between the parts are defined as
follows:
After the joints between parts have been defined, the kinematic behavior of the assembly must be
verified. The wrong choice of joints can cause either of the following conditions in the system:
• Overconstraint – More constraints exist than are required, leading to a stiff response of the system.
• Underconstraint – Fewer constraints exist than are required, leading to unnecessary deformation modes.
In either case, the result is inconsistent kinematic behavior, and the joints must be redefined to obtain
a proper kinematic response.
The free degrees of freedom for the digger-arm assembly are calculated as follows:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
288 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
The number of free degrees of freedom should be 2; therefore, the calculation shows that the Digger-
Arm model is severely overconstrained due to improper joint definitions.
The overconstraints can be minimized by redefining the joints between the various parts. Some exper-
imentation may be required. For example, consider the constraints in the joints between the following
part pairs (as shown in Figure 19.4: Connections Between Piston, Cylinder, and Arms (p. 289)):
It is apparent that Piston1 does not rotate with respect to Cylinder1 due to the rotational constraints
in the translational joint; therefore, the rotational constraints in the revolute joints between Piston1-
Arm1 and Piston1-Arm2 are not required. The relative displacement constraints between these parts
must be maintained, however, and so the revolute joint can be replaced by a spherical joint. The
spherical joint provides the same relative displacement constraints but allows the relative rotations to
be unconstrained.
In a similar way, other joints are replaced as necessary. The new joint definitions between the various
parts are as follows:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 289
Transient Dynamic Analysis of a Digger-Arm Assembly
The redefinition of joints outlined in the table is not necessarily unique; however, it meets the objective
of ensuring that a joint does not unnecessarily constrain motion in a particular direction which has
been otherwise constrained.
With the new joint definitions, the free degrees of freedom in the model are now calculated as follows:
The number of free degrees of freedom in the model matches the required number of degrees of
freedom. A transient dynamic analysis of the digger-arm assembly, with all parts rigid and joints redefined,
exhibits the correct kinematic response.
In this problem, the joints as defined above are used for all subsequent analyses.
For further information, see Connecting Multibody Components with Joint Elements in the Multibody
Analysis Guide and Constraints and Lagrange Multiplier Method in the Mechanical APDL Theory Reference.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
290 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
The two connecting rods are defined to be flexible for this analysis. The flexible parts are meshed with
SOLID185 elements. A total of 876 SOLID185 elements are used in the analysis. The following figure
shows the finite-element mesh:
For further information, see Modeling Flexible Bodies in a Multibody Analysis in the Multibody Analysis
Guide.
Following is the general process for generating and using a CMS superelement:
1. Prepare a full model of the flexible multibody system (including joint loads).
- Create a node component (the master component) that defines the master degrees of
freedom.
- Create a component of elements (the slave component) that are dependent on the master
degrees of freedom.
3. Generate a CMS substructure file (in the generation pass) characterizing the dynamic flexibility of the
body.
4. Use the CMS substructure information (in the use pass) in a standard analysis.
The CMS substructure information is used to define a CMS superelement representing the
flexible body.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 291
Transient Dynamic Analysis of a Digger-Arm Assembly
5. Expand the results of the analysis (in the expansion pass) to all elements in the flexible body to recover
its stress and deformation fields.
6. Postprocess the results for stress and deformation fields in the model.
For the analysis requiring a CMS superelement representation of the flexible part, the two connecting
rods are once again defined as flexible. The rods are then modeled as CMS superelements.
The following example input creates the master and slave components:
/com,*********** Create Components for CMS ***********
nsel,none ! create nodal component of master nodes
nsel,a,node,,Node#, ! add pilot node attached to this body
nsel,a,node,,Node#, ! add pilot node attached to this body
cm,COMPONENT_master,node
esel,s,elem,,COMPONENT
nsle ! select any nodes touching these bodies in the user-defined component
esln ! add in any touching elements such as contact and surf effect
cm,COMPONENT_slave,elem
allsel
Because the analysis phase considers both the lower and higher modes of vibration to be equally im-
portant, the free-interface method (CMSOPT,FREE) is used to generate the substructure files.
The following example input defines and uses the CMS superelement during the analysis phase:
/filnam, FILENAME
resume, ! resumes database
/prep7
cmsel,u,Component_slave ! unselect the flexible elements
cmsel,a,Component_master ! make sure the master nodes are selected
/solu
antype,trans
nlgeom, on
...
...
finish
The following example input shows expands the results to the full flexible body:
/filnam, FILENAME
resume
/solu
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
292 of ANSYS, Inc. and its subsidiaries and affiliates.
Boundary Conditions and Loading
expass,on
seexp,Component,FILENAME ! substructure name and the use pass jobname
numexp,all,,,yes ! Expand all time points
solve
finish
For further information, see Using Component Mode Synthesis Superelements in a Multibody Analysis
in the Multibody Analysis Guide.
The following example input applies the acceleration and displacement loading (ACEL and DJ):
acel,%_acelx%,%_acely%,%_acelz% ! apply acceleration loading via table
esel,s,type,, id1# ! select the first translational joint
dj,all,ux,%_load1% ! apply loads defined by table
esel,all
esel,s,type,, id2# ! select the other translational joint
dj,all,ux,%_load2% ! apply loads defined by table
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 293
Transient Dynamic Analysis of a Digger-Arm Assembly
In general, automatic time-stepping schemes are preferred over fixed-time stepping schemes, as the
program is allowed to cut back when the problem does not converge; when the convergence rate is
good, larger time increments are used to achieve faster convergence. For this problem, however, fixed
time-stepping with a time increment of 0.2 seconds is used so that the results at additional time points
can be captured. Three load steps, each having a 10-second duration, are used, for a total analysis time
of 30 seconds.
A default numerical damping of 0.1 is set (TINTP), and the HHT time-integration method is selected
(TRNOPT).
The following example input performs the essential tasks in the analysis process:
/solu
antype,4 ! transient analysis
nlgeom,on ! activate large-deformation effects
trnopt,full,,,,,hht ! HHT time-integration method
tintp,0.1 ! numerical damping
autots,off ! automatic time-stepping disabled
deltim,0.1,0.1,0.1 ! set time increments
time,10. ! set total time
timint,on ! activate time-integration effects
outres,all,all ! output all results at all time points
!
solve ! solve for 1st load step
time, 20.
solve ! solve for 2nd load step
time, 30.
solve ! solve for 3rd load step
finish
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
294 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The following figure shows the rotation of the frame with respect to ground for all three transient dy-
namic analysis cases. The motion of the frame (due to the activation of the piston-cylinder assemblies)
is the same in all three cases.
The following figure compares the relative rotation of the bucket with respect to the frame for the three
analysis cases. The rotations in the three cases are nearly identical.
Figure 19.9: Relative Rotation at General Joint Along Local Z Axis (Bucket-Frame)
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 295
Transient Dynamic Analysis of a Digger-Arm Assembly
Both relative-rotation figures shown above indicate that the kinematic behavior of the digger-arm as-
sembly, whether modeled as entirely rigid or as a rigid/flexible combination, exhibit the same behavior.
When flexibility is involved in an analysis, modeling the assembly as though it consisted of rigid parts,
and ensuring that the joints are defined correctly, allows the engineer to focus on other issues.
The stresses and deformation fields can be evaluated when the connecting rods are made flexible. The
following figure shows the deformation plots (USUM) of the connecting rods when the rods are modeled
as flexible:
The following figure shows the equivalent stresses in the connecting rods when the rods are modeled
using SOLID185 elements and as CMS superelements, respectively.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
296 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The time point is at 25 seconds (load step 3, substep 25). The equivalent stress plots show good
agreement with each other.
The following table shows the CPU time comparisons between the three analysis cases.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 297
Transient Dynamic Analysis of a Digger-Arm Assembly
In all cases, shared memory parallel with two CPUs is used. The results are reported for a 64-bit Linux
system.
The table and the other results discussed illustrate some essential analysis considerations:
• It is worthwhile to verify the kinematic behavior of the multibody system by assuming all parts as rigid,
as the time and resources required for the rigid analysis are small.
• Modeling the flexible bodies using CMS superelements results in significant time savings.
• Analysis of the assembly when the flexible part is modeled with 3-D finite elements necessarily requires
more time due to the increased number of degrees of freedom. Such an analysis, however, should be
performed only when the flexible part undergoes large deformations, either due to nonlinear material
behavior or to nonlinear geometric effects.
19.8. Recommendations
To perform a similar transient dynamic analysis of a multibody system, consider the following hints and
recommendations:
• Ensure that the motion and behavior of the model is kinematically consistent by modeling all parts as
rigid bodies connected to each other by joints.
• Relax overconstraints due to joints by selecting joints of similar behavior. Use a spherical or cylindrical
joint instead of a revolute joint if some constraints are deemed unnecessary.
• When modeling flexible parts, build the model incrementally and verify that the motion and behavior
are consistent with expected behavior.
• Use full modeling of the parts only when detailed information about the stress and deformation fields
is required.
• For reduced run times, model the flexible parts as CMS superelements whenever possible.
For additional recommendations, see Troubleshooting a Flexible Multibody Analysis in the Multibody
Analysis Guide.
• DiggerArm-Flexible.dat -- Input file for connecting rods modeled as flexible using SOLID185 elements.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
298 of ANSYS, Inc. and its subsidiaries and affiliates.
Input Files
• DiggerArm-CMS.dat -- Input file for modeling the flexible connecting rods as CMS superelements.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 299
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
300 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 20: Dynamic Simulation of a Printed Circuit Board Assembly
Using Modal Analysis Methods
This example demonstrates uses residual vectors to improve the solution accuracy in modal subspace
based analysis methods, such as modal superposition and power spectral density (PSD) analyses. The
problem includes a study of the computational efficiency of the results-expansion procedure used to
obtain the full model solution.
20.1. Introduction
Portable electronic devices such as digital cameras, mobile phones, and PDAs use printed circuit boards
(PCBs). Due to increased demands for convenience and multi-functionality, these devices are designed
with a focus on miniaturization to accommodate a higher density and smaller dimensions of integrated
circuit (IC) packages. These design constraints require smaller solder joints with finer pitch, contributing
to the vulnerability of interconnections at the board level. Exposure to harsh dynamic loading environ-
ments during transportation and customer use are a critical issue for PCBs. PSD analysis simulates the
random excitations with unknown loading encountered in these harsh conditions.
The modal superposition method efficiently solves a large linear dynamic system by transforming it
into a set of uncoupled equations using the normal modes of the system. The first step in the modal
superposition method is to obtain the eigen frequencies and eigen modes of the system through
modal analysis. The downstream modal transient, modal harmonic, and spectral analyses are then per-
formed.
In the modal analysis, only a subset of the lower frequencies is usually extracted, truncating the higher
frequency modes. As a result, the accuracy of the modal subspace based solution is not guaranteed,
though accuracy can be improved using residual vectors. The residual vectors are calculated and nor-
malized to the modes extracted, and can then be used in all the downstream analyses (modal transient,
modal harmonic, and spectral analyses).
The efficiency of the modal superposition expansion pass has been improved using a direct combination
approach of stress/strain modes. You can activate the expansion in the modal analysis by applying the
element results expansion option.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 301
Dynamic Simulation of a Printed Circuit Board Assembly Using Modal Analysis
Methods
20.2. Problem Description
The model below is a PCB assembly consisting of three PCBs stacked together. A PSD analysis with base
excitation using acceleration response spectrum is performed on this model. The aim is to determine
the 1-σ displacement solution and compare the accuracy of results with and without the residual vectors.
Improvement of the computational efficiency is verified via modal superposition expansion (MXPAND).
20.3. Modeling
This section describes the detailed modeling of the PCB assembly. The following modeling topics are
covered:
20.3.1. Modeling of the PCB Structure
20.3.2. Contact Modeling
BEAM188 uses Timoshenko beam theory, which includes shear deformation effects, and is one of the
most accurate beam elements for analyzing moderately stubby beam structures. A hexahedral dominant
mesh is used to mesh the board and the solids, resulting in each PCB having 14600 nodes. Both the
board and the IC packages are made of polyethylene material. The supporting columns are made of
aluminum alloy. The total number of nodes for the model is 44097 with 26046 elements.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
302 of ANSYS, Inc. and its subsidiaries and affiliates.
Boundary Conditions and Loading
• Support for lower and higher order elements on the contact and target surfaces. (SHELL181 is a linear element,
whereas SOLID186 is a quadratic element)
• No restrictions on the shape of the target surface. Surface discontinuities can result from manufacturing
flaws or mesh discretization.
For each board, fifteen contact pairs are identified, as seen in the figure below.
Figure 20.2: Bonded Contact Pairs Between IC Packages and Circuit Boards
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 303
Dynamic Simulation of a Printed Circuit Board Assembly Using Modal Analysis
Methods
Figure 20.3: PCB Boundary Conditions
The following example input shows how the fixed constraints are applied:
CM, N_BASE_EXCITE, NODE ! Group the set of nodes at the base of the columns as a component
D, ALL, ALL ! Apply displacement and rotation constraints to this set
N_BASE_EXCITE in the input above specifies the nodes at the junction of the supporting columns with
the base. For PSD analysis, the loading is applied in the form of base excitation on the set
N_BASE_EXCITE.
The PSD values for the input spectrum at different frequency points are shown in the figure below. The
input segment between 1.0E-02 and 1.0E+01 has 2 intermediate points in order to obtain a good fit
for the curve-fitting polynomial used in the PSD integration process.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
304 of ANSYS, Inc. and its subsidiaries and affiliates.
Analysis and Solution Controls
The modal analysis is performed using the Block Lanczos mode extraction method with the residual
vector. (For additional information on the residual vector method, refer to Residual Vector Method in
the Mechanical APDL Theory Reference and Residual Vector Method in Structural Analysis Guide)
Linear acceleration loading is applied in the y direction (ACEL), as it is needed to generate the residual
vector in conjunction with the downstream PSD analysis. Acceleration loading in the global y direction
is chosen to generate the residual vector because in the subsequent PSD analysis, base excitation is
applied in the same direction. The example input below shows how residual vectors are generated in
the modal analysis:
NUM_MODES=50 ! Number of modes to be extracted = 50
/SOLU
ANTYPE,MODAL ! Perform a modal analysis
MODOPT,LANB,NUM_MODES ! Use block Lanczos to extract the modes
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 305
Dynamic Simulation of a Printed Circuit Board Assembly Using Modal Analysis
Methods
RESVEC, ON ! Calculate the residual vectors
ACEL,0,1,0 ! Set the linear acceleration of the reference frame along the global Y axis
For power spectral density (PSD) analysis, base excitation is applied in the form of acceleration in the
vertical y direction, which is represented in terms of the acceleration due to gravity. The PSD values for
the input spectrum at different frequency points are given previously in Figure 20.4: PSD Base Excitation
Values (p. 305) and a damping ratio of 0.05 is selected.
Because of the improved convergence properties of this method, fewer eigenmodes are required from
the eigensolution. The dynamic response of the structure can be divided into two parts: contributions
from the lower modes, and contributions due to the higher modes, which can be expressed as a com-
bination of residual vectors.
The response power spectral density (RPSD) 1-σ displacement solutions are obtained with and without
the residual vectors. The residual vectors are calculated in the modal analysis, and subsequently used
in the PSD analysis with the same procedure. The sample input below shows how residual vectors are
used in a PSD analysis:
/SOLU
ANTYPE,SPECTR ! Perform a spectral analysis
RESVEC, ON ! Include the residual vectors
ACEL,0,0,0 ! Initialize the linear acceleration of the ref frame to zero
SPOPT,PSD ! Use power spectral density (PSD) as the spectrum type.
! Rest of the code is same as in the previous section
Full model modal analysis is carried out with and without storing the element results for the PSD ana-
lysis, and their efficiencies are compared.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
306 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The figure below shows the elapsed and CPU times with and without the residual vector method and
the expansion of the modes. Significant reduction in the solution time is achieved with the use of
MSUPkey = YES.
The solution time is slightly increased with the residual vector. For 50 modes, the elapsed time with
and without the residual vector are 143.00 and 137.42 seconds, respectively.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 307
Dynamic Simulation of a Printed Circuit Board Assembly Using Modal Analysis
Methods
Figure 20.6: Mode Shape of the Residual Vector
The residual vector method uses an additional vector computed in the modal analysis pass in addition
to the eigenvectors in the modal transformation. This increases the accuracy of the response solution.
RPSD displacement solutions with and without the residual vector are listed in the table below. For 50
modes without the residual vector, accuracy of the results is poor, as shown in column A in the following
table. Accuracy is significantly improved with the use of the residual vector, as shown in column B.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
308 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 309
Dynamic Simulation of a Printed Circuit Board Assembly Using Modal Analysis
Methods
The following figure shows a subset of these values. The figure also contains RPSD values for 300 modes
with the residual vector. As the frequency of this residual vector (1489 Hz) falls outside the range of
the input frequency, there is hardly any change in result the for 300 modes. Hence, this is taken as the
baseline for the analysis. Generally residual vectors should be used to increase accuracy and reduce the
number of modes whose frequency is most likely to fall within the input frequency range for spectral
analysis.
Figure 20.7: Improved Accuracy of RPSD Values with Residual Vectors for 50 Modes
20.8. Recommendations
To perform a similar type of analysis, consider the following hints and recommendations:
• If a subsequent MSUP analysis will be performed (for example, a modal superposition transient, harmonic,
spectrum or PSD analysis after modal analysis), write the element result to the mode file for use in the ex-
pansion pass in the subsequent analysis (MXPAND,,,,,,YES).
• The Residual Vector Method is recommended in the subsequent MSUP analysis if the excitation is known
prior to a modal analysis. The residual vector may be insignificant if sufficient modes are extracted.
20.9. References
The following references are used in this example problem:
1. Bo Zhang, Han Ding, XinJun Sheng. "Modal analysis of board-level electronic package." Microelectronic
Engineering. 85: 610-620. 2008.
2. Ying-Chih Lee, Bor-Tsuen Wang, Yi-Shao Lai, Chang-Lin Yeh, Rong-Sheng Chen. "Finite element model
verification for packaged printed circuit board by experimental modal analysis." Microelectronics Reliability.
48: 1837-1846. 2008.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
310 of ANSYS, Inc. and its subsidiaries and affiliates.
Input Files
• Electronics_PCB.cdb -- PCB model information for a single layer (called by other .dat files). The .dat
files generate the 3 three-layered PCB structure.
• psd_full_with_mxpand.dat -- Input file for the PSD analysis with the full model (300 modes) using
MSUPkey = YES in MXPAND.
• psd_full_No_mxpand.dat -- Input file for the PSD analysis with the full model (300 modes) using
MSUPkey = NO in MXPAND .
• psd_reduced_with_resvec.dat -- Input file for the PSD analysis using reduced model (50 modes)
with residual vector and MSUPkey = YES in MXPAND.
• psd_reduced_No_resvec.dat -- Input file for the PSD analysis using reduced model (50 modes) with
no residual vector and MSUPkey = YES in MXPAND.
• psd_full_with_resvec.dat -- Input file for the PSD analysis using full model (300 modes) with residual
vector and MSUPkey = YES in the MXPAND command.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 311
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
312 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 21: Buckling and Post-Buckling Analysis of a Ring-Stiffened
Cylinder Using Nonlinear Stabilization
This example problem is a nonlinear buckling and post-buckling analysis using nonlinear stabilization.
The problem uses a stiffened cylinder subjected to uniform external pressure to show how to find the
nonlinear buckling loads, achieve convergence at the post-buckling stage, and interpret the results.
21.1. Introduction
Buckling analysis is crucial to successful structure design and simulation, especially when thin structures
such as shells and beams are involved. While linear buckling analysis is comparatively straightforward,
it is limited by approximations and cannot simulate post-buckling phenomena. Nonlinear buckling
analysis does not have these limitations and is therefore preferred, even if it is a little more complicated
and requires some trial-and-error experimentation.
By analogy, it is also difficult in the physical world to determine the initiation of buckling. “From a sci-
entific and engineering point of view, the interesting phases of buckling phenomena generally occur
before the deformations are very large when, to the unaided eye, the structure appears to be undeformed
or only slightly deformed."[1] To perform a nonlinear buckling analysis, special nonlinear analysis tech-
niques are necessary to overcome the convergence difficulties, and a few trials are usually needed.
The following techniques are available for solving instability or buckling problems:
• Nonlinear stabilization
This capability deals with both local and global instabilities of buckling and can be used with any
other nonlinear technique except the arc-length method.
• Arc-length method
This method deals only with global instability or buckling when forces are applied, and can simulate
the negative slope region of a load-displacement curve.
This technique uses the dynamic effect to prevent divergence, but can be difficult to use.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 313
Buckling and Post-Buckling Analysis of a Ring-Stiffened Cylinder Using Nonlinear
Stabilization
For more information, see Unstable Structures in the Structural Analysis Guide.
This example uses a ring-stiffened cylinder under external hydrostatic pressure to demonstrate how to
predict buckling loads and simulate post-buckling phenomena with the aid of nonlinear stabilization.
The numerical simulation results are compared to the reference experimental results.[2]
The cylinder is subjected to a differential external pressure. The pressure causes a local buckling phe-
nomenon characterized by buckling of the skin between stiffening rings, leading eventually to collapse.
The buckling pressure, buckling and collapse pattern, number of waves into which the cylinder buckled,
and load-displacement curve are examined and compared to the reference results (p. 334).
The following table shows the dimensions of the cylinder, ring spacing, and the thicknesses of both
cylinder and Z-section rings. The dimensions chosen are based on the reference model.
21.3. Modeling
The finite element model of the ring stiffened cylinder is meshed with 26,796 SHELL281 elements with
an element size of 10 mm. The fine mesh is required for buckling analysis, and a full 360-degree model
is necessary because the deformation is no longer axisymmetric after buckling occurs.
All shell elements have uniform thickness. Five sections are created in the model with no offsets, so the
shell sections are offset to the midplane by default.
The following figure shows the mesh (a) and the expanded solid representation (b) of the shell mesh
(/ESHAPE):
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
314 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
a)
Meshed shell model (/ESHAPE,0)
b)
Expanded solid representation of the shell mesh (/ESHAPE,1)
This figure shows a cylinder sector of 45 degrees of the cylinder, plates and Z rings:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 315
Buckling and Post-Buckling Analysis of a Ring-Stiffened Cylinder Using Nonlinear
Stabilization
Figure 21.2: Inside View of a 45-Degree Sector
This figure shows the stiffening ring detail and its dimensions:
The plastic part of the material is modeled using the von Mises yield criteria coupled with an isotropic
hardening (TB,BISO), described by a bilinear stress-strain curve with the following constants:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
316 of ANSYS, Inc. and its subsidiaries and affiliates.
Analysis and Solution Controls
The tangent modulus is assumed, as no data is provided for this property in the reference information.
Because the reference [2] does not provide all necessary material data, some of the material data are
obtained from other publications detailing the properties of the same material.
For this numerical analysis, the only required boundary conditions are those to prevent the six rigid
body motions. A total of six displacements are therefore applied to three nodes located on the top
plate at 0, 90, and 270 degrees; the nodes are restricted so that all rigid translations and rotations are
not possible for the cylinder.
Loading consists of a uniformly distributed external differential pressure: Pext = 0.24 MPa
To run the linear buckling analysis, a static solution with prestress effects must be obtained, followed
by the eigenvalue buckling solution using the Block Lanczos method and mode expansion, as shown
in the following example input:
/solu
pstres,on ! Activate the prestress option
solve
finish
/solu
outres,all,all
antype,buckle ! Buckling analysis
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 317
Buckling and Post-Buckling Analysis of a Ring-Stiffened Cylinder Using Nonlinear
Stabilization
bucopt,lanb,10, ! Mode extraction is Block Lanczos
mxpand,10 ! Expand mode shapes
solve
finish
In this problem, the geometry, elements, and pressure are all axisymmetric. It is not possible, therefore,
to simulate nonaxisymmetric buckling with the initial model. To overcome this problem, small geometric
imperfections (similar to those caused by manufacturing a real structure) must be introduced to trigger
the buckling responses.
Although small perturbation loads can also be introduced to serve the same purpose, it is not an ideal
method because it is difficult to determine how large the loads should be and where to apply them.
Also, a perturbation load that is too large can change the problem completely.
The geometric imperfections can be generated either in the shape of the buckling mode or in a given
shape with a random amplitude:
• The imperfections in the buckling mode shapes are obtained by running a preliminary linear buckling
analysis (p. 317), then updating the geometry of the finite-element model to the deformed configuration.
This technique is done by adding the displacements of the mode shapes reduced by a scaling factor.
It is safe to add a few mode shapes to avoid any bias in the imperfections.
• Introduce pseudo-random imperfections by modifying the coordinates of the nodes with a random
number (using the RAND parametric function).
The imperfection magnitudes are generally dependent on the geometry and should be in the same
range as the manufacturing tolerance (typically less than one percent) so that they do not change the
problem during analysis.
For this problem, the imperfections are added as a sum of the first 10 modes shapes extracted in the
preliminary buckling analysis. (While random imperfections can also be used, their disadvantage is that
they cannot be repeated; therefore, the results would differ each time the analysis is run.)
Because the radius of the cylinder is 355.69 mm and the maximum displacement of a mode shape is 1
mm, a factor of 0.1 is applied when updating the geometry with mode shapes. The factor assumes the
manufacturing tolerance of the radius to be on the order of 0.1. The following input example shows
how the imperfections are added:
/prep7
*do,i,1,10
upgeom,0.1,1,i,buckling,rst ! Add imperfections as a tenth of each mode shape
*enddo
finish
To perform the analysis, the load must be allowed to increase using very small time increments so that
the expected critical buckling load can be predicted accurately.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
318 of ANSYS, Inc. and its subsidiaries and affiliates.
Analysis and Solution Controls
Some convergence difficulty at or after buckling is expected, so is useful to have all restart files available
for resuming the analysis with nonlinear stabilization. It is best to have the restart files at each substep
(RESCONTROL,DEFINE,ALL,1)--or at least for every few substeps--as it is virtually impossible to know
for certain when the buckling is initiated and therefore when to activate the stabilization via the restart.
Buckling initiation is difficult to detect by visual inspection but can be observed by plotting a load-dis-
placement curve or by a monitor file inspection. To detect the moment of buckling initiation, carefully
study the monitor file at this stage, as shown in this figure:
From the monitor file, several observations can be made to help determine if buckling has started to
occur and at what time, as follows:
• Difficulties in convergence occur. The program bisects the load-step increment and attempts a new
solution at a smaller load
• The maximum displacement monitored has an instantaneous change in value. This is a good indicator
of large displacement for a smaller load increment specific to buckling.
• The maximum displacement monitored has an instantaneous change in sign. This is another good in-
dication that buckling has begun to occur.
In this example, the change in time (or load) increment, and displacement value, occurs between substeps
10 and 11, which corresponds to TIME = 0.51781 and TIME = 0.53806 and to a pressure between 0.124
MPa and 0.129 MPa. It is therefore very likely that buckling occurred at this time; to be sure, the analysis
is continued. The goal is to verify the assessment made at this stage by obtaining the load-displacement
behavior over a larger range.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 319
Buckling and Post-Buckling Analysis of a Ring-Stiffened Cylinder Using Nonlinear
Stabilization
If the convergence difficulty is caused by buckling, resuming the analysis means starting a post-buckling
analysis. Because the post-buckling state is unstable, special techniques are necessary to compensate.
In a static analysis, nonlinear stabilization is the best option. When local buckling or time-dependent
material exists, it is the only option.
A substep experiencing convergence difficulty in the initial run (13 in this case) should generally be
avoided as the substep from which to restart; therefore, to continue with a post-buckling analysis, sta-
bilization is activated from substep 10. If convergence occurs at this substep, and assuming minimal
stabilization energy, this solution can be accepted.
Selecting which method to use, and then choosing the right value for the energy or damping factor,
are not trivial decisions. The best option and value to use depends on the type of instability, the type
and size of the elements used in the model, and the end time and number of substeps used for the
load step. In most cases, the decision is based on a trial-and-error correction process. The goal is to
achieve convergence with the smallest stabilization force, which can be controlled in turn by either an
energy ratio or damping factor.
For more information, see Using Nonlinear Stabilization in the Structural Analysis Guide.
The following topics describe the nonlinear stabilization methods tried in the ring-stiffened cylinder
problem:
21.6.3.1. Damping Method
21.6.3.2. Energy Method
• It has a specific range of values that can be used (between 0 and 1), and
• After running, it calculates a damping factor at the beginning of the next substep, providing a reference
value for the damping factors that can be used (if damping is chosen as the stabilization method).
The first energy value tried provided a damping factor of 0.1e-2, which helped convergence but without
producing significant buckling. The damping values were then gradually reduced to 0.1e-5 and, even
though in each case convergence was achieved, the collapse due to buckling did not occur. A value of
0.1e-6 did not result in convergence and the trials for applying the damping factor method ceased.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
320 of ANSYS, Inc. and its subsidiaries and affiliates.
Analysis and Solution Controls
The observed is due to the localized phenomena inherent in this problem. When damping is applied,
the specified value is used for all elements. When the applied damping value is too large, too much
stabilization force is applied to the structure, so the system is too stiff and converges easily without
much deformation. Conversely, when the applied damping is too small, the unstable elements do not
benefit from sufficient stabilization forces, so the solution diverges.
The conclusion is that the damping option is not ideal for problems involving significant local buckling.
After determining that this problem should use stabilization via the energy method, several trials were
performed using energy ratios in a range from 0.001 to 0.0001. The trials showed that larger energy
ratios result in convergence, but with no significant deformation and excessive stabilization energy,
and smaller energy ratios result in no convergence.
The smallest stabilization energy ratio to offer convergence was found to be 0.000143. With that value,
the analysis converges and the full loading is reached, while significant buckling is observed and global
stabilization energy is kept at the lowest possible level.
The following figure shows a time-history plot of the stiffness and stabilization energies:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 321
Buckling and Post-Buckling Analysis of a Ring-Stiffened Cylinder Using Nonlinear
Stabilization
Figure 21.5: Time-History Plot of Stiffness Energy (SENE) and Stabilization Energy (STEN)
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
322 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Figure 21.6: Load Factor and Mode Shape for the First Eigenvalue
The number of waves is more than the number obtained in the reference experiments, predicted by
theory, or obtained by the current nonlinear buckling analysis together with post-buckling analysis.
All extracted eigenvalues cluster around a critical value of 0.63, which is equivalent to a buckling pressure
of 0.15 MPa. This behavior is expected and is characteristic of a highly symmetrical system such as a
ring-stiffened cylinder under hydrostatic pressure. As is the case in this problem, a completely symmet-
rical system is known to be very sensitive to imperfections because many different modes are associated
with the same eigenvalue or closely spaced eigenvalues.[1]
Generally, the results from the nonlinear buckling analysis with stabilization agree with the reference
results, as follows:
21.7.1. Buckling Behavior Is Expected
21.7.2. Number of Buckles Decreases
21.7.3. Buckle Size and Behavior Is Consistent with Reference Results
21.7.4. Plasticity Affects Buckling Results Insignificantly
The results can be verified from the plots of applied pressure versus axial shortening for a node located
under the first ring. The following figure shows (a) the applied load versus axial shortening for one
node, and (b) the axial shortening with time for the same node but using a shorter range of strain (up
to 0.002):
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 323
Buckling and Post-Buckling Analysis of a Ring-Stiffened Cylinder Using Nonlinear
Stabilization
Figure 21.7: Applied Pressure and Axial Shortening
a)
Applied pressure vs. axial shortening
b)
Time vs. axial shortening
The shorter range of strain used in plot (b) in the figure above allows changes in the slope of the curve
to be easily observed and buckling time (pressure) to be easily predicted. Buckling occurs when the
slope of load-displacement curve (the pressure-axial shortening curve in this case) becomes zero or
reduces suddenly and significantly.
The displacement is almost constant between times 0.5178 and 0.559, where the load-vs.-displacement
slope is zero or slightly negative. As the load approaches the maximum value, the rate of deformation
increases very quickly until the cylinder fails.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
324 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
From TIME = 0.98 to TIME = 1, the total deformation increases from 11.2 to 47.7 mm, at which point
the rings and the skin between rings fails. The curves agree closely with the experimental results.
Note
Some material properties are not reported in the reference, and the cylinder used to obtain
the reference results may have had defects in addition to the geometric imperfections. Any
additional imperfections or differences in material data would of course affect the comparison
to the experimental results.[2]
The following figure offers a much better view of the buckling phenomenon, given by the applied
pressure vs. radial displacement curve for the node with maximum deformation:
A complex buckling mechanism is evident. Buckling clearly begins to occur at approximately 0.129 MPa,
when the displacement ceases to be linear and the slope inverses its direction. The displacement con-
tinues with changes in the slope sign (from negative to positive), corresponding to inward and outward
displacements of the buckles up to 0.22 MPa. At this point, the displacement increases dramatically and
leads to failure.
The following figure shows the total deformation at TIME = 0.5178 (buckling pressure 0.124 MPa) when
buckling is initiated; the deformation is hardly detectable by visual inspection, yet the buckling has
already occurred:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 325
Buckling and Post-Buckling Analysis of a Ring-Stiffened Cylinder Using Nonlinear
Stabilization
Figure 21.9: Total Deformation at the Time of Buckling Initiation
The following figure shows the deformation at TIME = 0.559 (pressure 0.134 MPa), corresponding to
the zero-slope plateau on the applied pressure-axial shortening curve (p. 324). The results are comparable
to the reference data and observations.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
326 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 327
Buckling and Post-Buckling Analysis of a Ring-Stiffened Cylinder Using Nonlinear
Stabilization
The number of buckles (20) is consistent with theory (20) and close to the reference results (19).
The size of the buckles between the first and second ring and the third and fourth ring is 140 mm,
which is in good agreement with the corresponding buckle size of 140 mm from the reference results.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
328 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The buckles formed on one side of the cylinder before the other side, which is consistent with the ref-
erence observations.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 329
Buckling and Post-Buckling Analysis of a Ring-Stiffened Cylinder Using Nonlinear
Stabilization
Figure 21.12: Another Total Deformation View
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
330 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Only localized areas of high stresses exist at the failure locations between the rings, as shown in this
figure:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 331
Buckling and Post-Buckling Analysis of a Ring-Stiffened Cylinder Using Nonlinear
Stabilization
Figure 21.14: Von Mises Stress Distribution at the End of the Analysis
21.8. Recommendations
When setting up a similar nonlinear buckling and post-buckling analysis, consider the following hints
and recommendations:
21.8.1. Buckling Analysis Guidelines
21.8.2. Nonlinear Stabilization Guidelines
A nonlinear buckling or post-buckling analysis is not a single-run analysis but a set of analyses. Some
advanced nonlinear techniques such as stabilization are necessary in the static analysis which may
require trial-and-error experimentation.
It is best to start with a linear buckling analysis to verify the correctness and completeness of the
model, and to get a general idea of how large the buckling load may be. In the linear analysis, the
mode shapes can be expanded to predict the buckled mode shapes (used for introducing geometric
imperfections later).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
332 of ANSYS, Inc. and its subsidiaries and affiliates.
Recommendations
Geometric imperfections introduced with modes from the linear buckling analysis are recommended.
The magnitude of imperfection should be within the range of manufacturing tolerances.
When nonlinear analysis stops due to convergence difficulty, the monitor file should be inspected
carefully. Many bisections in the substeps indicate instability, and a significant change in the displace-
ment values could mean that buckling is starting to occur.
To verify the nonlinear buckling load and to perform post-buckling analysis, the analysis should be
restarted using some advanced nonlinear technique to examine the load-displacement behavior for
a larger range of loads. In a static analysis, nonlinear stabilization is preferred and is the only option
if local buckling or time-dependent materials exist.
• Determine the optimal displacement to find the buckling load on the load-displacement curve.
The buckling load is the load at which the slope of the load-displacement curve reduces significantly.
For which node and for which direction the displacement should be chosen depends upon the
problem being solved, and a few trials may be necessary to find the optimal displacement. Some
displacements show the start of buckling more clearly than others.
The buckling type helps to determine which stabilization method to use (damping or energy). For
local buckling, the energy method is the only option. For global buckling, either the energy or
damping method can be used.
Obtain the damping factor value and use it as a reference value if damping is the desired method
to control the stabilization.
Energy is the preferred method in the post-buckling analysis. Because the energy ratio ranges only
from 0 to 1, however, damping can be used to obtain very fine control of stabilization forces if neces-
sary. The damping factor can be any positive number, but should not stray too far from the reference
value.
When different energy ratios are tried, do not change the number of substeps or time, as this will
affect your comparison and therefore your decision as to which stabilization method you should use.
The specified energy ratio or damping factor should be large enough to circumvent divergence and
small enough to avoid excessive stiffness. The best solution should be the one converged using the
smallest energy ratio or damping factor.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 333
Buckling and Post-Buckling Analysis of a Ring-Stiffened Cylinder Using Nonlinear
Stabilization
• Activate stabilization via a multiframe restart.
Although you can activate nonlinear stabilization at the beginning of the solution, in most cases it
is more efficient and accurate to activate stabilization in a multiframe restart.
If you do so, save the restart files and restart from the substep before the last converged substep. If
the substep has many bisections or a large number of iterations, try one substep before, or the substep
at which monitored displacement shows significant change (as it is likely the point where buckling
starts).
Accuracy can be affected if the stabilization damping or energy values are excessive. The stabilization
energy should be much less than the strain energy. Results checking can be done in either the POST1
(/POST1) or POST26 (/POST26) postprocessor.
21.9. References
The following references are used in this example problem:
2. Dow, Donaldson A. "Buckling and Postbuckling Tests of Ring-Stiffened Circular Cylinders Loaded by Uniform
External Pressure." NASA Technical Note NASA TN D-3111. Langley Research Center: November, 1965.
• buckling.cdb -- The common database file containing the model information for this problem (called
by buckling.dat).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
334 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 22: Modal and Harmonic Frequency Analyses of an
Automotive Suspension Assembly Using CMS
This example problem uses component mode synthesis (CMS) to generate dynamic superelements for
use in downstream linear dynamic analyses. The example demonstrates how CMS technology can greatly
reduce the use of computational resources and maintain a level of accuracy similar to that of the full
model in modal and harmonic analyses.
22.1. Introduction
Automotive suspension systems contribute to a car's handling and braking to improve safety, and keep
vehicle occupants comfortably isolated from road noise, bumps, and vibrations. When a car moves on
uneven terrain, the wheels are subjected to base excitation. For analysis purposes, this can be approx-
imated as harmonic excitation. Because the suspension forms part of the chassis in a car, response of
the suspension system to harmonic displacement excitation is a critical subject for analysis.
Component mode synthesis (CMS) is a form of substructure-coupling analysis frequently used in struc-
tural dynamics. CMS enables the behavior of the entire assembly to be derived from its constituent
components. First, the dynamic behavior of each component is formulated. Then, the dynamic charac-
teristics of the full system model are formed by enforcing equilibrium and compatibility along component
interfaces. CMS is more accurate than the Guyan reduction for modal, harmonic, and transient analyses
because CMS includes truncated sets of normal mode generalized coordinates, which capture the basic
modal characteristics of the components.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 335
Modal and Harmonic Frequency Analyses of an Automotive Suspension Assembly
Using CMS
22.2. Problem Description
The model used in this analysis is a slightly asymmetric automotive suspension assembly (p. 336). The
analysis calculates the response of two points attached to a link on the body due to a harmonic (sinus-
oidally varying) displacement excitation at the bottom of the wheels. The displacements of the response
points are determined for a specified frequency range. Modal and harmonic analyses are performed.
Results using the CMS generated models are compared to results for a full (non-substructure) model.
22.3. Modeling
The assembly shown in Figure 22.1: 3-D Suspension Model (p. 336) consists of two wheels and a body.
A wheel consists of 11 parts, and the wheels are attached to the body via shafts. The body consists of
53 parts, and the dimension of the overall bounding box is 0.79594 × 0.79324 × 2.5252 (in meters).
The model uses SOLID186, SOLID187, and BEAM188 elements. SOLID186 is a 3-D 20-node solid element
exhibiting quadratic displacement behavior suitable for modeling homogeneous structural solids. SOL-
ID187 is a 3-D 10-node element that exhibits quadratic displacement behavior, but is suitable for
modeling irregular meshes. BEAM188 is a 3-D 2-node linear element used for linear elements to create
general welds between different parts in contact. Bonded contact is assumed. This contact prevents
relative displacement between the different parts of the assembly.
The parts of the model are made of structural steel and have a combined mass of 1250.1 kg.
Tetrahedral and hexahedral dominant meshes are used to mesh the assembly. The model has 305,968
nodes and 165,901 elements. Figure 22.2: Meshed Suspension Model (p. 337) shows the meshed model
of the assembly.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
336 of ANSYS, Inc. and its subsidiaries and affiliates.
Boundary Conditions and Loading
For the harmonic analysis, a constant damping ratio of 5 percent is used to prevent possible large-re-
sponse amplitudes.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 337
Modal and Harmonic Frequency Analyses of an Automotive Suspension Assembly
Using CMS
Figure 22.3: Suspension Assembly with Boundary Conditions and Displacement Loading
The following example input shows how the boundary conditions are applied in the generation pass
of the CMS model:
/FILNAME, BODY
/SOLU
/COM, SELECT A SUBSET OF THESE NODES
CMSEL, S, BOLT1 ! Select a new subset of components named Bolt1
CMSEL, A, BOLT2 ! Additionally select a set Bolt2 and extend the current set
CMSEL, A, BOLT3 ! Additionally select a set Bolt3 and extend the current set
CMSEL, A, BOLT4 ! Additionally select a set Bolt4 and extend the current set
CMSEL, A, BOLT5 ! Additionally select a set Bolt5 and extend the current set
CMSEL, A, BOLT6 ! Additionally select a set Bolt6 and extend the current set
ESLN ! Select the elements attached to the selected nodes
NSLE ! Select the nodes attached to the selected elements
D, ALL, ALL ! Constrain the nodes for all degrees of freedom
The loading is applied in the form of harmonic displacement in the vertical y direction on a selected
set of nodes on boththe wheels. The sample input shows how the displacement loading is applied in
the full harmonic analysis:
/COM, NODES AT THE BOTTOM OF THE FIRST WHEEL FOR EXCITATION
N1=NODE(-0.13838, -0.28280, 1.0714) ! Node #191882
N2=NODE(-0.13733, -0.28293, 1.0501) ! Node #192365
N3=NODE(-0.13786, -0.28287, 1.0608) ! Node #210243
/SOLU
ANTYPE, HARMONIC ! Perform a harmonic analysis
HROPT, FULL ! Employ the full method
KBC, 1 ! Use the same value of the load for each substeps
DMPRAT, 0.05 ! Set a constant damping ratio of 5%
HARFRQ, 160, 200 ! Choose the frequency range between 160 and 200 Hz
NSUBST, 20 ! Number of substeps = 20
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
338 of ANSYS, Inc. and its subsidiaries and affiliates.
Analysis and Solution Controls
NSEL, A, NODE, , N2 ! Additionally select the node n2 and extend the current set
NSEL, A, NODE, , N3 ! Additionally select the node n3 and extend the current set
NSEL, A, NODE, , N4 ! Additionally select the node n4 and extend the current set
NSEL, A, NODE, , N5 ! Additionally select the node n5 and extend the current set
NSEL, A, NODE, , N6 ! Additionally select the node n6 and extend the current set
D, ALL, UY, 100 ! Define a displacement degree of freedom on these nodes in the y
! direction. This harmonic displacement amplitude = 100.
NSEL, ALL ! Restore the full set of nodes before solving
SOLVE ! Solve the full model for harmonic analysis
FINISH
22.6.1. Substructure
Substructuring is a procedure that condenses a group of finite elements into one element represented
as a matrix. The single-matrix element is called a superelement. The substructure analysis uses the
technique of matrix reduction to reduce the system matrices to a smaller set of degrees of freedom
(DOFs). The use pass of a substructure reduces the computer time and enables the solution of very
large problems with limited computer resources. The following figure shows how the suspension assembly
is divided into three superelements: Wheel1, Wheel2, and Body.
A substructure analysis involves three distinct steps called passes: the generation pass, the use pass,
and the expansion pass.
1. Generation pass
The objective of the generation pass is to condense a group of standard finite elements into a single
superelement. The elements are condensed by identifying a set of master degrees of freedom
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 339
Modal and Harmonic Frequency Analyses of an Automotive Suspension Assembly
Using CMS
(MDOFs) known as master nodes. Master nodes are used to define the interface between the super-
element and the other elements and to capture the characteristics for the dynamic analyses.
Proper choice of the interface master nodes is important to maintain continuity among the supere-
lements and the other parts of the model. To optimize solution time, select the interface with the
fewest nodes that still maintains continuity. For a 3-D model, select the interface at a region of
minimum cross-sectional area. For a 2-D model, you would select the interface at a region of minimum
length.
As shown in the following figure, the interface master nodes are defined on the shafts connecting
the wheels and the body because the cross-sectional areas of the shafts are minimal compared to
the other parts.
Figure 22.5: Interface Master Nodes Between the Superelements of the Suspension Assembly
The Master nodes are also defined where boundary conditions or constraint equations are applied,
as shown in the following figure:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
340 of ANSYS, Inc. and its subsidiaries and affiliates.
Analysis and Solution Controls
For the harmonic analysis, the points of load application are also defined as master nodes, as shown
in this figure:
In the use pass (described below), the master nodes are the only nodes available for a superelement.
The interface master nodes associated with Wheel1 and Body are named component “master1.”
Master nodes associated with Wheel2 and Body are named component “master2”. In the substructure
analysis, the superelement Wheel1 consists of the first wheel and the attached shafts up to “master1”
and is termed Wheel1_for_solve. Likewise, the superelement Wheel2 consists of the second wheel
and the attached shafts up to “master2” and is termed Wheel2_for_solve.
The following example input shows how the superelement of the first wheel is created after defining
the master nodes:
/FILNAME, FIRSTWHEEL ! Name of the Superelement is FirstWheel
/SOLU
CMSEL, S, MASTER1, NODE ! Select a new set of interface nodal components named master1
M, ALL, ALL ! Define them as master nodes
/COM, SELECT A SET OF NODES FOR POINT OF APPLICATION OF HARMONIC DISPLACEMENT ON THE FIRST WHEEL
NSEL, S, NODE, , N1 ! Select the node n1 as a new subset
NSEL, A, NODE, , N2 ! Additionally select the node n2 and extend the current set
NSEL, A, NODE, , N3 ! Additionally select the node n3 and extend the current set
M, ALL, ALL ! Define them as master nodes
ALLSEL, ALL
CMSEL, ALL
2. Use pass
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 341
Modal and Harmonic Frequency Analyses of an Automotive Suspension Assembly
Using CMS
The superelement is used to make it a part of the whole model. The entire model may be a supere-
lement, or the superelement may be connected to non-superelements.
The solution from the use pass consists of only the reduced solution for the superelement, which
is the DOF solution only at the MDOF, and the complete solution for non-superelements.
3. Expansion pass
The results at all DOFs in the superelement are calculated at the start of the expansion pass. If
multiple superelements are used in the use pass, a separate expansion pass will be required for each
superelement.
The reduced solution obtained from the use pass is applied to the model as displacement boundary
conditions, and the complete solution within the superelement is solved.
The expansion pass logic for substructuring analysis searches for the superelement .LN22 file and,
if found, uses the sparse solver to perform a back-substitution.
You can specify a load step and a substep for expanding a particular frequency (EXPSOL).
The generated substructure information in CMS is in the .sub file, which is all that is required in the
use pass. Because the internal details of the structure are not exposed using CMS, specialized teams
can work on the same structure without having to provide detailed or proprietary information about
the component. The superelements created in the generation pass are combined in the use pass with
knowledge of master nodes. The superelements are given a new element type: MATRIX50.
The following sample input shows how the superelements are combined in the use pass.
/FILNAME, USE
/PREP7
*GET, MAX_ETYPE, ETYP, , NUM, MAX ! Retrieve the value of the maximum element type number
! and store it as MAX_ETYPE
ET, MAX_ETYPE+1, 50 ! Define Matrix50 as the next element type number
TYPE, MAX_ETYPE+1 ! Assign this type number to the elements
MAT, 1 ! Assign 1 as the material number to subsequently defined
! elements
REAL, MAX_ETYPE+1 ! Set the element real constant set attribute pointer
SE, FIRSTWHEEL ! Define the superelement associated with the Wheel1
SE, BODY ! Define the superelement associated with the Body
SE, SECONDWHEEL ! Define the superelement associated with the Wheel2
FINISH
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
342 of ANSYS, Inc. and its subsidiaries and affiliates.
Analysis and Solution Controls
Figure 22.8: Superelement of the Suspension Model After the Use Pass
A CMS substructure can be advantageous over a basic substructure because it is more accurate than a
Guyan reduction for model, harmonic, and transient analyses. CMS includes truncated sets of normal
modal generalized coordinates that characterize the behavior of the components of the structural
model.
A typical use of CMS involves modal analysis of a large, complicated structure (such as an aircraft or
nuclear reactor) where various teams are in charge of the design of a component of the structure. With
CMS, design changes to a single component affect only that component; therefore, additional generation
passes are necessary only for the modified substructure.
• Fixed-interface (CMSOPT,FIX), where interface nodes are constrained during the CMS superelement gener-
ation pass.
• Free-interface (CMSOPT,FREE), where interface nodes remain free during the CMS superelement generation
pass.
• Residual-flexible free-interface (CMSOPT,RFFB), where interface nodes remain free during the CMS supere-
lement generation pass.
The fixed-interface CMS method is preferable for most analyses. The free-interface method and the re-
sidual-flexible free-interface method are useful when an analysis requires more accurate eigenvalue
computation at the mid to high end of the spectrum. As in substructuring, a component mode synthesis
analysis involves three distinct steps or passes. In CMS generation pass, a group of finite elements are
condensed into a single CMS superelement that includes a set of master degrees of freedom (MDOFs)
and truncated sets of normal mode generalized coordinates. The MDOFs serve to define the interface
between the superelements and the other elements. The CMS use pass and expansion pass utilize the
same procedure as a basic substructure analysis.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 343
Modal and Harmonic Frequency Analyses of an Automotive Suspension Assembly
Using CMS
As multiple superelements were used for the present model in the use pass, a separate expansion pass
would usually be required for each of them. However, since the response nodes belong to the supere-
lement Body, expansion of this superelement is sufficient.
Modal analysis is a linear analysis aimed at finding the eigensolution; therefore, no force is involved in
the modal analysis. Nonlinearities, such as plasticity and contact (gap) elements, are ignored even if
they are defined. In this model, the Block Lanczos method is used to extract the modes with the
boundary conditions shown in Figure 22.6: Master Nodes Defined at Fixed Bolts (p. 340).
This analysis technique calculates only the steady-state, forced vibrations of a structure. The transient
vibrations, which occur at the beginning of the excitation, are not accounted for in a harmonic analysis.
Harmonic analysis is typically linear. Some nonlinearities, such as plasticity are ignored, even if they are
defined; however, unsymmetric system matrices (such as those encountered in a fluid-structure interaction
problem ) can be accommodated in the harmonic analysis.
In the present model, base excitation is applied in the form of sinusoidal displacement in the vertical
y-direction at selected nodes on both wheels. Those nodes are defined as master nodes in the generation
pass (as shown in Figure 22.3: Suspension Assembly with Boundary Conditions and Displacement
Loading (p. 338) and Figure 22.7: Master Nodes Defined at Points of Application of Harmonic Displace-
ment (p. 341)).
The response is calculated at the response nodes specified during the expansion pass, as shown in this
figure:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
344 of ANSYS, Inc. and its subsidiaries and affiliates.
Analysis and Solution Controls
Figure 22.9: Non-Master Nodes for Response Calculation in the Expansion Pass
As the response nodes belong to only one of the superelements (Body), the results file of Body is used
for the response calculation. The following example input performs the expansion pass with postpro-
cessing:
/CLEAR, NOSTART
/FILNAME, BODY ! Change the jobname to Body
RESUME, , ! Resume the database file Body.db
/SOLU
EXPASS, ON ! An expansion pass will be performed
OUTRES, ALL, ALL ! Write the solution of the specified solution results item for
! every substep in the database
SEEXP, BODY, USE ! Specify the name of the dsub file as use, containing the
! superelement degree-of-freedom (DOF) solution
NUMEXP, ALL, , , ! Expand all substeps with element results
SOLVE
SAVE
FINISH
/post26
/SHOW, CLOSE
FINISH
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 345
Modal and Harmonic Frequency Analyses of an Automotive Suspension Assembly
Using CMS
22.7. Results and Discussion
Both the modal and harmonic analyses are performed using full and CMS models, and their solution
times are noted. Significant improvement in the solution time is observed for the CMS model with very
little loss of accuracy.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
346 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The following table shows the elapsed and CPU times for modal analysis using the full and CMS models
for 100 frequencies. Significant improvement in solution time is achieved during the use pass via the
CMS method.
Table 22.2: Comparison of CPU and Elapsed Times for Modal Analysis
Table 22.3: Comparison of Response Amplitudes for Full and CMS Models
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 347
Modal and Harmonic Frequency Analyses of an Automotive Suspension Assembly
Using CMS
Node # 63121 Node # 63134
Freq (Hz) Full CMS % Diff Freq (Hz) Full CMS % Diff
Model Fixed Model Fixed
Interface Interface
198 34.8474 34.7009 0.42 198 26.3318 25.9327 1.52
200 33.7753 33.7011 0.22 200 24.6321 24.3226 1.26
The following table shows the elapsed and CPU times for harmonic analysis using full and CMS models
for 20 substeps. The use pass includes the expansion. Significant improvement in solution time is
achieved via the CMS method.
Table 22.4: Comparison of CPU and Elapsed Times for Harmonic Analysis
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
348 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Figure 22.10: Nodal Responses Due to Harmonic Displacement Excitation at the Bottom of the
Wheels
The peaks are explained by observing the mode shape at the undamped natural frequency of 187.22
Hz (Figure 22.11: Mode Shapes at the Undamped Natural Frequency of 187.22 Hz (p. 349)). The figure
shows the configuration of the structure at two extreme deflections. The mode shape at this frequency
has tilting vibrations of the wheels (in phase) about the x axis with the associated deflection of part of
the body. Harmonic displacement at the bottom of the wheels in the Y-Direction excites this mode,
resulting in peaks at nearby frequencies.
Mode shapes for nearby higher frequencies (223.63 Hz and 237.48 Hz) do not involve significant tilting
of the wheels or deflection of the links attached to the response nodes. Figure 22.12: Mode Shapes at
the Undamped Natural Frequency of 223.63 Hz (p. 350) and Figure 22.13: Mode Shapes at the Undamped
Natural Frequency of 237.48 Hz (p. 350) show the mode shapes at these frequencies.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 349
Modal and Harmonic Frequency Analyses of an Automotive Suspension Assembly
Using CMS
Figure 22.12: Mode Shapes at the Undamped Natural Frequency of 223.63 Hz
22.8. Recommendations
To perform a similar type of analysis, keep the following hints in mind:
• The computational benefits gained from CMS analysis can be increased by reducing the number of interface
degrees of freedom defined. For maximum benefit, define as few as possible.
• CMS analysis can take advantage of repetitive geometry or parts to make the analysis more efficient
22.9. References
The following references are used in this example problem:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
350 of ANSYS, Inc. and its subsidiaries and affiliates.
Input Files
1. Craig, R.R. and M. D. D. Bampton.“Coupling of Substructures for Dynamic Analysis.” AIAA Journal. 6.7 (1968):
1313-1319.
2. Craig, R.R.“A Review of Time Domain and Frequency Domain Component Mode Synthesis Methods.” Inter-
national Journal of Analytical and Experimental Modal Analysis. 2.2 (1987): 59-67.
• CMSModel.dat -- Modal and harmonic analysis on the suspension assembly using CMS method.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 351
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
352 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 23: Modal Analysis of a Wind Turbine Blade Using Beam
Elements
This example problem demonstrates how to use current-technology beam elements to simulate the
behavior of a slender composite structure such as a wind turbine blade.
• The meshed beam-section subtype for conveniently representing the built-up sections with complex geometry
and nonuniform materials.
• The nonaveraged results file format for storing results from an analysis involving composite materials.
For related information, see Beam and Pipe Cross Sections in the Structural Analysis Guide.
23.1. Introduction
Wind turbines play an important role in harvesting alternative sources of energy. The turbine blade is
the critical component in a wind turbine. An optimal blade design is crucial to the ultimate efficiency
and strength of the turbine.
For their excellent formability and density to strength ratios, fiber-reinforced composite materials are
now extensively used in the blade construction. Compared to their conventional counterparts, composite
materials introduce a number of additional design parameters, such as the matrix and fiber material
properties, laminar thicknesses, and fiber orientations.
Each blade design must be carefully verified. For example, to avoid a catastrophic failure, a design must
not lead to a natural frequency close to any of the resonance frequencies. Without the aid of an effective
simulation tool for design verification, the design process of a composite blade can be excessively time-
consuming.
Because of the complex blade geometry, a typical strategy is to create a 3-D finite-element model of
the turbine blade using shell or solid elements. With a detailed 3-D model, both global and local
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 353
Modal Analysis of a Wind Turbine Blade Using Beam Elements
mechanical responses of the blade can be adequately predicted with shell or solid elements. The disad-
vantage of using shells or solids becomes apparent when design changes are necessary.
A small variation in the design can lead to partial or even complete reconstruction of the 3-D model.
Because 3-D models are generally difficult to build and modify, frequent modifications during the design
process may be impractical.
In some circumstances, especially during the preliminary design stage, only the global mechanical re-
sponses of the blade are sought. A simplified model, such as a 1-D beam model, is more desirable for
such situations. The current-technology beam elements, with their advanced beam-section modeling
capabilities, can be used to accurately predict natural frequencies of a typical composite turbine blade
with minimal modeling effort and computational costs.
In an actual application, all three components of the blade can be made using layered composites. For
illustrative purposes, this problem adopts a three-layered construction for the skin only, with homogen-
eous and orthotropic materials used for the spar caps and shear webs.
The blade geometry is defined via representative cross-sections at eight locations along the longitudinal
axis, as shown in this figure:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
354 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
The geometry is assumed to be linearly tapered between each pair of adjacent cross-sections. The
natural frequencies and modes of the blade are analyzed using three comparative models:
23.3. Modeling
The following models for the wind turbine blade are described:
23.3.1. Wind Turbine Blade: BEAM188 Models
23.3.2. Wind Turbine Blade: SHELL281 Reference Model
The root part is contained within the first two round sections (the area between sections #1 and #2).
The root part is made of the skin material only.
Two sections (sections #3 and #4) are defined at the same location. While section #4 and the remain-
ing sections consist of all three section components (skin, spar caps, and shear webs), section #3
contains only the skin component. Because sections #2 and #3 are topologically consistent, a
transition part is defined between sections #2 and #3, connecting the root part to the airfoil part.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 355
Modal Analysis of a Wind Turbine Blade Using Beam Elements
The section-area geometry is meshed using MESH200 elements in a quadrilateral shape with eight
nodes (KEYOPT(1) = 7), and appropriate material property types are assigned to different components
within the sections.
The following example input stores the mesh information for section #1:
CMSEL, s, area1 ! Select area component of first blade cross-section.
ALLSEL, below, area ! Select everything below the selected area
SECWRITE, mesh1,sect,,1 ! Output mesh to file "mesh1.sect"
3. Define constant beam sections types(BEAM), each with a meshed subtype (MESH).
Beam sections with a mesh subtype are defined (SECTYPE), and the section meshes from the prior
step are restored (SECREAD).
The following example input reads the mesh information contained in the mesh1.sect file and
associates it with beam section #1:
SECTYPE,1,beam,mesh ! Specify beam section 1 with mesh subtype
SECREAD,mesh1,sect,,mesh ! Import mesh from file “mesh1.sect”
Using the constant previously defined beam sections, define seven beam sections (SECTYPE) of
section type TAPER.
The sections are assigned later to beam elements in the various blade parts (root, transition, and
airfoil).
The following example input defines section #11 for the transition part:
SECTYPE,11,taper ! Specify tapered-beam section type
SECDATA,2,2.0 ! Put BEAM section 2 at global X location 2.0 as the starting section
SECDATA,3,4.0 ! Put BEAM section 3 at global X location 4.0 as the ending section
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
356 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
Line segments end where the constant beam sections are defined so that no elements are split by
the sections.
Two BEAM188 meshes are generated, one a fine mesh with the linear interpolation option (KEYOPT(3)
= 0), the other a coarse mesh with the cubic interpolation option (KEYOPT(3) = 3). The following
two figures show both models, respectively:
Figure 23.4: BEAM188 Blade Model with Fine Mesh and Linear Interpolation
Figure 23.5: BEAM188 Blade Model with Coarse Mesh and Cubic Interpolation
The elements are visualized with the expanded solid display option (/ESHAPE) to show the 3-D
section details.
The following example input meshes line #3, representing the transition part:
SECNUM,11 ! Set the current section as section 12
LMESH,3 ! Mesh line number 3
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 357
Modal Analysis of a Wind Turbine Blade Using Beam Elements
Material properties (p. 358) and boundary conditions (p. 358) in this model are identical to those of the
BEAM188 models.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
358 of ANSYS, Inc. and its subsidiaries and affiliates.
Boundary Conditions and Loading
Extensive constraint equations (CEs) are used for properly assembling the three structural components
(indicated by the green areas in the figure above).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 359
Modal Analysis of a Wind Turbine Blade Using Beam Elements
As a result of the simplified model, an abrupt section-property change exists from the three-component
airfoil section to the skin-only transition section. Significant section deformation is expected at this
connection location.
While the SHELL281 reference model can simulate the section deformation, the same capability in the
BEAM188 models is greatly limited due to basic beam characteristics. To create a comparable reference
model, an artificially rigid plane is inserted at this location, indicated by the red symbols in the figure
above. Similarly, another rigid plane is inserted at the connection of the root part and the transition
part.
This technique does not invalidate the beam modeling, as these types of sudden section-property
changes do not exist in actual blade designs.
The first five frequencies predicted by the three comparative models are consistent within a range of
five percent. The difference increases in higher modes.
The following figure shows the mode shapes of the first five matching frequencies:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
360 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Figure 23.9: Comparison of Mode Shapes Predicted by BEAM188 and SHELL281 Models
The current-technology beams are shown to be effective for simulating slender structures. Compared
to 3-D modeling using shell and solid elements, beam modeling greatly reduces model-creation time
and effort while generally leading to acceptable overall results.
The applicability of current-technology beam elements, however, is bound by the first-order shear-de-
formation (or Timoshenko) theory. Beam elements are therefore not suitable for extremely stubby
structures or structures that may undergo large section deformation (as shown in the higher modes
predicted in the SHELL281 model).
With the aid of the mesh (MESH) section subtype, current-technology beam elements can be used for
certain types of composite structures. Consider the applicability of beam elements carefully before ad-
opting them for composite-structure simulation. When modeled with beam elements, composite
structures introduce these restrictions:
• The material properties should not exhibit great variation within the section
• Because of beam-theory limitations, some material properties such as Poisson’s ratio, and the coupling
between axial stiffness and twisting/transverse shear stiffness, are ignored in beam elements.
The tapered-beam section type (TAPER) can be used with all standard (and even user-defined) beam
sections, making it a powerful and flexible tool for modeling structures with complex and varying cross-
section geometry.
The cubic interpolation option enables the BEAM188 model to generate an equally accurate solution
with far fewer elements than the BEAM188 model with the linear option. Cubic interpolation is also
required for modeling partially distributed or pointed loads within the element.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 361
Modal Analysis of a Wind Turbine Blade Using Beam Elements
The typical beam-bending stress results (Sxx) with different results-file formats (KEYOPT(15) in current-
technology beam and pipe elements) are shown in the following two figures:
With the averaged result format (KEYOPT(15) = 0), continuous bending stress is observed even across
the interfaces between two different materials.
For beams with built-up sections, the nonaveraged format (KEYOPT(15) = 1) is appropriate. As shown
in the following figure, the stress discontinuity in heterogeneous media is properly captured using the
nonaveraged format:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
362 of ANSYS, Inc. and its subsidiaries and affiliates.
Input Files
23.8. Recommendations
To perform a similar analysis for a slender composite structure, consider the following hints and recom-
mendations:
• When applicable, use current-technology beam elements (such as BEAM188) to achieve better computational
efficiency and to simplify model creation and modification. Verify that all modeling requirements, such as
slenderness ratio and material orientation, are satisfied.
• Use the meshed beam-section subtype (SECTYPE,,BEAM,MESH) and/or the tapered-beam section type
(SECTYPE,,TAPER) for modeling structures with complex section geometry and profiles.
• Select higher-order interpolation options in BEAM188 if nonuniform deformation along the length is expected.
Because higher-order interpolation options introduce invisible internal elements nodes, however, carefully
inspect boundary conditions and loading to avoid inconsistency.
• Select the nonaveraged results file format for proper postprocessing of the simulation results.
• blade_188_linear.inp -- Input file for the modal analyses of the wind turbine blade modeled with
linear BEAM188 elements.
• blade_188_linear.cdb -- Common database file containing the model data for the wind turbine blade
modeled with linear BEAM188 elements (called by blade_188_linear.inp).
• blade_188_cubic.inp -- Input file for the modal analyses of the wind turbine blade modeled with cubic
BEAM188 elements.
• blade_188_cubic.cdb -- Common database file containing the model data for the wind turbine blade
modeled with cubic BEAM188 elements (called by blade_188_cubic.inp).
• blade_281.inp -- Input file for the modal analyses of the wind turbine blade modeled with SHELL281
elements.
• blade_281.cdb -- Common database file containing the model data for the wind turbine blade modeled
with SHELL281 elements (called by blade_281.inp).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 363
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
364 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 24: Hydrostatic Fluid Analysis of an Inflating and Rolling
Tire
This example problem demonstrates how to model a fluid that is fully enclosed by a solid (the container).
The problem shows how loading on the container, and the resulting container deformation, affect the
pressure, volume, density and mass of the contained fluid.
• Reinforcing
A transient analysis of an inflated and rolling tire shows the tire deformation through various load steps.
24.1. Introduction
Hydrostatic fluid elements are well suited for calculating fluid volume and pressure for coupled problems
involving fluid-solid interaction. Modeling with a hydrostatic fluid allows one to study the changes in
fluid behavior when it is contained within a solid upon which various loads are imposed.
Such an analysis is useful in the example problem presented here, where it is possible to examine the
changing pressure, density and volume of the air inside of a tire during inflation and rolling. Another
example application involves the volume and pressure changes of a fluid contained within a cylinder
in which a piston is moving.
The automobile industry strives to improve fuel economy and to lessen power loss, both of which are
affected by the rolling resistance of a vehicle's tires. To do so, it is necessary to accurately predict the
changing fluid conditions within a rolling, deformed tire.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 365
Hydrostatic Fluid Analysis of an Inflating and Rolling Tire
with a hydrostatic fluid element, and the pressure, volume and density is monitored as loads are applied
to the tire.
The tire is inflated to 36 psi and a mass of 1 ton is added at the axle to simulate the portion of the
vehicle's mass resting on that axle.
1. Apply a gravity load and set the reference temperature for the air.
Load steps 1 through 4 are performed statically, while load step 5 is performed as a transient analysis
to study the loading effect on vertical acceleration.
24.3. Modeling
To simulate actual conditions, the tire dimensions approximate those of a P215/65R16 car/minivan tire.
An incompressible hyperelastic material model is used for the tire. Reinforcing elements are embedded
into the solid elements to model steel reinforcing within the tire structure.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
366 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
The following figure shows the fluid elements generated over the solid elements:
Because the fluid elements are pyramid shaped with common vertices at each pressure node ID, the
fluid elements cover some undesired volumes as well. So that fluid elements exist only in the region
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 367
Hydrostatic Fluid Analysis of an Inflating and Rolling Tire
where air should be present, fluid elements having a negative volume in the undesired region are used,
as shown in this figure:
This technique involves manually generating fluid elements with the nodes ordered in a clockwise dir-
ection.
The following input example shows how to model the air inside of the tire using hydrostatic fluid element
HSFLD242:
! Defining hydrostatic fluid element
Et,3,242 ! Hydrostatic fluid element
keyopt,3,5,1 ! Fluid mass calculated based on the volume
! of the fluid element
type,3
mat,2 ! Gas material model used to model the inside fluid
real,3
r,3,0.10156 ! Initial air pressure (atmospheric) = 0.10156 N/mm^2
n,1,0.,340.,0. ! Define pressure node
For more information about modeling, see Modeling Hydrostatic Fluids and Example Model Using
Hydrostatic Fluid Elements in the Structural Analysis Guide.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
368 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 369
Hydrostatic Fluid Analysis of an Inflating and Rolling Tire
The following input example shows how to model tire reinforcing using REINF265:
! Section definition for reinf1
sectype,13,reinf,smear,reinf13 ! Define section type and
! sub type(smear), 13 is section id number
secdata,4,4,2,,,ELEf,3,0.5 ! Define section data
! For more information, see SECDATA command documentation
!
vsel,s,volu,,5,11 ! Select base element
vsel,a,volu,,20,26
eslv,s
esel,r,type,,2
secn,13 ! Assign section to reinforcing element
ereinf ! Generate reinforcing element
!
! Reinforcing elements are generated similarly in other sections
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
370 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
Node-to-surface contact elements (CONTA175) are used for the rim nodes and a target element
(TARGE170) is used to define a pilot node at the center of the tire, as shown in the following figure:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 371
Hydrostatic Fluid Analysis of an Inflating and Rolling Tire
The CONTA174 element is used to model the part of the tire's surface which comes in contact with the
road. The road is modeled as a rigid target (TSHAP,QUAD). Impact constraints (KEYOPT(7) = 4) are
specified for the contact element.
Tire Rubber
Young's Modulus 6894.8 MPa
Poisson's Ratio 0.5
Density 2.67E-9 Kg/mm3
Mooney-Rivlin Material Model Constants
C10 0.551584 MPa
C01 0.137896 MPa
D 0
Air is used as the inside fluid, modeled using a compressible gas model.
Internal Air
Initial Density 1.225E-12 Kg/mm3
Reference 20°C
Temperature
Temperature Offset 274°C
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
372 of ANSYS, Inc. and its subsidiaries and affiliates.
Boundary Conditions and Loading
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 373
Hydrostatic Fluid Analysis of an Inflating and Rolling Tire
A comparison to the expected results (obtained via an analytical solution) validates the simulation of
the air using the hydrostatic fluid element HSFLD242.
Density: ρf = ρo * Pf / Po
Pressure: Pf = Po * Vo / Vf when mass is constant during load step
Mass: mass = density * volume
The following table shows the expected results at each load step:
The following table shows the simulation results at each load step:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
374 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 375
Hydrostatic Fluid Analysis of an Inflating and Rolling Tire
Most importantly, the figures indicate a good agreement with the expected results, validating the use
of hydrostatic fluid element HSFLD242 to model the air inside of the tire.
24.8. Recommendations
To perform a similar analysis of a fluid that is fully enclosed within a solid container, consider the fol-
lowing hints and recommendations:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
376 of ANSYS, Inc. and its subsidiaries and affiliates.
Input Files
After creating the fluid elements (ESURF), note whether they have positive or negative volumes.
Fluid elements have negative volumes if nodes are ordered in a clockwise direction, or if the normal
of the container is oriented outward from the pressure node.
• Introduce the fluid mass based on the volume of the fluid element.
Set KEYOPT(5) = 1 for the hydrostatic fluid. No mass is added if the volume becomes negative.
Set KEYOPT(6) = 0 for the hydrostatic fluid to model compressible fluid (default), or KEYOPT(6) = 1
to model incompressible fluid.
The pressure node can be located anywhere within the fluid volume. If the fluid volume is bounded
by one or more symmetry lines or planes, however, the pressure node must be on the line or plane
of symmetry, or on the intersecting corner or line of multiple symmetry lines or planes, and it must
have symmetrical boundary conditions.
For contact, specify impact constraints (KEYOPT(7) = 4). This option includes automatic time-increment
adjustments for standard (default) or rough contact (KEYOPT(12) = 0 or 1, respectively) in a transient
dynamic analysis.
• tire_hsfld242.cdb -- The common database file containing the model information for this problem
(called by tire_hsfld242.dat).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 377
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
378 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 25: Cardiovascular Stent Simulation
This example problem shows how to simulate stent-artery interaction during and after stent placement
in an occluded artery.
• Contact
• Nonlinear stabilization
25.1. Introduction
A bare metal stent, as shown in the following figure (courtesy of Lakeview Center) is an effective device
for opening atherosclerotic arteries and other blockages:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 379
Cardiovascular Stent Simulation
The success of stenting depends largely on how the stent and the artery interact mechanically. In both
the stent-design process and in pre-clinical patient-specific evaluations, computer simulation using finite
element analysis (FEA) has become an accepted tool for studying stent-artery interaction.
A viable stent-artery finite element model must properly reflect the nonlinear nature of the phenomenon,
such as the biological tissue properties, large arterial wall deformation, and the sliding contact between
the stent and the artery wall. This example problem shows how to use appropriate technologies to
overcome these modeling challenges.
The artery is simplified as a two-layered straight cylinder, with one layer representing the artery wall
and the other representing the calcified plaque.
The following figure shows the general dimensions of the artery and stent:
3. Contract the artery using mean blood pressure and creating contact between the stent and the artery wall.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
380 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
25.3. Modeling
Cardiovascular stent modeling involves three components:
25.3.1. Stent Modeling
25.3.2. Artery and Plaque Modeling
25.3.3. Stent-Plaque Contact Modeling
For modeling simplicity and computational efficiency, beam elements are preferred over solid elements.
The stent assembly has a 3.5 mm diameter, a 15 mm length, and 8 crowns. The wire for constructing
the stent has a circular cross-section with an outer diameter of 0.1 mm.
Although Nitinol material is commonly used for the stent, the nonlinear material behavior of Nitinol
requires a separate discussion. For the purposes of this problem, therefore, the model uses linear
elastic 316L steel instead.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 381
Cardiovascular Stent Simulation
The artery layer is meshed with 9,000 SOLID185 layered structural solid elements with the simplified
enhanced strain formulation (KEYOPT(2) = 3). Mixed u-P formulation (KEYOPT(6) = 1) is used to overcome
the volumetric locking typically associated with incompressible biological tissue.
The plaque layer is also meshed with 9,000 SOLID185 elements. Full integration with the method is
used for the plaque elements, as the material of the calcified plaque is considered to be linear elastic.
A coincident mesh is created at the artery-plaque interface to enforce a secure bond between the artery
and the plaque.
Based on St. Venant’s principle, both the artery and plaque are extended by 3 mm to reduce end effects.
Fine elements are used near the two ends to mitigate any convergence difficulty caused by large localized
deformation.
A Lagrangian multiplier method on contact normals and penalty tangent method on target normals is
used (KEYOPT(2) = 3), along with automatic bisection (KEYOPT(7) = 1) and standard contact behavior
(KEYOPT(12) = 0).
The inner plaque wall surface is meshed with TARGE170 target elements. Zero-friction behavior is assumed
[1].
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
382 of ANSYS, Inc. and its subsidiaries and affiliates.
Boundary Conditions and Loading
Figure 25.5: Standard Line-to-Surface Contact Between Stent and Inner Plaque Wall
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 383
Cardiovascular Stent Simulation
MPC pilot nodes (TARGE170) are fixed in all six degrees of freedom. The boundary conditions allow for
radial arterial expansion, while adequately preventing rigid body motion of the artery.
MPC pilot nodes (TARGE170) are fixed in all six degrees of freedom.
The following figure illustrates the load surface and load history:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
384 of ANSYS, Inc. and its subsidiaries and affiliates.
Analysis and Solution Controls
Figure 25.8: Uniform Pressure Loading on the Inner Plaque Wall (a) and Load History (b)
(a) (b)
Load Step 1
During the first load step, an elevated blood pressure of 0.1 N/mm2 is applied to the inner surface of
the plaque wall to cause sufficient radial wall expansion for subsequent stent placement.
Stent contact elements (CONTA177) are killed (EKILL) to remove the effects of the stent.
This load step uses a maximum of 20 substeps with 20 initial substeps (NSUBST,20,20).
The following figure shows the effects of the first load step:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 385
Cardiovascular Stent Simulation
Figure 25.9: Cross-Sectional View of Artery and Stent After Balloon Angioplasty (Load Step 1)
Load steps 2 and 3 use three total substeps to allow the Newton-Raphson residuals (from the nonlinear
expansion in load step 1) to equilibrate after the stent contact elements are reactivated (EALIVE).
Load Step 4
In the fourth load step, blood pressure is ramped to a magnitude of 0.0133 N/mm2, which represents
the mean arterial blood pressure (100 mmHg).
Under this reduced load, the atherosclerotic artery collapses onto the stent scaffold.
This load step uses 200 initial substeps, 2000 maximum substeps, and 20 minimum substeps to obtain
contact convergence (NSUBST,200,2000,20).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
386 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The positive effect of stenting is evident in the following figure, which shows the artery wall configura-
tions before and after stent placement:
Figure 25.10: Arterial Wall Deformation During Balloon Angioplasty (a) and After Stent Placement
(b)
(a) (b)
The following figure clearly shows the expected tissue prolapse (tissue extension into the gaps in the
stent):
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 387
Cardiovascular Stent Simulation
This figure shows the detailed stress distribution on the inner artery wall, with an expected pattern
matching the stent geometry:
Finally, the stent retraction under compressive load from the occluded artery wall is shown in this figure:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
388 of ANSYS, Inc. and its subsidiaries and affiliates.
Recommendations
The simulation results agree well with those in the published literature [1].
FEA-based simulation is capable of quickly generating accurate and detailed information about stent-
artery interaction. Finite element modeling is being used not only to develop state-of-the-art stent in-
novations, but also for pre-clinical patient-specific assessment and customization.
25.8. Recommendations
To perform a similar stent-artery interaction analysis, consider the following hints and recommendations:
• Compared to surface-to-surface contact with a full solid model, line-to-surface contact can provide similar
results using significantly less solution time.
• The choice of units is critical for avoiding numerical difficulties. For biological problems, millimeters-micron
units are preferred.
• To achieve faster solutions, coincident nodes and surfaces are preferred over bonded contact.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 389
Cardiovascular Stent Simulation
25.9. References
The following reference is used in this example problem:
1. Lally, C., F. Dolan, P. J. Pendergrast. [Journal of Biomechanics]. “Cardiovascular stent design and vessel
stresses: a finite element analysis.” 38 (2005): 1574-1581.
• stent.cdb -- The common database file containing the model information for this problem (called by
stent.dat).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
390 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 26: Nonlinear Analysis of a Rubber Boot Seal
This example problem demonstrates the capabilities and advantages of the surface-projection-based
contact method in a highly nonlinear problem. This analysis of a rubber boot seal model includes geo-
metric, material, and changing status nonlinearities (contact).
• Surface-projection-based contact
• Geometric nonlinearities
26.1. Introduction
Rubber boot seals are used in many industrial applications to protect the flexible joint between two
bodies. In the automotive industry, rubber boot seals cover constant velocity joints on the drive shaft
to protect them from the outside elements (dust, humidity, mud, etc.). These rubber boots are designed
to accommodate the maximum possible swing angle of the joints and to compensate for changes in
the shaft length.
This rubber boot seal example demonstrates geometric nonlinearities (large strain and large deformation),
nonlinear material behavior (rubber), and changing status nonlinearities (contact). The objective of this
example is to show the advantages of the surface-projection-based contact method and to determine
the displacement behavior of the rubber boot seal, stress results, and location of the contact point on
the outer surface and inner surface of the boot during the shaft motion.
Surface-projection-based contact can be defined by setting KEYOPT(4) = 3 for the contact elements.
This option enforces contact constraints on an overlapping region of contact and target surfaces rather
than on individual contact nodes or gauss points. The contact penetration/gap is computed over the
overlapping region in an average sense.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 391
Nonlinear Analysis of a Rubber Boot Seal
Figure 26.1: 3-D Model of Rubber Boot Seal and Rigid Shaft
26.3. Modeling
Due to the symmetry of the structure, only half of the rubber boot is modeled. For the rubber boot,
the hyperelastic material model is used. The shaft is considered as a rigid body.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
392 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
Figure 26.2: Half Symmetry Model of the 3-D Rubber Boot Seal
• Rigid flexible contact between the rigid cylindrical shaft and the inner surface of the rubber boot.
• Self contact at the inner surface of the rubber boot using the surface-projection-based contact method.
• Self contact at the outer surface of the rubber boot using the surface-projection-based contact method.
26.3.2.1. Rigid-flexible Contact Pair between Rigid Shaft and Rubber Boot
The rigid cylindrical shaft is modeled by the TARGE170 element with the TSHAP,CYLI command. The
radius of the cylindrical shaft is 14 mm. This rigid cylinder is in initial interference with the internal
surface of the rubber boot.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 393
Nonlinear Analysis of a Rubber Boot Seal
The following contact settings are used for the contact elements, CONTA174:
• KEYOPT(4) = 0 to set the location of the contact detection point at the gauss integration point
The following commands illustrate the contact settings of the rigid flexible contact pair:
et,2,170 ! Target element to define rigid cylinder
et,3,174 ! Contact element for rigid-flexible contact pair
keyopt,3,9,2 ! Include interference with ramped effects
keyopt,3,4,0 ! Gauss point contact detection algorithm
keyopt,3,10,2 ! Update contact stiffness at each iteration
r,2
rmodif,2,1,14 ! Radius of the cylinder = 14 mm
rmodif,2,3,1.0
26.3.2.2. Self Contact Pairs at Inner and Outer Surfaces of Rubber Boot
To model a self contacting pair, both the target and contact surfaces are the same. KEYOPT(4) = 3 is
used to define surface-projection-based contact, and KEYOPT(10) = 2 is used to update contact stiffness
at each iteration.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
394 of ANSYS, Inc. and its subsidiaries and affiliates.
Boundary Conditions and Loading
In addition, the coefficient of friction for the rubber material is set to 0.2.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 395
Nonlinear Analysis of a Rubber Boot Seal
Figure 26.6: Boundary Conditions at the Bottom of the Rubber Boot (in the Axial Direction)
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
396 of ANSYS, Inc. and its subsidiaries and affiliates.
Analysis and Solution Controls
Figure 26.7: Boundary Conditions at the Bottom of the Rubber Boot (in the Radial Direction)
The load is applied in terms of displacements and rotations through different load steps.
• Load Step 1
Base node (pilot node) at the end of the shaft’s center axis is constrained in all direction as follows:
RIGID_BASE=1
RIGID_TOP =2
n,RIGID_BASE,0,0,0 ! Base node at the end of the
! shaft’s center axis
n,RIGID_TOP,0,97,0 ! Top node at the shaft’s center axis
d,RIGID_BASE,all
• Load Step 2
Boot seal gets compressed when the shaft moves down. The vertical movement of the shaft is
governed by the displacement applied to the base node (pilot node) at the end of the shaft’s
center axis. The following displacement is applied at the pilot node for the 2nd load step.
d,RIGID_BASE,uy,-10
• Load Step 3
Shaft is rotated by giving certain amount of rotation about z-axis to the base node (pilot node)
at the end of the shaft’s center axis, as follows:
d,RIGID_BASE,rotz,0.55
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 397
Nonlinear Analysis of a Rubber Boot Seal
• Compared to other methods, the surface-projection-based contact method (KEYOPT(4) = 3) requires the
least number of cumulative iterations to solve this problem.
• Convergence of the problem is less sensitive to the normal contact stiffness factor (FKN) when using the
surface-projection-based contact method (KEYOPT(4) =3).
Figure 26.8: Comparison of Cumulative Iterations for Different Contact Detection Methods
The surface-projection-based contact method (KEYOPT(4) = 3) produces smoother contact forces than
other contact detection algorithms. It is less sensitive to the magnitude of the contact stiffness.
For smaller values of FKN (0.1 to 1.0), the variation in cumulative iterations is less for all contact detection
methods. But for the higher values of FKN (1 to 10), the variation in cumulative iterations for the surface-
projection-based method is much less compared to the other contact detection methods. FKN = 1 is
optimal for nonlinear analysis. This can be observed in Figure 26.8: Comparison of Cumulative Iterations
for Different Contact Detection Methods (p. 398).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
398 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
than for other contact detection algorithms. The overall performance using the surface projection
method turns out to be better. This can be observed in the table below:
Figure 26.9: Displacement Vector Sum at the Maximum Shaft Angle (p. 399) shows the displacement
position of the boot at maximum shaft angle.
Figure 26.10: Von Mises Stress at the Maximum Shaft Angle (p. 400) illustrates the location of the critical
zones where the maximum stress occurs in the boot at maximum shaft angle. In spite of the fact that
it is low, and considering the fatigue effects of the material after a given number of cycles, it is clear
that these areas are the most likely to fail under fatigue loads.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 399
Nonlinear Analysis of a Rubber Boot Seal
The following figures show total strain at different points of contact occurring in the rubber boot at
different shaft angels. The first instance of contact occurs between the outer plaits of the rubber boot.
After that, stretching is noticed in the first contact zone, and the second contact occurs between the
inner plaits of the rubber boot. Finally, the third contact occurs between the inner plaits of the rubber
boot and the shaft.
Figure 26.11: Total Strain at First Contact between Outer Plaits of Rubber Boot
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
400 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Figure 26.12: Total Strain at Second Contact between Inner Plaits of Rubber Boot
Figure 26.13: Total Strain at Third Contact between Inner Plaits of Rubber Boot and Shaft
The following animation shows the von Mises stress, reaction force in the Y direction, and moment
about the Z-axis through all load steps.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 401
Nonlinear Analysis of a Rubber Boot Seal
26.8. Recommendations
The following are recommendations for performing this type of analysis:
• Note that the surface-projection-based method (KEYOPT(4) = 3) is not used to define contact between
the rigid shaft and the rubber boot because this method does not support rigid surfaces defined by
primitive target segments.
• For problems such as an interference fit, over-penetration is expected. These problems often have
convergence difficulties if the initial penetration is step-applied in the first load step. You may overcome
convergence difficulties by ramping the total initial penetration over the first load step using KEYOPT(9)
= 2.
• bootseal_3d.dat -- Input file for the nonlinear analysis of the rubber boot seal using surface-projection-
based contact.
• bootseal_3d.cdb -- Common database file containing the model information for this problem (called
by bootseal_3d.dat).
• bootseal3d.anf -- File containing geometry information for the 3-D rubber boot seal model.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
402 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 27: Hot-Rolling Structural Steel Analysis with 3-D Rezoning
This example problem uses rezoning in a 3-D large-deformation problem. In large-deformation analyses
(such as rolling, forging, and extrusion), elements are subject to excessive deformation, which can
eventually cause the analysis to terminate.
For more information about rezoning and the methods for repairing or replacing a distorted mesh, see
Rezoning in the Advanced Analysis Guide.
27.1. Introduction
Hot-rolling is a metal-forming process occurring above the recrystallization temperature of the material.
Many types of hot-rolling processes exist, including structural shape rolling, where a component is
passed through rollers to achieve the desired shape and cross section.
Structural steel is the most common hot-rolled material. Common shapes for structural steel include I-
beams, H-beams, T-beams, U-beams, and channels. I-beams have an I-shaped cross-section. The hori-
zontal elements of the cross section are called flanges, and the vertical element is called the web.
In this example problem, the hot-rolling process to form the I-beam is simulated statically using
rezoning.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 403
Hot-Rolling Structural Steel Analysis with 3-D Rezoning
In the unsteady phase, the billet (rectangular bar of steel) comes into contact with the rollers and fills
the gap between the rollers before moving through the rollers. When the billet begins to move through
the rollers, the process is considered to be in a steady state until the end face of the billet comes into
contact with the rollers.
The static analysis is performed in two load steps: the first builds up the rolling process, and hot-rolling
occurs in the second.
In the first load step, the billet moves toward rigid rollers to establish contact with the rollers and to
fill the gap between the rollers. In order to build up the rolling process, the billet should partially fill
the gap between rollers so that when rollers begin to rotate, they can pull the billet in via friction.
In the second load step, the rollers pull the billet in and eventually shape the rectangular billet into an
I-section block.
In this example, the simulation terminates at near the end of the first load step due to mesh distortion.
A rezoning operation repairs the distorted mesh, and the analysis resumes and continues to completion
using the new mesh.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
404 of ANSYS, Inc. and its subsidiaries and affiliates.
Problem Description
Horizontal cylindrical rollers pull the block from the top and bottom to increase the width and reduce
the depth of the block. The rollers control the width of the flange parts of the I-shape; they are
modeled using rigid target elements.
• Side rollers
Vertical cylindrical rollers with small fillets at either end. The fillets are necessary to ensure smooth
material flow. The side rollers pull the block from the sides to create the I-shaped cross section. They
control the width of the web part of the I-shape. As with the top and bottom rollers, the side rollers
are modeled using rigid target elements.
The following figure shows that the problem is symmetrical about two planes (XZ and YZ):
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 405
Hot-Rolling Structural Steel Analysis with 3-D Rezoning
To reduce modeling and computational time, therefore, only one quarter of the model is analyzed. After
the analysis has completed, however, the results are viewable in the full model by performing symmetry
expansion about two planes of symmetry.
The simulation is performed statically in two load steps. In the first load step, the block is moved towards
fully constrained rigid rollers in order to build up the rolling process. In the second load step, hot-rolling
is performed by rotating the rollers about their axes of rotation, and the block is free to move in the
horizontal direction (Z).
As the rollers rotate, the block passes through them due to the high level of friction between the rollers
and the block. Eventually, the full block passes through the rollers to achieve the I-shaped cross section.
In such large-deformation problems, however, mesh distortion is common, leading to convergence
difficulties or eventual analysis termination.
The analysis diverges in the first load step due to excessive distortion in a few elements. A rezoning
operation repairs the distorted mesh and allows the analysis to continue.
27.3. Modeling
The following figure shows one-quarter of the model, all that is necessary for this analysis:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
406 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
The geometry of the block is 1/4 of the full geometry of the block shown in Figure 27.1: Hot-Rolling
Model (p. 405). Appropriate geometries for the rollers are also considered in the quarter model.
Notice that the block has small fillets at either end of the top face. The fillet helps to establish contact
with the top roller. Without the fillet, the sharp corner of the block would cause local singularities, and
the analysis would diverge.
Use this method if the shape of the target surface matches any of the available primitives (circle,
cylinder, cone, and sphere). You can also combine primitive segments with general segments (lines,
parabolas, triangles, and quadrilaterals) to define a target surface.
You must first specify the shape of the target elements (TSHAP), then define the radius via real con-
stants. Create nodes (N) and elements (E) directly. The elements will have their shapes already defined.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 407
Hot-Rolling Structural Steel Analysis with 3-D Rezoning
Use this method when target surfaces cannot be modeled with primitives. Because the side rollers
have fillets at either end, they cannot be modeled using primitives. In this case, CAD geometry of
the side roller is created in DesignModeler and then meshed with TARGE170 elements to create the
contact pair with the block.
For more information, see Defining the Target Surface in the Contact Technology Guide.
The rigid top roller is modeled using one TARGE170 element and primitives (TSHAP, CYLI).
One pilot node is created and associated with the top roller (the rigid target surface). The pilot node
governs the motion of the entire target surface. (Think of a pilot node as a sort of handle for the rigid
target surface.) Forces/moments or rotations/displacements for the entire target surface should generally
be specified via the pilot node. The location of the pilot node is important when rotation or moment
loading is required. In this case, the pilot node is created at the center of the mass of the top roller, as
the roller must be rotated in the second load step.
The augmented Lagrangian algorithm is used in this contact pair. The following example input shows
how the contact pair is modeled:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
408 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
et,2,CONTA174
KEYOPT,2,10,2 ! Select option to update contact stiffness in each iteration
et,3,TARGE170
! Create and attach pilot node to this rigid target surface (top roller)
type,3
real,3
n,9999,-0.5,1.85,0.30
tshap,pilot
e,9999
allsel,all
Contact surfaces are modeled with CONTA174 elements and the rigid side roller is modeled with
TARGE170 elements. As with the previous contact pair (p. 408), one pilot node is created at the center
of the mass of the side roller. The pilot node governs the motion of the side roller.
The following example input shows how the contact pair is modeled:
et,4,CONTA174
KEYOPT,2,10,2 ! Select option to update contact stiffness in each iteration
et,5,TARGE170
cmsel,s,tn2.cnt ! Select the nodal component of 2 faces of the block
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 409
Hot-Rolling Structural Steel Analysis with 3-D Rezoning
nplot
type,4 ! Select element type 4 (CONTA174)
real,5 ! Use Real Constant 5
esurf ! Generate contact elements
allsel,all
! Model rigid side roller by meshing the area of side roller
type,5 ! Select element type 5 (TARGE170)
real,5 ! Use same real contact 5
aatt,1,5,5 ! Assign attributes to the areas of side roller
amesh,all ! Mesh geometry of the side roller with TARGE170 elements
allsel,all
esel,s,type,,5 ! The target normals are not oriented correctly, so we must
!reverse the direction of normals
esurf,,reverse ! Reverse the direction of the normals on target elements
allsel,all
! Create and attach pilot node to this rigid target surface (side roller)
type,5
real,5
n,20999,-0.4461,0.4337,0.0456
tshap,pilot
e,20999
allsel,all
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
410 of ANSYS, Inc. and its subsidiaries and affiliates.
Boundary Conditions and Loading
To showcase the usefulness of rezoning in 3-D problems, this problem is solved statically in two load
steps:
27.5.1. Load Step 1: Establish Contact with Rollers
27.5.2. Load Step 2: Hot-Rolling
Both the rigid rollers (top and side) are constrained in all directions with the use of pilot nodes. No
friction is used.
The following input applies the boundary conditions and loading used in the first load step:
!Load Step 1
!Boundary Conditions
cmsel,s,ux_node ! Select node on symmetry plane (YZ plane)
d,all,ux ! Apply symmetry boundary conditions
cmsel,s,uy_node ! Select node on symmetry plane (XZ plane)
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 411
Hot-Rolling Structural Steel Analysis with 3-D Rezoning
Because the top and side rollers are different sizes, different rotations are given to the rollers in order
to maintain continuity. Based on the size of top roller, the friction coefficient, and the block length, one
full rotation is applied to the top roller (as shown in Figure 27.8: Loadings (p. 411)).
No forward or backward slip is considered while calculating the rotation value of the side roller. Rotation
of the side roller is calculated using this formula:
Rotation of side roller = Rotation of top roller * (Radius of top roller / Radius of side
roller)
The following input applies the boundary conditions and loading used in the second load step:
!Load Step 2
cmsel,s,uz_node ! Select the nodes of the left end face of the block
ddel,all,uz ! Free the block to move in Z direction
nsel,s,,,9999
d,all,rotx,-6.28 ! Rotate top roller
nsel,s,,,20999
d,all,roty,-10.47 ! Rotate side roller
mp,mu,1,0.6 ! Use high value of friction
allsel,all
Restart files are saved at each substep, as rezoning requires a restart file, and the substep at which
rezoning is required is still unknown. Results items are also stored at each substep.
For illustrative purposes, restart files and results items are saved at each and every substep,
despite the considerable memory requirements for doing so. In most cases, it is sufficient to
save them at every few substeps instead.
The following input applies the solution settings for the run:
/solu
time,1 ! Define time for first load step
nlgeom,on ! Enable nonlinear geometry effects
OUTRES,ALL,ALL ! Save the results data for each substep
RESCONTRL,DEFINE,ALL,ALL,0 ! Save restart files for each substep
NSUBST,100,1000000,20 ! Define initial, maximum and minimum substeps
solve ! Solve the first load step
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
412 of ANSYS, Inc. and its subsidiaries and affiliates.
Analysis and Solution Controls
The initial run diverges in the first load step at TIME = .7718. The following figure shows the deformed
shape of the model at the last converged substep:
Element shape checking of the deformed model suggests that the mesh is overly distorted at the location
indicated.
Rezoning repairs the distorted mesh and allows the analysis to continue. Rezoning is applied to this
problem as described in the following topics:
27.6.1. Rezoning Initiated at the 30th Substep
27.6.2. Distorted Mesh Replaced by an Imported New Mesh
27.6.3. Solution Items Mapped from Original Mesh to New Mesh
27.6.4. Analysis Resumes Using the New Mesh
The analysis diverges after the 40th substep at TIME = .7718. Very little time (from TIME = .77 to .7718)
occurs between the 31st to the 40th substeps, however, indicating that severe distortion of the mesh
begins to occur at the 31st substep.
Because rezoning should be attempted at one or more substeps before the substep where mesh deform-
ation occurs, the 30th substep is a logical choice at which to initiate rezoning. Shape-checking (SHPP
or CHECK) of the deformed mesh at the 30th substep confirms the choice, as the deformed mesh violates
no error limits.
The following figure shows the deformed mesh at the 30th substep:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 413
Hot-Rolling Structural Steel Analysis with 3-D Rezoning
Figure 27.10: Deformed Mesh of the Block at 30th Substep of First Load Step
The new mesh can be generated in Workbench or by other third-party software by first creating the
deformed geometry from the deformed mesh, then meshing the deformed geometry with new settings
to obtain a new, good mesh. The new mesh must be better than the original mesh; otherwise, rezoning
cannot improve convergence.
The following figure shows the original, deformed mesh and the new, good mesh:
The following input initiates rezoning at the 30th substep and reads in the new mesh:
/solu
rezone,manual,1,30 ! Start rezoning at the 30th substep
remesh,start ! Start “remeshing” process
remesh,read,Rezone_Mesh,cdb ! Read new mesh in “cdb” format
remesh,finish ! Transfer boundary conditions and loadings from the
! old mesh to the new mesh, and recreate contact
! pairs (if any)
After remeshing (REMESH,FINISH), the program transfers surface loads, forces, boundary conditions,
and contact pairs (if any) from the original, deformed old mesh to the new, good mesh. It is good
practice to check the model after remeshing to verify that all such transfers to the new mesh were
successful.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
414 of ANSYS, Inc. and its subsidiaries and affiliates.
Analysis and Solution Controls
The following figure shows the contact pressure results for the top roller contact pair, first from the
30th substep (where rezoning is performed), and after mapping variables and balance residuals from
the original mesh to the new mesh:
Figure 27.12: Comparison of Contact Pressure Results Before and After Rezoning
The mapping operation concludes the rezoning process, and a standard multiframe restart resumes
solution processing using the new mesh.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 415
Hot-Rolling Structural Steel Analysis with 3-D Rezoning
Following the successful rezoning, the following figure shows the deformation plot (USUM) of the
model after building up the rolling process (at the end of the first load step):
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
416 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The hot-rolling process is performed in the second load step by rotating the rollers and using high-
friction contact between the rollers and the billet. The following figure shows the deformation plot
(USUM) of the model at the end of the second load step:
Figure 27.15: USUM Plots: End of Analysis (After Second Load Step)
To view the full-model results, two symmetric expansions (/EXPAND) are performed. The following input
performs the symmetric expansion:
CS,11,0,28752,29076,28734,1,1 ! Create local Cartesian coordinate system (using three nodes) on YZ symmetry plane
/EXPAND,1,LRECT,HALF,,.000001,,1,LRECT,HALF,,,.000001, ,RECT,FULL
The following figure shows the deformation plot in the fully expanded model, where the rectangular
block becomes an I-shaped beam:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 417
Hot-Rolling Structural Steel Analysis with 3-D Rezoning
The following figure shows the variation of the X component of the moment along with the time for
the top roller:
The plot suggests that rolling moment varies little from TIME = 1.25s to TIME = 1.6. The hot-rolling
process during this time range can be considered to be steady state.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
418 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Similarly, the following figure shows the variation of the force components of the top roller:
The plot also shows that forces vary little from TIME = 1.25 to TIME = 1.6s. As would be expected from
the top roller, the Y component of the force (that is, the downward force) is dominant in the rolling
process.
The following figure shows the equivalent plastic strain plot in the fully expanded model:
Figure 27.19: Equivalent Plastic Strain Plot in the Full Expanded Model
As expected, the high plastic strain region is observed at the location where the web and flange com-
ponents of the I-beam are connected.
For some duration in the rolling process, the plastic strain varies little over time in the high plastic strain
region. This behavior occurs during the steady-state (rolling) phase of the simulation.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 419
Hot-Rolling Structural Steel Analysis with 3-D Rezoning
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
420 of ANSYS, Inc. and its subsidiaries and affiliates.
Recommendations
27.8. Recommendations
To perform a similar 3-D simulation using rezoning, consider the following hints and recommendations:
• The hot-rolling process can be simulated via static analysis in two load steps. The first load step pushes the
billet until it establishes contact with the rollers, and the second pulls the billet by rotating the rollers.
• Before rezoning, back up results and restart files associated with the initial run in a separate directory.
Rezoning updates results and restart files, so the original files are no longer available should you wish to try
rezoning at another substep.
• If rezoning is performed at a substep where the original mesh is too distorted (where shape-checking [SHPP
or CHECK] indicates errors), then rezoning will not work. Rezoning should therefore be performed at an
earlier substep.
• A new mesh that is too fine as compared to the original mesh may cause mapping (MAPSOLVE) errors. The
primary requirement for the new mesh is that it should properly capture the outer surface geometry of the
deformed model.
• Check the model after remeshing (REMESH,FINISH) to verify that all boundary conditions, contact pairs, and
loadings have transferred correctly from the original mesh to the new mesh.
• After rezoning, if the analysis diverges again after passing the initial run's diverged time, multiple rezonings
may be necessary. If the analysis diverges again before passing the initial run’s diverged time, then either
the new mesh is of insufficient quality, or other problems unrelated to mesh distortion (such as geometry
and material instabilities) exist.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 421
Hot-Rolling Structural Steel Analysis with 3-D Rezoning
• Hot_Rolling_Model.cdb -- Common database file containing the hot-rolling model information (called
by Hot_Rolling_Model.dat).
• Initial_Mesh.cdb -- Common database file for the original mesh used in the hot-rolling problem.
• Rezone_Mesh.cdb -- Common database file for the new mesh used for rezoning in the hot-rolling problem.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
422 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 28: Friction Stir Welding (FSW) Simulation
This example problem shows how to simulate the friction stir welding (FSW) process. Several character-
istics of FSW are presented, including tool-workpiece surface interaction, heat generation due to friction,
and plastic deformation. A nonlinear direct coupled-field analysis is performed, as thermal and mechan-
ical behaviors are mutually dependent and coupled together during the FSW process.
Because it is often difficult to find a full set of engineering data to simulate the FSW process, the
problem emphasizes the simulation rather than the numerical results. A simplified version of the model
created by Zhu and Chao [1 (p. 444)] illustrates the FSW simulation method.
28.1. Introduction
Friction stir welding (FSW) is a solid-state welding technique that involves the joining of metals without
filler materials. A cylindrical rotating tool plunges into a rigidly clamped workpiece and moves along
the joint to be welded. As the tool translates along the joint, heat is generated by friction between the
tool shoulder and the workpiece. Additional heat is generated by plastic deformation of the workpiece
material. The generated heat results in thermal softening of the workpiece material. The translation of
the tool causes the softened workpiece material to flow from the front to the back of the tool where
it consolidates. As cooling occurs, a solid continuous joint between the two plates is formed. No melting
occurs during the process, and the resulting temperature remains below the solidus temperature of
the metals being joined. FSW offers many advantages over conventional welding techniques, and has
been successfully applied in the aerospace, automobile, and shipbuilding industries.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 423
Friction Stir Welding (FSW) Simulation
Thermal and mechanical behaviors are mutually dependent during the FSW process. Because the tem-
perature field affects stress distribution, this example uses a fully thermo-mechanically coupled model.
The model consists of a coupled-field solid element with structural and thermal degrees of freedom.
The model has two rectangular steel plates and a cylindrical tool. All necessary mechanical and thermal
boundary conditions (p. 430) are applied on the model. The simulation occurs over three load steps,
representing the plunge, dwell, and traverse phases (p. 431) of the process.
The temperature rises at the contact interface due to frictional contact between the tool and workpiece.
FSW generally occurs when the temperature at the weld line region reaches 70 to 90 percent of the
melting temperature of the workpiece material [3 (p. 444)]. The temperature obtained around the weld
line region in this example falls within the range reported by Zhu and Chao [1 (p. 444)] and Prasanna
and Rao [3 (p. 444)], while the maximum resulting temperature is well below the melting temperature
of the workpiece.
The calculated frictional heat generation and plastic heat generation show that the friction between
the tool shoulder and workpiece is responsible for generating most of the heat. A bonding temperature
is specified at the contact interface of the plates to model the welding behind the tool. When the
temperature at the contact surface exceeds this bonding temperature, the contact is changed to bonded.
The model used in this example is a simplified version of the thermo-mechanical model developed
by Zhu and Chao [1 (p. 444)] for FSW with 304L stainless steel. Zhu and Chao presented nonlinear
thermal and thermo-mechanical simulations using the finite element analysis code WELDSIM. They
initially formulated a heat-transfer problem using a moving heat source, and later used the transient
temperature outputs from the thermal analysis to determine residual stresses in the welded plates
via a 3-D elastoplastic thermo-mechanical simulation.
A direct coupled-field analysis is performed on a reduced-scale version of the Zhu and Chao model
[1 (p. 444)]. Also, rather than using a moving heat source as in the reference model, a rotating and
moving tool is used for a more realistic simulation.
The tool pin is ignored. The heat generated at the pin represents approximately two percent of the
total heat and is therefore negligible.
The simulation welds two 304L stainless steel plates (workpiece) with a cylindrical shape tool, as shown
in the following figure:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
424 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
The FSW process generally requires a tool made of a harder material than the workpiece material being
welded. In the past, FSW was used for soft workpiece materials such as aluminium. With the development
of tools made from super-abrasive materials such as polycrystalline cubic boron nitride (PCBN), FSW
has become possible with high-temperature materials such as stainless steel [4 (p. 445)]. A cylindrical
PCBN tool is modeled in this case.
The workpiece sides parallel to the weld line are constrained in all the directions to simulate the
clamping ends. The bottom side of the workpiece is constrained in the perpendicular (z) direction to
simulate support at the bottom. Heat losses are considered on all the surfaces of the model. All
boundary conditions (p. 430) are symmetric across the weld centerline.
The simulation is performed in three load steps, each representing a respective phase (plunge, dwell,
and traverse (p. 431)) of the FSW process.
28.3. Modeling
Modeling is a two-part task, as described in these topics:
28.3.1. Workpiece and Tool Modeling
28.3.2. Contact Modeling
The plate size is 3 x 1.25 x 0.125 in (76.2 x 31.75 x 3.18 mm). The tool shoulder diameter is 0.6 in (15.24
mm).
Plate thickness remains the same as that of the reference model, but the plate length and width are
reduced. The plate width is reduced because the regions away from the weld line are not significantly
affected by the welding process, and this example focuses primarily on the heat generation and tem-
perature rise in the region nearest the weld line.
The height of the tool is equal to the shoulder diameter. Both the workpiece (steel plates) and the tool
are modeled using coupled-field element SOLID226 with the structural-thermal option (KEYOPT(1) =
11).
A hexahedral mesh with dropped midside nodes is used because the presence of midside nodes (or
quadratic interpolation functions) can lead to oscillations in the thermal solution, leading to nonphysical
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 425
Friction Stir Welding (FSW) Simulation
temperature distribution. A hexahedral mesh is used instead of a tetrahedral mesh to avoid mesh-ori-
entation dependency. For more accurate results, a finer mesh is used in the weld-line region. The fol-
lowing figure shows the 3-D meshed model:
The surface-projection-based contact method (KEYOPT(4) = 3 for contact elements) is defined at the
contact interface. The surface-projection-based contact method is well suited to highly nonlinear
problems that include geometrical, material, and contact nonlinearities.
The problem simulates welding using the bonding capability of contact elements. To achieve continuous
bonding and simulate a perfect thermal contact between the plates, a high thermal contact conductance
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
426 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
(TCC) of 2E06 W/m2 °C is specified. (A small TCC value yields an imperfect contact and a temperature
discontinuity across the interface.) The conductance is specified as a real constant for CONTA174 elements.
The maximum temperature ranges from 70 to 90 percent of the melting temperature of the workpiece
material. Welding occurs after the temperature of the material around the contacting surfaces exceeds
the bonding temperature (approximately 70 percent of the workpiece melting temperature). In this
case, 1000 °C is considered to be the bonding temperature based on the reference results. The bonding
temperature is specified using the real constant TBND for CONTA174. When the temperature at the
contact surface for closed contact exceeds the bonding temperature, the contact type changes to
bonded. The contact status remains bonded for the remainder of the simulation, even though the
temperature subsequently decreases below the bonding value.
The following example input defines the contact settings of the contact pair:
.
et,6,TARGE170
et,7,CONTA174
keyopt,7,1,1 ! Displacement and Temp DOF
keyopt,7,4,3 ! To include Surface projection based method
Two real constants are specified to model friction-induced heat generation. The fraction of frictional
dissipated energy converted into heat is modeled first; the FHTG real constant is set to 1 to convert all
frictional dissipated energy into heat. The factor for the distribution of heat between contact and target
surfaces is defined next; the FWGT real constant is set to 0.95, so that 95 percent of the heat generated
from the friction flows into the workpiece and only five percent flows into the tool [2 (p. 444)].
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 427
Friction Stir Welding (FSW) Simulation
A low TCC value (10 W/m2 °C) is specified for this contact pair because most of the heat generated
transfers to the workpiece. Some additional heat is also generated by plastic deformation of the workpiece
material. Because the workpiece material softens and the value of friction coefficient drops as the
temperature increases [3 (p. 444)], a variable coefficient of friction (0.4 to 0.2) is defined (TB,FRIC with
TBTEMP and TBDATA).
The following example input specifies the settings for the contact pair:
et,4,TARGE170
et,5,CONTA174
keyopt,5,1,1 ! Displacement & Temp DOF
keyopt,5,5,3 ! Close gap/reduce penetration with auto cnof
keyopt,5,9,1 ! Exclude both initial penetration or gap
keyopt,5,10,2 ! Contact stiffness update each iteration based
The following contact settings are used for the CONTA174 elements:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
428 of ANSYS, Inc. and its subsidiaries and affiliates.
Material Properties
It is assumed that the plastic deformation of the material uses the von Misses yield criterion, as well as
the associated flow rule and the work-hardening rule (Zhu and Chao [1 (p. 444)]). Therefore, a bilinear
isotropic hardening model (TB, BISO) is selected.
1. The TBDATA command defines the yield stress and tangent modulus.
The fraction of the plastic work dissipated as heat during FSW is about 80 percent [2 (p. 444)]. Therefore,
the fraction of plastic work converted to heat (Taylor-Quinney coefficient) is set to 0.8 (MP,QRATE) for
the calculation of plastic heat generation in the workpiece material.
To weld a high-temperature material such as 304L stainless steel, a tool composed of hard material is
required. Tools made from super-abrasive materials such as PCBN are suitable for such processes
[4 (p. 445)], and so a cylindrical PCBN tool is used here. The material properties of the PCBN tool are
obtained from the references ([5 (p. 445), 6 (p. 445)].
The following table shows the material properties of the PCBN tool:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 429
Friction Stir Welding (FSW) Simulation
Available data suggest that the value of the convection coefficient lies between 10 and 30 W/m2 °C
[1 (p. 444), 2 (p. 444), 3 (p. 444)] for the workpiece surfaces, except for the bottom surface. The value of
the convection coefficient is 30 W/m2°C for workpiece and tool. This coefficient affects the output
temperature. A lower coefficient increases the output temperature of the model. A high overall heat-
transfer coefficient (about 10 times the convective coefficient) of 300 W/m2 °C is assumed for the con-
ductive heat loss through the bottom surface of the workpiece. As a result, the bottom surface of the
workpiece is also treated as a convection surface for modeling conduction losses. Because the percentage
of heat lost due to radiation is low, radiation heat losses are ignored. An initial temperature of 25 °C is
applied on the model. Temperature boundary conditions are not imposed anywhere on the model.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
430 of ANSYS, Inc. and its subsidiaries and affiliates.
Boundary Conditions and Loading
28.5.3. Loading
The FSW process consists of three primary phases:
2. Dwell -- Friction between the rotating tool and workpiece generates heat at the initial tool position until
the workpiece temperature reaches the value required for the welding.
3. Traverse (or Traveling) -- The rotating tool moves along the weld line.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 431
Friction Stir Welding (FSW) Simulation
During the traverse phase, the temperature at the weld line region rises, but the maximum temperature
values do not surpass the melting temperature of the workpiece material. As the temperature drops, a
solid continuous joint appears between the two plates.
For illustrative purposes, each phase of the FSW process is considered a separate load step. A rigid
surface constraint is already defined for applying loading on the tool.
The following table shows the details for each load step.
The tool plunges into the workpiece at a very shallow depth, then rotates to generate heat. The depth
and rotating speeds are the critical parameters for the weld temperatures. The parameters are determined
based on the experimental data of Zhu and Chao. The tool travels from one end of the welding line to
the other at a speed of 2.7 mm/s.
FSW simulation includes factors such as nonlinearity, contact, friction, large plastic deformation, struc-
tural-thermal coupling, and different loadings at each load step. The solution settings applied consider
all of these factors.
The first load step in the solution process converges within a few substeps, but the second and third
load steps converge only after applying the proper solution settings shown in the following table:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
432 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
To allow for a faster solution, automatic time-stepping is activated (AUTOTS,ON). The initial time step
size (DELTIM) is set to 0.1, and the minimum time step is set to 0.001. The maximum time step is set
as 0.2 in load steps 2 and 3. A higher maximum time-step size may result in an unconverged solution.
The time step values are determined based on mesh or element size. For stability, no time-step limitation
exists for the implicit integration algorithm. Because this problem is inherently nonlinear and an accurate
solution is necessary, a disturbance must not propagate to more than one element in a time step;
therefore, an upper limit on the time step size is required. It is important to choose a time step size
that does not violate the subsequent criterion (minimum element size, maximum thermal conductivity
over the whole model, minimum density, and minimum specific heat).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 433
Friction Stir Welding (FSW) Simulation
The deflection causes high stresses to develop on the workpiece beneath the tool, as shown in this
figure:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
434 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Following load step 1, the temperature remains unchanged (25 °C), as shown in this figure:
As the tool begins to rotate at this location, the frictional stresses develop and increase rapidly. The
following two figures show the increment in contact frictional stresses from load step 1 to load step 2:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 435
Friction Stir Welding (FSW) Simulation
All frictional dissipated energy is converted into heat during load step 2. The heat is generated at the
tool-workpiece interface. Most of the heat is transferred to the workpiece (p. 427) (FWGT is specified to
0.95). As a result, the temperature of the workpiece increases rapidly compared to that of the tool.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
436 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The maximum temperature on the workpiece occurs beneath the tool during the last two load steps.
Heat generation is due to the mechanical loads. No external heat sources are used. As the temperature
increases, the material softens and the coefficient of friction decreases. A temperature-dependent
coefficient of friction (0.4 to 0.2) helps to prevent the maximum temperature from exceeding the ma-
terial melting point.
The observed temperature rise in the model shows that heat generation during the second and third
load steps is due to friction between the tool shoulder and workpiece, as well as plastic deformation
of the workpiece material.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 437
Friction Stir Welding (FSW) Simulation
The melting temperature of 304L stainless steel is 1450 °C. As shown in the following figure, the max-
imum temperature range at the weld line region on the workpiece beneath the tool is well below the
melting temperature of the workpiece material during the second and third load steps, but above 70
percent of the melting temperature:
Figure 28.16: Maximum Temperature (on Workpiece Beneath the Tool) Variation with Time
The two plates can be welded together within this temperature range.
The following figure shows the temperature distributions on the top surface of the workpiece along
the transverse distance (perpendicular to the weld line):
Figure 28.17: Temperature Distribution on the Top Surface of Workpiece at Various Locations
As shown in the following figure and table, the temperature plots indicate the temperature distribution
at various locations on the weld line when the maximum temperature occurs at those locations:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
438 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The following figure shows the temperature distribution in the thickness direction at location 1:
As expected, the highest temperature caused by heat generation appears around the weld line region.
By comparing the above temperature results with the reference results, it can be determined that the
temperatures obtained at the weld line are well below the melting temperature of the workpiece ma-
terial, but still sufficient for friction stir welding.
The following table and figure show the time-history response of the temperature at various locations
on the weld line:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 439
Friction Stir Welding (FSW) Simulation
The sticking portion of the interface shows the bonding or welding region of the plates. If the bonding
temperature was assumed to be 900 °C, then the welding region would increase, as shown in this figure:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
440 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The CONTA174 element's FDDIS (SMISC item) output option is used to calculate frictional heat generation
on the workpiece. This option gives the frictional energy dissipation per unit area for an element. After
multiplying this value with the corresponding element area, the friction heat-generation rate for an
element is calculated. By summing the values from each CONTA174 element of the workpiece, the total
frictional heat generation rate is calculated for a given time.
It is possible to calculate the total frictional heat-generation rate at each time-step (ETABLE). The fol-
lowing figure shows the plot of total frictional heat generation rate on the workpiece with time:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 441
Friction Stir Welding (FSW) Simulation
The plot indicates that the frictional heat starts from the second load step (after 1 second).
The element contact area can be calculated using the CONTA174 element CAREA (NMISC, 58) output
option.
/post26
file,fsw,rst
numvar,200
solu,191,ncmit ! Solution summary data per substep to be
! stored for cumulative no. of iterations.
store,merge ! Merge data from results file
filldata,191,,,,1,1
realvar,191,191
parres,new,'fsw','parm'
vput,fht,11,,,fric_heat
plvar,11 ! Plot of frictional heat rate against time
finish
A similar calculation is performed to check the heat generation from plastic deformation on the work-
piece. The SOLID226 element's output option PHEAT (NMISC, 5) gives the plastic heat generation rate
per unit volume. After multiplying this value with the corresponding element volume, the plastic heat
generation rate for an element is calculated. By summing the values from each element (SOLID226) of
the workpiece, the total plastic heat generation rate is calculated for a particular time.
It is possible to calculate the total frictional heat generation rate at each time-step (ETABLE). The fol-
lowing figure shows the plot of the total plastic heat-generation rate with time.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
442 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
/post26
file,fsw,rst
numvar,200
solu,191,ncmit !solution summary data per substep to be
!stored for cumulative no. of iterations.
store,merge !Merge data from results file
filldata,191,,,,1,1
realvar,191,191
parres,new,'fsw','parm'
vput,pha,10,,,pheat_nmisc
plvar,10 !Plot of Plastic heat rate against time
finish
Figure 28.23: Total Frictional Heat Rate Variation with Time (p. 441) and Figure 28.24: Total Plastic Heat
Rate Variation with Time (p. 443) clearly show that friction is responsible for generating most of the heat
needed, while the contribution of heat due to plastic deformation is less significant. Because the tool-
penetration is shallow and the tool pin is ignored, the plastic heat is small compared to frictional heat.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 443
Friction Stir Welding (FSW) Simulation
28.8. Recommendations
To perform a similar FSW analysis, consider the following hints and recommendations:
• FSW is a coupled-field (structural-thermal) process. The temperature field affects the stress distribution
during the entire process. Also, heat generated in structural deformation affects the temperature field. The
direct method of coupling is recommended for such processes. This method involves just one analysis that
uses a coupled-field element containing all necessary degrees of freedom. Direct coupling is advantageous
when the coupled-field interaction involves strongly-coupled physics or is highly nonlinear.
• A nonlinear transient analysis is preferable for simulations where the objective is to study the transient
temperature and transient heat transfer.
• The dynamic effects of different physics should be controlled. In this problem, for example, the dynamic
effects of the structural degrees of freedom are disabled as they are unimportant.
• Separating the solution process into three load steps helps you to understand the physics and solve the
problem.
• The contact between the two plates must be nearly perfect to maintain temperature continuity. For a perfect
thermal contact, specify a high thermal contact conductance (TCC) coefficient between workpiece plates.
A high coefficient results in temperature continuity across the interface.
• Because the problem is nonlinear, proper solution settings are required. Set the following analysis controls
to the appropriate values to achieve the converged solution: LNSRCH, CUTCONTROL, KBC, NEQIT, NROPT,
and AUTOTS.
• Convergence at the second and third load steps is difficult to achieve. The depth of penetration of the tool
on the workpiece (uz), rotational speed of the tool (rotz), and time-step size play crucial roles in the conver-
gence of the second load step. Use a very small time-step size if the rotational speed is higher than 60 RPM.
• A symmetric mesh (about the joint line) is preferred to capture the exact outputs and their effects on the
workpiece. A hex mesh with dropped midside nodes is recommended for the workpiece as well as the tool.
This approach helps to maintain symmetry and prevent the temperature from reaching negative values
during the simulation.
• A minimum of two element layers is required in the thickness direction. A fine sweep mesh near the weld
line yields more accurate results; however, too fine a mesh increases computational time. A fine mesh is
unnecessary on the tool side. To minimize computational time, the tool can be considered to be rigid with
no temperature degrees of freedom.
28.9. References
The following references are used in this example problem:
1. Zhu, X. K., and Y. J. Chao.“Numerical Simulation of Transient Temperature and Residual Stresses in Friction
Stir Welding of 304L Stainless Steel.” Journal of Materials Processing Technology. 146.2 (2004): 263-272.
2. Chao, Y.J., X. Qi, and W. Tang.“Heat Transfer in Friction Stir Welding - Experimental and Numerical Studies.”
Journal of Manufacturing Science and Engineering-Transactions of the ASME. 125.1 (2003): 138-145.
3. Prasanna, P., B. S. Rao, and G. K. Rao.“Finite Element Modeling for Maximum Temperature in Friction Stir
Welding and its Validation.” Journal of Advanced Manufacturing Technology. 51 (2010): 925-933.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
444 of ANSYS, Inc. and its subsidiaries and affiliates.
Input Files
4. Sorensen, C.D., and T. W. Nelson.“Friction Stir Welding of Ferrous and Nickel Alloys.” Friction Stir Welding
and Processing. Ed. M. W. Mahoney and R. S. Mishra. Materials Park: ASM International, 2007. 111-121.
5. Ozel, T., Y. Karpat, and A. Srivastava.“Hard Turning with Variable Micro-Geometry PcBN Tools.” CIRP Annals
- Manufacturing Technology. 57 (2008): 73-76.
6. Mishra, R. S. Friction Stir Welding and Processing. Ed. R. S. Mishra and M. W. Mahoney. Materials Park, OH:
ASM International, 2007.
• fsw.dat -- Input file for the friction stir welding (FSW) problem.
• fsw.cdb -- Common database file containing the FSW model information (called by fsw.dat)
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 445
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
446 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 29: Rocket Nozzle Extension Simulation: Operation
This is the second of two example problems that simulate a rocket nozzle:
• The first example, Rocket Nozzle Extension Simulation: Fabrication (p. 149), demonstrates how to simulate
the thermal stresses during the manufacturing stage of the nozzle.
• The second example problem that follows demonstrates how to simulate thermal stresses induced during
the operation of the rocket nozzle.
While both examples are based on the same geometry, the assumptions are different, therefore leading
to different meshes and different element types.
For this problem, it is assumed that the rocket has been launched and hot gases are flowing through
the nozzle, subjecting the inside and outside of the nozzle body to convection heat loading. The heat
loading leads to a significant thermal gradient through the thickness of the body that manifests as high
thermal stresses.
Solid thermal and structural elements accurately simulate the multiphysics of the problem. While a fully
coupled element could solve the problem, a loose coupling method is used instead. Because the body
material could be homogeneous or a layered composite, the simulation requires a solid element type
with both homogeneous and layered material capabilities.
29.1. Introduction
The shell element model gives accurate stresses in most regions; however, through-the-thickness stresses
are not as accurate, especially where the reinforcement joins with the nozzle body. Solid elements are
used for this analyses to improve the accuracy of through-the-thickness stresses. This problem therefore
demonstrates some of the features of the solid layered thermal elements (SOLID279). The geometry for
this example problem has already been meshed and stored in a cdb file.
For this example, it is assumed that the material behavior is orthotropic (both structurally and thermally).
As such, it is important to define material properties along certain orthogonal directions within the
elements. This underscores the need to define an element coordinate system within each element.
While there is a good argument for defining material directions independent of the underlying elements,
this is currently restricted by the available technology.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 447
Rocket Nozzle Extension Simulation: Operation
All elements have default element coordinate systems, but these defaults may not always be convenient.
Material directions could be misaligned with respect to the element coordinate system (ESYS) and, as
such, you may need to modify them. You can typically accomplish this with the following steps:
1. Define the element coordinate system - Due to rapidly changing curvature, each element in this model
must have its own element coordinate system defined. (Consider using LOCAL and EMODIF commands
for a given brick mesh.) As a result, the element z axis is aligned with the thickness direction, and the element
x axis is aligned with the curvature. This makes it very convenient to define material properties along pre-
ferred directions.
2. Adjust the element connectivity - Because solid elements are being used, you must adjust the element
connectivity so that the IJKL face is aligned with the element coordinate z axis. This ensures that the layer
definition is parallel to face 1 (the IJKL face normal n) of the element and is normal to the ESYS z axis.
(Consider using EORIENT to accomplish this for any arbitrary mesh with a defined element coordinate
system.)
Both of these steps have already been accomplished for the model provided in this example.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
448 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
Thermal stresses can be obtained using an element with TEMP and DISP degrees of freedom (DOFs)
that are fully coupled (strong coupling). Alternately, stresses can be obtained using a thermal run and
then a structural run (loose coupling). The advantages and disadvantages of these methodologies are
not fully detailed here. Instead, this example uses the loose coupling method to demonstrate the flex-
ibility that it offers.
In this example a static thermal analysis is run using element SOLID279, and then temperatures are
transferred to a static structural analysis using the LDREAD command. For the structural run, SOLID279
is converted into SOLID186. For the thermal run, it is assumed that the material can be homogeneous
or layered. A similar assumption is made for the structural run, therefore allowing you to use the tem-
peratures from either a homogeneous or layered run for a structural run, as illustrated in the figure
below:
A material could be treated as "homogeneous" for a thermal run and "layered" for a structural run. This
allows you the flexibility to mix and match the runs at your own discretion. A strongly coupled solution
would not provide this level of flexibility and freedom.
A thermal analysis is performed using a homogeneous material option, followed by an analysis using
the 4-layer material option. The solutions from these analyses serve as the loads for a structural analysis
(both homogeneous and 4-layer material option). A total of four structural solutions will be obtained.
To ensure an axisymmetric solution for the thermal run, the TEMP is constrained to be the same along
the theta direction (CE command). Axisymmetry DOF constraints are specified to ensure an axisymmetric
solution for the structural run (DSYM). While there are many different ways to accomplish axisymmetry
for a sector model, this method was selected for its ease of implementation.
29.3. Modeling
The nozzle extension consists of a main curved wall and a reinforcing ring close to the jet exhaust, as
shown in Figure 29.4: Nozzle Extension Geometry (p. 450)
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 449
Rocket Nozzle Extension Simulation: Operation
Both the wall and the reinforcing ring are made of layered composite materials. The model uses a single
orthotropic material for all layers; however, the material orientation varies from layer to layer. The ma-
terial is assumed to be homogeneous for both the thermal and mechanical analysis. This simplification
could speed up the computations and provide a reasonable solution.
Because the model exhibits symmetry, optimal computational efficiency is achieved by modeling only
a single 1° sector (the base sector) of the entire nozzle extension. The geometry model of the 1° sector
is shown in Figure 29.5: Nozzle Extension 1° Base Sector Geometry (p. 450), where (a) is the global geo-
metry and (b) is a detail view of the reinforcing ring.
Since the thermal solution exhibits cyclic symmetry for temperature, CE can be used to accomplish this.
The CE command could be used for the structural cyclic symmetry, but in this example, the DSYM
command is used to demonstrate the different ways to accomplish a task in ANSYS.
The reduced 1° geometry model is meshed with SOLID279 elements. Figure 29.6: Ring Element Plot (p. 451)
and Figure 29.7: Expanded Layer Representation of the Solid Mesh (p. 451) respectively show the mesh
and the layer solid representation of the solid mesh (/ESHAPE).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
450 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
Using the SOLID279 element it is possible to make the element behave like a homogeneous material
or a layered material (KEYOPT(3) = 0 or 1). Shell sections define the layers of the composite material.
(The SECTYPE command specifies the section type, and the SECDATA command specifies the material,
thickness, material orientation, and the number of integration points of each layer.) Two sections are
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 451
Rocket Nozzle Extension Simulation: Operation
created: one for the main wall and another for the reinforcing ring. The following table summarizes the
shell section properties:
Assume that the reinforcing ring is securely bonded to the outer surface of the main extension wall.
The secure bonding is easily simulated (CE/CP commands).
The material is assumed to be linear elastic and temperature-independent. The material has different
thermal expansion coefficients in the in-plane (X and Y) and through-the-thickness (Z) directions.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
452 of ANSYS, Inc. and its subsidiaries and affiliates.
Boundary Conditions and Loading
• Three different film convection boundary conditions for different regions of the model
A realistic thermal model may have film coefficients that were derived from empirical correlations.
Constant values are used here for simplicity. It is possible to specify heat flux boundary conditions instead
of convection boundary conditions.
For the structural model, three types of boundary conditions are used (see Figure 29.9: Structural
Boundary Conditions (p. 454)):
In addition to the aforementioned boundary conditions, thermal loads are applied to the structural
model (LDREAD). This accomplishes loose coupling, as described in the introduction.
For the structural analysis, other mechanical loads could be active that have been ignored (such as
pressure and shearing stresses due to gaseous flows). A realistic simulation would account for all possible
loads that could be major contributors to stress.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 453
Rocket Nozzle Extension Simulation: Operation
This is followed by a linear static analysis for the structural model using the homogeneous and layered
options (4 different analyses). As mentioned in the introduction, temperatures must be transferred from
the first two analyses to the last four analyses (LDREAD) when the loosely coupled method is used.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
454 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 455
Rocket Nozzle Extension Simulation: Operation
An additional complication to consider when postprocessing elements with layers, is that while regular
postprocessing commands can be used, the results will be based on the corner node values, and the
values within each layer will be ignored. For example, the PLNSOL,TEMP command will create a contour
plot based on the 8 nodes of the element, but it does not reveal how the temperature varies within
the 4 layers. To postprocess layer by layer, activate each individual layer, then issue the relevant post-
processing command.
The TEMP in layer 3 of the reinforcing ring is shown in Figure 29.13: Layer 3 Temperature Results (p. 457).
You can generate this by issuing the following command in /POST1:
Layer,3 !activate the layer for postprocessing
presol,bfe,temp ! plot the element stored layer temp
You can also force the layer TEMP onto the 8 corner nodes of the element and then contour the element
solution (LAYER). This ensures that the element solution plotted is the actual layer solution in the figure.
To view the layer heat flux and temp gradient issue the PLESOL command after the LAYER command.
By default the layer value is 0. This implies that the postprocessed quantities are top of top layer and
bottom of bottom layer.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
456 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The following figures show the displacement solution for the four analyses of the reinforcing ring. These
are all node-based plots. The maximum displacement in all cases occurs where the reinforcing ring is
bonded to the nozzle body.
The displacement solution for the layered cases do not vary significantly. As a result, using homogeneous
or layered thermal loading for the layered mechanical problem does not appear to have a significant
effect. A similar conclusion can be drawn for the displacement solution of the homogeneous cases;
however, the same conclusion may not be valid for stresses. Do not draw general conclusions based
on this model; instead, analyze each model based on its loading conditions and assumptions.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 457
Rocket Nozzle Extension Simulation: Operation
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
458 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Figure 29.18: Layer 1 Equivalent Stresses (p. 460) shows the equivalent stresses in layer 1. The figure
shows that stresses do not peak near the bonded region for layer 1 as might be expected. This under-
scores the need to analyze the solution layer by layer rather than element by element. Consider changing
the ring shape, material, number of layers, or layer orientations to shift the peak stresses to acceptable
levels.
When the equivalent stresses in layer 2, 3, and 4 are plotted, the location of peak stresses noticeably
shifts. This underscores the need to study each layer carefully and refrain from drawing immediate
conclusions about other layers. For example, peak stresses in layer 4 shift to the bonded region, which
could not have been predicted from the layer 1 solution alone.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 459
Rocket Nozzle Extension Simulation: Operation
29.8. Recommendations
To perform a similar type of analysis, consider the following hints and recommendations:
• Consider refining the mesh near bonded regions of the mesh, as well as near high curvature regions. This
recommendation is based on the observed peak stress regions of all the layers.
• By default all layer information is written to the results file. For large meshes it would be wise to select the
proper KEYOPT(8) setting to minimize storage requirements with the layered option for SOLID279 or SOLID186
elements. The downside of reducing the file size is that it will not be possible to postprocess the individual
layers. You will need to decide what is appropriate for your analysis.
• For a mesh that has not been swept, it is difficult to see if the element coordinate system z axis is normal to
the IJKL face of the layered solid SOLID279/SOLID279 or SOLID185/SOLID186 element. Consider using the
EORIENT command to accomplish this. This is emphasized in the introduction. Also keep the following
points in mind when performing a layered analysis:
– The element coordinate system z axis must also be normal to the IJKL face (face 1) .
• For complex material fibres, it is convenient to align the element coordinate system axis with the fibre dir-
ection. In this case, consider defining a unique element coordinate system for each element using a combin-
ation of LOCAL and EMODIF,ALL,ESYS commands as detailed in the introduction. In Figure 29.19: Single
Element Coordinate System (p. 461) below, only one LOCAL command is required. In Figure 29.20: Multiple
Element Coordinate Systems (p. 461) below, each element must have a distinct element coordinate system,
implying multiple LOCAL commands.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
460 of ANSYS, Inc. and its subsidiaries and affiliates.
Input Files
Y X
Y
Y
X
X
• nozzle.dat -- Input file for the rocket nozzle extension operation simulation.
• nozzle_cdbfile.cdb -- Common database file containing the rocket nozzle extension model information
(called by nozzle.dat)
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 461
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
462 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 30: Acoustic Analysis of a Small Speaker System
This example problem demonstrates the use of acoustic elements coupled with structural elements to
examine the performance of a speaker assembly.
For more information about acoustic fluid-structure interaction, see Accounting for Acoustic Fluid-
Structure Interaction (FSI) in the Acoustic Analysis Guide.
30.1. Introduction
A dynamic loudspeaker includes a speaker connected to a frame via a suspension system. The speaker
has a cone and voice coils that are excited by a driver magnet. The suspension system includes the
surround, which holds the cone centered, and the spider, which provides a restoring mechanism of the
cone to the frame.
When designing a speaker, material selection (material properties), component thickness, and other
parameters can be changed to alter performance characteristics over a frequency range. Ideally, constant
performance is obtained in the operating frequency range (bass, midrange, and treble).
The movement of the structure (speaker) imparts energy into a fluid (air) and generates acoustic waves.
The impedance of the fluid influences the structural behavior in turn. To properly characterize the
acoustic response, a coupled structural-acoustic analysis is performed, wherein structural and acoustic
equations are solved simultaneously.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 463
Acoustic Analysis of a Small Speaker System
The magnetic force characterization and the air inside the enclosure are ignored. Attention is focused
on the acoustic response of the system.
Components include the cone, surround, spider, frame, magnet, and voice coil. A 1/4 symmetry model
is used. The air inside the cabinet is ignored.
As shown in the following figure, the acoustic fluid elements are outside of the cabinet:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
464 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
Perfectly matched layer (PML) elements are used on the outermost region to absorb outgoing acoustic
waves.
30.3. Modeling
The 1/4 3-D model of the structure is created in ANSYS DesignModeler and meshed with SOLID186
elements. Flexible components such as the surround, cone, and spider are meshed with finer elements
because they are of high importance. Stiffer components such as the frame, magnet, pole piece, and
enclosure are meshed with coarser elements.
The acoustic elements used are FLUID220 and FLUID221. Both unsymmetric and symmetric formulations
are available for fluid-structure interaction (controlled via KEYOPT(1)). The symmetric algorithm (KEYOPT(1)
= 2) is used because it requires less memory and computational time. (If using the symmetric formulation,
all acoustic fluid elements must have this setting, regardless of whether the elements are adjacent to
structural elements.)
The PML region is specified using acoustic elements with KEYOPT(4) = 1. Approximately six elements
are included in the PML region. The region between the PML region and the structure, or the “buffer”
region, also includes approximately six elements. A minimum of four elements should exist in both the
buffer region and the PML region. Using more elements improves accuracy (at an increase in computa-
tional cost).
Vibroacoustic problems should use shared nodes between the structural and acoustic domains. Avoid
using dissimilar meshes at the interface and contact elements.
The force applied on the voice coil is modeled using SURF154 surface effect elements. A tangential load
per unit area in the element’s coordinate system is applied (KEYOPT(2) = 1).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 465
Acoustic Analysis of a Small Speaker System
In the following example input, an FSI flag is applied to predefined nodal component “FSI” to designate
acoustic faces that have fluid-structure interaction calculations activated:
d,PML,pres,0
esel,s,ename,,220,221
sf,FSI,fsi
allsel,all
cmsel,all
The acoustic elements must be selected first, as the FSI flag is applied only to acoustic element faces,
and not structural element faces that share the same nodes.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
466 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The sparse direct solver was chosen for this model. Vibroacoustic harmonic problems can also be effi-
ciently solved with the Quasi-Minimal Residual (QMR) iterative solver.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 467
Acoustic Analysis of a Small Speaker System
Note the peak response at 800 Hz, which coincides with the structural mode discussed in Structural-
Only Modes (p. 467). Because the frequency spacing is 20 Hz, additional frequency points would need
to be solved to obtain the actual peak in the harmonic analysis.
The SPL at approximately 50 mm from the speaker at 0° and 30° off-axis are shown this figure:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
468 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The acoustic response shows a maximum value at the structural resonance; however, the SPL drops
suddenly at 860 Hz.
The reference pressure for SPL calculations is taken from the real constant value, which defaults to 20
μPa. SPL in the time-history postprocessor (/POST26) can be plotted (NSOL).
SPL can be listed (PRNSOL command) or plotted (PLNSOL command) in the general postprocessor.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 469
Acoustic Analysis of a Small Speaker System
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
470 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Displacement of the cone should ideally generate the acoustic waves. Figure 30.3: Modal Analysis at
807 Hz (p. 467) shows that the structural resonance at 807 Hz involves the surround rather than the
cone, which explains why the 800 Hz SPL results show larger pressure values near the surround in Fig-
ure 30.6: SPL at 800 Hz (p. 470).
The resulting nonuniform frequency response plot shown in Figure 30.4: Displacement Frequency Re-
sponse Plot (p. 468) indicates that the surround design should be modified or the speaker should not
operate at this frequency.
In the following figure, the velocity magnitude is displayed as a contour plot, where the bottom of
acoustic elements adjacent to the cone are displayed (and only a subset of the fluid domain is shown
for the sake of clarity):
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 471
Acoustic Analysis of a Small Speaker System
The following figure shows a polar plot (PLFAR,PRES,SPLP) of SPL at 800 Hz:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
472 of ANSYS, Inc. and its subsidiaries and affiliates.
Recommendations
The plot of directivity can be used to determine how focused the acoustic signal is and to evaluate
speaker performance. Directivity influences how on-axis or off-axis listeners hear the sound generated
by the speaker.
30.8. Recommendations
To perform a similar type of acoustic analysis, consider the following hints and recommendations:
30.8.1. Harmonic Sweep and Mesh Generation
30.8.2. GPU Accelerator
If a large frequency range is solved, the lowest frequency solved has a large wavelength and the highest
frequency solved has a small wavelength. To solve for a large frequency range and satisfy the two re-
quirements stated above, it may be necessary to use very large PML and buffer regions (to address low
frequency requirements) meshed with very small elements (to satisfy high frequency requirements).
This can be computationally expensive.
Because the element size depends on wavelength, acoustic analyses benefit greatly from parametric
models where each frequency is solved with a mesh that is optimal for that configuration. Parametric
models can be generated using the APDL command language, or they can be generated within ANSYS
Workbench using ANSYS DesignModeler and Workbench Mechanical.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 473
Acoustic Analysis of a Small Speaker System
In this case, use of GPU Accelerator resulted in a solution time 1.76 times faster for two cores using this
particular hardware configuration.
• speaker.cdb -- Common database file for the original mesh used in the speaker analysis (called by
speaker.dat).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
474 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 31: Fitting Parameters for a Chaboche Kinematic Hardening
Model
This example problem demonstrates how to determine material parameters for a third-order Chaboche
kinematic hardening model using the curve-fitting tool.
A method is presented to estimate the initial parameters and obtain a least-squares best fit to the data.
The fitted parameters are validated by conducting uniaxial simulations using a single element and
comparing the results with the experimental data.
31.1. Introduction
The Chaboche model is a nonlinear kinematic hardening model commonly used to model the plastic
deformation of metals. A phenomenon associated with kinematic hardening is ratcheting, the progressive
accumulation of plastic strain during cyclic loading.
Ratcheting is observed, for example, in high-pressure piping systems in nuclear power plants subject
to seismic loading. A primary concern is that the accumulated plastic strain could lead to catastrophic
failure, even though the loading amplitude from individual cycles is not expected to cause failure.
31.2. Modeling
Commonly used plasticity models for ratcheting are based on the von Mises yield criterion and a kin-
ematic hardening rule. The von Mises yield criterion for kinematic hardening is:
(31.1)
where is the isotropic hardening yield stress, and is the center of the yield surface which is a
function of the stress tensor and back stress tensor :
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 475
Fitting Parameters for a Chaboche Kinematic Hardening Model
The back stress tensor results in a shifting of the yield surface in stress space. A bias in this shift over
repeated loading causes the progressive accumulation of plastic strain (ratcheting (p. 475)).
Experimental data and the curve-fitting tool are used to determine a set of material parameters for the
Chaboche kinematic hardening model. A third-order Chaboche kinematic hardening model is used, as
it provides sufficient variation to calibrate the nonlinear behavior of the material and can account for
ratcheting behavior. More parameters can be used but, for the data used in this problem, a third order
is sufficient.
(31.2)
(31.3)
where:
Each of the back stress terms of the Chaboche model have the form of an Armstrong-Frederick rule,
where the parameters represent a plastic modulus and serve as the parameters for history depend-
ence. For accurate ratcheting predictions, specify at least three components [3].
For a stable hysteresis of strain-controlled cyclic loading, the solution for the back stress in the uniaxial
direction is:
(31.4)
(31.5)
where:
The uniaxial yield stress is the sum of the initial yield stress and the back stress component in the uni-
axial direction, expressed as:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
476 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
(31.6)
Using Equation 31.4 (p. 476) and Equation 31.5 (p. 476) for plastic strains at or near the strain limit ,
Equation 31.6 (p. 477) for a third-order Chaboche model becomes:
(31.7)
The parameters must be determined so that the model closely matches the material behavior. One
method for doing so is to obtain data from stabilized strain-controlled and stress-controlled ratcheting
experiments and use that data with the curve-fitting tool to determine material parameters that minimize
the error between the data and the model predictions.
The nonlinear curve-fitting method used by the curve-fitting tool is an iterative process and requires
initial values for each of the material parameters. The success and quality of the fit depends on how
far the initial values are from the values that give a good fit. A procedure is developed using stabilized
hysteresis stress-strain data to determine a good estimate for the initial parameters.
Stabilized hysteresis stress-strain data is obtained by cycling a specimen under strain-controlled loading
with symmetric tension-compression amplitude. A cyclically stabilized hysteresis response is achieved
when a cycle of stress-strain response remains unchanged through subsequent cycles.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 477
Fitting Parameters for a Chaboche Kinematic Hardening Model
Three distinct regions are noted. Part 1 is the initial onset of yielding, Part 2 is the knee of the hysteresis
curve, and Part 3 is the constant modulus segment.
For fitting purposes, the individual components of back stress for the model (α1, α2, and α3) are chosen
to represent the three regions of the strain-controlled stable hysteresis loop. In each region, the corres-
ponding Ci is chosen to approximately match the plastic modulus. The γi parameters are chosen to ac-
commodate the history dependence defined by Equation 31.4 (p. 476) and the relationship defined by
Equation 31.7 (p. 477).
Using the described heuristic method, the initial material parameters can be estimated as follows:
2. C1 is the slope in Part 1 of the stress-strain curve at the transition from elastic to plastic deformation. This
value is approximately the plastic modulus at yielding.
3. Using Equation 31.4 (p. 476), γ1 should be large enough that the exponential term quickly diminishes so
that α1 is approximately constant outside of Part 1.
5. γ2 is calculated from the chosen C2 and the ratio C2 / γ2 that satisfies Equation 31.7 (p. 477).
7. γ3 is not included in Equation 31.4 (p. 476) through Equation 31.7 (p. 477) and is assigned a small positive
value.
The procedure was developed from a trial-and-error method used to fit the material parameters [4] and
usually gives a set of initial parameters that allows the curve-fitting tool to obtain a good fit.
The initial yield stress of the material is generally known, and in this problem it is fixed and does not
affect the error minimization performed by the curve-fitting tool.
γ3 is chosen as a small positive value because it does not enter into the closed-form equations, and
experience indicates that this is generally a good choice. While the fitting procedure iterates to a value
of γ3 that best fits the data, this value is often difficult to estimate based on stabilized hysteresis strain-
controlled data. The value of this parameter affects the amount of ratcheting in each cycle of a stress-
controlled experiment and is best determined using that data.
Although the procedure for estimating the initial material parameters was developed for stabilized
hysteresis strain-controlled data, it can be applied to a single cycle of stress-controlled data; however,
the quality of the resulting fit should be checked carefully.
The curve-fitting tool provides a comparison of the data used for fitting and the model's prediction. It
is good practice, however, to compare the experimental data with a simulation of the entire experimental
load history to validate the fitted parameters, as shown in this figure:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
478 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
To verify that the parameters from the curve-fitting tool give a good match to experimental data over
the entire loading history, uniaxial single-element tension-compression simulations are performed.
Because the initial yield stress for the material is well known, the value is fixed.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 479
Fitting Parameters for a Chaboche Kinematic Hardening Model
The value for γ3 does not significantly affect the model predictions for the stable hysteresis loop, and
it is best fit to cyclic stress ratcheting data. In this problem, the value is fixed because allowing it to
vary in the fitting procedure produced a poor set of material coefficients.
The following table shows the initial parameters and those determined via the curve-fitting tool:
Table 31.1: Initial and Fitted Parameters for the Strain-Controlled Experiment
31.3.1.2. Simulation
To simulate the loading history over the entire uniaxial strain-controlled experiment, a single SOLID185
element is used with quarter symmetry boundary conditions and uniaxial displacement in the Y direction.
Elastic properties are a Young’s modulus of 26300 ksi and a Poisson’s ratio of 0.302. The nodes are
given a displacement value to simulate a strain cycle with an amplitude of 0.01.
The following figure shows the loading history for the entire experiment:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
480 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The following table compares the simulated results at the maximum negative plastic strain to the ex-
perimental data:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 481
Fitting Parameters for a Chaboche Kinematic Hardening Model
Figure 31.5: Plot Comparing Strain-Controlled Experimental Data with Simulation Data
The model predictions indicate that little evolution occurs in the stress-strain plot after the initial cycle
and the hysteresis loop has stabilized. If the hysteresis continues to evolve over several loading cycles,
the model prediction from the curve-fitting tool will not correspond to the stabilized hysteresis predicted
by the model.
To alleviate the inconsistency, try inputting multiple copies of the same experimental data. The curve-
fitting algorithm can then minimize the error from the model prediction over multiple cycles.
Table 31.3: Initial and Fitted Parameters for the Stress-Controlled Experiment
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
482 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
While the heuristic procedure for determining the initial parameters was developed for stabilized hys-
teresis strain-controlled data, using the procedure on a single load cycle of stress-controlled data often
gives an adequate set of initial parameters. It may be necessary to adjust the initial parameters, however,
if a good fit is not evident after comparing the results from the fitted model to the experimental data.
31.3.2.2. Simulation
The same finite element model is used as that for the simulation of the strain-controlled experiment.
The loading is changed to a uniaxial stress cycle between 38.68 ksi and -25.45 ksi. The results from the
single element simulation are compared with the experimental data in the following figure:
Figure 31.6: Plot Comparing Stress-Controlled Experimental Data with Simulation Data
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 483
Fitting Parameters for a Chaboche Kinematic Hardening Model
Table 31.4: Initial and Fitted Parameters for Fit to Both Data Sets
The comparison of the experimental data sets and the model predictions data for the fit to multiple
data sets are shown in the following two figures:
Figure 31.7: Plot Comparing Stabilized Hysteresis Data to Fit of Multiple Data Sets
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
484 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Figure 31.8: Plot Comparing Stress Data to Fit of Multiple Data Sets
The stabilized hysteresis comparison shows a slightly better fit in the unloading knee and a change in
the data fit for the loading knee. The predictions at the maximum and minimum plastic strain points
are not significantly affected, however. The stress-controlled data comparison also shows a shift in the
reloading knee region, but the plastic strain prediction in the initial unloading point still shows significant
error.
The following figure shows data from Bari and Hassan [4] that illustrates the accumulation of plastic
strain over 45 cycles of loading:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 485
Fitting Parameters for a Chaboche Kinematic Hardening Model
Figure 31.9: Plot Comparing Ratcheting Strain Data to Prediction from Fitted Chaboche Model
The load cycle is the same as that used for the stress-controlled data (p. 482). Also shown are predictions
from the model using the parameters from the fit to multiple data sets, except that three different values
for γ3 are used. The plot shows that the rate of plastic strain accumulation after about 10 cycles of
loading is nearly linear with the value of γ3 controlling this rate of ratcheting.
For a value γ3 = 25, the rate of ratcheting approximately matches the experimental data; however, the
total amount of plastic strain from the model does not match the experimental data. One reason is that
the plastic strain in the early part of the load history is affected by the isotropic hardening behavior,
which quickly saturates after a few loading cycles.
Selecting a value for γ3 to match the ratcheting rate is a manually iterative process, as a fixed value is
set and a new fit to the other Chaboche parameters is obtained via the curve-fitting tool.
31.4. Recommendations
When fitting a Chaboche nonlinear kinematic model to experimental data using the curve-fitting tool,
consider the following recommendations:
• Use stabilized hysteresis strain-controlled data for estimating initial parameters. Refer to the heuristic pro-
cedure described in Determining Material Parameters (p. 477).
• Stress-controlled data can be used for estimating initial parameters, but the resulting fit should be carefully
compared to the experimental data. When using stress-controlled data, omitting the elastic-plastic transition
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
486 of ANSYS, Inc. and its subsidiaries and affiliates.
Input Files
region helps to limit the effect of isotropic hardening in the experimental data and can result in a better fit
for the kinematic hardening parameters.
• The experimental data should be arranged such that the first point is the initial (0,0) stress-strain point. Al-
though this value is not required in the experimental data, the fitting algorithm simulates the experimental
data starting from the (0,0) stress-strain point.
• Use a small nonzero value for γ3 for the initial estimate. If stress-controlled ratcheting data are available, γ3
can be manually adjusted or automatically fit to control the rate of ratcheting.
• The initial values of the constants C1, γ1, C2, γ2, C3, γ3 should be of decreasing order.
• Compare the results of the fitted model to the experimental data to verify the fit. If the fit is poor, modify
the values for the initial parameters and try the fit again.
• If the model cannot be made to fit the entire range of experimental data, it may be necessary to sacrifice
the fit quality in some regions of the loading history in order to fit other regions of the loading history.
• Use caution when extrapolating model predictions to loading histories outside of the available experimental
data, as the results may be of limited value.
31.5. References
The following references are used in this example problem:
1. Chaboche, J.L..“Constitutive Equations for Cyclic Plasticity and Cyclic Viscoplasticity.” International Journal
of Plasticity 5:3 (1989): 247-302.
2. Chaboche, J.L..“On Some Modifications of Kinematic Hardening to Improve the Description of Ratcheting
Effects.” International Journal of Plasticity 7 (1991): 661-678.
3. Rezaiee-Pajand M. and S. Sinaie.“On the Calibration of the Chaboche Hardening Model and a Modified
Hardening Rule for Uniaxial Ratcheting Prediction.” International Journal of Solids and Structures. 46 (2009):
3009-3017.CRC, (2005).
4. Bari, S. and T. Hassan.“Anatomy of Coupled Constitutive Models for Ratcheting Simulation.” International
Journal of Plasticity 16 (2000): 381-409.
• fit_chaboche_strain.dat -- Input file to fit the model to stabilized hysteresis strain-controlled exper-
imental data and simulate a uniaxial, strain-controlled experiment.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 487
Fitting Parameters for a Chaboche Kinematic Hardening Model
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
488 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 32: Anterior Cruciate Ligament (ACL) Simulation
This example problem simulates the response of an anterior cruciate ligament (ACL) of a human knee
subjected to tension, flexion, and rotation. The problem uses an anisotropic hyperelastic material
model with viscoelasticity.
32.1. Introduction
Sports and automobile accidents are significant causes of ligament injury, and injury to the anterior
cruciate ligament (ACL) is the most common ligament injury [1]. Of the many ligaments holding the
human knee joint together, the ACL is of special importance because it maintains knee stability and
helps to prevent damage to the knee by limiting joint motion [2].
The ACL is composed mostly of densely packed collagen fibers that connect the femur and tibia [6].
The ACL lies inside the knee joint, as shown in this figure:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 489
Anterior Cruciate Ligament (ACL) Simulation
Because of large deformations and rotations, the ACL exhibits highly nonlinear and anisotropic behavior,
and shows a clear strain rate dependency [1]. The anisotropic behavior is caused by collagen fiber
families arranged in a matrix of soft material.
Ligaments are viscoelastic and therefore display time-dependent and load-history-dependent mechan-
ical behavior [2]. The viscoelastic behaviors for ligaments have clinical significance as they help to prevent
fatigue failure of the ligaments.
The problem presented here uses an anisotropic hyperelastic material model with viscoelasticity to
model the anisotropy, nonlinear behavior, large strains, high compliance for bending loading and in-
compressibility. The exponential strain energy potential is used for anisotropic hyperelasticity with a
Prony series viscoelastic model.
As shown in Figure 32.1: Human Knee Components (p. 490), the ACL attaches to the knee-end of the
femur (at the back of the joint) and passes down through the knee joint to the front of the flat upper
surface of the tibia. It passes across the knee joint in a diagonal direction.
The primary objective of this simulation is to investigate the behavior of the ACL in relation to the
movement of the femur; therefore, a section of the ACL is constructed. The femur is modeled as a rigid
surface. The length of the ACL is specified as 31 mm. The ACL is assumed to exhibit visco-anisotropic
hyperelastic behavior. Several analyses are performed to simulate various femur movements.
The ACL has no stress-free state. At all knee flexions, stress is present in the ligament [2]. The precise
experimental distribution of the stress field is unknown; therefore, the simulation is performed without
prestressing the ACL.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
490 of ANSYS, Inc. and its subsidiaries and affiliates.
Material Properties
A constant strain rate is applied with displacement at the top of the ACL, while the lower part is fixed
at the bottom. The behavior of the ACL under knee flexion of 45 degrees and rotation of 11.25 degrees
is also examined.
32.3. Modeling
The ACL model geometry is similar to the geometry used by the Open Knee Project [8], but with different
dimensions. It is an approximate model with nonuniform geometry, where the midsection cross-sectional
area is smaller than that of the insertions. A typical ACL has bundles of collagen fibers, but the model
used here is a solid model with anisotropic material properties.
A tetrahedral mesh of the ACL is built with SOLID187 elements. A rigid surface is modeled to represent
the action of the femur. A standard contact behavior (KEYOPT(12) = 0) is defined between the femur
surface and the upper part of the ACL. A pilot node is also defined for the rigid surface to give the ro-
tations.
The exponential strain energy potential is used for characterizing the isochoric part.
where:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 491
Anterior Cruciate Ligament (ACL) Simulation
The constants a1, c1, and c2 are taken from Peña, et al. [1], and the compressibility parameter d is
considered to be small.
G
τ1 0.3 (sec)
α 2 0.4
G
G
τ2 9.0 (sec)
TB,AHYPER,1,,10,EXP
TBDATA,1, A1,A2,A3,B1,B2,B3
TBDATA,7, C1,C2,E1,E2
TB,AHYPER,1,,,PVOL
TBDATA,1, d
! Viscoelasticity
alpha1=0.30
alpha2=0.4
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
492 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
tau1=0.3
tau2=9.0
TB,PRONY,1,,2,SHEAR
TBDATA,1,alpha1,tau1
TBDATA,3,alpha2,tau2
The preferred orientation of the collagen fibers of the ACL induces the transversely isotropic symmetry
of the ligament [5]; therefore, only the direction vector aligned with the loading (-y) axis is considered
for performing the uniaxial extension simulation.
The femur is modeled with a rigid contact surface to enable modeling of the bending and twisting
behavior of the ACL. In the analyses, the tibia (bottom) side of the ACL model remain fixed and is con-
strained with all degrees of freedom.
The knee joint works in flexion, extension, and rotation. The simulations therefore show the behavior
of the ACL under the uniaxial flexion, tension, and rotation.
In the first case, a uniaxial displacement loading is applied at the top surface of the ACL with total
stretch of 1.12. A high constant strain rate of 50 percent per second with an elongation rate of 15.5
mm/sec is maintained. The following example input specifies the loading:
Ly = 31 ! length of the ACL (mm)
erat = 15.5 ! elongation rate (mm/s)
ts = 1.12 ! total stretch
dl = (ts-1)*Ly ! total displacement (mm)
tt = dl/erat ! total time (sec)
In the third case, a rotation of 11.25 degrees is applied using the femur surface in a small time period
of 1.5 seconds.
Large-deflection effects are included in the analysis (NLGEOM,ON). Automatic time stepping is in effect,
with an initial time increment of 0.001 seconds.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 493
Anterior Cruciate Ligament (ACL) Simulation
uniaxial tension at constant strain rate of 50 percent per second for anisotropic visco-hyperelastic and
Neo-Hookean viscoelastic models:
Figure 32.3: Von Mises Stress (Anisotropic Visco-Hyperelastic and Neo-Hookean Viscoelastic)
In the anisotropic visco-hyperelastic model, the peak stresses in the ACL are seen near the insertion
sites, suggesting that this area could be susceptible to injury.
Ligaments are highly anisotropic due to their fibrous structure [7]. The collagen fibers in the cruciate
ligaments of the knee are highly aligned with the long axis; therefore, an anisotropic model is required
to show the behavior of the ACL under deformation.
The strain rate during injury is a significant factor in determining the magnitude of the lesion. The stress-
strain behavior of the ligament is therefore essential in order to examine the importance of the strain
rates during the ACL movement.
The following figure shows the stress-strain behavior of the ACL under uniaxial tension for visco-aniso-
tropic, visco-hyperelastic, and Neo-Hookean viscoelastic material models along with the experimental
results of Pioletti et al. [3]:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
494 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The stress and strain values are the average values on the cross-sectional area of the ACL.
The anisotropic visco-hyperelastic ACL model used in this problem exhibits a response closer to the
experimental data when the ACL is subjected to uniaxial loading.
The following figure shows the stress- strain behavior of the ACL under uniaxial tension for the aniso-
tropic visco-hyperelastic model with differing strain rates:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 495
Anterior Cruciate Ligament (ACL) Simulation
At high strain rates, the results are closer to the experimental results. It is known that damage in mature
ligaments typically occurs at high load rates. The variation of ligament stress-strain behavior with strain
rate is an indicator of the viscoelastic nature of the tissue.
An important viscoelastic characteristic of ligaments is stress relaxation (a decline in stress over time
under a constant deformation), and so a relaxation test is simulated. A uniaxial deformation is applied
up to a stretch ratio of 1.12 for a very small time period (0.24 seconds) and is then allowed to relax for
50 seconds. The following figure demonstrates viscoelasticity in stress relaxation for the ACL model:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
496 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The following figure shows the stress variation in the ACL under knee flexion:
The femur surface rotates about the z axis, and the tibia side (the bottom part of the ACL) remains
fixed. During knee flexion, an increase in von Mises stress occurs near the minimum cross sectional area
of the ACL.
The femoral and tibial insertion zones are not considered, as this model emphasizes only the ACL beha-
vior irrespective of the insertion zones. If one also considers the insertion zones, more accurate stresses
near the femoral insertion zone can be expected.
The knee joint enables leg rotation along its longitudinal axis. The cruciate ligaments limit the internal
rotation [2]. The following figure shows the stress variation in the ACL under the knee rotation:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 497
Anterior Cruciate Ligament (ACL) Simulation
The femur surface rotates about the vertical y axis, and the tibia side remains fixed. The stress in the
ligament is increased with the internal rotation of the femur.
32.8. Recommendations
To perform a similar type of analysis using anisotropic hyperelasticity and viscoelasticity, consider the
following recommendations:
• The ACL has irregular and complex geometry; therefore, the ideal element type for ACL simulation is the
10-node tetrahedral element SOLID187. It is well suited to modeling irregular geometries (such as those
created using CAD/CAM systems).
• The fiber directions in the ACL can be defined using the vectors (A,B) in the anisotropic hyperelastic mater-
ial model. The ligaments are highly anisotropic because of their strongly unidirectional fibrous structure
and provide primary resistance to the tensile loading; therefore, only the direction vector aligned with the
long axis is considered.
• Although a simplified model of ACL is used in this problem, femur and tibia sides with insertion zones can
also be considered in an ACL simulation. In such a case, a fine mesh yields more accurate results. Too fine
a mesh, however, increases computational time; to minimize the computational cost, the femur and tibia
sides can be considered to be rigid.
32.9. References
The following references are used in this example problem:
1. Peña, E. et al.“An Anisotropic Visco-Hyperelastic Model for Ligaments at Finite Strains. Formulation and
Computational Aspects.” International Journal of Solids and Structures 44 (2007): 760-778.
2. Pioletti, D.P. (1997) Viscoelastic Properties of Soft Tissues: Application to Knee Ligaments and Tendons . Ph.D.
Thesis. École Polytechnique Fédérale de Lausanne: Switzerland.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
498 of ANSYS, Inc. and its subsidiaries and affiliates.
Input Files
3. Pioletti, D. et al.“Viscoelastic Constitutive Law in Large Deformations: Application to Human Knee Ligaments
and Tendons.” Journal of Biomechanics. 31 (1998): 753-757.
4. Song, Y. et al.“A Three Dimensional Finite Element Model of the Human Anterior Cruciate Ligament: A
Computational Analysis with Experimental Validation.” Journal of Biomechanics 37 (2004): 383-390.
5. Limbert, G., M. Taylor, J. Middleton.“Three Dimensional Modeling of the Human ACL: Simulation of Passive
Knee Flexion with a Stressed and Stress-Free ACL.” Journal of Biomechanics 37:11 (2004): 1723-1731.
6. Borotikar, B.S. (2009) Subject Specific Computational Models of the Knee to Predict Anterior Cruciate Ligament
Injury . Ph.D. Thesis. Cleveland State University: USA.
7. Haghpanahi, M. and F. Jalayer.“Three-Dimensional Finite Element Analysis of the Human ACL.” Paper
presented at the 1st WSEAS International Conference on Biomedical Electronics and Biomedical Informatics:
Rhodes, Greece (August, 2008).
8. Erdemir, A. and S. Sibole.“Open Knee: A Three-Dimensional Finite Element Representation of the Knee
Joint.” Computational Biomodeling Core and Department of Biomedical Engineering, Lerner Research In-
stitute, Cleveland Clinic: Cleveland, USA.
• acl_tension.dat -- Input file for the analysis of the ACL under uniaxial tension loading.
• acl_tension.cdb -- Database file for the finite-element model of the ACL under uniaxial tension (called
by acl_tension.dat).
• acl_flexion.dat -- Input file for the analysis of the ACL under flexion.
• acl_flexion.cdb -- Database file for the finite-element model of the ACL under flexion (called by acl_
flexion.dat).
• acl_rotation.dat -- Input file for the analysis of the ACL under rotation.
• acl_rotation.cdb -- Database file for the finite-element model of the ACL under rotation (called by
acl_ rotation.dat).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 499
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
500 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 33: Analysis of a Piezoelectric Flextensional Transducer in
Water
This example problem couples structural, piezoelectric, and acoustic elements to analyze the acoustic
response of a flextensional transducer to voltage excitation.
• Use of the Robin boundary condition to absorb outgoing acoustic waves in 3-D
33.1. Introduction
Flextensional transducers usually operate in the low- or mid-frequency range and are attractive because
they can generate high-power output. A Class IV type of flextensional transducer, shown in the following
figure, contains a stack of piezoelectric ceramics in an elliptically-shaped shell that is covered with
rubber to isolate it from the surrounding water:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 501
Analysis of a Piezoelectric Flextensional Transducer in Water
Voltage applied to the piezoelectric ceramics causes motion along the x axis (major axis), which in turn
causes amplified motion along the y axis (minor axis). This behavior produces large-volume displacement
and therefore high-power output. The Class IV flextensional transducer produces a roughly omnidirec-
tional radiation pattern, although there reduced output occurs along the minor axis.
Transducer design can include changing material properties or dimensions to obtain the desired resonant
frequency or response amplitude (quadrupole bending mode).
For this example, the flextensional transducer is idealized as a 2-D planar model and excited at 1400
Hz. Results of 2-D analyses are compared to a 3-D (2.5-D) model. The term 2.5-D describes a single layer
of extruded 3-D elements used to mimic 2-D planar behavior. The example also uses advanced 3-D
capabilities, such as wave-absorption conditions and far-field postprocessing, to model 2-D analyses.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
502 of ANSYS, Inc. and its subsidiaries and affiliates.
Problem Description
The quarter-symmetry model has 15 piezoelectric stacks (and 30 in the full model). The polarization
direction is defined as the element’s x axis (aligned with the global X axis), and the element coordinate
system x-axes alternate between each piezoelectric ceramic.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 503
Analysis of a Piezoelectric Flextensional Transducer in Water
33.3. Modeling
The 2-D (p. 504) and 3-D (p. 505) transducer models are created in ANSYS DesignModeler and meshed
in the ANSYS Mechanical application. The major axis length is 450 mm and the minor axis length is 210
mm.
The surrounding water is a circular region meshed with FLUID29 acoustic elements with a distance of
1.1 m. The truncated boundary defined by the 1.1 m circular arc is meshed with FLUID129 infinite fluid
elements to absorb outgoing acoustic waves.
The following figure shows the final mesh with an average element size of 10 mm:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
504 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
A rectangular region 300 mm from the sides of the transducer represents the surrounding water. The
model is extruded 10 mm in the out-of-plane (z-axis) direction to create a 2.5-D representation.
The following figure shows the final mesh with an average element size of 10 mm:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 505
Analysis of a Piezoelectric Flextensional Transducer in Water
Material
Material Property
Aluminum Rubber
-2
Young's Modulus (Nm ) 68.9E9 30E6
Poisson's Ratio .3 .49
-3
Density (kg/m ) 2710 1100
The water density is assumed to be 1000 kg/m3, and its speed of sound is 1500 m/s.
Following are the material properties of the piezoelectric ceramic (PZT8), where the polarization direction
is along the element x axis:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
506 of ANSYS, Inc. and its subsidiaries and affiliates.
Boundary Conditions and Loading
To active the vibroacoustic coupling, the FSI flag is specified to the nodal component N_FSI .
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 507
Analysis of a Piezoelectric Flextensional Transducer in Water
The direct sparse solver is used for this model. Because the number of degrees of freedom in the
model is not large due to the 2-D simplification of the transducer, the solution completes in a few
seconds.
Notice that while the maximum SPL is at the surface of the transducer’s minor axis, the SPL near the
truncated boundary is lower along the minor axis.
This SPL plot along an arc 1 m from the center shows the direction of the arc from the x axis towards
the y axis:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
508 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The following figures show the 2-D and 3-D plots followed by the two plots overlaid:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 509
Analysis of a Piezoelectric Flextensional Transducer in Water
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
510 of ANSYS, Inc. and its subsidiaries and affiliates.
References
The plot overlay indicates good correlation between the 2-D and 3-D models.
Note
The meshes for both cases were not identical, so some variation in the results was expected.
33.8. Recommendations
To perform a similar type of piezoelectric flextensional transducer analysis, consider the following hints
and recommendations:
• For both the FLUID129 infinite fluid element (2-D) and the Robin boundary condition (3-D), include at least
a quarter-wavelength of fluid in a given direction, where the wavelength is based on the lowest frequency
of interest.
• For linear (lower-order) fluid elements, use approximately 14-16 elements per wavelength, where the
wavelength is based on the highest frequency of interest.
33.9. References
Theory behind the analysis of a flextensional transducer can be found in this resource:
1. Kang, K. and Y. Roh (1980).“Optimization of Structural Variables of a Flextensional Transducer by the Stat-
istical Multiple Regression Analysis Method.” Journal of the Acoustical Society of America. Ph.D. Thesis. Uni-
versity of California, Berkeley: USA.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 511
Analysis of a Piezoelectric Flextensional Transducer in Water
• xducer2d.cdb -- Common database file for the 2-D mesh (called by xducer2d.dat).
• xducer3d.cdb -- Common database file for the 3-D mesh (called by xducer3d.dat).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
512 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 34: Dynamic Simulation of a Nuclear Island
This example problem is a power spectral density (PSD) analysis of the nuclear island component of a
nuclear power plant. The problem shows the analysis methods available to simulate the response of
the nuclear island to a seismic event, accounting for motion incoherency and wave-passage effects.
34.1. Introduction
Power spectral density (PSD) analysis is a widely used method for studying the responses of structures
subjected to a random input. It is a statistical measure defined as the limiting root-mean square (RMS)
value of a structure’s response. In this analysis approach, the magnitudes of the response can be specified
only by probability distribution functions that show the probability of the magnitude taking a particular
value. It is assumed that the random input has a zero mean and that its values are Gaussian distributed.
The problem presented here uses PSD analysis to simulate the response of a nuclear island (NI) com-
ponent of a nuclear power plant (NPP) to a seismic event, taking into account the partial correlation
between ground motions occurring at various locations of the basemat. This effect is known as ground
motion incoherency (GMI), and is of particular significance in cases where nearby faults generate short-
duration, high-frequency waves.
GMI consists of spatial variation of both horizontal and vertical ground motions. The horizontal spatial
variation of seismic ground motion is the result of the combination of three phenomena:
• Wave-passage effect, which is the difference in the arrival times of seismic waves at different locations.
• The incoherence effect, resulting from reflection and refraction of waves through the soil during their
propagation, as well as the superposition of waves arriving from an extended source at various locations.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 513
Dynamic Simulation of a Nuclear Island
Ground motions in recorded earthquake events exhibit spatial incoherency in high-frequency con-
tents.[1][2] It is therefore important to account for these effects to accurately predict the response of
the structure. By considering rock-like soil condition, the spatial variation due to both the wave-passage
effect and the incoherency effect can be observed.
The method applied here could also be used to estimate structure-borne sound radiated into buildings
such as concert halls when located within the immediate vicinity of railway lines. More generally, the
same method can be applied to any simulation involving an incident field and/or wave with known
statistical properties (involving frequency and space), such as an automotive windshield or a rocket
nozzle.
For more information, see the following topics in the Mechanical APDL Theory Reference:
• Spatial Correlation
• Wave Propagation
A nuclear island (NI) component of a typical modern nuclear power plant (NPP) is modeled. The model
consists of one reactor building and four auxiliary buildings, all located on a single basemat. Each
building encompasses five floor levels. Overall plan dimensions are 100 x 100 m2. The building height
range is 40-50 m.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
514 of ANSYS, Inc. and its subsidiaries and affiliates.
Problem Description
The structure is excited by a typical base motion generated by a nearby fault. The event generates a
high-amplitude, short-duration, high-frequency quake. Such loading is commonly applied in the form
of a ground motion response spectrum (GMRS), which provides the peak response of a series of single-
degree-of-freedom (SDOF) systems. The GMRS selected by Xu and Samaddar [1 (p. 533)] is used here
and is shown in the following figure:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 515
Dynamic Simulation of a Nuclear Island
The GMRS typical of eastern United States sites is characterized by a high energy content in the high-
frequency range (>10 Hz).
The GMI effects on responses can be incorporated using either the stochastic approach or the determ-
inistic approach [6 (p. 533)]. The deterministic approach is used here. The following coherency functions
describe the relationship between ground motions at separate locations as a function of the separation
distance and the frequency of the ground motion [2]:
The plane-wave coherency function is used for this example, as it applies to rock-like soil conditions.
The function is expressed as:
where γpw is the plane wave coherency representing the random horizontal spatial variation of ground
motion, F is the ground motion frequency, ξ is the separation distance between locations in meters,
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
516 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
and tanh is the hyperbolic tangent function. All other parameters are given in the following table for
horizontal and vertical ground motion:
Coherency functions are plotted as a function of frequency for various separation distances, as shown
in Figure 34.5: EPRI-TR 1014101 -- Plane Wave Coherency Model for Horizontal Direction (p. 520).
34.3. Modeling
The entire building is meshed using four-node structural shell elements (SHELL181). Shell thickness is
uniform (1 m) for the wall and slab, and the basemat thickness is 2 m.
Mesh size is based on the shortest wavelength to be resolved. In this case, the flexural wavelengths
can be predicted using the classical formula [3 (p. 533)]:
where λb is the flexural wavelength, B is the plate bending inertia per unit-length, ω is the circular fre-
quency, and m” is the plate mass per unit-length.
A cutoff frequency of 50 Hz is specified, which gives a bending wavelength of about 12 m. Using six
elements per wavelength is sufficient and results in a mesh size of 2 m.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 517
Dynamic Simulation of a Nuclear Island
Figure 34.3: NPP Structures FEM (Mesh Size Adapted to 50 Hz Cutoff Frequency)
To simplify the problem in order to better illustrate the method, the following assumptions apply:
Ignored effects include the effects of waves reflected back to the soil/structure interface, and waves
radiated following the structural motion. This assumption is conservative for very stiff, rock-like soil
situations. The loading can be modeled as an enforced displacement (or acceleration) at the bound-
aries.
A rigorous treatment would require a distinction between prestressed (external shell and dome) and
reinforced concrete parts damping. Because most of the structure is made of reinforced concrete,
uniform damping is applied. This assumption is valid because the high-frequency seismic events are
not expected to damage the building significantly. High-frequency events are less harmful to large,
ductile concrete structures, but more detrimental to smaller, stiffer objects such as secondary com-
ponents.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
518 of ANSYS, Inc. and its subsidiaries and affiliates.
Boundary Conditions and Loading
34.5.1. Loading
The probable SDOF system peak responses (response spectrum) is related to the PSD of the input signal
using the following relationship [5 (p. 533)]:
where G is the one-sided ground-motion acceleration PSD ((m/s2)2/Hz), ξ is the damping ratio, F is the
excitation frequency, Sa(F,ξ) is the acceleration response spectrum, and T0 is the earthquake duration
(s).
The acceleration PSD is required to run a single point (coherent) PSD analysis. In the case of multiple
partially coherent excitations, it is necessary to perform these additional tasks:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 519
Dynamic Simulation of a Nuclear Island
• Calculate the direct and cross participation of each pair of regions to the overall response.
A suitable region size should be selected based on the central frequency and the spatial coherency
model used. The coherence curve shown in the following figure is used to determine the minimum size
of the region:
Figure 34.5: EPRI-TR 1014101 -- Plane Wave Coherency Model for Horizontal Direction
By definition, the coherence function is unity for coherent input. The figure shows that
the coherence function is near unity at low frequencies (less than 5 Hz), and that it de-
creases with frequency and separation distance between the observations points in the
free-field. A decrease in response amplitudes for high-frequency vibration modes can
therefore be expected.
At a given frequency, two points experience correlated motion if the coherence value is close to unity
(or approximately above 0.6).
At a frequency of 15 Hz (the peak of PSD curve), two points distant by more than 5 m would exhibit a
coherency below 0.6. To be rigorous, a region size of 5 m should be used; for a 100 x 100 m square
basemat, 20 x 20 = 400 regions would therefore be required. Here, a 10 m region size is used to mod-
erate the problem size while capturing the primary phenomena.
To capture wave-passage effects, a region size smaller than the wavelength should be used. To simplify
the analysis, it is assumed that the NPP is excited by a wave exhibiting a horizontally propagating dir-
ection, with the wavelength equal to:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
520 of ANSYS, Inc. and its subsidiaries and affiliates.
Analysis and Solution Controls
where λ is the wavelength, Fcut is the cutoff frequency, and Cs is the soil shear wave velocity.
With a cutoff frequency of 50 Hz and a soil shear wave velocity of 3000 m/s, the resulting dominant
wavelength is 60 m. Based on a minimum of six regions per wavelength, the region size is input as 10
m.
Because no element results are needed, only nodal solutions (nodal displacements and rotations) are
requested.
As a general rule, plate-type structures exhibit constant modal density in the medium frequency range.
As shown in the following figure, the number of modes within a given frequency range is approximately
constant above 22 Hz:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 521
Dynamic Simulation of a Nuclear Island
/solu
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
522 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
For each region (elementary excitation), a unit nodal displacement is input (D), and then modal parti-
cipation factors are evaluated (PFACT). This is most easily achieved by using an APDL do-loop.
Elementary excitation must be defined for each region using PSD tables. The magnitude-vs.- frequency
data can be input with a PSD table (via PSDFRQ, which defines the frequency points of an input spectrum,
and PSDVAL, which defines the PSD values of an input spectrum). The PSD curve as defined in Fig-
ure 34.4: Ground Motion Power Spectral Density (Acceleration PSD) (p. 519) must be used as an input
spectrum. Additionally, the input motion has a unit of acceleration (defined via PSDUNIT).
For cross excitations, PSD co-spectral values are obtained by multiplying the original (direct) PSD values
with the plane-wave coherence value, as defined in Figure 34.5: EPRI-TR 1014101 -- Plane Wave Coherency
Model for Horizontal Direction (p. 520), frequency by frequency (COVAL).
For cases where wave-passage effects must be considered, cross PSDs have both a real part (defined
via COVAL) and an imaginary part (defined via QDVAL). Both terms are equal to the previous cross-
PSD value multiplied by a phasor of unit amplitude, and have an argument equal to the phase shift
between the two locations considered. The relationship can be expressed as [2]:
where PSDWavePassage is the co-spectral values of cross PSDs with wave-passage effect, PSDPlaneWave is
the spectral values of cross PSDs with no wave-passage effect, and dij is the distance between the two
locations i and j.
The distance dij depends on whether the incoherency or the wave-passage effects are being considered:
• For incoherency, dij is the Euclidian distance between the two locations (that is, the conventional distance).
• For wave-passage effects, dij is the distance along the propagation (orientation) vector. It is the dot product
of the vector dij.N, where dij is the vector between nodes i and j, and N is the unitary orientation vector de-
fining the propagation direction).
Because the absolute acceleration result is the only quantity of interest, only this single calculation is
requested (PSDRES).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 523
Dynamic Simulation of a Nuclear Island
(A) and (B) show that the motion incoherency effects on the nodal solutions are significant, and that
the overall response has been reduced. (B) and (C) reveal that the consideration of wave-passage
effects further reduces the overall response.
(A) Fully Coherent Input with No Wave Passage Effect
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
524 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 525
Dynamic Simulation of a Nuclear Island
The plots provide information about the overall response reduction only. To derive the frequency-by-
frequency reduction ratio, the acceleration response PSDs at various locations of the structure using
both coherent and incoherent inputs are compared.
The following figure shows a comparison of response PSDs obtained at the building centerline at five
different elevations (RPSD):
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
526 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 527
Dynamic Simulation of a Nuclear Island
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
528 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
• In Eastern United States rock sites, ground motions generally exhibit high energy content for frequencies
10 Hz and above. When high-frequency ground motion is applied to the NPP model, it is expected that out-
of-phase high-frequency modes will be randomly distributed across the foundation footprint in accordance
with the specified coherency function. The incoherency effect should therefore reduce the high-frequency
responses.
• The motion incoherency effects reduce the responses for all frequency ranges, especially in the high-frequency
range above 10 Hz. For frequencies below 10 Hz, the responses are virtually identical for incoherent and
coherent ground inputs.
• The combined effects of motion incoherency and wave-passage effects are significant, and they further reduce
the responses in the high-frequency range.
• The points situated farthest from the excitation location are affected more by incoherent input as a result
of the smearing effect due to numerous contributors.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 529
Dynamic Simulation of a Nuclear Island
Fully Coherent Input with Partially Coherent Input Partially Coherent Input
No Wave Passage Effect with No Wave Passage with Wave Passage Effect
Effect
Solution PSD Post- Solution PSD Post- Solution PSD Post-
Time Table Processing Time Table Processing Time Table Processing
(Sec) Time (Sec) Time (Sec) Time
(Sec) (Sec) (Sec)
R14.0 12 1 10 1785 21 1908 27852 68 31519
R14.5 10 1 6 927 21 723 14196 68 14196
The following figure shows the solution time (the covariance matrix calculation) required for the PSD
analyses with partially coherent input motion with and without the wave-passage effect:
The following figure shows the time required for the response PSD calculation (the POST26 elapsed
time) or the PSD analyses with partially coherent input motion with and without the wave-passage effect:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
530 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Figure 34.11: Comparison of Wall Time for RPSD Calculation (POST26 Postprocessing Time)
Both figures show that the elapsed time for solution and postprocessing is substantially lower in
Mechanical APDL R14.5.
It is of interest to evaluate the scalability of SMP on Mechanical APDL R14.5 using various numbers of
cores, with a maximum of 8 cores.
The following table shows the wall time of the PSD analyses with partially coherent input motion with
wave-passage effect for various numbers of cores:
Table 34.3: Comparison of Wall Time for Various Numbers of Cores (R14.5)
The following figure shows the solution time (covariance matrix calculation) required using various
numbers of cores:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 531
Dynamic Simulation of a Nuclear Island
Figure 34.12: Comparison of Wall Time of Solution for Various Numbers of Cores
The same trend is observed for the POST26 postprocessing wall time.
Computational time spent on solution and postprocessing for four cores is nearly half that of two cores,
meaning that a performance increase up of about 2x is achieved. Similar comparisons have also been
observed with eight-core and four-core runs. It can therefore be concluded that the scalability of the
SMP-based solver is good.
34.8. Recommendations
The following recommendations are suggested for PSD analyses that account for the effects of motion
incoherency and wave-passage effect:
• Select proper input ground motion based on site conditions. The input ground motion considered here is
for rock-like soil conditions.
• Choose coherency functions based on the knowledge of time lags between the locations per the site condi-
tions.
• Select a suitable region size based on the coherency model used, central frequency, and the problem type.
The size of regions should be such that they capture motion incoherency and wave-passage effects.
• Include a suitable number of modes in the analysis based on the cutoff frequency and the study of modal
density.
• Execute a few test runs using a minimal number of modes, as the cost of the analysis can grow rapidly. The
growth ratio is governed by the number of regions (or more specifically, the number of PSD tables) to the
power of 2, as well as the number of modes, also to the power of 2. The total analysis duration is therefore
predictable after a single test run. Parallelization is recommended.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
532 of ANSYS, Inc. and its subsidiaries and affiliates.
Input Files
34.9. References
The following references are used in this example problem:
1. Xu, J. and S. Samaddar.“Case Study: Effect of Soil-Structure Interaction and Ground Motion Incoherency
on Nuclear Power Plant Structures.” ASME 2009 Pressure Vessels and Piping Conference. 26-30 July 2009,
Prague. Vol. 8. Seismic Engineering, 2009. 369-377.
2. Abrahamson, N.“Program on Technology Innovation: Spatial Coherency Models for Soil-Structure Interac-
tion.” Electric Power Research Institute. TR 1012968. 2006.
3. Cremer, L., M. Heckl, and E.E. Ungar.Structure Borne Sound. 2nd ed. Berlin: Springer-Verlag, 1990.
4. American Society of Civil Engineers. Seismic Analysis of Safety-Related Structures and Commentary. 1999.
5. Preumont, A. Random Vibrations and Spectral Analysis. Dordrecht; Boston: Kluwer, 1994.
• EPRI_TR1014101.MAC -- Macro for calculating plane-wave coherency values, per EPRI-TR 101401
Tables 3-4 and 3-5.
PSD Analyses:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 533
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
534 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 35: Elastoplastic Creep Analysis of Lead-Free Solder Bumps
This example problem is a thermo-mechanical analysis of lead-free solder bumps subjected to cyclic
thermal loading.
35.1. Introduction
Creep is a rate-dependent material nonlinearity in which the material continues to deform under a
constant load. Creep occurs as a result of long-term exposure to a high level of stress that does not
exceed the yield strength of the material. Creep is more severe in materials subjected to heat for long
periods. The creep strain rate can be a function of stress, time, temperature, and neutron flux level.
Under constant load, the uniaxial strain-vs.-time creep behavior is illustrated in the following figure:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 535
Elastoplastic Creep Analysis of Lead-Free Solder Bumps
In the primary stage, which tends to occur over a comparatively short period, the strain rate decreases
with time. The secondary stage exhibits an associated constant strain. In the tertiary stage, the strain
rate increases rapidly until failure (rupture). Typically, the primary and secondary stages of creep are
usually of greatest interest.
In static or transient structural analysis, creep can be simulated via implicit creep models. Compared to
other creep methods, implicit creep is preferred as it is computationally faster and more accurate. The
primary and secondary stages of creep can be simulated using different implicit creep material models,
as shown in the following table:
The temperature fluctuations endured by flip chip packages can cause progressive damage in solder
joints. Damage accumulation beyond certain limits results in electrical failure. Such failures are typically
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
536 of ANSYS, Inc. and its subsidiaries and affiliates.
Problem Description
the result of thermal-expansion mismatches between the materials used [1]. The mismatches lead to a
complex deformation behavior, and are associated with irreversible, temperature- and strain-rate or
time-dependent inelastic characteristics, producing viscoplastic deformation in and around the solder
joints. Deformation behavior can be simulated via viscoplastic material models, or by a creep model
used with a plasticity material model.
In the electronics industry, a primary goal of thermo-mechanical analysis is to simulate the stress and
strain responses of the solder joint to better predict its service reliability. The problem presented here
is a thermo-mechanical analysis of flip chip packages using creep and plasticity material models.
Because the harmful health-related effects of lead has forced electronic manufacturers to reduce their
use of toxic heavy metals (including lead) in solders, thermo-mechanical analysis for solder joints becomes
increasingly important as reliable alternative solders to replace lead-based solders are explored [2]. This
example uses a lead-free solder 96.5Sn-3.5Ag.
Figure 35.2: Distribution of Solder Bumps in Full Flip Chip Model and Size of Solder Bump
The size of the chip is 6.95 mm × 6.95 mm ×0.6 mm. The size of the substrate is 14 mm ×14 mm × 0.65
mm. The distance between chip and substrate is 0.075 mm. The diameter of the upper pad, the diameter
of the lower pad, and the width of the solder bump are 0.12 mm, 0.144 mm, and 0.1507 mm, respectively.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 537
Elastoplastic Creep Analysis of Lead-Free Solder Bumps
To minimize the number of degrees of freedom, and therefore the computational resources required
for the thermo-mechanical analysis, a 1/8 symmetry of the full model is used. To determine the cyclic
thermal load effect on the solder joint, a generalized Garofalo creep model is used for the secondary
stage of the creep analysis.
35.3. Modeling
The following modeling work is necessary for this example problem:
35.3.1. Flip Chip Modeling
35.3.2. Contact Modeling
The completed model contains 57,134 nodes and 11,558 3-D 20-node hexahedron structural solid ele-
ments.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
538 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
Figure 35.4: Flip Chip Mesh with SOLID90 for Thermal and SOLID186 for Structural Analysis
Temperature profiles at different time steps, obtained from transient thermal analysis, are used in
nonlinear structural analysis to predict residual creep strains in solder bumps. The analysis requires the
same mesh used for thermal analysis. It is therefore necessary to convert thermal element SOLID90 to
structural element SOLID186 (ETCHG).
Contact and target elements CONTA174 and TARGE170, respectively, are compatible with structural
element SOLID186, so those elements are not changed. Contact element options can be modified if
necessary (KEYOPT).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 539
Elastoplastic Creep Analysis of Lead-Free Solder Bumps
Similarly, contact pairs are created between other parts of the flip chip.
The generalized Garofalo model is used to simulate creep behavior. The creep strain rate of the model
is expressed as:
where is the change in equivalent creep strain with respect to time, is the equivalent stress, T is
the temperature, C1 through C4 are the constants, t is the time at the end of the substep, and e is the
natural logarithm base.
The following experimental data are used with the curve-fitting tool for creep materials to find the
creep model constants:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
540 of ANSYS, Inc. and its subsidiaries and affiliates.
Material Properties
Figure 35.6: Creep Experimental Data [4-6] at Different Temperatures Used for Curve Fitting
The curve-fitting method is nonlinear, so an iterative process is used to obtain a good fit with the ex-
perimental data; therefore, initial values for each material parameter are required. The following input
finds the generalized Garofalo material constants:
/COM, Read experimental data saved in dat files
TBFT,EADD,1,CREEP,temp218.dat ! Read experimental data at 218 K temperature
TBFT,EADD,1,CREEP,temp273.dat ! Read experimental data at 273 K temperature
TBFT,EADD,1,CREEP,temp298.dat ! Read experimental data at 298 K temperature
TBFT,EADD,1,CREEP,temp348.dat ! Read experimental data at 348 K temperature
TBFT,EADD,1,CREEP,temp398.dat ! Read experimental data at 398 K temperature
Table 35.2: Generalized Garofalo Creep Constants from Curve Fitting and Reference
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 541
Elastoplastic Creep Analysis of Lead-Free Solder Bumps
For more information, see Creep Material Curve-Fitting in the Material Reference.
The following material properties [3] are used for the thermo-mechanical analysis of the flip chip:
Elastic and perfectly plastic material behavior is assumed. Bilinear isotropic hardening is used to define
the solder plasticity to compare results.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
542 of ANSYS, Inc. and its subsidiaries and affiliates.
Boundary Conditions and Loading
The temperature on the free surface of the chip is increased at a constant rate for 150 seconds to 398
K.
The same constant temperature is maintained in the next load step for 300 seconds, then cooled to
218 K in 300 seconds.
The following figure illustrates each heating phase and each cooling phase, and the constant temperature
maintained for 300 seconds between each phase:
Temperature degree-of-freedom constraints are used to apply loading on the chip surface, as shown
in the following input:
NSEL,S,LOC,Y,1.325 ! Select free surface nodes of chip
D,ALL,TEMP,218 ! Applied temperature
ALLSEL,ALL
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 543
Elastoplastic Creep Analysis of Lead-Free Solder Bumps
Issue the LDREAD command at every substep to read the temperature at the appropriate time. Change
the TIME value to the final solution time of a given substep. A linear variation in the solder temperature
between load steps is observed; therefore, the thermal load is applied at the end of each load step
rather than to each substep.
Rigid body motion is constrained with displacement of all degrees of freedom at the center of a substrate
bottom node.
The following input shows the solution commands used for the thermal analysis:
TRNOPT,FULL ! Specifies transient analysis options.
TIME,150 ! Final time for solution.
OUTRES,ALL,ALL ! Saved results at each substep.
SOLVE
The following shows the solution commands used for the structural analysis:
! RAMP (LOAD STEP 2)
NLGEOM,ON ! Includes large-deflection effects
RATE,ON ! Perform implicit creep analysis
TIME,150 ! Set load step end time
KBC,0 ! Loads are ramped
LDREAD,TEMP,,,TIME,0,'THERMAL','RTH'
SOLVE
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
544 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Figure 35.8: Temperature Distribution in Flip Chip at 3150 Seconds and 3750 Seconds
Temperature on the chip is reduced from 398 K to 218 K in the 12th load step, and increased from 218
K to 398 K in the 14th load step.
The temperature distribution is nearly uniform in all load steps, due to assumed perfect thermal contact
between parts of the flip chip and a high thermal contact conductance value. These result are used in
the structural analysis as thermal loading.
The reference structural analysis [3] results using the generalized Garofalo creep model constants are
compared to the structural analysis results with creep curve-fitting constants. The following two figures
show the displacement vector sum of both analyses:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 545
Elastoplastic Creep Analysis of Lead-Free Solder Bumps
Figure 35.9: USUM Results from Creep Curve Fitting Constants and Reference Constants at the
End of 12th Load Step (3150 Seconds)
Figure 35.10: USUM Results from Creep Curve Fitting Constants and Reference Constants at the
End of 14th Load Step (3750 Seconds)
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
546 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
All time-history results are plotted for solder element results at node 4112, positioned at the end of
outmost solder and near the chip, as shown in the following figure:
The equivalent stress result at the node is compared to the reference results. The sign for Von Mises
stress above room temperature is changed to represent compressive stresses developed due to resistance
in thermal expansion. Results from curve-fitting constants matching well with the reference result, as
shown in this figure:
Figure 35.12: Time History Post Process Results of Equivalent Stress: Comparison with Wen-Ren
Jong [3]
Due to unavailable thermal contact conductance and plasticity behavior used in the reference analysis,
the following assumptions apply to this problem:
• Elastic and perfectly plastic behavior with isotropic hardening in the structural analysis.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 547
Elastoplastic Creep Analysis of Lead-Free Solder Bumps
The results based on the curve-fitting creep constants are therefore compared to another analysis using
the reference creep constants and run with the same assumptions, as shown in this figure:
In the following figure, von Mises creep strain results from the curve-fitting constants match closely
with the material constants used in the reference solution. Elasticity and plasticity constants in both
models are identical, so von Mises elastic strain and von Mises plastic strain match exactly, as shown
in the following two figures:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
548 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The following animations show the von Mises creep strain and von Mises stress animation (respectively)
along with the time-history postprocessing of the results at node 4112.
Figure 35.16: Von Mises Creep Strain Animation of Solder with Time-History Results at Node
Number 4112
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 549
Elastoplastic Creep Analysis of Lead-Free Solder Bumps
Figure 35.17: Figure 17: Von Mises Stress Animation of Solder with Time-History Results at Node
Number 4112
One can observe the increase in creep strain with cyclic load steps.
35.8. Recommendations
To perform a similar elastoplastic creep analysis, consider these recommendations:
• Save the thermal analysis results (OUTRES) at the desired time steps. Use the temperature results for input
loading in the subsequent structural analysis.
• For better results, the stress/strain rate and temperature range of the application should fall within the
considered experimental data range. Using too much additional experimental data outside the required
range also leads to more complexity in finding material constants and reduces accuracy within the desired
range of strains. Engineering judgment is therefore necessary to select an experimental data range for curve
fitting.
• Because the curve-fitting procedure is a numerical technique for obtaining material constants without
considering the physics of the material model, material model knowledge is required to judge the constants
obtained from curve fitting in order to avoid solution inconsistency.
• For more information, see Rate-Dependent Plasticity (Viscoplasticity) in the Material Reference and Rate-
Dependent Plasticity (Including Creep and Viscoplasticity) in the Mechanical APDL Theory Reference.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
550 of ANSYS, Inc. and its subsidiaries and affiliates.
Input Files
35.9. References
The following references are used in this example problem:
1. Kornain, Z., A.Jalar, N.Amin, R.Rasid, C.S.Foong. "Comparative Study of Different Underfill Material on Flip
Chip Ceramic Ball Grid Array Based on Accelerated Thermal Cycling." American Journal of Engineering and
Applied Sciences 3.1 (2010): 83-89.
2. Shangguan, Dongkai.“Lead-Free Solder Interconnect Reliability.” Materials Park, OH: ASM International,
2005.
3. Jong, Wen-Ren, Chien-Chia Chiu, Hsin-Chun Tsai, Hsiu-Tao Chung.,“Elastic-Plastic-Creep Analysis of Lead
and Lead-Free Solder Bumps in FC.” Annual Technical Conference (ANTEC) of the Society of Plastics Engineers
(2005): 1665-1669.
4. Darveaux, Robert and Corey Reichman.“Mechanical Properties of Lead-Free Solders.” 57th Electronic
Components and Technology Conference (2007): 695-706.
5. Clech, John Paul.“Sn-Ag Properties and Creep Data.” Material Measurement Laboratory. Web. 15 January
2013. <http://www.metallurgy.nist.gov/solder/clech/Sn-Ag_Bulk.htm>.
6. Lau, John H. and Stephen H. Pan.“Creep Behaviors of Flip Chip on Board with 96.5Sn-3.5Ag and 100In Lead-
Free Solder Joints.” International Journal of Microcircuits and Electronic Packaging24 (2001): 11-18.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 551
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
552 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 36: VCCT-Based Crack-Growth Simulation of a Composite
Laminated T-Joint
This example problem demonstrates the virtual crack closure technique (VCCT) to simulate the interfacial
crack growth of a laminated T-Joint. The simulation involves growing an existing crack along a predefined
path using interface elements.
• Crack-growth simulation
36.1. Introduction
Composite T-joints are used in a wide variety of applications, from fuselage stiffeners in aerospace
structures to load-transmission joints between the hull and bulkhead in oceangoing vessels [1].
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 553
VCCT-Based Crack-Growth Simulation of a Composite Laminated T-Joint
The T-Joint consists of laminated panels connected via composite overlaminates and deltoids (gap
fillers).
Because the overlaminates and the deltoid make up the load-transmission path between a vessel’s hull
and bulkhead, the strength of the joints depends on the strength of both components. Due to the
variable quality of the interfaces and the presence of defects, overlaminates can be the primary source
of delamination [2].
The T-Joint simulation shows delamination propagation under a mechanical pull-off load case using
VCCT-based crack growth. The energy-release-rate calculations using VCCT are based on the assumption
that the energy needed to separate a surface is the same as the energy needed to close the same surface,
and therefore that the crack-growth criteria is the energy-release rate. Because cracks generally grow
along interfaces, VCCT-based crack-growth simulation is an effective technique for simulating interface
delamination of laminate composites.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
554 of ANSYS, Inc. and its subsidiaries and affiliates.
Problem Description
The dominant delamination occurs along the interface between the deltoid and overlaminate in the
curve area first and then grows up to the end of laminate. During the first delamination process, a
second delamination also starts between two panels which completely separate the panels. Based on
the study of T-Joint failure, therefore, two critical locations of higher interlaminar stresses are considered
as the initial failure points. For the simulation, two small cracks are defined at these critical locations,
as shown in this figure:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 555
VCCT-Based Crack-Growth Simulation of a Composite Laminated T-Joint
The fracture energy and interlaminar material strength are based on reference results [3].
• Defining the crack-growth set, fracture criterion, crack-growth path, and solution-control parameters
(CGROW).
36.3. Modeling
Tensile test simulations are performed on the models having the dimensions shown in part (a) of the
following figure:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
556 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
Figure 36.4: T-Joint Dimensions (a) and Cracks with Crack Tip IDs (b)
(a) (b)
Dimensions are based on the specimen used in the reference study [4].
Because plane strain is simulated, the finite element model is created using the 2-D structural solid
element PLANE182 (KEYOPT (1) = 2, KEYOPT(3) = 2).
The various components are bonded to each other using standard bonded contact definitions, except
along the crack paths.
Crack paths are defined using the interface elements INTER202. To avoid crack surface interpenetrations,
contact is also defined along the crack paths. The mesh around the crack paths is finer than around
other sections of the model, as shown in the following figure:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 557
VCCT-Based Crack-Growth Simulation of a Composite Laminated T-Joint
Figure 36.5: T-Joint Finite Element Model with Initial Cracks and Crack Paths
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
558 of ANSYS, Inc. and its subsidiaries and affiliates.
Material Properties
For VCCT-based crack-growth simulation, isotropic, orthotropic or anisotropic materials can be used
with several fracture criteria (including a user-defined option). Multiple cracks are also allowed.
For this simulation, a linear fracture criterion is defined for crack onset and subsequent crack growth.
The linear option assumes that the fracture criterion is a linear function of the Mode I (GI), Mode II (GII),
and Mode III (GIII) energy-release rates, expressed as:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 559
VCCT-Based Crack-Growth Simulation of a Composite Laminated T-Joint
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
560 of ANSYS, Inc. and its subsidiaries and affiliates.
Analysis and Solution Controls
• Grow the crack when the assumed fracture criterion is satisfied (p. 561) (CGROW).
For more detailed information, see VCCT Energy-Release Rate Calculation in the Fracture Analysis Guide
and VCCT-Based Crack-Growth Simulation in the Fracture Analysis Guide.
VCCT calculation requires the finite element mesh to be in the crack-extension direction. The crack tip
component, normal of crack plane, and crack-extension direction provide the data necessary for the
calculation:
Crack-plane and crack-extension directions are defined in relation to the global Cartesian coordinate
system by assuming the flat crack plane.
The following commands specify the solution controls for crack growth:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 561
VCCT-Based Crack-Growth Simulation of a Composite Laminated T-Joint
Initially, the stresses become high around the crack tips, especially near the crack (located at the curve
area). As the stress increases, the energy-release rate at the crack tip also increases and reaches the
critical value. Fracture occurs when the following fracture criterion is met:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
562 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
where fc is the fracture-criterion ratio. The recommended ratio is 0.95 through 1.05. The default is 1.0.
The following table shows the energy-release-rate values for the crack tips in the model at two different
time steps, before and at the time of crack growth:
(default =
1)
1 0.030178 7.17E-02 0.43018 0.66920333 f < fc
0.035178 0.79123 2.4774 5.11483333 f > fc
2 0.030178 0.073927 0.37991 0.62633333 f < fc
0.035178 0.45969 1.9227 3.455 f > fc
3 0.030178 0.055529 0.43103 0.61612667 f < fc
0.035178 0.29733 0.9005 1.8916 f > fc
4 0.030178 0.023272 0.67118 0.74875333 f < fc
0.035178 0.07832 1.6736 1.93466667 f > fc
The table data show that crack growth occurs first at crack tips 1 and 2, whereas crack growth at crack
tips 3 and 4 occurs a little later. This behavior indicates that the dominant delamination occurs along
the interface between the deltoid and overlaminate in the curve area first, then between the UD braid
and the deltoid. Also for crack tip 1, 2 and 4, the fracture criteria ratio is much higher than after crack
growth is initiated, indicating an unstable crack growth. To avoid overpredicting the maximum loading,
use a smaller minimum time step (DTMIN) value.
The following figure shows the energy-release rate increasing until failure occurs:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 563
VCCT-Based Crack-Growth Simulation of a Composite Laminated T-Joint
The figure shows the variation of reaction force with the displacement in the Y direction. The response
of the T-joint model at times t1, t2 and t3 indicates that the reaction force increases first and reaches
the maximum value before crack-growth initiation. After time t3, almost all cracks begin to grow. When
crack growth begins, the force drops suddenly and decreases steadily.
Delamination is easily observed by the deformation of interface elements or by hiding the interface
elements from the model in postprocessing.
The following figures show the delamination of the panels at various time steps:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
564 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 565
VCCT-Based Crack-Growth Simulation of a Composite Laminated T-Joint
The crack growth from crack tips 1 and 3 eventually merge, thereby separating the laminated panels.
The boundary conditions and loading are the same as in the VCCT model, and predefined crack models
are used with the same crack sizes.
The following table shows the input parameters for the CZM model:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
566 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The following figures show the Y-component stresses in the VCCT and CZM models with identical
meshes.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 567
VCCT-Based Crack-Growth Simulation of a Composite Laminated T-Joint
Similar to the VCCT model, the debonding in CZM starts from the curved crack portion, then merges
with the debonding from the horizontal cracked portion, thereby separating the laminated panels.
The following figure shows the force-deflection (Y-force vs. Y-displacement) response of delamination
for the two models:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
568 of ANSYS, Inc. and its subsidiaries and affiliates.
Recommendations
Figure 36.13: T-Joint Y-Force vs. Y-Displacement: VCCT and CZM Models
For both models, the force increases with the applied displacement and peaks quickly before the crack
begins to grow. The reaction force then decreases rapidly at the initial phase of crack growth, then
slows with the subsequent crack growth.
A slight difference in the results is observed because the fracture criterion used for VCCT crack growth
is linear and critical fracture energy based only, while the delamination for CZM model is based on in-
terlaminar strengths along with critical energy. An artificial damping coefficient also affects the conver-
gence of the CZM model. Furthermore, in the VCCT model, the nodes separate instantaneously when
the fracture criterion is reached, implying that the load-carrying capacity drops more rapidly than it
does in the CZM model.
36.8. Recommendations
When setting up a VCCT-based crack-growth analysis, consider the following hints and recommendations:
• Differences in the size of the elements ahead of and behind the crack tip/front affect the accuracy of the
energy-release-rate calculation. Use equally sized meshes for elements along the predefined crack path
whenever possible.
• The mesh size itself also affects the solution. Examine mesh-size convergence before attempting the finite
element solution.
• To ensure accuracy of the energy-release-rate calculation, define the crack extension carefully.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 569
VCCT-Based Crack-Growth Simulation of a Composite Laminated T-Joint
– The strain energy released when a crack advances by a small amount is the same as the energy required
to close the crack by the same amount.
– The crack tip field/deformation at the crack tip/front location is similar to that which occurs when the
crack extends by a small amount.
Neither assumption applies when crack growth approaches the boundary, or when
the two cracks approach each other; therefore, when using the VCCT calculation
for problems such as T-joints where the two cracks approach each other, examine
the analysis results carefully.
• In the case of unstable crack growth or a rapid extension of the crack, specify smaller DTMAX and DTMIN
values (CGROW) to allow time for load rebalancing.
36.9. References
The following references are used in this example problem:
1. Dharmawan, F., S. Rodney, et al.“Geometry and Damage E ects in a Composite Marine T-Joint.” Composite
Structures 66 (2004): 181-187.
2. Dharmawan, F. (2008) Structural Integrity and Damage Tolerance of Composite T-Joints in Naval Vessels . Ph.D.
Thesis. School of Aerospace, Mechanical and Manufacturing Engineering, RMIT University: Melbourne,
Victoria, Australia.
3. Chen, J., Eric Ravey.“Prediction of Delamination in Braided Composite T-Piece Specimens.” Journal of
Composites Science and Technology 69 (2009): 2363–2367.
4. DiBella, G., C. Borsellino.“Experimental and Numerical Study of Composite T-Joints for Marine Applications.”
International Journal of Adhesion and Adhesives 30 (2010): 347-358.
• vcct2d.dat -- Input file for the VCCT-based 2-D analysis of the T-joint.
• vcct2d.cdb -- Common database file for the VCCT-based 2-D finite-element T-joint model (called by
vcct2d.dat).
• czm2d.dat -- Input file for the CZM-based 2-D analysis of the T-joint.
• czm2d.cdb -- Common database file for the CZM-based 2-D finite-element T-joint model (called by
czm2d.dat).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
570 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 37: Bolt Thread Simulation
This example problem demonstrates a simplified modeling technique for bolt thread simulation via the
bolt section method. The method delivers approximate results with the near accuracy of the true
threaded bolt model, but without the need for detailed thread geometry and refined mesh discretization.
The bolt section method also offers a significant savings in simulation time.
• Modeling of 2-D/3-D bolt thread by assigning the bolt section to contact elements.
37.1. Introduction
A bolted joint is used to hold two or more parts together to form an assembly of a mechanical structure.
In order to achieve expected physical behaviors of a bolted joint structure, a detailed three-dimensional
bolt model, which fully includes the bolt pretension effect and the frictional behavior at contact interfaces,
is desirable. However, for a large, complex structure, detailed modeling of bolted joints is difficult because
of problem size restrictions and computational costs associated with analyzing the entire structure.
The bolt thread modeling technique available for 2-D and 3-D contact elements offers simplified mod-
eling with accuracy close to that of a true threaded bolt model. Bolt threads can be simulated by assign-
ing a bolt section (defined by the SECTYPE command) to contact elements that are overlaid on a
smooth cylindrical bolt surface. To approximate the behavior of the bolt, calculations are performed
internally based on the user-specified thread geometry data and the end points of the bolt axis (input
via the SECDATA command).
The bolt thread modeling technique is useful for system level modeling where the primary bolt function
is to transfer a load. The absence of geometric details and refined mesh discretization in the thread
region also makes it a computationally inexpensive method. This technique can be applied to 3-D
models and 2-D axisymmetric models
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 571
Bolt Thread Simulation
A pressure load of 50 MPa (which is less than the equivalent pretension load) is applied to the upper
surface of the cover plate after applying pretension to the bolt. The resulting bolt shank stress (stress
in the region between the bolt head and the bolt thread) due to the pretension load and the inclusion
of frictional contact behavior are the major concerns during the bolt simulation.
The objective of this problem is to show that the bolt section method simplifies the modeling of this
bolt joint and produces approximate thread behavior and shank stress that are comparable to the true
threaded bolt model.
This method is the most accurate bolt simulation. The detailed modeling of threads provides accurate
thread behavior in the model. Very refined mesh discretization is required in the thread region,
which makes this method computationally expensive.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
572 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
In this method, the bolt thread is simulated by assigning a bolt section to the contact elements
overlaid on the smooth cylindrical bolt surface. (No detailed thread geometry is required.) Calculations
are performed internally based on the thread parameters given by the SECDATA command to ap-
proximate the behavior of the bolt. This method is computationally inexpensive.
In this method, MPC bonded behavior is defined in the thread region. (No detailed thread geometry
is required.) This method is very fast computationally, but thread behavior can be lost.
Both 2-D axisymmetric and 3-D models are used to compare these three methods. The 2-D model setup
for all three methods is shown in the figure below.
Figure 37.2: 2-D Axisymmetric Problem Setup for (a) True Thread, (b) Bolt Section, (c) MPC Bonded
Method
37.3. Modeling
An M120 structural steel bolt with standard thread dimensions is modeled with a cover plate and a
base plate of reasonable dimensions. Both 2-D and 3-D bolt thread modeling is performed. A bilinear
isotropic plastic material model is used for the bolt and the plates.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 573
Bolt Thread Simulation
37.3.1. Model the Bolt with Cover Plate and Base Plate
Two models, one with a threaded bolt surface and another with a smooth bolt surface, are created to
demonstrate the simplicity and advantage of the bolt section method over the true thread simulation
method.
37.3.1.1. Model the True Threaded Bolt with Cover Plate and Base Plate
The 3-D threaded bolt model represents a single start M120 bolt with a cover plate and a base plate.
The bolt has a maximum diameter of 120 mm, a pitch diameter of 116 mm, a 6 mm pitch, and a half
thread angle of 30 degrees (according to the standard thread profile).
The model is meshed with SOLID186 and SOLID187 elements. Mesh refinement is performed in the
thread region.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
574 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
Figure 37.4: 3-D Half Model of a Threaded Bolt Joint with Mesh Details
The 2-D surface geometry is extracted from the 3-D thread geometry. The 2-D model is meshed with
PLANE183 axisymmetric elements (KEYOPT(3) = 1).
Figure 37.5: 2-D Axisymmetric Model of a True Threaded Bolt Joint with Mesh Details
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 575
Bolt Thread Simulation
37.3.1.2. Model the Simplified Bolt with Cover Plate and Base Plate
Instead of modeling a true threaded bolt, smooth cylindrical surfaces are created for both the bolt and
base plate.
A 2-D axisymmetric model is created and meshed with PLANE183 axisymmetric elements (KEYOPT(3)
= 1).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
576 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
Figure 37.7: 2-D Axisymmetric Model of a Simplified Bolt Joint with Mesh Details
The 3-D model is generated with SOLID186 elements by extruding the 2-D axisymmetric model about
the Y-axis with the EEXTRUDE command.
eextrude,,30,,,360 !creates 3-D model by rotating 2-D axisymmetric model about y-axis
Figure 37.8: 3-D Half Model of a Simplified Bolt Joint with Mesh Details
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 577
Bolt Thread Simulation
The following steps and commands illustrate the procedure to simulate pretension in the bolt:
1. Mesh the bolt, then cut the mesh and insert PRETS179 elements to form the pretension section. The PSMESH
command is used to create pretension elements in the bolt at y = 260 mm.
ESEL,S,TYPE,,<type number> !selection of bolt elements by element type number
PSMESH,,,,ALL,,,y,260,,,,ECOMP,NCOMP !automatically generate the pretension elements @ y = 260
2. The SLOAD command is used to apply the pretension load. In the first load step, the pretension load is
applied as a force to node K. In the second load step the force locks, allowing additional loads. The effect
of the initial load is preserved as a displacement after it is locked.
SLOAD,1,PL01,lock,FORC,2544690,1,2
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
578 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
37.3.3.1.1. Contact in the Thread Region for the True Thread Simulation Method
A frictional contact pair is defined between the bolt thread surface (the contact surface) and the base
plate thread surface (target surface). The surface projection contact detection method (KEYOPT(4) = 3)
is used because it provides more accurate contact tractions and stresses for underlying elements com-
pared with other contact detection methods.
Figure 37.10: Contact and Target Surfaces at the Bolt Thread and the Base Plate Thread
37.3.3.1.2. Contact in the Thread Region for Bolt Section Simulation Method
Contact elements (CONTA174 for the 3-D case, CONTA172 for the 2-D case) are generated on the smooth
cylindrical bolt surface. A bolt section is assigned to the contact elements to simulate the thread. The
contact region is computed internally based on the user-specified thread geometry data and the two
end points of the bolt axis. Target elements (TARGE170 for the 3-D case, TARGE169 for the 2-D case)
are overlaid on the smooth cylindrical hole of the base plate.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 579
Bolt Thread Simulation
Figure 37.11: Contact and Target Surfaces at the Bolt Surface and Base Plate
The SECTYPE and SECDATA section commands are used to define the bolt section for contact elements.
The format for these commands is shown below.
SECTYPE,SECID,CONTACT,BOLT
SECDATA,Dm,P,ALPHA,N,X1,Y1,Z1,X2,Y2,Z2
where
Dm = Mean pitch diameter, dm
P = Pitch distance, p
ALPHA = Half-thread angle, α
N = Number of starts (defaults to 1)
X1, Y1, Z1, X2, Y2, Z2 = Two end points of the bolt axis in global Cartesian coordinates
The following commands illustrate the procedure for bolt thread modeling using the bolt section
commands:
et,100,174 ! Define contact element type
keyopt,100,4,3 ! Define surface projection based contact detection method
keyopt,100,10,2 ! Update stiffness at each iteration
mp,mu,100,.15 ! Frictional co-efficient
sectype,5,contact,bolt
secdata,116,6,30,,0,0,0,0,162,0
et,101,170 ! Define target element type
r,100,
cmsel,s,bolt_thread,node
type,100
mat,100
real,100
secn,5 ! Assign the section to contact elements
esurf
cmsel,s,bottomplate_thread,node
type,101
mat,100
real,100
secn,0
esurf
allsel,all
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
580 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
37.3.3.1.3. Contact in the Thread Region for the MPC Simulation Method
In this method, MPC bonded contact behavior is defined between the smooth cylindrical bolt surface
and the smooth base plate. There is no thread behavior defined. To define MPC contact, the following
contact elements KEYOPT settings are used:
37.3.3.2. Contact Between the Bolt Head and the Cover Plate
A frictional contact pair is defined between the bolt head (the contact surface) and the cover plate (the
target surface). The surface projection contact detection method (KEYOPT(4) = 3) is used for this contact
pair.
Figure 37.12: Contact and Target Surfaces at the Bolt Head and Cover Plate
37.3.3.3. Contact Between the Cover Plate and the Base Plate
A frictional contact pair is defined between the cover plate (the contact surface) and the base plate
(the target surface). The surface projection contact detection method (KEYOPT(4) = 3) is used for this
contact pair.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 581
Bolt Thread Simulation
Figure 37.13: Contact and Target Surfaces at the Cover Plate and Base Plate
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
582 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
• In the first load step, pretension loading of 2544690 N is applied to the pretension section through the
SLOAD command.
sload,1,p101,lock,forc,2544690,1,2 ! pretension load of 2544690 N
• In the second load step, a pressure load of 50 MPa (which is less than the equivalent pretension load) is
applied to the upper surface of the cover plate.
sf,sfload_nn,pres,-50 ! pressure load of 50 MPa on upper surface of cover plate
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 583
Bolt Thread Simulation
Figure 37.15: UY Displacement in the 2-D Axisymmetric Model for all Three Methods
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
584 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Figure 37.16: von Mises Stress in the 2-D Axisymmetric Model for all Three Methods
The stress plot of the thread region for the bolt section method closely matches the stress plot for the
true thread simulation, and the MPC method plot varies. The comparison between stress plots
demonstrates that accurate simulation of thread behavior and shank stress in the bolt can be achieved
by using the bolt section method.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 585
Bolt Thread Simulation
Figure 37.17: Linearized Stress in the Bolt Shank Along a Path at y = 280 for all Three Methods.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
586 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Figure 37.18: Contact Status in the Thread Region for all Three Methods
Figure 37.19: UY Displacement in the 3-D Half Model for all Three Methods
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 587
Bolt Thread Simulation
Figure 37.20: von Mises Stress in the 3-D Half Model for all Three Methods
Figure 37.21: von Mises Stress in the Bolt for all Three Methods
The stress plot of the thread region for the bolt section method closely matches the stress plot for the
true thread simulation, and the MPC method plot varies. This validates the conclusion that the bolt
section method achieves accuracy near to that of the true thread simulation.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
588 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Figure 37.22: von Mises Stress in the Bolt Shank at Section y = 280 for all Three Methods
Table 37.3: Simulation Time and Cumulative Iterations for 2-D Models
Table 37.4: Simulation Time and Cumulative Iterations for 3-D Models
Comparisons of von Mises stress and computation time demonstrate that accurate simulation of thread
behavior and shank stress results in the bolt can be achieved by the bolt section method, with the added
benefit of significant savings of simulation time.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 589
Bolt Thread Simulation
37.8. Recommendations
The following points are important for bolt thread modeling:
• The bolt cylindrical surface should be the contact surface, and the bolt section should be assigned to the
contact elements.
• Coordinates of the axis data points (required for SECDATA) should be on the bolt axis and defined in the
global coordinate system.
• In order to achieve better accuracy, the mesh size in the thread region should be less than the pitch distance.
• This technique is valid only for standard, straight threads. It is not applicable for nonstandard threads such
as tapered threads or buttress threads.
• The maximum stress in the thread region may vary with mesh density. However, the overall stress distribution
remains similar in pattern.
• boltSimulation_2-D — Input file for 2-D axisymmetric analysis of bolt thread simulation for all three
methods.
• boltSimulation_3-D — Input file for 3-D full model analysis of bolt thread simulation for all three
methods.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
590 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 38: Large-Deformation Neo-Hookean Analysis (via UserMat
Subroutine)
This problem formulates a 3-D large deformation, hyperelastic material to demonstrate the user-defined
material capability in nonlinear geometry analyses. Details are given for stress and material tangent
calculations and formulation in a co-rotated frame, as well as conversion of tensor quantities to Voigt
notation. The problem also serves as a usage example of the UserMat subroutine, a user-programmable
feature for developing a custom, constitutive model to define the stress-strain behavior of a material.
38.1. Introduction
The UserMat subroutine is a user-programmable feature for developing a custom, constitutive
model to define the stress-strain behavior of a material. User-defined materials can be used with current-
technology elements that have a displacement-based element formulation.
Called at each element integration point, the incremental and total deformations are passed into the
UserMat subroutine, along with stress and state variables from the previous increment. You must de-
velop an algorithm which updates the material state and stress at the end of the increment, as well as
calculates the algorithmic material tangent. In analyses with nonlinear geometry (NLGEOM,ON), the
user-defined material is formulated in a co-rotational frame, and the stress and algorithmic tangent
must be calculated in this co-rotated frame.
To demonstrate the user-defined material model, the shearing of a 1 x 1 x 1 block is simulated, and the
effective stress versus effective strain is reported. The model is recast in the co-rotated frame (required
by the user-defined material subroutine), and an algorithm for the determination of the co-rotated
frame rotation vis polar decomposition of the deformation gradient is also presented. The stress, mater-
ial tangent, and change of basis operations from are converted from full tensor form to Voigt notation.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 591
Large-Deformation Neo-Hookean Analysis (via UserMat Subroutine)
38.3. Modeling
The following modeling topics are available:
38.3.1. Neo-Hookean Model
38.3.2. Co-rotational Frame
38.3.3. Voigt Notation
where:
(38.3)
(38.4)
where:
(38.5)
where:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
592 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
(38.6)
38.3.1.1. Stress
Substituting the 1st-order derivatives of the potential from Equation 38.5 (p. 592) into Equation 38.2 (p. 592),
the 2nd Piola-Kirchhoff stress becomes:
(38.7)
(38.8)
where:
Substituting the 1st- and 2nd-order derivatives from Equation 38.5 (p. 592) and Equation 38.6 (p. 593) into
Equation 38.3 (p. 592), the material tangent stiffness is:
(38.9)
(38.11)
where:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 593
Large-Deformation Neo-Hookean Analysis (via UserMat Subroutine)
where:
The stress and tangent stiffness tensors are returned in the co-rotated frame.
where IU, IIU, and IIIU are the principle invariants of U. The stretch tensor is given in closed form by:
(38.15)
The eigenvalues of U are the square root of the eigenvalues of C. The principal invariants of U in terms
of the eigenvalues of C are:
where λ1, λ2, and λ3 are the eigenvalues of C given by the roots of the characteristic polynomial:
(38.16)
C is a symmetric, positive-definite, rank 3 matrix. The characteristic polynomial has three positive real
roots given by:
(38.17)
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
594 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
38.3.2.2. Stress
The constitutive model defined by the Neo-Hookean hyperelastic material gives the stress in the spatial
configuration by direct evaluation of Equation 38.8 (p. 593). No rate form is used and no integration is
required. The Cauchy stress in the co-rotated frame is returned by the user-defined material subroutine.
where:
The superimposed dot indicates the time derivative in the reference configuration.
Using the Jaumann rate, the constitutive response in the co-rotated frame is given by:
(38.20)
where cJ is the Jaumann tangent stiffness tensor and is implicitly defined by:
(38.21)
(38.22)
In component form:
(38.23)
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 595
Large-Deformation Neo-Hookean Analysis (via UserMat Subroutine)
where:
Minor symmetry with respect to the index pairs is required for this conversion. The 4th-order identity i
and the Jaumann tangent stiffness lack minor symmetry. To ensure minor symmetry, they are redefined
as:
(38.24)
(38.25)
38.3.3.1. Stress
The Cauchy stress in Voigt notation is:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
596 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
(38.26)
(38.27)
(38.28)
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 597
Large-Deformation Neo-Hookean Analysis (via UserMat Subroutine)
(38.29)
(38.30)
For an arbitrary change of basis given by the orthogonal tensor T, the components of the 2nd-order
stress tensor and the 4th-order tangent stiffness tensor transform as:
(38.31)
(38.32)
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
598 of ANSYS, Inc. and its subsidiaries and affiliates.
Analysis and Solution Controls
Comparing the tensor and Voigt notation transformations, the change of basis tensor Q is given by:
(38.33)
where:
For a change of basis from the spatial frame to the co-rotated frame, T = RT.
Material Properties
Young's Modulus (Nm-2) 30 x 106
Poisson's Ratio 0.3
The shear is simulated by displacing the nodes at Y = 1 in the X-direction by a distance of 1 unit and
preventing any displacement in the Y-direction for these nodes. The nodes are then returned to their
original position to simulate one cycle of shear deformation.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 599
Large-Deformation Neo-Hookean Analysis (via UserMat Subroutine)
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
600 of ANSYS, Inc. and its subsidiaries and affiliates.
References
38.8. Recommendations
To perform a similar analysis, consider the following hints and recommendations:
• For constitutive models that require a stress integration algorithm, ensure that the algorithm is accurate.
Exercise the user-defined material in simple simulations to determine if it is producing accurate stress results
for the types of simulations that will use the user-defined material routine. The accuracy of the stress update
directly affects the accuracy of the finite element solution.
• The material tangent returned by the user-defined material routine should be the algorithmic tangent in
the co-rotated frame. For a Newton-Raphson solution, the material tangent affects the convergence rate
and an algorithmic material tangent will generally give a quadratic convergence rate for iterations that are
close to the solution.
• The user-defined material routine is available only for displacement-based element formulations. Because
it is not available for mixed-uP formulations, materials that are nearly incompressible must use a penalty
formulation for the volumetric behavior. The penalty formulation can cause convergence difficulty and errors
in the pressure calculation when the volumetric behavior is very stiff; therefore, avoid a user-defined mater-
ial subroutine for nearly incompressible materials when the volumetric behavior is important.
38.9. References
The following reference is used in this example problem:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 601
Large-Deformation Neo-Hookean Analysis (via UserMat Subroutine)
• sample_usr.dat -- Input file for a shear simulation with the Neo-Hookean user-defined material.
The stressstrain_usr.log output file is also included to verify that the UserMat subroutine and
custom executable are producing the expected Neo-Hookean results.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
602 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 39: Wire Bonding Ultrasonic Transducer
This example problem simulates electrical excitation of an ultrasonic transducer used for wire bonding
applications. The model includes piezoelectric material definition, and prestressed modal and harmonic
response analyses.
39.1. Introduction
Wire bonding is the most commonly-used process for creating interconnects between the integrated
circuit (IC) and its package using fine metal (such as gold or aluminum) wires [1, 2]. In wedge bonding,
ultrasonic energy, pressure, and heat are applied to form a bond; the process avoids introducing impur-
ities and provides flexibility in material choice. For larger diameter wires, the frequency is often around
50-60 kHz whereas, for finer diameter wires, the frequency is much higher, up to as much as 200 kHz.
The design of the transducer includes examining the natural frequencies associated with its longitudinal
motion. Changes in geometry, for example, can influence the vibration and electrical characteristics of
the device.
In piezoelectric ceramics, an applied voltage induces strain (displacement) in the material and vice-versa,
demonstrating the coupling of the electric and structural fields. Piezoelectric ceramics are very brittle
in tension, so a preload is needed to keep the ceramics in a compressive state of stress in operation.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 603
Wire Bonding Ultrasonic Transducer
The bonding tool, made of alumina, is shown at the far left. It is connected to the titanium horn by
small screws (not modeled). The horn is connected to the piezoelectric driver assembly. The driver as-
sembly consists of piezoelectric rings sandwiched between aluminum front and back plates, which are
connected together with a steel bolt that provides prestress. The transducer is mounted to the machine
via a steel holder.
The holder should be placed at the nodal point of the transducer for optimal performance. A modal
analysis is performed without the holder, and the first longitudinal mode is determined.
As shown in the following figure, a plot of the z displacement with contour range -1 to 1 provides an
idea of the appropriate location to position the holder:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
604 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
39.3. Modeling
The 3-D model of the transducer is created in ANSYS DesignModeler and meshed in ANSYS Mechanical,
as shown in the following figure:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 605
Wire Bonding Ultrasonic Transducer
The piezoelectric elements are meshed with SOLID226, and the other parts are meshed with SOLID186
and SOLID187 elements. The total number of elements is 67,756 and the total number of nodes is
115,414.
The coupled-field element SOLID226 supports many physics types. In this case, KEYOPT(1) = 1001 specifies
piezoelectric behavior. Piezoelectric elements are orthotropic, so each of the piezoelectric rings assume
z-axis polarization with alternating z-axis orientation.
Although copper terminals lie between the piezoelectric rings, they are omitted here for simplicity.
Other details and features (such as small screws or wire holes) are also omitted, as they have no effect
on the overall response.
The parts are connected via shared nodes along the interfaces. Although contact elements can be used
for piezoelectric analyses, they are unnecessary for such a simple geometry in this case.
A bolt connects the top and bottom plates of the driver, as shown in the following figure:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
606 of ANSYS, Inc. and its subsidiaries and affiliates.
Material Properties
The fastener is split in half with PRETS179 pretension elements connecting the two halves together.
The amount of preload or adjustment is controlled via the pretension node.
Piezoelectric materials typically have orthotropic material properties, although isotropic materials are
assumed in this example, as they are defined in that manner in the reference.
The piezoelectric materials properties (taken from Berlincourt [3]) are as follows:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 607
Wire Bonding Ultrasonic Transducer
A system damping ratio of 0.1 percent is assumed for the model, as the ultrasonic transducer has very
little damping.
A static analysis is performed to calculate the prestress due to the tightening of the bolt to the driver
assembly. A preload force of 50 N is applied to the pretension node. For the subsequent dynamic analyses,
the adjustment is changed to 0.0.
Although the actual pretension adjustment (change in grip length) is non-zero, the static analysis is
performed only to create the stress-stiffening matrix that influences the overall stiffness matrix. In a
harmonic response analysis, all loads are applied sinusoidally, so imposing a preload force or non-zero
adjustment creates a harmonically-varying bolt load, which is incorrect. The effect of the bolt preload
to the stiffness of the driver is considered in the linear perturbation analysis, so the actual adjustment
is not required to be applied in the dynamic analysis.
Because the terminal is equipotential, all of the voltage degrees of freedom for each terminal are coupled,
leaving two independent voltage degrees of freedom at the terminal locations. One voltage is specified
as ground (voltage of 0).
In the modal analysis, the positive terminal is left unconstrained. In the harmonic response analysis, a
voltage of 5V constant with respect to frequency is applied to the positive terminal.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
608 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The linear perturbation method for modal analyses is described in Applying Prestress Effects in a Modal
Analysis in the Structural Analysis Guide.
The linear perturbation method for full harmonic response analyses is described in Prestressed Harmonic
Analysis in the Structural Analysis Guide.
In the solver output, the participation factors in the z (longitudinal) direction are listed at the end of
the modal analysis, as follows:
***** PARTICIPATION FACTOR CALCULATION ***** Z DIRECTION
CUMULATIVE RATIO EFF.MASS
MODE FREQUENCY PERIOD PARTIC.FACTOR RATIO EFFECTIVE MASS MASS FRACTION TO TOTAL MASS
1 1947.47 0.51349E-03 0.13955E-03 0.000974 0.194730E-07 0.469851E-06 0.391044E-06
2 1948.16 0.51331E-03 0.39207E-05 0.000027 0.153722E-10 0.470222E-06 0.308693E-09
3 4427.05 0.22588E-03 0.90019E-05 0.000063 0.810341E-10 0.472177E-06 0.162727E-08
4 4427.86 0.22584E-03 -0.46883E-05 0.000033 0.219797E-10 0.472708E-06 0.441383E-09
5 9304.53 0.10747E-03 -0.22162E-05 0.000015 0.491135E-11 0.472826E-06 0.986266E-10
6 9320.90 0.10729E-03 -0.96822E-03 0.006760 0.937453E-06 0.230920E-04 0.188253E-04
7 11382.1 0.87857E-04 0.17053E-05 0.000012 0.290807E-11 0.230921E-04 0.583979E-10
8 11385.3 0.87833E-04 0.53578E-05 0.000037 0.287058E-10 0.230928E-04 0.576451E-09
9 14535.4 0.68797E-04 0.94644E-06 0.000007 0.895743E-12 0.230928E-04 0.179877E-10
10 15206.9 0.65760E-04 -0.14972E-05 0.000010 0.224167E-11 0.230929E-04 0.450157E-10
11 19129.1 0.52276E-04 -0.72228E-04 0.000504 0.521688E-08 0.232187E-04 0.104762E-06
12 19211.2 0.52053E-04 -0.17198E-01 0.120072 0.295763E-03 0.715948E-02 0.593931E-02
13 21192.5 0.47187E-04 -0.13350 0.932096 0.178230E-01 0.437200 0.357911
14 23711.4 0.42174E-04 0.15719E-04 0.000110 0.247096E-09 0.437200 0.496203E-08
15 23712.1 0.42172E-04 0.20055E-03 0.001400 0.402200E-07 0.437201 0.807671E-06
16 26594.1 0.37602E-04 -0.14323 1.000000 0.205145E-01 0.932181 0.411958
17 28705.8 0.34836E-04 0.42418E-05 0.000030 0.179930E-10 0.932181 0.361324E-09
18 29154.9 0.34300E-04 0.10645E-01 0.074323 0.113319E-03 0.934915 0.227559E-02
19 30305.6 0.32997E-04 0.10060E-04 0.000070 0.101211E-09 0.934915 0.203245E-08
20 30312.8 0.32989E-04 0.49382E-04 0.000345 0.243859E-08 0.934915 0.489702E-07
21 34346.4 0.29115E-04 -0.52235E-05 0.000036 0.272853E-10 0.934915 0.547926E-09
22 36793.2 0.27179E-04 0.51304E-02 0.035820 0.263215E-04 0.935550 0.528571E-03
23 44385.1 0.22530E-04 0.22402E-05 0.000016 0.501871E-11 0.935550 0.100782E-09
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 609
Wire Bonding Ultrasonic Transducer
Modes having high participation factors in the z direction are candidates for evaluation as desirable
longitudinal modes. Also examine the mode shapes to determine whether excessive transverse motions
exist, as those modes should not be excited during transducer operation. Upon examination of the
results in this case, modes 16, 30 and 39 are the modes of interest, as shown in the following three
figures.
This second mode of interest is the one to be investigated in the subsequent harmonic response analyses:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
610 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
It is worth noting that if the transducer were to be used for a higher-frequency application, the third
mode of interest is at 87.3 kHz:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 611
Wire Bonding Ultrasonic Transducer
In all modes, the tip of the bonding tool has little motion in the x and y directions as compared to the
z direction, necessary for proper wire bonding to occur. Also, the frequencies of the second and third
modes are roughly twice and thrice that of the first mode, as expected.
The “reaction force” for voltage degrees of freedom is charge. In the POST26 time-history postprocessor
(/POST26), the charge Q is retrieved at the terminal. As current and , then .
This operation can be performed via the CFACT and PROD commands to calculate current based on
the charge. Impedance is calculated and plotted as follows:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
612 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The tip x, y and z displacements are output directly in POST26 and plotted as follows:
As indicated, the lateral motion (x and y) is much less than the longitudinal motion (z). The displacement
for this applied voltage is a little more than 0.1 micron.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 613
Wire Bonding Ultrasonic Transducer
39.8. Recommendations
To perform a similar analysis, consider the following hints and recommendations:
• For piezoelectric materials, ensure that the polarization direction (defined by the element coordinate system)
is correct. Regardless of the element axis (x, y or z) chosen as the polarization direction, all orthotropic ma-
terial and piezoelectric constants must be defined accordingly.
• For piezoelectric constants, the piezoelectric stress matrix ([e] form) or the piezoelectric strain matrix ([d]
matrix) can be used. If using the piezoelectric strain matrix, however, the elastic properties must be input
via the TB,ANEL command. Also, IEEE standards list the coefficients based on (x, y, z, yz, xz, xy) ordering,
while ANSYS input requires (x, y, z, xy, yz, xz) ordering. For more information, see Piezoelectric Matrix in the
Coupled-Field Analysis Guide.
• When calculating the electrical impedance, the current is calculated from the “reaction” charge, as indicated
in Results from Prestressed Full Harmonic Response Analysis (p. 612). Each node with voltage degrees of
freedom constrained returns a portion of the “reaction” charge. Instead of summing all “reaction” charges
for the terminal nodes, however, if the voltage degrees of freedom are coupled, then the total “reaction”
charge can be obtained at the master degree of freedom (MDOF). This method provides much simpler
postprocessing, as results from only a single node (MDOF) need to be queried.
• In a harmonic response analysis, all loads and boundary conditions are applied in sinusoidally. Consequently,
if pretension elements (PRETS179) are present, constrain their adjustment to zero during the harmonic re-
sponse portion of the analysis.
• For linear perturbation analyses, use the RESCONTROL,LINEAR command in the base linear static analysis
to generate the files needed for the multiframe restart (.Rnnn , .RDB, .ESAV). Linear perturbation analyses
require use of multiframe restart files, but the restart files are not created for linear static analyses unless
the RESCONTROL command is issued.
39.9. References
The following references are used in this example problem:
1. Yan, T-H et al.“Design of a Smart Ultrasonic Transducer for Interconnecting Machine Applications.” Sensors.
9.6 (2009): 4986-5000.
2. Or, S.W. et al. "Performance Study of an Ultrasonic Transducer Used for Wire Bonding." Electron Devices
Meeting. Hong Kong: 1998.
3. Berlincourt, D. and H. H. A. Krueger. Rev C. Near.“Properties of Morgan Electro Ceramic Ceramics.” Properties
of Piezoelectricity Ceramics. Technical Publication TP-226.
• wirebond2.dat -- Input file for static and linear perturbation modal analysis.
• wirebond3.dat -- Input file for static and linear perturbation full harmonic response.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
614 of ANSYS, Inc. and its subsidiaries and affiliates.
Input Files
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 615
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
616 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 40: Shape Memory Alloy (SMA) with Thermal Effect
This example problem presents two shape memory alloy (SMA) simulations: a spinal spacer implant and
a spring actuator.
40.1. Introduction
A shape memory alloy (SMA) is a material that, after being subjected to mechanical loading/unloading
cycles, is able to undergo large deformations without showing residual strains (pseudoelasticity) or that
can recover from large deformations via temperature change (shape memory effect).
Pseudoelasticity and the shape memory effect are material characteristics especially useful for aeronaut-
ical, biomedical, and structural engineering applications. Although much progress has been made with
SMA material analysis and design, many challenges still exist for precisely controlling SMAs due to highly
nonlinear hysteretic transformation, material degradation, and thermo-mechanical fatigue. Finite element
analysis has been widely used to simulate SMA material and provides a valuable tool for designing
products using SMA materials.
where εL is a maximum value norm of E in in the phase transformation after fully transformed.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 617
Shape Memory Alloy (SMA) with Thermal Effect
γ is defined by
where the limit function F is given in terms of the transformation stress Xtr and the elastic domain ra-
dius R in the form of the Prager-type limit function:
(40.5)
where:
Thus, the governing equations for the phase transformation are expressed as:
(40.6)
In addition to the Young’s modulus and Poisson’s ratio of martensite and austenite, six other parameters
are defined: M, R, h, T0, β, and εL.
The spinal vertebrae spacer is simulated via SOLID187 elements, and the spring actuator is simulated
via BEAM188 and SOLID185 elements.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
618 of ANSYS, Inc. and its subsidiaries and affiliates.
SMA Thermal Effect Simulations
Spinal spacers restore disc space height, alignment, and the spine’s ability to bear weight, any or all of
which can be lost due to IVD degeneration. Finite element analysis of implant function can help improve
the design and quality of the spinal spacer.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 619
Shape Memory Alloy (SMA) with Thermal Effect
40.3.1.2. Modeling
A 3-D geometry of the spinal spacer is created in Unigraphics, using dimensions found in Petrini 2005
[2]. The geometry is imported into Mechanical APDL and meshed with SOLID187 elements. Because the
spacer is symmetrical, only 1/4 of the spacer is studied.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
620 of ANSYS, Inc. and its subsidiaries and affiliates.
SMA Thermal Effect Simulations
In step 4, the temperature is increased from 297 K to 311 K. Convergence is achieved quickly as this
temperature is below T0.
In step 5, the temperature is again increased from 311 K to 324 K. The major phase transformation does
not occur in this step, so convergence is again achieved quickly.
In step 6, the temperature is increased above 324 K, and the shape memory effect occurs, so convergence
is slower.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 621
Shape Memory Alloy (SMA) with Thermal Effect
The following figure shows the deformation of the spacer at each step:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
622 of ANSYS, Inc. and its subsidiaries and affiliates.
SMA Thermal Effect Simulations
In step 2, the displacement is 4.5 mm and the stress is 1994 MPa. After elastic recovery, the peak dis-
placement decreases to 3.0 mm and the stress is 579 MPa. In the final step, displacement and stress
approaches zero, indicating that the spacer has returned to its original shape.
The simulation accurately depicts the spacer under load (step 2), during elastic recovery (step 3), and
at full recovery due to SMA thermal effects (step 6).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 623
Shape Memory Alloy (SMA) with Thermal Effect
In this problem, a vertical helical spring is simulated to repeat its two-way motion due to the shape
memory effect. The following related topics are available:
40.3.2.1. Problem Description
40.3.2.2. Modeling
40.3.2.3. Material Properties
40.3.2.4. Boundary Conditions and Loading
40.3.2.5. Analysis and Solution Controls
40.3.2.6. Results and Discussion
The spring is loaded by a weight of 1830 N in the martinsite state at a temperature of 250 K, then
heated to 400 K. At the increased temperature, the spring lifts the weight. The spring is then cooled
back to 250 K and stretches again. A repeatable, two-way motion occurs, as shown in this figure:
40.3.2.2. Modeling
The geometry of the spring actuator is created in Mechanical APDL with a wire diameter of 4 mm, a
spring external diameter of 24 mm, a pitch size of 12 mm, with two coils, and an initial length of 28
mm, as shown in the following figure:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
624 of ANSYS, Inc. and its subsidiaries and affiliates.
SMA Thermal Effect Simulations
The corresponding finite element model is created using BEAM188 elements. A 3-D model is generated
by extruding the initial finite element model and meshing with SOLID185 elements.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 625
Shape Memory Alloy (SMA) with Thermal Effect
The results from the BEAM188 and SOLID185 models are compared.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
626 of ANSYS, Inc. and its subsidiaries and affiliates.
SMA Thermal Effect Simulations
The maximum displacement is 43 mm, greater than the original length of 28 mm.
In step 2, after heating with the shape memory effect, the spring actuator recovers to a maximum dis-
placement of 10 mm. The deformation is in the martinsite state to support the weight, as shown here:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 627
Shape Memory Alloy (SMA) with Thermal Effect
In step 3, after cooling to 250 K, the spring actuator stretches back to its original length:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
628 of ANSYS, Inc. and its subsidiaries and affiliates.
References
The displacement history indicates that the BEAM188 and SOLID185 models have similar results. The
BEAM188 model is much more efficient, however, requiring about an hour to complete. In comparison,
the SOLID185 model requires more than eight hours to complete.
40.4. Recommendations
To perform similar types of analyses involving SMAs, consider the following:
• The stiffness of the material model is generally unsymmetrical, but a symmetric matrix is used in the solution
by default. If convergence difficulty occurs during the solution, specify the unsymmetric solver option
(NROPT,UNSYM).
• SMA phase transformation involves a transformation stage and the saturated transformation. Because con-
vergence is more difficult to achieve during the transformation stage, break the transformation loadstep
up into smaller ones and use smaller time steps.
• The superelasticity option (TBOPT = SUPE) of the material model (TB,SMA) supports only 3-D, plane strain,
and axisymmetric stress states. The memory option (TBOPT = MEFF) of the material model supports most
of the stress states including beam, shell, plane strain, axisymmetric, and 3-D stress states. The LINK180
element is not supported for either material option.
40.5. References
The following references are used in this example problem:
1. Paremer, A., S. Fumer, D. P. Rice. Musculoskeletal Conditions in the United States. 1st ed. Park Ridge: American
Academy of Orthopaedic Surgeons, 1992.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 629
Shape Memory Alloy (SMA) with Thermal Effect
2. Petrini, L., F. Migliavacca, et al.“Computational Studies of Shape Memory Alloy Behavior in Biomedical
Applications.” Journal of Biomedical Engineering. 127 (2005): 716-725.
3. Arghavani, J., F. Auricchio, R. Naghdabadi.“A finite strain kinematic hardening constitutive model based
on Hencky strain: General framework, solution algorithm, and application to shape memory alloys.” Inter-
national Journal of Plasticity. 27 (2011): 940-961.
• actuator_beam_input.dat -- Input file for the spring actuator portion of the analysis (using beam
elements).
• td-40b.cdb -- Common database file containing the spring actuator model (using beam elements).
• actuator_solid_input.dat -- Input file for the spring actuator portion of the analysis (using solid
elements).
• td-40c.cdb -- Common database file containing the spring actuator model (using solid elements).
...
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
630 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 41: Acoustic Analysis of a Viscothermal Resonator
This example problem uses acoustic elements and viscothermal losses (comparing a boundary layer
impedance [BLI] model and a low reduced frequency [LRF] model) to analyze the noise-reduction of
sound-absorbing trim panels with quarter-wave resonators.
• Ports definition
41.1. Introduction
Development of systems for minimizing noise is an ongoing process, as noise is a factor that can quickly
and adversely affect comfort. In an aircraft cabin, for example, the turbulent boundary layer surrounding
the fuselage is a primary source of noise in the mid- to high-frequency range (500-2000 Hz). To reduce
the sound pressure level in the cabin, an effective solution consists of quarter-wave resonator panels
made of an assembly of tubes of varying diameters and lengths. The absorption capability of the reson-
ator panel is the result of a combined effect of the quarter-wave phenomenon and viscothermal loss.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 631
Acoustic Analysis of a Viscothermal Resonator
The problem presented here calculates the absorption coefficient of an example resonator panel.
The following figure shows the geometry of the resonator model used in this simulation:
This figure shows the resonator composed of tubes with variable diameters and lengths, resulting in
an optimized absorption for all frequencies:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
632 of ANSYS, Inc. and its subsidiaries and affiliates.
Problem Description
Radii and length for each tube are defined with the following values [1]:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 633
Acoustic Analysis of a Viscothermal Resonator
41.3. Modeling
The 3-D model of the structure is created in ANSYS DesignModeler and meshed with FLUID220 and
FLUID221 3-D acoustic elements, with pressure as a unique degree of freedom (KEYOPT(2) = 1).
A normal velocity and a nonreflective radiation boundary are applied on the impedance tube inlet,
simulating the sound generated experimentally by a speaker.
Boundary layer impedance (BLI) and low reduced frequency (LRF) models are successively applied on
the resonator tubes to analyze viscous and thermal effects.
The element size chosen ensures at least six elements per wave length for the highest frequency of
interest.
All bodies are grouped in a single part for node connectivity at the body’s interfaces.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
634 of ANSYS, Inc. and its subsidiaries and affiliates.
Boundary Conditions and Loading
A normal surface velocity is applied on the exterior. A transparent port and a radiation boundary are
also applied on the same face:
CMSEL,S,source
ESEL,S,ENAME,,220
ESEL,A,ENAME,,221
SF,ALL,SHLD,-1, 0
SF,ALL,PORT,1
SF,ALL,INF
Another port, used to calculate the absorption coefficient, is defined on the extremities of the tubes:
CMSEL,S,ends
ESEL,S,ENAME,,220
ESEL,A,ENAME,,221
SF,ALL,PORT,2
Two models are available for simulating the viscothermal losses in narrow tubes:
• A boundary layer impedance (BLI) model is defined on the tube faces in order to account for the visco-
thermal effects:
CMSEL,S,res
ESEL,S,ENAME,,220
ESEL,A,ENAME,,221
SF,ALL,BLI
• A low reduced frequency (LRF) model is defined for each tube via the TB,AFDM,,,,CIRC material model:
TB,AFDM,1,,,CIRC
TBDATA,1,Rtube
Generally, the LRF model applies to layers and tubes below the cut-off frequency, and the BLI model
applies to geometries that are large compared to the viscothermal boundary layer thickness.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 635
Acoustic Analysis of a Viscothermal Resonator
To analyze the absorption coefficient, the boundary layer impedance (BLI) and low reduced frequency
(LRF) models are compared, as shown in the following figures:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
636 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The results show that (except for a lower initial absorption at 1000 Hz) this resonator is efficient, as the
absorption coefficient is close to 1 over the frequency range.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 637
Acoustic Analysis of a Viscothermal Resonator
41.8. Recommendations
To perform a similar analysis involving an acoustic viscothermal model, consider the following recom-
mendations:
• Ensure at least six elements per wave length for the highest frequency of interest by adapting the mesh
element size (p. 634) to the frequency range of the analysis.
• Select one of the available viscothermal models: the low reduced frequency (LRF) model applies to layers
and tubes below the cut-off frequency, and the boundary layer impedance (BLI) model applies to geometries
that are large compared to the viscothermal boundary layer thickness.
• Issue the PLAS,ALPHA command to postprocess the absorption coefficient of the system.
41.9. References
The following reference is used in this example problem:
1. Hannink, M. H. C., Y.H. Wijnant, A. de Boer.“Optimised Sound Absorbing Trim Panels for the Reduction of
Aircraft Cabin Noise.” 11th International Congress on Sound and Vibration. ICSV 11. St. Petersburg, Russia.
(2004)
• visco_resonator_BLI.dat -- Input file for the viscothermal resonator analysis using the BLI model.
• visco_resonator_BLI.cdb -- Common database file for the original mesh used in the BLI model (called
by visco_resonator_BLI.dat).
• visco_resonator_LRF.dat -- Input file for the viscothermal resonator analysis using the LRF model.
• visco_resonator_LRF.cdb -- Common database file for the original mesh used in the LRF model
(called by visco_resonator_LRF.dat).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
638 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 42: Wire Crimping Modeled with General Contact
This example problem demonstrates the ease of contact modeling via the general contact method. This
method offers automated contact creation and requires minimal input. The general contact procedure
is especially useful when a large number of contacting surfaces are involved in the model and the
geometry makes it difficult to determine contact pairs.
• Contact-property definitions
• Rigid-flexible and flexible-flexible contact modeled by general contact, including surface-to-surface, edge-
to-surface, and edge-to-edge contact configurations
42.1. Introduction
In this example, a multi-strand wire is joined to an electrical terminal (connector) through a mechanical
deformation process known as crimping. The U-shaped portion of the connector (the grip) is folded
around the wires by a rigid punch, forming a B-shaped crimp which provides connectivity between the
wires and the electric terminal.
Due to the complexity of this model, defining all possible contact surfaces by the pair-based contact
method would be a difficult and time-consuming task. By using the general contact method, contact
surfaces are created automatically; only a limited number of contact surfaces require the specification
of non-default contact properties. Both flexible-flexible and rigid-flexible contact are modeled.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 639
Wire Crimping Modeled with General Contact
A transient analysis is performed to 3.4e-4 seconds in order to capture the occurrence of all possible
contact (surface-to-surface, edge-to-surface, and edge-to-edge) defined by general contact.
42.3. Modeling
The 3-D crimp joint model consists of a 0.5 mm thick grip and seven stranded wires, with each wire
having a 0.725 mm diameter. The grip and wires are made of a copper alloy which is modeled by the
multi-linear isotropic hardening plasticity material model. The model also includes a rigid punch and a
rigid support.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
640 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
Figure 42.2: 3-D Meshed Model of Grip with Seven Stranded Wires
The rigid target surfaces are taken from a pair-based contact model and converted to general contact.
To include these target surfaces in the general contact definition, a zero real constant ID and a zero
material ID are assigned via the EMODIF command with I1 = GCN. A unique section ID (SECNUM) and
a unique element type ID (TYPE) are also assigned via the EMODIF command. The same ID number is
used for the section and element type IDs.
! Defining the rigid punch as a general contact surface
Alternatively, you could generate these target surfaces via the AMESH command if you were modeling
this structure from scratch.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 641
Wire Crimping Modeled with General Contact
The value 20 is used for feature angle in order to achieve the desired split of the contact surfaces and
edges between them. In this model, only the top edges of the grip are considered, and the rest are
deleted.
allsel,all
gcgen,,20,1, ! Automatically generate CONTA174 elements on
! exterior faces & CONTA177 on feature edges
! Only the top edges of the grip are considered, and the rest are deleted
keyopt,95,3,0
keyopt,96,3,0
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
642 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
keyopt,109,3,0
keyopt,110,3,0
keyopt,111,3,0
keyopt,112,3,0
Figure 42.4: Surface Contact Elements (a) and Edge Contact Elements (b)
Next, the contact interactions are defined via the GCDEF command. By default, frictionless standard
contact is assumed among all general contact surfaces. To override this default, the command
GCDEF,AUTO,ALL,ALL,100,100 is used to define frictional contact for all contact surfaces. Additional
GCDEF commands are used to exclude some surfaces that may cause spurious contact.
! Define interactions between general contact surfaces via the GCDEF command
r,100
rmodif,100,3,1 ! Define FKN for real ID 100
gcdef,exclude,grip_zp_face,all_face
gcdef,exclude,grip_zn_face,all_face
gcdef,exclude,grip_zp_face,all_edge
gcdef,exclude,grip_zn_face,all_edge
gcdef,exclude,gripTop_xn_face,grip_inner_face
gcdef,exclude,75,77
gcdef,exclude,gripTop_xp_face,grip_inner_face
gcdef,exclude,76,79
allsel,all
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 643
Wire Crimping Modeled with General Contact
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
644 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
(a) Displacement of the rigid punch and deformed shape of the grip
(b) Equivalent plastic strain; the maximum is at the top edge of the grip as it comes in contact
with the rigid punch
(c) Contact status plots of both surface-to-surface and edge-to-surface contact during the
solution
Figure 42.6: Results at 2.5e-4 Seconds: (a) Displacement, (b) Equivalent Plastic Strain, (c) Contact
Status
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 645
Wire Crimping Modeled with General Contact
(a) Displacement of the rigid punch and the final deformed shape of the grip
(b) Equivalent plastic strain; the maximum is at the grip
(c) Contact status plots of surface-to-surface, edge-to-surface, and edge-to-edge contact during
the solution
Figure 42.7: Results at 3.4e-4 Seconds: (a) Displacement, (b) Equivalent Plastic Strain, (c) Contact
Status
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
646 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Figure (a) below shows the cross-sectional view of the B-shaped crimp which is finally formed at the
end of the deformation. Observe that the grip is completely folded around the wires and is in contact
with the wires. Figure (b) shows the final deformed shape of the wires and their out-of-plane extrusion.
The following animation shows the deformation over time of the grip as it is folded around the wires
by the rigid punch to form a B- shaped crimp.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 647
Wire Crimping Modeled with General Contact
42.8. Recommendations
When setting up a general contact analysis, consider the following hints and recommendations:
• Carefully consider the feature angle used on the GCGEN command. Choose a value that will produce
the desired contact surfaces and edges.
• When using the exclude option on the GCDEF command along with node components (to identify
surfaces), use the _FACE component name extension to exclude only surface contact. Otherwise, both
surface and edge elements will be excluded by default.
• Always check the normal direction of the rigid surface target elements.
• wirecrimp.dat -- Input file for wire crimping process modeled with the general contact method.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
648 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 43: Contact Surface Wear Simulation
This example problem simulates wear at a contact surface. The wear occurs at the interface between a
hemispherical ring rotating over a flat ring. Wear characteristics demonstrated include removal of ma-
terial due to wear, changes in contact pressure and area due to wear, and a continuous decrease of the
wear rate in steady-state conditions.
• User-defined wear
43.1. Introduction
Wear is the progressive loss of material from the surface of a solid body when in contact with another
body. The program approximates this loss of material by repositioning the contact nodes at the contacting
surface. The new node locations are determined by a wear model which calculates how much and in
what direction a contact node is to be moved to simulate wear based on the contact results.
This example shows use of the Archard Wear model and also demonstrates the user-defined subroutine
for modeling wear (userwear).
Since wear involves material removal, the element quality of solid elements underlying the contact
elements becomes progressively worse with increasing wear. Remeshing is required to successfully
simulate large amounts of wear. This example demonstrates how manual rezoning or mesh nonlinear
adaptivity can be used to improve mesh quality when a model undergoes large amounts of wear.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 649
Contact Surface Wear Simulation
The hemispherical ring is subjected to a pressure load of 4000 N/mm2 and is rotating with a frequency
of 100,000 revolutions/sec. Sliding of the hemispherical ring on the flat ring causes wear in the rings.
43.3. Modeling
A 2-D axisymmetric model representing the two rings is meshed and loaded as shown in the figure.
The rings are meshed with 2-D axisymmetric PLANE182 elements (KEYOP(3) = 1). Frictionless contact
is modeled between the two rings by overlaying the surfaces with contact elements (CONTA172) and
target elements (TARGE169).
Two versions of this model are created, one with asymmetric contact and the other with symmetric
contact. The first model uses asymmetric contact to simulate wear only on the top hemispherical ring;
contact elements are defined on the top ring and target elements are defined on the bottom ring. In
the second model, wear is simulated on both rings by defining symmetric contact between them;
contact and target elements are defined on the surfaces of both rings.
Since wear can be modeled only on surfaces that have contact elements, the first example exhibits
wear only on the hemispherical ring and the second example exhibits wear on both rings.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
650 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
• Location of the contact detection point: nodal point, normal to target surface (KEYOPT(4) = 2)
Since simulating wear requires repositioning of the contact nodes, the contact detection point must
be at nodes (KEYOPT(4) = 1 or 2), or projection-based contact can be used (KEYOPT(4) = 3).
Contact elements must be defined on the surface undergoing wear. A wear model is then associated
with those contact elements via a TB,WEAR material definition. The following command is used to
define the Archard wear model:
TB,WEAR,MATID,,,ARCD
The Archard wear model is specified by inputting constants C1 through C4 on the TBDATA command.
These constants represent the wear coefficient (K), material hardness (H), the contact pressure exponent
(m), and the sliding velocity exponent (n).
The wear coefficient K can sometimes be scaled to simplify modeling. As an example, consider this ring-
on-ring problem in which the rings are rotating at constant speed. The only effect of this rotation/sliding
at the contact surface is to produce wear (friction is absent). The wear coefficient K can be scaled such
that the rotation is not explicitly modeled, but its effect is included in the computation of wear. This
greatly reduces the simulation time and effort.
More specifically, if a linear dependence of wear rate on the sliding velocity is assumed, the wear
coefficient K can be scaled by the sliding velocity. In this example, sliding velocity is 2πN*R, where N =
100,000 revolutions/sec and R is the distance from the axis of rotation. Scaling K by 2πN*R results in
the wear rate being linearly dependent upon the sliding velocity without explicitly modeling the sliding.
The distance from the axis of rotation (R) is assumed to be constant for all points and is taken as 100
mm (the distance of the center of the ring from the axis of rotation).
To more accurately model the sliding velocity, you can program a user-defined wear model (via the
userwear subroutine) that includes sliding velocity based on the distance from the axis of rotation.
Three example inputs are prepared to demonstrate different approaches to modeling wear:
43.3.1.1. Wear on One Contact Surface (Asymmetric Contact)
43.3.1.2. Wear on Both Contact Surfaces (Symmetric Contact)
43.3.1.3. User-Defined Wear (userwear Subroutine)
The material data for wear is defined using TBDATA commands. The wear properties for the copper
ring are as follows:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 651
Contact Surface Wear Simulation
To initiate wear after a steady state has been reached with respect to loading, TB,WEAR is used in
conjunction with TBFIELD,TIME. The problem is simulated in two load steps. In the first load step,
pressure is ramped to the desired level and wear is inactive during this load step. To achieve this
TB,WEAR is defined as follows:
TB,WEAR,3,,,ARCD ! Mat #3 = material for wear of contact element on copper ring
TBFIELD,TIME,0 ! Time at the beginning of load step 1
TBDATA,1,0,1,1,0,0 ! C1 = 0 results in no wear for load step 1
TBFIELD,TIME,1 ! Time at the end of load step 1
TBDATA,1,0,1,1,0,0 ! C1 = 0 results in no wear for load step 1
In the second load step, the pressure is held constant and wear is defined as follows:
TBFIELD,TIME,1.01 ! Time at the beginning of load step 2
TBDATA,1,kcopper,1,1,0,0 ! Value of wear coefficient resulting in initiation of wear
TBFIELD,TIME,4 ! Time at the end of load step 2
TBDATA,1,kcopper,1,1,0,0 ! Wear coefficient kept constant during load step 2
The wear properties for the copper ring are as presented in the asymmetric example described above.
The wear properties for the steel ring are as follows:
A symmetric contact definition between dissimilar meshes and materials can lead to a non-smooth
contact pressure distribution. Thus, the option to use nodal stresses of the underlying solid element to
calculate the wear increment (C5 = 1 on TBDATA) is recommended and is used in the symmetric example.
Wear for the copper ring is defined as in the asymmetric example with the addition of C5 = 1:
TB,WEAR,3,,,ARCD ! Mat #3 = material for wear of contact element on copper ring
TBFIELD,TIME,0 ! Time at the beginning of load step 1
TBDATA,1,0,1,1,0,1 ! C1 = 0 results in no wear for load step 1; C5 = 1 = nodal stress
TBFIELD,TIME,1 ! Time at the end of load step 1
TBDATA,1,0,1,1,0,1 ! C1 = 0 results in no wear for load step 1; C5 = 1
TBFIELD,TIME,1.01 ! Time at the beginning of load step 2
TBDATA,1,kcopper,1,1,0,1 ! Value of wear coefficient resulting in initiation of wear; C5 = 1
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
652 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
The userwear subroutine uses five input properties defined via TBDATA. C1 through C4 are the same
as for the Archard wear law, and C5 passes in the rotational speed (10,000 rotations/sec). The example
userwear subroutine calculates position-dependent sliding velocity using the distance of the contact
point from the axis of rotation (R) and defines wear increment accordingly, therefore avoiding explicit
modeling of the sliding motion.
Asymmetric contact is considered in this example. The wear is defined as follows for the copper ring:
ang_velocity=2 1e5 ! angular velocity to be passed into userwear to calculate
! accurate sliding velocity
TB,WEAR,3,,5,USER ! Mat #3 = material for wear of contact element on copper ring
TBFIELD,TIME,0 ! Time at the beginning of load step 1
TBDATA,1,0,1,1,0,ang_velocity ! C1 = 0 results in no wear for load step 1
TBFIELD,TIME,1 ! Time at the end of load step 1
TBDATA,1,0,1,1,0, ang_velocity ! C1 = 0 results in no wear for load step 2
TBFIELD,TIME,1.01 ! Time at the beginning of load step 2
TBDATA,1,kcopper,1,1,0,ang_velocity ! Value of wear coefficient resulting in initiation of wear
TBFIELD,TIME,4 ! Time at the end of load step 2
TBDATA,1,kcopper,1,1,0, ang_velocity ! Wear coefficient kept constant during load step 2
One way to improve the mesh is to use the mesh nonlinear adaptivity feature. A wear-based contact
criterion triggers mesh nonlinear adaptivity whenever the mesh is distorted. The critical ratio between
the amount of wear and the underlying solid element’s height is user-defined. When the criterion is
reached, mesh nonlinear adaptivity is triggered.
Both the asymmetric example and the symmetric example use mesh nonlinear adaptivity, which requires
the following steps:
• Create a component that contains the contact elements that are undergoing wear.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 653
Contact Surface Wear Simulation
In both examples, the component name is “conreal”, and the following command triggers mesh adaptivity:
NLADAPTIVE,conwearel,add,contact,wear,0.50 ! Wear criterion is 50% of element height.
In this case adaptivity occurs whenever wear at any contact point exceeds 50% of the average height
of the solid element underlying the contact element. Each time the criterion is reached, the analysis is
stopped, the mesh quality is improved by morphing the mesh, history-dependent variables and
boundary conditions are mapped, and the analysis is restarted with an improved mesh. This process is
done automatically.
Manual Rezoning
Manual rezoning is another method that enables you to remesh the distorted mesh and continue the
wear simulation with an improved mesh. This method, which requires more user intervention, is
demonstrated in the user-defined wear example input file (see Input Files (p. 662)).
Contact elements are used to define a rigid surface constraint between the nodes on the top surface
of the hemispherical ring and a pilot node which is constrained in the X direction. A force is applied
on the pilot node that is equivalent to a pressure of 4000 N/mm2.
The equivalent pressure is ramped during the first load step from 0 to 4000 N/mm2 and is kept constant
at 4000 N/mm2 during the second load step. Wear is activated in the second load step.
The repositioning of contact nodes during wear can result in changing contact status. If the wear incre-
ment is too large, all contact elements may go from a closed status to an open status, resulting in rigid
body motion. To prevent this, a very small time increment is used so that the wear increment is also
small and changes in contact status are minimized.
The following DELTIM command is used to set a small time increment size in the second load step:
DELTIM,0.01,1e-6,0.02
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
654 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
For the axisymmetric contact example, the contact condition is similar to the classical Herzian contact
at the end of load step 1 (see figure below). Wear in the hemispherical ring is activated in the second
load step. The figure shows the contact pressure at the end of 300,000 rotations (3 seconds) of the
hemispherical copper ring over the steel ring.
Since the amount of wear is proportional to the contact pressure, the regions with initially high contact
pressure wear more and local curvature is reduced, thereby reducing the contact pressure in those re-
gions. This also leads to increasing area of the ring coming in contact and increases contact pressure
in regions that had low contact pressure at the beginning of the wear.
Thus the simulation captures the physical process of wear and results in an increased contact area and
more uniform contact pressure; maximum contact pressure goes down with wear and minimum contact
pressure goes up with wear. That is, wear makes contact pressure more uniform as shown in the figure
below.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 655
Contact Surface Wear Simulation
Figure 43.2: Contact Pressure Before and After Wear of Hemispherical Ring (Axisymmetric Contact)
For the symmetric contact example, wear occurs on the top hemispherical ring and on the bottom flat
ring. As shown in the figure below, the wear results in similar effects, making contact pressure more
uniform on both sides of the contact surfaces and increasing contact area.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
656 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Figure 43.3: Contact Pressure Before and After Wear of Both Rings (Symmetric Contact)
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 657
Contact Surface Wear Simulation
Figure 43.4: Stress in the Loading Direction (Y) Before and After Wear of Hemispherical Ring
Mesh nonlinear adaptivity improves the mesh quality during wear in this analysis. The specified wear-
based criterion is 0.50, which means mesh morphing occurs whenever wear causes the underlying solid
element to lose 50% of its height.
The axisymmetric contact example undergoes mesh nonlinear adaptivity four times. The figure below
shows the stress in the loading direction and the element quality before and after the first mesh non-
linear adaptivity process.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
658 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Figure 43.5: Stress in the Loading Direction (Y) and Element Shapes Before and After Mesh
Morphing
The following animation provides a visualization of the changing stress distribution throughout the
second load step, as well as the remeshing of elements due to mesh morphing.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 659
Contact Surface Wear Simulation
Figure 43.6: Animation of Stress in the Loading Direction During Load Step 2
The stress in the loading direction along the contact interface before and after wear is shown in the
figure below. Note that the wear makes the stress more uniform.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
660 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Figure 43.7: Stress in the Loading Direction (Y) Along the Contact Interface Before and After Wear
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 661
Contact Surface Wear Simulation
Figure 43.8: Time Evolution of Wear in Y Direction at the Initial Point of Contact
43.8. Recommendations
When performing wear simulations, consider the following recommendations:
• Use one of the following contact algorithms: augmented Lagrangian or penalty function (KEYOPT(2) = 0 or
1). Modeling wear with the pure Lagrangian contact algorithm can result in convergence problems and is
not recommended.
• Use very small substeps so that the wear increment is small. A large wear increment can abruptly change
the contact status and cause convergence difficulties.
• In general, you should use asymmetric contact to model wear on only one side of the contact interface.
However, you can use symmetric contact if wear is desired on both sides of the interface. In this case, define
contact elements on both sides of the interface and use the option for the nodal-stress-based wear calculation
(C5 of Archard wear model = 1 on TBDATA) to achieve better results.
• Simulating a large amount of wear can result in severe mesh distortions. In such cases, use the wear-based
nonlinear adaptivity criterion to improve the mesh quality via mesh morphing.
• RingOnRingWear_Asym.dat -- Input file using the Archard wear model and asymmetric contact.
• RingOnRingWear_Asym.cdb -- Database file for the asymmetric contact analysis (called by RingOn-
RingWear_Asym.dat).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
662 of ANSYS, Inc. and its subsidiaries and affiliates.
Input Files
• RingOnRingWear_Sym.dat -- Input file using the Archard wear model with nodal stresses for wear and
symmetric contact.
• RingOnRingWear_Sym.cdb -- Database file for the symmetric contact analysis (called by RingOnRing-
Wear_Sym.dat).
• RingOnRingWear_User.cdb -- Database file for the analysis demonstrating a user-defined wear model
(called by RingOnRingWear_User.dat).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 663
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
664 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 44: C*-integral Evaluation for 3-D Surface Flaws
This example problem shows how to evaluate the C*-integral for cracks in structural components. C*-
integral characterizes the crack tip conditions in homogenous materials undergoing a secondary (steady-
state) creeping deformation.
Analyses of a simple semicircular surface flaw in a rectangular block and a warped flaw along a tubular
joint are discussed.
• Evaluating the C* integral for a warped semi-elliptical surface flaw in a tubular joint.
• Analysis controls for secondary creep deformation dominating the crack tip domain.
For more information, see C*-integral Calculation in the Fracture Analysis Guide.
44.1. Introduction
Structural components operating at high temperatures may fail due to the slow extension of a pre-ex-
isting crack in the structure over a period of time. For an elastic-secondary power law creeping material
model, the stress and strain singularities at the crack tip can be controlled by a time-dependent loading
parameter.[1] For long-term loading, the C*-integral parameter can be path-independent and applies
to cracks only when the body with the crack undergoes extensive steady-state creep.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 665
C*-integral Evaluation for 3-D Surface Flaws
A torus is created around the crack front to control the mesh at the crack front. Following is the semi-
circular crack and torus along the crack front used to create a sweep mesh with the SOLID186 3-D 20-
node structural solid (brick) element:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
666 of ANSYS, Inc. and its subsidiaries and affiliates.
Problem Description
Figure 44.2: Semicircular Crack and Torus Along the Crack Front
Figure 44.3: X-Joint Pipe Full Model with Warped Surface Flaw at Welded Joint
The problem consists of two tubular members attached to each other by a welded joint. The tubular
members (labeled Tube 1 and Tube 2) have outer diameters of 323.85 mm (D1) and 219.08 mm (D2),
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 667
C*-integral Evaluation for 3-D Surface Flaws
and thicknesses of 15.88 mm (t1) and 8.18 mm (t2), respectively. The semi-elliptical surface crack lies
on a plane parallel to the radial direction of the heavier running pipe.
A torus is created around the crack front to control the mesh at the crack front. The semi-elliptical surface
crack at the weld toe is warped along the welded joint and it is perpendicular to the outer surface of
the 323.85 mm diameter pipe in the thickness direction.
The warped crack profile at the welded joint is created via the interaction of a rotated semi-ellipse on
the 323.85 mm diameter pipe and an extruded circle normal to the inner surface of the same pipe at
the welded joint:
Figure 44.5: Two-plane Symmetry of X-joint Pipe with Warped Surface Flaw Geometry
44.3. Modeling
For 3-D models, the recommended element type near the crack front is SOLID186, the 3-D 20-node
structural solid (brick).
The rectangular block and x-joint pipe models are meshed with SOLID187 3-D 10-node tetrahedral
structural solid elements (except for the regions around the crack tip).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
668 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
Figure 44.6: Rectangular Block Model with Boundary Conditions and Loading Applied
Because of the two-plane symmetry inherent to the X-joint problem, a quarter model is considered for
analysis:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 669
C*-integral Evaluation for 3-D Surface Flaws
Figure 44.8: Two-plane Symmetry of X-joint Pipe with Warped Surface Flaw at Welded Joint
If the crack surface is not normal to any component in the global coordinate system, a local coordinate
system should be created in such a way that one component of the coordinate system is perpendicular
to the crack surface. For example, in the X-joint model, a local cylindrical coordinate system is created
such that one of its components is normal to the crack surface:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
670 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
Figure 44.9: X-joint Pipe with Warped Surface Flaw at Welded Joint in Thickness Direction
In the X-joint model, the warped crack in the thickness direction is perpendicular to the inner surface
of the 323.85 mm diameter pipe.
A torus is created around the crack front to obtain a good sweep mesh (VSWEEP). A common area
separates the torus and the remaining volume at the interface.
In 3-D models, a fracture has two surfaces with a common edge at the crack front; the same is true for
the torus. The two torus surfaces are used to create a clean sweep mesh around the crack front. One
surface is defined as the source, the other as the target. In this procedure, the sweep mesh generates
brick elements with one layer of prism elements around the crack front. Following is the sweep mesh
with SOLID186 around the crack front in the x-joint pipe model, with the source and target areas for
the sweep mesh indicated in the deformed body:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 671
C*-integral Evaluation for 3-D Surface Flaws
Figure 44.10: Sweep Mesh with SOLID186 Around the Crack Front
The warped surface contains a sweep mesh and a set of nodes, along with source and target areas in
the deformed torus body. The source and target areas existed at the same location in the non-deformed
structure.
A line component of the crack front lines (CM) is created. The component is useful for selecting nodes
attached to the crack front (NSLL). The nodal component of those nodes is used to define crack tip
node components (CINT,CTNC), as shown in the following figures:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
672 of ANSYS, Inc. and its subsidiaries and affiliates.
Material Model and Material Properties
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 673
C*-integral Evaluation for 3-D Surface Flaws
*SET,C4,0
TB,CREEP,1,,,1
TBDATA,1,C1,C2,C3,C4
Analysis of the X-joint model is performed with two-sided symmetry. Two-plane symmetric boundary
conditions are applied and one midside node at the horizontal plane is constrained in the opposite
direction to restrict rigid-body motion. A pressure load of -10 MPa is applied on the top of the small-
diameter tube, as shown in Figure 44.8: Two-plane Symmetry of X-joint Pipe with Warped Surface Flaw
at Welded Joint (p. 670).
*SET,HOUR,2000
AUTOTS,ON
NSUB,50,100,10
TIME,HOUR
OUTRES,ERASE
OUTRES,ALL,ALL
RATE,ON
NLGEOM, ON
SOLVE
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
674 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 675
C*-integral Evaluation for 3-D Surface Flaws
The creep strain is approximately 100 times larger than the elastic strain in the secondary creep stage,
which dominates the entire specimen at the end of the simulation.
Following is the maximum Von Mises stress occurring at the crack tip:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
676 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The following figure shows the C*-integral along the crack front for different contours, where the path-
independence appears after path 3:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 677
C*-integral Evaluation for 3-D Surface Flaws
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
678 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 679
C*-integral Evaluation for 3-D Surface Flaws
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
680 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The creep strain is approximately 30 times larger than the elastic strain in the secondary creep stage,
which dominates the local region of specimen at the end of the simulation.
The following shows the C*-integral values along the crack front, which gradually becomes path-inde-
pendent after path 2.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 681
C*-integral Evaluation for 3-D Surface Flaws
Figure 44.20: C*-integral Along the Crack Front -- X-Joint Pipe with Warped Flaw
44.8. Recommendations
When setting up a C*-integral calculation, consider the following hints and recommendations:
• The recommended element type for 3-D fracture models along the crack tip is the 3-D 20-node structural
solid (brick) element SOLID186.
• A fine sweep mesh along the crack front yields more accurate results.
• When contour results begin with the first ring of elements around crack tip nodes (PATH 1 in Figure 44.16: C*-
integral Along the Crack Front -- Rectangular Block (p. 678) and Figure 44.20: C*-integral Along the Crack
Front -- X-Joint Pipe with Warped Flaw (p. 682)), the first contour result is discarded. In such cases, experience
has shown that the first contour it is less accurate than other contours due to highly concentrated local
deformation.
• The C*-integral could be path-independent only when steady-state creep deformation dominates the integ-
ration domain around the crack tip.
44.9. References
The following references are used in this example problem:
1. Goldman, H.L., J. W. Hutchinson. "Fully Plastic Crack Problems: The Center-Cracked Strip Under Plane Strain."
International Journal of Solids and Structures. 11 (1975): 575-591.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
682 of ANSYS, Inc. and its subsidiaries and affiliates.
Input Files
2. Riedel, H. "Creep Deformation at Crack Tips in Elastic-viscoplastic Solids." Journal of the Mechanics and
Physics of Solids. 29 (1981): 35-49.
3. Riedel, H., J. R. Rice. "Tensile Cracks In Creeping Solids." Fracture Mechanics: Twelfth Conference. ASTM STP
700 (1980): 112-130.
4. Kumar, V., M. D. German, C. F. Shih. "An Engineering Approach For Elastic-plastic Fracture Analysis." Report
NP-1931, 1981.
5. Kanninen, M. F., C. H. Popelar. Advanced Fracture Mechanics. New York: Oxford University Press, 1985.
• xjoint_pipe_with_warped_flaw.dat -- Input file for the fracture analysis of the x-joint model.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 683
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
684 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 45: Forced Response Analysis of a Mistuned Bladed Disk
with Aerodamping
This example problem demonstrates the free vibration and forced response analyses of a tuned and
mistuned NASA Rotor 67 fan using cyclic modeling methods, linear perturbation, and aero coupling
solution approaches.
The problem includes modal analyses, prestressed modal analyses using linear perturbation, and
prestressed mode-superposition harmonic analyses using linear perturbation. Pressures acting on the
model because of unsteady flow are imported from ANSYS CFX.
The results of the cyclic symmetry analysis are verified against reference results obtained from an ana-
lysis of the full 360° model.
45.1. Introduction
Turbine engine rotors and bladed disks, which play a crucial role in the energy supply and mobility in-
dustries, are dynamic systems that are known to suffer from severe vibration problems. These vibrations
may be introduced by aerodynamic forces. Hence, it is important to study the aeroelastic behavior of
the blades in turbomachinery design.
Aeroelastic phenomena can be classified into two categories: forced response and flutter. Typically,
flutter is an asynchronous self-excited vibration, generally occurring at a frequency corresponding to
one of the lower blade or coupled blade-disk natural frequencies. On the other hand, forced response
resonance of rotor blades generally results from periodic aerodynamic forcing functions with frequencies
of integer multiples of the natural frequency of the system.
In general, a bladed disk is designed to have identical blades, but there are always random deviations
among individual blades due to manufacturing tolerances, wear, and other causes. This is known as
mistuning.
Though mistuning is typically small, mistuned bladed disks can have drastically larger forced response
levels than the ideal tuned design, which can cause blades to fail prematurely due to high cycle fatigue
(HCF). HCF is a major cost, safety, and reliability issue for gas-turbine engines. It is clearly of great interest
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 685
Forced Response Analysis of a Mistuned Bladed Disk with Aerodamping
to be able to predict, and ultimately to limit, the increases in maximum blade response caused by
mistuning.
Comprehensive modeling, analysis, and understanding of bladed disk vibration is therefore critical to
reduce the occurrence of HCF and to improve the performance and reliability of turbomachinery.
The following figure shows a cyclically symmetric sector of the Rotor 67 fan:
The model consists of a disk and a fan blade with a sector angle of 16.364 degrees. The full model
consists of 22 fan blades, as shown:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
686 of ANSYS, Inc. and its subsidiaries and affiliates.
Problem Description
• Modal
The perturbed modal and perturbed mode-superposition harmonic cyclic symmetry analyses include
the initial prestressed condition from nonlinear static analysis. The initial stress state from the nonlinear
static analysis is produced by considering the rotational effects on the fan and the thermal loading
applied on all nodes of the fan model.
In addition, the perturbed mode-superposition harmonic cyclic symmetry analysis also includes unsteady
flow pressure that is acting on the fan blades. This unsteady flow pressure is calculated for the 3-D
viscous flow of air modeled as an ideal gas for a fan running at 16043 RPM with a 10% inlet distortion
using ANSYS CFX.
Modal, perturbed modal, and perturbed mode-superposition harmonic analyses of the full 360° model
are also performed to validate the accuracy of the cyclic-sector model results.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 687
Forced Response Analysis of a Mistuned Bladed Disk with Aerodamping
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
688 of ANSYS, Inc. and its subsidiaries and affiliates.
CFD Modeling
For the CFX analysis only two blades are considered, as shown below:
Figure 45.4: CFX Model Showing the Domain of the NASA Rotor 67 Fan
The Fourier Transformation Inlet Disturbance method is used in this analysis. This method belongs to
the ANSYS Transient Blade Row (TBR) family of transient methods for obtaining a full wheel represent-
ation of the solution data by solving on a few passages per blade row, thereby providing substantial
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 689
Forced Response Analysis of a Mistuned Bladed Disk with Aerodamping
savings in solution time. The Fourier Transformation Inlet Disturbance method used in this study employs
a Double Passage strategy, where the Fourier coefficients are collected on a sampling plane at the in-
terface between the two rotors. Though this increases the domain size, it has been found to provide a
faster convergence than the single passage method.
A steady-state solution is carried out before the transient analysis is performed, and the results from
the steady-state solution are used to provide a good initial condition for the transient case. The transient
solution is carried out for a few revolutions of the blade until the periodicity of the solution is achieved.
The real and imaginary pressures on the blade surfaces are exported at the required engine order in a
.CSV format, which is read into ANSYS Mechanical for carrying out the forced response analysis.
The solution is deemed to be converged when the monitor plots attain periodicity. In addition, it is also
important to check the conservation of mass, momentum, and energy before accepting the final solution.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
690 of ANSYS, Inc. and its subsidiaries and affiliates.
Structural Modeling
These unsteady pressures define the aerodynamic forcing due to the fluid forces. These aerodynamic
data can be written to a file in the .CSV format at a specified engine order to perform further analysis
using ANSYS Mechanical.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 691
Forced Response Analysis of a Mistuned Bladed Disk with Aerodamping
The disk and blade geometries are meshed separately so that the blade elements and the node interface
separate the blade from the disk.
SURF154 elements are created on the surface of the fan blade to apply the pressure load.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
692 of ANSYS, Inc. and its subsidiaries and affiliates.
Structural Material Properties
Figure 45.8: Bonded Contact Pair Between the Disk and the Fan Blade
The contact surface is meshed with CONTA174 elements. The target surface is meshed with TARGE170
elements.
Units
Unit System Metric (m, kg, N, s, V, A)
Angle Degrees
Rotational Velocity rad/s
Temperature Celsius
Pressure Pa
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 693
Forced Response Analysis of a Mistuned Bladed Disk with Aerodamping
The following loads are considered for the cyclic symmetry analyses:
• Unsteady flow pressure applied on the fan blade (imported from CFX)
The rotational velocity (CGOMGA,0,0,1680) is applied along the global Z-axis. The bladed disk is modeled
with a coefficient of thermal expansion of 1.2 x 10-5 °C-1. The reference temperature is held at 22 °C
and an operating temperature of 50 °C is applied on all nodes of the model to generate the thermal
load vector.
The thermal load vector generated from the base static solve can be ignored in the subsequent analysis
(THEXPAND).
The unsteady flow pressures, which are imported from ANSYS CFX, are generated for EO = 2 engine
order excitation at a rotational frequency of 534.76 Hz. These pressure data are then mapped to the
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
694 of ANSYS, Inc. and its subsidiaries and affiliates.
Boundary Conditions and Loading
structural rotor fan blade model in ANSYS Mechanical APDL using its mapping capabilities via the /MAP
processor.
The following example input shows the steps involved in this mapping procedure:
/map ! Mapping processor
target, nbf1 ! Specifies target nodes for mapping pressures onto
! surface effect elements
FTYPE, cfxtbr,1 ! Specifies file and pressure type for the
! subsequent import of source points and pressures
READ,CFX_ExportResults_FT_10P_EO2.csv' ! Read coordinate and pressure data from a file
allsel,all,all
csys,0
/show,jpeg,rev
plgeom ! Plots source and target geometries
MAP,,2,,1, ! Map pressures from source points to target surface elements
plmap,target ! Plot target and source pressure
plmap,target,,,1
/show,close
WRITEMAP,'mapped_on_cyclic_model.dat' ! Write interpolated pressure
! data to a file
finish
To illustrate the step-by-step mapping procedure, the complete workflow in Mechanical APDL is shown
below:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 695
Forced Response Analysis of a Mistuned Bladed Disk with Aerodamping
Using the PLGEOM command, the target and source geometries can be plotted as shown:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
696 of ANSYS, Inc. and its subsidiaries and affiliates.
Boundary Conditions and Loading
Figure 45.10: Target and Source Geometries of the Rotor Fan Blade
The real and imaginary components of mapped target pressure are plotted using the PLMAP command,
as shown:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 697
Forced Response Analysis of a Mistuned Bladed Disk with Aerodamping
Figure 45.11: Mapped Target Pressure Plots on the Rotor Fan Blade Surface
The procedure for computing aerodynamic coefficients (stiffness and damping terms) for each interblade
phase angle is described in Aero Coupling. The AEROCOEFF command uses unsteady pressures obtained
from a CFD flutter analysis to compute an aerodynamic coefficient array. This aerodynamic coefficient
array is compatible with the CYCFREQ,AERO command. If all interblade phase angles are not available
from the CFD computation, the coefficients can be interpolated between nodal diameters and inserted
into the aerodynamic coefficient array.
In the case of a mistuned bladed disk, blades are subjected to engine order (EO) excitation. Engine order
excitation is the effective traveling wave excitation that a bladed disk experiences as it passes through
disturbances in the flow-field for each revolution. For example, one disturbance in the flow field results
in an EO = 1 excitation; two disturbances in the flow field result in an EO = 2 excitation.
In Mechanical APDL, the engine order excitation is applied using the CYCFREQ command with Option
= EO. Internally, the program computes the aliased engine order, including its sign. The aliased engine
order is determined from the input engine order as follows:
Engine Order C
Aliased Engine Order
N Even N Odd
C≤N/2 C ≤ (N - 1) / 2 C
N/2<C≤N-1 2(N + 1) ≤ C ≤ N N-C
N ≤ C ≤ 3N / 2 N ≤ C ≤ (3N - 1) / 2 C-N
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
698 of ANSYS, Inc. and its subsidiaries and affiliates.
Analysis and Solution Controls
Engine Order C
Aliased Engine Order
N Even N Odd
3N / 2 ≤ C ≤ 2N (3N +1) / 2 ≤ C ≤ 2N 2N - C
… … …
For the case of mistuned forced response analysis, the forcing/excitation frequency expressed in terms
of EO is as follows:
In general, mistuning in bladed disks is typically modeled as small, random perturbations to the stiffness
of the blade DOF. The deviation in stiffness for blade n is expressed as follows:
The mistuning parameters are provided in an array parameter of size N × 1, where N is the number of
blades. These mistuning parameters are defined using the CYCFREQ command with Option = MIST
and Value1 = K.
For the perturbed mode-superposition harmonic analysis, the unsteady flow pressures applied as per
the engine order excitation on the fan blade are treated as a harmonically varying load.
For procedure and solution controls for the modal cyclic symmetry (p. 182) analysis and nonlinear
prestessed modal cyclic symmetry analysis with linear perturbation (p. 184), refer to Analysis and Solution
Controls (p. 182) for the Centrifugal Impeller Analysis Using Cyclic Symmetry and Linear Perturba-
tion (p. 177).
For general information about linear perturbation, see Linear Perturbation Analysis in the Structural
Analysis Guide.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 699
Forced Response Analysis of a Mistuned Bladed Disk with Aerodamping
To perform a nonlinear prestressed mode-superposition harmonic cyclic symmetry analysis with linear
perturbation, you must first prestress the structure with nonlinear effects in a static solution. The per-
turbed procedure is used to perform a modal analysis from the prestressed state, followed by a mode-
superposition harmonic analysis.
For this analysis, the EO = 2 engine order excitation is applied using the CYCFREQ command with Op-
tion = EO and Value1 = 2.
The frequency range of excitation, 513.76 - 538.76 Hz with 50 substeps is chosen based on the modal
frequencies for the mode-superposition harmonic analysis. With this frequency range, the first few
modes of the bladed disk are excited.
csys,0
antype,static ! Perform Static analysis
nlgeom,on ! Include large deformation effects
rescontrol,define,all,1 ! Enable the file writing in multiframe restart
nsub,10,10,10 ! Number of substeps = 10
allsel,all,all
nsle
bf,all,temp,50 ! Define temperature = 50
allsel,all,all
csys,1
cgomega, 0, 0, 1680, ! Centrifugal load
time,1.0
solve
finish
/solu
antype,static,restart,,,perturb ! Perform a static restart with perturb from the last ! substep of the previous st
perturb,modal,,,DZEROKEEP ! Set the analysis options for perturbed modal analysis
solve,elform
modopt,lanb,50,0,14000
mxpand,,,,yes,,yes
cycopt,hindex,2,2,2
/input,'mapped_on_cyclic_model',dat ! Apply mapped pressure data to generate a
! load vector
solve
fini
/solu
antype,harmonic
hropt,msup,,,yes ! MSUP Harmonic, Write modal coordinates
outres,all,all
harfrq,513.76,538.76
nsubs,50,50,50
cycfreq,eo,2 ! EO = 2
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
700 of ANSYS, Inc. and its subsidiaries and affiliates.
Analysis and Solution Controls
solve
finish
To perform this analysis with linear perturbation, the procedure is identical to the one described in
Nonlinear Prestressed Mode-Superposition Harmonic Cyclic Symmetry Analysis with Linear Perturbation
- Tuned Response (p. 699), except that mistuning parameters are defined. These mistuning parameters
are specified as an array input using the CYCFREQ command with Option = MIST. For this analysis,
the following mistuning profile is considered:
An EO = 2 engine order excitation is applied using the CYCFREQ command with Option = EO and
Value1 = 2.
The cyclic mode-superposition harmonic analysis with mistuning requires a CMS reduction on the blade
alone. For this, you need to provide the blade information using the CYCFREQ command with Option
= BLADE. This blade information is composed of the name of the nodal component containing the
blade boundary nodes at the blade-to-disk interface, the name of the element component containing
the blade elements, and the number of blade modes to include in the CMS reduction.
The following figure shows the element component of the fan blade and the blade boundary nodes at
the blade-to-disk interface:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 701
Forced Response Analysis of a Mistuned Bladed Disk with Aerodamping
Figure 45.13: (a) Element Component of Fan Blade; (b) Blade Boundary Nodes at the Blade-to-Disk
Interface
To determine modal frequencies and corresponding modal damping ratios, damped modal analyses
for tuned and mistuned models are performed on the reduced systems using the CYCFREQ command
with Option = MODAL, and Value1 = ON.
For the forced response with aerodamping and mistuning, a frequency range of excitation of 503.76 -
553.76 Hz (with 50 substeps) is considered.
The following example input shows the different steps in this analysis:
!-----------------------------------------------
! DAMPLED MODAL SOLVE (MISTUNED)
!-----------------------------------------------
/output,scratch
/solu
antype,harm
hropt,msup
thexpand,off ! Ignore thermal strains
kbc,1
harfrq,503.76,553.76
nsubst,50,50,50
dmpstr,0.01 ! Structural damping
outres,all,all
hrout,on
lvscale,1,1
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
702 of ANSYS, Inc. and its subsidiaries and affiliates.
Analysis and Solution Controls
cycfreq,default
cycfreq,eo,2
cycfreq,blade,blade_interface,BladeElem,2
*dim,kmist,array,22,1
kmist(1,1) = 0.01,-0.02,0.01,0.04,-0.03,-0.01,-0.02,0.04,0.05
kmist(10,1) = -0.03,-0.02,0.03,-0.04,-0.01,-0.04,0.02,-0.03,0.01,
kmist(19,1) = -0.05,-0.03,0.01,0.01
cycfreq,mist,k,kmist
cycfreq,aero,fileAeroArray
cycfreq,modal,on,100,500,600
allsel,all
/output
solve
/output,scratch
fini
/output
!-----------------------------------------------
! CYCLIC MSUP HARMONIC SOLVE
!-----------------------------------------------
/output,scratch
/solu
antype,harm
hropt,msup
thexpand,off ! Ignore thermal strains
kbc,1
harfrq,503.76,553.76
nsubst,50,50,50
dmpstr,0.01 ! Structural damping
outres,all,all
hrout,on
lvscale,1,1
cycfreq,default
cycfreq,eo,2
cycfreq,blade,blade_interface,BladeElem,2
cycfreq,mist,k,kmist
cycfreq,aero,fileAeroArray
allsel,all
solve
Note
For an ideal/tuned cyclic bladed disk, engine order excitation will excite only those modes
with a number of nodal diameters that match the harmonic index of the excitation. For a
mistuned bladed disk, the modes have multiple harmonic contents so that many modes will
be excited by engine order excitation.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 703
Forced Response Analysis of a Mistuned Bladed Disk with Aerodamping
Mode No. Cyclic Sector Model (Hz) Full 360° Model (Hz) Error (%)
1 370.17 370.17 0.00
2 370.17 370.17 0.00
3 376.70 376.70 0.00
4 376.70 376.70 0.00
5 377.12 377.12 0.00
6 377.12 377.12 0.00
7 377.41 377.41 0.00
8 378.93 378.92 0.00
9 378.93 378.93 0.00
10 380.42 380.42 0.00
11 380.42 380.42 0.00
12 381.65 381.65 0.00
13 381.65 381.65 0.00
14 382.65 382.65 0.00
15 382.65 382.65 0.00
The nonlinear (NLGEOM,ON) prestressed modal frequencies of the cyclic sector obtained from the linear
perturbation analysis also show strong agreement with the full-model results, as shown in the following
table:
Mode No. Cyclic Sector Model (Hz) Full 360° Model (Hz) Error (%)
1 519.09 519.09 0.00
2 521.05 521.05 0.00
3 521.05 521.05 0.00
4 522.69 522.70 0.00
5 522.70 522.70 0.00
6 526.38 526.39 0.00
7 526.39 526.39 0.00
8 528.12 528.12 0.00
9 528.12 528.12 0.00
10 529.65 529.65 0.00
11 529.65 529.65 0.00
12 531.07 531.07 0.00
13 531.07 531.07 0.00
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
704 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Mode No. Cyclic Sector Model (Hz) Full 360° Model (Hz) Error (%)
14 532.29 532.29 0.00
15 532.29 532.29 0.00
The prestressed modal frequencies can be plotted as a function of the number of harmonic index
(nodal diameter) using the PLZZ command:
Modes dominated by blade motion tend to appear in a tight frequency range in the plot. In contrast,
the modal stiffness of a disk increases rapidly as the harmonic index (nodal diameter) increases, and so
disk-dominated system modes appear as slanted dotted lines in the plot. This plot is also known as a
SAFE diagram or a ZZENF diagram and indicates potential frequencies where resonance may occur.
To postprocess the results of a cyclic symmetry mode-superposition harmonic analysis, you must first
issue the CYCFILES command in /POST1.
For the nonlinear prestressed tuned response analysis, the following figures compare the postprocessing
results of the cyclic-sector model to the full 360° model:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 705
Forced Response Analysis of a Mistuned Bladed Disk with Aerodamping
Figure 45.15: Total Deformation Pattern (Real Only) at Frequency of 522.76 Hz in the Tuned
Analysis
Figure 45.16: Equivalent Stress Pattern (Real Only) at Frequency of 522.76 Hz in the Tuned Analysis
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
706 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Figure 45.17: Equivalent Strain Pattern (Real Only) at Frequency of 522.76 Hz in the Tuned Analysis
These results show strong agreement between the cyclic-sector model and the full 360° model. The
nodal solution plots (NSOL) show the amplitude of a nodal degree-of-freedom (DOF) value with respect
to the frequency of excitation:
Figure 45.18: Nodal Solution Plots with Respect to the Frequency of Excitation in the Tuned
Analysis
The plots indicate the occurrence of a resonance condition at 522.70 Hz. The displacement amplitude
of nodes at symmetric angular positions on sectors 1 through 10 is plotted. The plots show strong
agreement between the cyclic and full-model results.
You can also retrieve a maximum result value which occurs among all sectors of the model and all fre-
quency points using CYCSPEC and CYCCALC. With these commands, you can extract tables of displace-
ment, stress, and/or strain data for all computed harmonic solutions and for all sectors. The extracted
data can be plotted using PLCFREQ and PLCHIST.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 707
Forced Response Analysis of a Mistuned Bladed Disk with Aerodamping
For the tuned response analysis, the maximum directional deformation at the blade tip along the Z-
direction is plotted using PLCFREQ, as shown in the following figure:
Figure 45.19: Directional Deformation at the Blade Tip Along the Z-Direction in the Tuned Analysis
The plot clearly indicates the sector numbers which contain the maximum and minimum values for the
directional deformation along the Z-direction.
PLCHIST plots a histogram of the directional deformation along the Z-direction at the requested fre-
quency of 522.76 Hz; one bar corresponds to one sector:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
708 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Figure 45.20: CYCSPEC Histogram Response Plot at Frequency 522.76 in the Tuned Analysis
For the nonlinear prestressed mistuned response analysis with aerodamping, the following diagram
shows the variation of aerodynamic stiffness and damping values as a function of nodal diameter:
Figure 45.21: Aerodynamic Stiffness and Damping Coefficients vs. Nodal Diameter
The following figure shows plots of modal damping ratio versus modal frequencies obtained from the
damped modal analyses (with aerodamping) of the tuned and mistuned models, respectively:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 709
Forced Response Analysis of a Mistuned Bladed Disk with Aerodamping
Figure 45.22: Modal Damping Ratio vs. Modal Frequency for Tuned and Mistuned Models
The above figure shows that the frequency range in the mistuned case has increased compared to the
tuned case, while the modal damping ratio range has decreased.
The following figures show the postprocessing (/POST1) results for the nonlinear prestressed harmonic
analysis considering both aerodamping and mistuning:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
710 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Figure 45.23: Total Deformation Pattern (Real Only) at Frequency of 522.76 Hz in the Mistuned
Analysis
Figure 45.24: Equivalent Stress Pattern (Real Only) at Frequency of 522.76 Hz in the Mistuned
Analysis
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 711
Forced Response Analysis of a Mistuned Bladed Disk with Aerodamping
Figure 45.25: Equivalent Strain Pattern (Real Only) at Frequency of 522.76 Hz in the Mistuned
Analysis
The amplitudes of a nodal degree-of-freedom (DOF) value with respect to the frequency of excitation
are shown in the following nodal solution plots (NSOL):
Figure 45.26: Nodal Solution Plots with Respect to the Frequency of Excitation in the Mistuned
Analysis
The displacement amplitude of nodes at symmetric angular positions in sectors 1 through 10 is plotted.
Note that both aerodamping and mistuning effects are considered here.
The maximum directional deformation at the blade tip along the Z-direction is plotted using PLCFREQ,
as shown in the following figure:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
712 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Figure 45.27: Directional Deformation at the Blade Tip Along the Z-Direction in the Mistuned
Analysis
The mistuned result in Figure 45.27: Directional Deformation at the Blade Tip Along the Z-Direction in
the Mistuned Analysis (p. 713) shows several peaks as compared to a single peak in the tuned result in
Figure 45.19: Directional Deformation at the Blade Tip Along the Z-Direction in the Tuned Analysis (p. 708).
The sector numbers which contain the maximum and minimum values for the directional deformation
along the Z-direction are indicated. The added aerodamping and mistuning also result in reduced
maximum amplitude.
PLCHIST plots a histogram of the directional deformation along the Z-direction at the requested fre-
quency of 522.76 Hz; one bar corresponds to one sector:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 713
Forced Response Analysis of a Mistuned Bladed Disk with Aerodamping
Figure 45.28: CYCSPEC Histogram Response Plot at Frequency 522.76 Hz in the Mistuned Analysis
45.9. Recommendations
To perform a similar type of analysis, consider the following recommendations:
• The bladed disk must be meshed such that the blade elements and the node interface separating the blade
from the disk and/or the platform are readily accessible.
• For good interpolation of the pressure data while mapping between the fluid-structural domains, you should
define the mesh density of the structural blade closely to the mesh density of the CFD model.
• Be sure to extract all modes that may contribute to the harmonic response. As a general guideline, modes
contributing to the harmonic response fall in the range to , where is the harmonic frequency
(HARFRQ) used in the subsequent harmonic solution. If mistuned, a range of to is recommended.
• For the case of a mistuned response analysis, all harmonic indices must be included.
45.10. References
The following references are used in this example problem:
1. Lim, S.H., et al.“Compact, Generalized Component Mode Mistuning Representation for Modeling Bladed
Disk Vibration.” AIAA Journal. 45.9 (2007): 2285-2298.
2. Singh, M.P., et al.“SAFE Diagram: A Design and Reliability Tool for Turbine Blading.” 17th Turbomachinery
Symposium. Dallas, TX (1988).
3. Castanier, M.P., C. Pierre.“Modeling and Analysis of Mistuned Bladed Disk Vibration: Status and Emerging
Directions.” Journal of Propulsion and Power. 22.2 (2006).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
714 of ANSYS, Inc. and its subsidiaries and affiliates.
Input Files
Rzadkowski, R., et al. “Forced Response of the Mistuned First Stage Compressor Bladed Disc of an
Aircraft Engine - Experimental and Numerical Results.” 13th World Congress in Mechanism and Machine
Space: Guanajuato, Mexico (2011).
4. Elder, R., S. Patil, W. Homes.“Investigation of Efficient CFD Methods for the Prediction of Blade Damping.”
Proceedings of ASME Turbo Expo 2013: San Antonio, TX, (2013).
5. Zhang, C., Z. Ye, F. Liu.“Numerical Researches on Aeroelastic Problem of a Rotor Due to IGV/Fan Interaction,”
AIAA Aerospace Sciences Meeting Including the New Horizons Forum and Aerospace Exposition: Orlando, FL
(2009).
Cyclic-Sector Model:
• sector_FE_model.dat – The common input file for the cyclic-sector model of the rotor fan blade.
• cyc_modal_analysis.dat – Input file for the cyclic symmetry modal analysis of the rotor fan blade.
• cyc_pre_modal_analysis.dat – Input file for the cyclic symmetry nonlinear prestressed perturbed
modal analysis of the rotor fan blade.
• full_360_FE_model.dat – The common input file for the full model analysis of the rotor fan blade.
• full_360_FE_model_mist.dat – The common input file for the full model analysis of the mistuned
rotor fan blade.
• full_modal_analysis.dat – Input file for the modal analysis of the full model rotor fan blade.
• full_pre_modal_analysis.dat – Input file for the nonlinear prestressed perturbed modal ana-
lysis of the full model rotor fan blade.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 715
Forced Response Analysis of a Mistuned Bladed Disk with Aerodamping
For more information, see Obtaining the Input Files (p. xxxvii).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
716 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 46: Surface Subsidence Caused by Reservoir Depletion
This example analysis predicts solid deformation caused by coupling pore-pressure diffusion and solid-
matrix deformation. The demonstration case consists of surface subsidence occurring on a disc-shaped
compartmentalized reservoir in a homogeneous poro-elastic continuum.
46.1. Introduction
A hydrocarbon-bearing reservoir is buried deep in the subsurface and typically exists in a high-pore-
pressure state. The weight of the formation deposited above the reservoir (overburden) is supported
by both the rock matrix and the fluid pore pressure. During oil or gas production, the pore pressure
depletes and the overburden compacts the reservoir, causing the surface to subside.
To obtain the necessary legal permits to begin production from a reservoir, surface subsidence must
be limited in accordance with the requirements of the given jurisdiction. Minimal subsidence is especially
critical when the reservoir is located within the vicinity of a populated area.
The example simulation presented here predicts the surface subsidence caused by a uniform pressure
depletion of a completely compartmentalized disc-shaped reservoir. The problem considers the effects
of reservoir depth, radius, and height on compaction and surface subsidence. The correlation of maximum
surface subsidence to geometric parameters is shown to closely match the trends of the reference
analytical solution.[1]
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 717
Surface Subsidence Caused by Reservoir Depletion
To model the effect of subsurface as a half-space, the width of the domain is approximately four times
the radius of the reservoir. The reservoir is depleted uniformly.
46.3. Modeling
The computational domain is modeled as a fully saturated poro-elastic continuum. The continuum is
discretized with finite elements with pore pressure and displacement degrees of freedom using coupled
pore-pressure-thermal mechanical solid ( CPT nnn) elements.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
718 of ANSYS, Inc. and its subsidiaries and affiliates.
Boundary Conditions and Loading
The interface between the reservoir and overburden formation is modeled using surface-based contact
with both displacement and pore-pressure degrees of freedom. The contact is modeled as bonded and
impermeable with respect to displacement and pore-pressure degrees of freedom, respectively.
The following input modifies the real constants to model impermeable contact:
! REAL CONSTANT MODIFICATION (IMPERMEABLE CONTACT)
RMODIF,6,37,0
RMODIF,7,37,0
RMODIF,8,37,0
RMODIF,9,37,0
RMODIF,10,37,0
RMODIF,11,37,0
Property Value
-2
Young's Modulus (Nm ) 25.0
Poisson's Ratio 0.25
Solid Bulk Modulus (GPa) 37.0
Solid Specific Weight (g/cc) 2.65
Biot Coefficient 1.0
Porosity 0.3
Fluid Specific Weight (g/cc) 1.0
Permeability (mD) 100
Pressure Depletion (MPa) 10
The following input creates node-set components for all boundary faces except the top face of the
domain:
! CREATING THE NODE SETS FOR BOUNDARY CONDITIONS
NSEL,S,LOC,Z,-DEPTH
CM,BOTTOM,NODE
ALLSEL
NSEL,S,LOC,X,WIDTH/2
CM,POSX,NODE
ALLSEL
NSEL,S,LOC,X,-WIDTH/2
CM,NEGX,NODE
ALLSEL
NSEL,S,LOC,Y,WIDTH/2
CM,POSY,NODE
NSEL,S,LOC,Y,-WIDTH/2
CM,NEGY,NODE
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 719
Surface Subsidence Caused by Reservoir Depletion
CMSEL,A,NEGX
D,ALL,UX,0
ALLSEL
CMSEL,S,POSY
CMSEL,A,NEGY
D,ALL,UY,0
ALLSEL
CMSEL,S,BOTTOM
D,ALL,UZ,0
ALLSEL
To model reservoir depletion, the following input decreases the pore pressure by DP (=10 MPa):
! DEPLETION
CMSEL,S,RESERVOIR
NSLE,S,CORNER
D,ALL,PRES,-DP
As shown in the following two figures, the normal surface displacement for various values of η(=D/R)
and H are plotted as a function of radial distance normalized with the radius of the reservoir:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
720 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 721
Surface Subsidence Caused by Reservoir Depletion
The maximum subsidence occurs at origin and decays rapidly to zero at a distance of about four times
the radius of the reservoir.
Geertsma[1] (Eq. 6) shows analytically that the maximum subsidence is a quadratic function of η(=D/R)
and a linear function of reservoir height (H):
As shown in the following figure, the numerical results of this example problem match the analytical
trends very well:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
722 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Figure 46.5: Maximum Subsidence as a Quadratic Function of η(=D/R) and a Linear Function of
Reservoir Height (H)
The results indicate that subsidence increases rapidly with a decrease in reservoir depth as compared
to an increase in reservoir height.
The following two figures show typical contour plots for surface subsidence and reservoir compaction,
respectively:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 723
Surface Subsidence Caused by Reservoir Depletion
46.8. Recommendations
To perform a similar type of surface-subsidence analysis, consider the following recommendations:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
724 of ANSYS, Inc. and its subsidiaries and affiliates.
Input Files
• If possible, use sweep meshing in the reservoir and in the domain above and below it.
• Because the global stiffness matrix is unsymmetric, use the Newton-Raphson solver with the unsymmetric
option (NROPT,UNSYMM) to avoid convergence issues.
46.9. References
The following reference is used in this example problem:
1. Geertsma, J. "Land Subsidence Above Compacting Oil and Gas Reservoirs." Journal of Petroleum Technology.
Society of Petroleum Engineers. SPE-3730-PA. 25.6 (1973): 734.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 725
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
726 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 47: Electromigration in a Solder Ball
This example problem is a transient electromigration analysis of a solder ball. The finite element solution
calculates the deviation in atomic concentration from an initial unit value due to the combined effect
of diffusion, electromigration, stress migration, and thermomigration.
47.1. Introduction
Electromigration is a process of mass transport in metal interconnects induced by high density electrical
currents. It is a key failure mechanism in integrated circuits where current densities are high due to
miniaturization. The mass transport of metal atoms can result in the formation of hillocks, whiskers, and
voids, all leading to the electrical failure of the circuit.
The properties affecting electromigration are highly temperature dependent, and the four fields in-
volved—structural, electrical, thermal, and diffusion—are coupled in many ways. For example, electrically-
driven metal diffusion and thermal expansion due to Joule heating cause compression in the conductor
(back stress) which can retard and ultimately stop electromigration.
The coupled-field element used in this example employs strong (matrix) coupling, which is essential to
obtaining convergence of the four fields. By modeling the four fields simultaneously, you can conveniently
specify all the needed material properties and coupling effects in a single analysis.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 727
Electromigration in a Solder Ball
The distance between the conductors is 450 μm. The diameter of the solder ball is 760 μm, and its
width where it meets the conductors is 612 μm. These dimensions approximately correspond to the
ball grid array (BGA) structure [1 (p. 737)].
The conductors are 40 μm thick, 800 μm wide (400 μm in the half symmetry model), and 1000 μm long.
47.3. Modeling
The simple geometry for this example is created and meshed in Mechanical APDL.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
728 of ANSYS, Inc. and its subsidiaries and affiliates.
Material Model and Contact Properties
The model is meshed with the SOLID226 coupled-field element. CONTA174 contact elements are defined
between the solder ball and the copper conductors. Both the SOLID226 and CONTA174 elements have
the following degrees of freedom needed for this analysis: concentration (CONC), temperature (TEMP),
voltage (VOLT), and displacement (UX, UY, UZ). For the SOLID226 elements, KEYOPT(1) = 100111 activates
these degrees of freedom. For the CONTA174 elements, KEYOPT(1) = 12 activates these degrees of
freedom.
The constants and material coefficients in μMKSV units are used as described in the tables below.
Most of the material properties for copper and SAC were taken from [2 (p. 737)]. Some copper material
properties, such as the pre-exponential diffusivity coefficient, the activation energy of diffusion, and
the change number, were selected from [3 (p. 737)].
It is assumed that the electrical resistivity of the solder and the copper plates is not affected by temper-
ature or concentration. The proper direction of atomic flux with respect to the electric current density
is specified by the negative sign of the charge numbers. The coefficients of diffusion expansion are
specified as positive to ensure the correct sign of the diffusion strain for back stress calculation.
Constants
Boltzmann constant kB (pJ/K) 1.3806488E-11
Boltzmann constant kB_eV (eV/K) 8.6173324E-5
Universal gas constant R (J/(K*mol)) 8.31445
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 729
Electromigration in a Solder Ball
Constants
Reference temperature for thermal 25
strain calculation (°C)
Reference normalized concentration 1.0
for diffusion strain calculation
Copper
Property Value MP Label
Young’s modulus (MPa) 127.7E3 EX
Poisson’s ratio 0.31 PRXY
Electrical resistivity (TOhm*μm @ 2.38E-14 RSVX
200°C)
Pre-exponential diffusivity ((μm)2/s) 7.8E7 DXX
Thermal conductivity 393E6 KXX
(pW/(μm*K))
Specific heat (pJ/(kg*K) 385.2E12 C
Density (kg/(μm)3) 8900E-18 DENS
Coefficient of thermal 17.1E-6 ALPX
expansion (1/°C)
Coefficient of diffusion 1E-5 BETX
expansion
Property Value TBDATA Constant
Activation energy of diffusion 210E3 C1 = Qa/R
Qa (J/(K*mol))
Atomic volume V1 (μm)3 1.182E-11 C2 = V1/kB
Charge number Z1 -4 C4 = Z1/kB_eV
SnAgCu (SAC)
Property Value MP Label
Young’s modulus (MPa) 26.2E3 EX
Poisson’s ratio 0.35 PRXY
Electrical resistivity (TOhm*μm 20.75E-14 RSVX
@ 200°C)
Pre-exponential diffusivity 4.1E7 DXX
2
((μm) /s)
Thermal conductivity (pW/(μm*K)) 57E6 KXX
Specific heat (pJ/(kg*K)) 219E12 C
3
Density (kg/(μm) ) 7390E-18 DENS
Coefficient of thermal 23E-6 ALPX
expansion (1/°C)
Coefficient of diffusion expansion 1E-5 BETX
Property Value TBDATA Constant
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
730 of ANSYS, Inc. and its subsidiaries and affiliates.
Boundary Conditions and Loading
SnAgCu (SAC)
Activation energy of diffusion Ea 0.8 C1 = Ea/kB_eV
(eV)
Atomic volume V2 (μm)3 2.71E-11 C2 = V2/kB
Heat of transport Q (eV) 0.0094 C3 = Q/kB_eV
Charge number Z2 -23 C4 = Z2/kB_eV
The input listing below demonstrates how to use the above properties to define the migration model
for solder.
kB=1.3806488e-23*1.e12 ! Boltzmann constant, pJ/K
kB_eV=8.6173324e-5 ! Boltzmann constant, eV/K
Following are the contact element properties defined via CONTA174 real constants.
Contact Properties
Property Real Value
Constant
Structural stiffness FKN 1.0 times underlying element stiffness
2
Thermal conductance (pW/((μm) *°C)) TCC 1E6
Electrical conductance ECC 1E14
(pA/((μm)2*Volt))
Diffusivity coefficient ((μm)3/s) DCC 1E6
• Electrical The end of one lead is grounded while a current of (2.85x1012)/2 pA (for the half model) is step-
applied to the end of the other lead. The VOLT degree of freedom is coupled to distribute the current evenly.
This produces a current density of 8.9x107 pA/(μm)2 (or 8.9x107 A/m2) in the copper leads. This is a typical
current density for components vulnerable to electromigration.
nsel,s,loc,x,600 ! Ground the nodes at the end of one conductor
d,all,volt,0
nsel,s,loc,x,-600 ! Apply a uniform current at the end of the other conductor
cp,1,volt,all ! Couple the VOLT degree of freedom
nd=ndnext(0)
alls
• Thermal A convection boundary condition with a film coefficient of 20 pW/(μm)2(°C) to a bulk temperature
of 50°C is specified for all surfaces except the symmetry plane. This film coefficient is the same in units of
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 731
Electromigration in a Solder Ball
Watts and meters; it is a low value and is two or three times what is produced by natural convection. A more
realistic way to model heat dissipation out of the solder ball would be to include thermal conduction through
the materials that surround the solder ball.
An initial temperature of 50°C is specified. The temperature offset from absolute zero to zero is set
to 273 degrees.
nsel,s,ext ! Select nodes on the surface of the model
nsel,u,loc,z,0 ! Unselect nodes on the symmetry plane
sf,all,conv,20,50 ! hc in pW/(µm)(°C), tbulk in deg C
nsel,all
toffs,273 ! Temperature offset from absolute zero to zero
...
ic,all,temp,50 ! Model temperature at start of transient
The UY displacements at the top and bottom surfaces of the model are constrained to simulate resistance
to thermal and diffusion expansion by the passivation and other interface layers.
nsel,s,loc,y,-265 ! Select nodes at top and bottom of conductors
nsel,a,loc,y,265
d,all,uy,0 ! Fully constrain the normal DOF to simulate the
! the structural restraint by the adjacent layers
The UX displacements are constrained at the lower left end of the copper plate.
nsel,s,loc,x,-600
d,all,ux,0
• Diffusion An initial unit normalized concentration is specified. No concentration sink is specified at the
ends of the copper conductors. Diffusion through the copper is very low, and electromigration essentially
occurs at the blocked interface between the solder and the copper.
ic,all,conc,1 ! Specify the initial normalized concentration to be 1
Structural
Hydrostatic stress results are in units of MPa. The gradient of hydrostatic stress produces diffusion from
high to low "pressure". The stresses are due to the constraint on the top and bottom surface of the
model and thermal strain incompatibility between the solder and the copper.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
732 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The figure below was produced with the command PLNSOL,NL,HPRES. A large negative hydrostatic
stress occurs at the singularity produced by the sharp re-entrant corner at the edge of the solder/copper
interface.
Thermal
Because the model is very small and the materials have high thermal conductivity, the temperature
reaches the steady-state in a few seconds and remains constant throughout the simulation. Therefore,
the gradient of temperature does not contribute to atomic diffusion. The uniform temperature increase
does affect diffusion by producing stress gradients due to constrained thermal expansion. The figure
below was produced with the command PLNSOL,TEMP.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 733
Electromigration in a Solder Ball
Electrical
Current density is in units of pA/(μm)2 or A/m2. The figure below was produced by the command
PLNSOL,JC,SUM. Note the increase in current density (current crowding) at the entrance and exit of the
solder ball. This is the location where metal depletion has been observed in solder balls.
Diffusion
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
734 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Regions with concentration values less than 1.0 may produce voids. Concentrations greater than 1.0
could produce hillocks or protrusions of metal from the surface. The figure below was produced with
the command PLNSOL,CONC for the solder joint nodes.
Figure 47.8: Minimum Concentration vs. Time (p. 736) and Figure 47.9: Maximum Concentration vs.
Time (p. 736) show the evolution of minimum and maximum concentrations with time, respectively.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 735
Electromigration in a Solder Ball
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
736 of ANSYS, Inc. and its subsidiaries and affiliates.
Input Files
The concentration plots show that the steady state occurs after 1 year (or around time = 3x107 sec).
47.8. Recommendations
To perform a similar type of analysis, consider the following hints and recommendations:
• The elements CONTA174 (3-D surface-to-surface contact) and TARGE170 (3-D surface-to-surface target)
support the TEMP, VOLT, and CONC degrees of freedom. Use these contact elements to connect dissimilar
meshes, to model contact resistance, or to model imperfect contact.
• Use large time steps in a transient analysis. For example, in this problem time steps of 1x106 seconds and
larger are used. Small steps are unnecessary and may contribute to spatially varying concentration oscillations.
A steady-state (static) analysis is also possible.
• Use the postprocessing commands PLVECT,CG (concentration gradient) and PLVECT,DF (diffusion flux) to
observe atomic divergence.
• Due to very low atomic diffusivity, materials such as copper may show spatially varying concentration oscil-
lations. In such cases, a tetrahedral mesh of these regions may be preferred.
47.9. References
The following references are used in this example problem:
1. H.Liu, C. Yu, P. Li, and J. Chen,“Current crowding and its effects on electromigration and interfacial reaction
in lead-free solder joints,” J. Electron. Packaging, vol.130, pp.59-63, 2008.
2. S. Wang and L. Liang,“Solder joint reliability under electromigration and thermal-mechanical load,” Proc.
IEEE Electronic Components and Technology Conference (ECTC 07), pp. 1074-1083, 2007.
3. B. Chao, S.-H. Chae, X. Zhang, K.-H. Lu, J. Im, and P.S. Ho “Investigation of diffusion and electromigration
parameters for Cu-Sn intermetallic compounds in Pb-free solders using simulated annealing,” Acta Mater.,
vol. 55, pp. 2805-2814, 2007.
• solder_ball_contact.dat -- Input file for the transient analysis of the solder ball model.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 737
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
738 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 48: Active and Passive Lateral Earth-Pressure Analysis
This example problem simulates soil behavior under active and passive earth-pressure loadings. The
problem shows how the nonlinear plastic behavior of soil can be modeled with a Mohr-Coulomb ma-
terial.
• Geomechanical materials
• Multiframe restart
48.1. Introduction
Soil-structure interaction occurs when part of an engineering structure exchanges stresses with the
surrounding ground. Typical structures for which such interaction must be considered include ground
foundations, piles, and retaining walls. By assessing the stress exerted on an engineering structure by
the ground, and the stress exerted on the ground by the structure, a safety evaluation based on the
results can be performed during or after construction [1].
Initially, the in situ stress state is applied by considering the gravitational acceleration of the soil. After-
wards, two stress states are generated.
An active earth-pressure state is created by moving the retaining wall away from the soil. A passive
earth-pressure state is produced by moving the retaining wall towards the soil. Both stress states are
strongly influenced by the plastic deformation of the retained soil.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 739
Active and Passive Lateral Earth-Pressure Analysis
Figure 48.1: Soil Geometry and Active/Passive Movement of Left Retaining Wall
48.3. Modeling
A 2-D model is considered. The element formulation is based on linear quadrilateral PLANE182 elements
using an enhanced strain formulation under plane strain conditions (KEYOPT(1) = 2, KEYOPT(3) = 2).
The element edge size is fixed to 1 m.
The active and passive earth-pressure states are generated in subsequent load steps via a multiframe
restart.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
740 of ANSYS, Inc. and its subsidiaries and affiliates.
Analysis and Solution Controls
The retaining wall on both sides of the soil is assumed to be frictionless; therefore, only a fixed horizontal
movement is necessary.
The bottom movement of the soil structure is supported via rolling boundary conditions.
In the first load step, an in situ earth-pressure state is generated by applying a gravitational acceleration
of on the soil in the vertical direction.
In a following load step, an active earth-pressure state is generated by horizontally moving the soil
nodes of the left boundary away from the soil using a specified displacement of ux = -0.075 m.
Alternatively, a passive earth-pressure state is produced by horizontally moving the soil nodes of the
left boundary towards the soil using a specified maximum displacement of ux = 0.5 m.
The active and passive pressure-state loads are applied via 10 initial and 100 maximum substeps with
automatic time-stepping enabled.
Figure 48.2: Vertical Displacement for Gravity Load Step (Initial Stresses Not Applied)
Generally, the soil exists in an already consolidated state. Initial displacements due to at-rest loads are
therefore unnatural and should be minimized.
The vertical stress state varies linearly with the depth of the soil, expressed as:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 741
Active and Passive Lateral Earth-Pressure Analysis
where:
= soil density
= gravitational acceleration
= vertical height of eachelement
The coefficient of lateral earth pressure is the ratio of horizontal to vertical stress components. For a
horizontally retained non-overconsolidated soil under elastic loading conditions, it is defined (via Poisson’s
ratio ) as:
The following figure shows that the initial at-rest pressure state is correctly applied, while the soil
structure retains its initial shape:
Figure 48.3: Vertical Displacement for Gravity Load Step (Initial Stresses Applied)
1. Determine the initial stress state via a standard finite element calculation.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
742 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
3. Recalculate the initial state using the initial calculation results in the .ist file.
Example 48.1: Recalculating Initial Stress State Using Results from a Prior Calculation
/solu ! Enter the solution processor
time,1
! Solve the first load step and finish the solution processor
solve
finish
ANTYPE,STATIC,NEW
time,1
! Activate the restart option for use in the active/passive pressure load steps
rescontrol,define,1,last
! Solve the first load step again and finish the solution processor
solve
finish
Both the active and passive pressure-state conditions are generated via a multiframe restart (ANTYPE,,RE-
START) of the initial in situ stress step.
time,2
! Specify a restart at the last substep of the first load step
antyp,,restart,1,last
…
! Solve the second load step and finish the solution processor
solve
finish
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 743
Active and Passive Lateral Earth-Pressure Analysis
The coefficient of lateral earth pressure for the in situ stress state is consistent with the relationship
using Poisson’s ratio (p. 741) as:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
744 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Figure 48.5: Horizontal Displacement for Active Pressure Load Step (Negative Displacement of
Retaining Wall in x-Direction)
For this type of problem, the coefficient of lateral earth pressure for an active stress state is solely a
function of the friction angle [3], so that:
The result of the nonlinear soil structure analysis is comparable to this value. Also, in most of the
domain, consistent with theoretical assumptions. The failure mode is shearing-dominated.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 745
Active and Passive Lateral Earth-Pressure Analysis
Figure 48.6: Equivalent Plastic Strains for Active Pressure Load Step
Passive earth pressure is the amount of stress that a soil structure can exert on an actively loaded
structure that interacts with the soil. It is generally much higher than the in situ stress state. The following
figure shows the horizontal movement of a typical passive loading condition:
Figure 48.7: Horizontal Displacement for Passive Pressure Load Step (Positive Displacement of
Retaining Wall in x-Direction)
Using failure-state analysis [3], the coefficient of lateral earth pressure for this passive stress state is
again a function of the friction angle, so that:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
746 of ANSYS, Inc. and its subsidiaries and affiliates.
Recommendations
For the passive stress state, therefore, the horizontal stress components are approximately 3x the ver-
tical in situ stress along the retaining wall:
Figure 48.8: Equivalent Plastic Strains for Passive Pressure Load Step
The soil undergoes plastic deformation to a depth of approximately 10 m. Due to the large hydrostatic
stress state in the bottom region, no plastic strains develop.
The results of the at-rest pressure step, the active pressure step, and the passive pressure step are
consistent with theoretical assumptions. The analysis correctly predicts the complex stress state of this
soil-structure interaction problem.
48.8. Recommendations
When setting up a solid-structure interaction analysis, consider the following recommendations:
• Recommended element types for 2-D and 3-D soil analyses are:
• Applying an in situ stress state should result in an elastic deformation state. If this is not the case, the loading
conditions and material properties may be incompatible.
• To ensure conformity of the specified stress state and outside loading, always apply initial state (INISTATE)
in a single substep.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 747
Active and Passive Lateral Earth-Pressure Analysis
48.9. References
The following references are used in this example problem:
1. DIN EN 1997-1:2014-3 (E), "Eurocode7: Geotechnical Design – Part 1: General Rules." 2014.
2. Lazebnik, G. E., G. P. Tsinker. "Lateral Earth Pressure 'At Rest.'" Monitoring of Soil-Structure Interaction: Instru-
ments for Measuring Soil Pressures. 165-183. New York: Springer, 1998.
3. Rankine, W.J.M. "On the Stability of Loose Earth." Philosophical Transactions of the Royal Society of London.
147 (1857): 9-27.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
748 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 49: Load-Limit Analysis of a Reinforced Concrete Slab
This example problem is a load-limit analysis of a reinforced concrete slab. The problem uses a modified
Drucker-Prager material to model the nonlinear plastic behavior of concrete.
• Geomechanics
49.1. Introduction
Reinforced concrete, widely used in the construction industry, is a composite material of concrete (as
the base matrix material) and steel reinforcing structures.
The concrete base material can resist large compressive stresses but fails quickly under tensile loading.
Steel reinforcing absorbs the tensile stresses so that the composite material can better withstand
compressive and tensile loadings.
The engineering design process for concrete structures includes determining the load limits at which
a given structure fails. The problem presented here demonstrates a load-limit analysis for a reinforced
concrete slab under surface-pressure loading [1].
The structure is affected by both its dead load and a surface-pressure load. The surface-pressure load
is increased until failure occurs.
The border of the structure is supported in the vertical y direction. Due to symmetry conditions, it is
possible to simplify the model. The finite element model includes the symmetry boundary conditions
on the cutting planes of the resulting quarter model:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 749
Load-Limit Analysis of a Reinforced Concrete Slab
Figure 49.1: Quarter Model: Concrete Slab and Depiction of Dirichlet Boundary Conditions
The slab is composed of concrete as the base matrix material incorporating a steel grid reinforcing
structure to absorb stresses that might develop under bending movement.
The reinforcing grid lies at a depth of hr = 0.15 m. The armoring has an equivalent distribution area of
Asx = 3.39cm2/m in the x direction, and Asz = 1.57cm2/m in the orthogonal z direction. The distance
between successive rebar elements is Δdx = 0.8m for elements acting in the global x direction and Δdz
= 0.6m for elements acting in the global z direction.
The influence of the reinforcing is incorporated into the simulation model via reinforcing elements using
a smeared continuum mechanics approach (REINF265). The placement and orientation of the armoring
is incorporated using mesh-independent reinforcing based on mesh-only elements (MESH200).
To account for the boundary singularities resulting from the idealized vertical support conditions, the
first element layer that connects to the support is defined as linearly elastic (shown in red).
49.3. Modeling
The element formulation for the concrete is based on SOLID185 linear brick elements with enhanced
strain formulation (KEYOPT(2) = 2). The solid geometry is meshed with 30 elements in each horizontal
direction and 4 elements over the depth of the concrete slab.
The steel armoring is modeled using a smeared continuum approach and shares the same nodes as
the surrounding concrete:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
750 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
The reinforcing elements are modeled using two additional discrete geometrical areas located at depth
hr. For each additional area, a surface mesh is generated (MESH200).
The smeared reinforcing area is defined (SECTYPE,REINF,SMEAR), followed by the associated equivalent
thickness and reinforcing spacing data (SECDATA).
The local directions of the reinforcing fibers are defined via local element reference coordinate systems
(LOCAL) for the generated MESH200 elements.
Finalizing the defined reinforcing (EREINF) generates smeared rebar reinforcing of the specified REINF265
section based on the MESH200 elements and the existing SOLID185 elements.
Additional reinforcing in the z direction is defined in the same way with another local element coordinate
system.
After defining the mesh-only elements (MESH200), the smeared reinforcing elements are generated
automatically (EREINF) for both reinforcing directions.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 751
Load-Limit Analysis of a Reinforced Concrete Slab
The reinforcing elements residing on top of the bottom layer are in the form of triangular elements.
The working direction of each reinforcing section corresponds to the x direction of each local coordinate
system (indicated by a red line).
The in situ dead load is generated by applying a gravitational acceleration of g = 9.81 m/s2 in the y
direction. A load-limit analysis is then performed.
A surface-pressure load is applied on the top-surface of the concrete slab and subsequently increased
until the load limit is reached.
The load limit is marked by a loss of numerical convergence, also identifiable as the horizontal
tangent of the resulting load/displacement curve.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
752 of ANSYS, Inc. and its subsidiaries and affiliates.
Boundary Conditions and Loading
The symmetry planes xy and yz are specified using normal rolling boundary conditions (that is, fixing
the z direction for all boundary nodes on the xy plane and fixing all boundary nodes in the x direction
on the yz-plane).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 753
Load-Limit Analysis of a Reinforced Concrete Slab
The dead load is calculated after applying the gravitational acceleration (ACEL).
A solid surface-pressure load is applied (SFA) with a maximum pressure value of pmax = 25 kPa at the
end of the second time step. The additional surface pressure results in an additional total load of 600
kN on top of the dead load of approximately 118 kN.
The load limit is calculated using an initial time step size of 0.05. To closely capture the loss of structural
integrity, the time increment can be reduced to 0.0001.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
754 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
A maximum total force of approximately 610 kN can be applied until the structure collapses at a max-
imum deflection of approximately 5.6 mm. The loss of structural integrity can be identified at this load
limit by the horizontal tangent of the force/displacement curve.
Figure 49.5: Vertical Displacements: Concrete Slab After Appplying Dead Load and Load Limit
(Top View)
The application of the load limit results in an ultimate displacement approximately ten times larger
than the at-rest deformation state. For both loading conditions, the maximum displacement is found
in the center, consistent with theoretical assumptions.
The following two figures show that the large structural deformations for the load-limit step result in
high internal stresses for the concrete base matrix:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 755
Load-Limit Analysis of a Reinforced Concrete Slab
Figure 49.6: Horizontal Stress Component SX: Concrete Slab at Ultimate Limit Load (Top and
Bottom View)
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
756 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Figure 49.7: Horizontal Stress Component SZ: Concrete Slab at Ultimate Limit Load (Top and
Bottom View)
Bending movement leads to compressive stresses on the top side of the concrete slab, and tensile
stresses in the bottom region.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 757
Load-Limit Analysis of a Reinforced Concrete Slab
Figure 49.8: Stress Components SX: Steel Reinforcing at Ultimate Load Limit (Top View)
The steel reinforcing supports the composite structure by carrying part of the loading.
The increasingly large tensile stresses in the concrete domain lead to crack formation, as indicated by
the equivalent plastic strains:
Figure 49.9: Equivalent Plastic Strains: Concrete Slab at Ultimate Limit Load (Bottom View)
The crack pattern forms in the center and spreads toward the outermost edge.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
758 of ANSYS, Inc. and its subsidiaries and affiliates.
Recommendations
The loss of structural integrity due to crack formation leads to the collapse of the structure for the in-
dicated load limit of 610 kN.
49.8. Recommendations
When setting up a load-limit analysis for a reinforced concrete model, consider the following recom-
mendations:
• Make use of symmetry conditions whenever possible to stabilize the numerical model.
• Competing crack growth can lead to bifurcation problems and, therefore, to a loss of numerical conver-
gence before the load limit is reached. Avoid the problem by defining custom weak points in the sim-
ulation model so that cracks form in well-defined regions.
• Use an initial Newton-Raphson nonlinear solution method to better capture the point of instability.
• Compared to a load-controlled analysis, the structural behavior after loss of stiffness can be more easily
traced via a displacement-controlled analysis; however, if the region of instability is of concern and a
load-controlled analysis is necessary, consider using the arc-length method (ARCLEN).
Using the arc-length method, the region of instability is identified at a load of approximately
610 kN and a deflection of 5.6 mm. The results correspond to those of the analysis shown in
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 759
Load-Limit Analysis of a Reinforced Concrete Slab
Figure 49.4: Reaction Force/Displacement Curve: Load-Limit Analysis (p. 754), validating the calcu-
lated load limit.
49.9. References
The following reference is used for this example problem:
1. DIN EN 1992-1:2011-1 (E). "Eurocode 2: Design of Concrete Structures – Part 1-1: General Rules for Buildings."
2011.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
760 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 50: Normal-Pressure Hydrocephalus Analysis Using
Hyperelastic Material
This example problem is a nonlinear analysis of normal-pressure hydrocephalus under combined
structural-thermal loading using coupled pore-pressure-thermal elements. The problem shows how to
solve for displacement, pore pressure, and temperature via the coupled elements using a hyperelastic
material under combined normal-pressure and thermal loading.
• Porous media
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 761
Normal-Pressure Hydrocephalus Analysis Using Hyperelastic Material
Pressure and thermal loading will be applied to the ventricle. The soil analysis option (ANTYPE,SOIL)
and consolidation option (SSOPT,CONSOLIDATION) are performed to determine the pore-pressure and
temperature distributions.
50.2. Modeling
The cross-section is modeled with coupled pore-pressure-thermal CPT212 elements:
et,1,212
keyopt,1,3,2 ! Plane strain
keyopt,1,12,1 ! Pressure degree of freedom enabled
keyopt,1,11,1 ! Temperature degree of freedom enabled
The brain cross-section is modeled as a plane strain condition. Temperature degrees of freedom enable
the thermal loading effect.
where:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
762 of ANSYS, Inc. and its subsidiaries and affiliates.
Analysis and Solution Controls
For the outer surface of the cross-section, the X and Y degrees of freedom are fixed. Pore pressure is
zero, and the temperature is 37 °C (average human temperature).
For the inner surface, a thermal boundary of temperature equal to 39 °C and a pressure loading of
666.61 Pa are applied.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 763
Normal-Pressure Hydrocephalus Analysis Using Hyperelastic Material
Figure 50.2: Temperature Distribution Contour of Brain Cross-section at Time 0.02 (a) and Time
1 (b)
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
764 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Pore-pressure distribution differs in some ways from the reference [1] results:
• The formulation used in this simulation (intrinsic to the coupled pore-pressure-thermal elements) differs
from that of the reference problem.
• This simulation uses a hyperelastic material. The reference problem uses an elastic material.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 765
Normal-Pressure Hydrocephalus Analysis Using Hyperelastic Material
• The geometry in this simulation is captured using digitized points from the reference.
For the coupled pore-pressure-thermal elements, the mass-balance equation for the pore-fluid phase
is:
where:
= volumetric strain
= compressibility parameter
= thermal expansion coefficient
= temperature
= flow source
In our problem, , , , and can be eliminated by appropriately defining the material property,
boundary conditions, and loadings. The term cannot be eliminated; therefore, the model used in
this simulation and the reference model can be inherently different. Generally, however, our simulation
gives reasonable pore-pressure distribution.
The stress field around the ventricle corners can be accurately calculated due to the mesh refinement
around those areas:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
766 of ANSYS, Inc. and its subsidiaries and affiliates.
References
Figure 50.5: Deformation Contour of the Last Time Step (a) and Deformed / Undeformed Shapes
(b)
50.7. Recommendations
In this type of analysis, the material properties are essential for obtaining good simulation results. Define
reasonable material properties, especially porous-media and thermal properties.
50.8. References
The following reference is used in this example problem:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 767
Normal-Pressure Hydrocephalus Analysis Using Hyperelastic Material
1. Shahim, K. et al. "Finite Element Analysis of Normal Pressure Hydrocephalus: Influence of CSF Content and
Anisotropy in Permeability." Applied Bionics and Biomechanics. 7.3 (2010): 187-197.
• brain.cdb -- The common database file for the brain cross-section model (called by Brain.dat).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
768 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 51: Multi-Filar Coil Inside a Tube Modeled with
Beam-to-Beam Contact
This example problem demonstrates three methods to model a multi-filar coil inside of a tube. Each
model uses a different contact scenario: surface-to-surface, beam-to-surface, or beam-to-beam. A com-
parison shows that the beam model using beam-to-beam contact offers the best advantage in terms
of simplified modeling and reduced computation time.
51.1. Introduction
Multi-filar (multi-wire) coils and multi-strand cables are mainly used in medical devices and in the
automotive industry. One example is an implantable lead, which might be part of a medical device such
as a cardiac defibrillator.
Bending analyses to model cables and wire bundles are typically performed to simulate actual physical
behaviors at the coil filar or cable strand level. Using solid elements to analyze these types of structures
can be computationally expensive. On the other hand, beam models with beam-to-beam contact offer
fast and accurate solutions with simplified modeling.
Two contact pairs are defined in each model: one self-contact pair for the coil's filar-to-filar contact, and
one contact pair between the coil and the tube. To apply the bending boundary condition, one end of
the tube and coil is fixed, and the other end is rotated about the Y axis by 1.2 radians.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 769
Multi-Filar Coil Inside a Tube Modeled with Beam-to-Beam Contact
Figure 51.1: 3-D Model of a Five-Filar Metal Coil Inside a Polymer Tube
The specific element types and contact models used are as follows:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
770 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
51.3. Modeling
Modeling of this problem involves the following steps:
51.3.1. Model the Five-Filar Coil
51.3.2. Model the Tube
51.3.3. Model the Contact Pairs
Case 1: Five layers of solid helical coils are created and meshed with SOLID186 elements; see (a) in the
figure below.
Case 2 and Case 3: A line model of helical coils is created and meshed with BEAM189 elements; see
(b) in the figure below.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 771
Multi-Filar Coil Inside a Tube Modeled with Beam-to-Beam Contact
Case 1 and Case 2: A full three-dimensional model of the tube is created and meshed with SOLID186
elements; see (a) in the figure below.
Case 3: A line model of the tube is created and meshed with PIPE289 elements; see (b) in the figure
below.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
772 of ANSYS, Inc. and its subsidiaries and affiliates.
Modeling
Figure 51.4: Tube Models: (a) Solid Elements, (b) Line Elements
Case 1: Surface-to-surface contact is used. The outer surface of the multi-filar coil is meshed with
CONTA174 contact elements, and the inner surface of the tube is meshed with TARGE170 target elements;
see (a) in the figure below.
Case 2: Line-to-surface contact is used. The multi-filar coil is meshed with CONTA177 contact elements,
and the inner surface of the tube is meshed with TARGE170 target elements; see (b) in the figure below.
Case 3: Line-to-line (beam-to-beam) contact is used. The multi-filar coil is meshed with CONTA177
contact elements, and the tube is meshed with TARGE170 target elements; see (c) in the figure below.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 773
Multi-Filar Coil Inside a Tube Modeled with Beam-to-Beam Contact
Figure 51.5: Contact Pair Between Coil and Tube for Case 1, Case 2, and Case 3
– KEYOPT(3) = 2 to include all contact scenarios, including internal crossing beam-to-beam contact and
beam-to-surface contact.
– KEYOPT(14) = 2 to define multiple target segments interacting with each contact detection point.
Case 1: Self contact between the coil filars is modeled as surface-to-surface contact. The outer surface
of the multi-filar coil is meshed with both CONTA174 contact elements and TARGE170 target elements;
see (a) in the figure below.
Case 2 and Case 3: Self contact between the coil filars is modeled as parallel line-to-line contact. The
multi-filar coil is meshed with both CONTA177 contact elements and TARGE170 target elements; see
(b) in the figure below.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
774 of ANSYS, Inc. and its subsidiaries and affiliates.
Material Properties
Figure 51.6: Self Contact Between Filars: (a) Case 1, (b) Case 2 and Case 3
– KEYOPT(14) = 2 to define multiple target segments interacting with each contact detection point.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 775
Multi-Filar Coil Inside a Tube Modeled with Beam-to-Beam Contact
Figure 51.7: MPC-Based Rigid Constraints (a) and Boundary Conditions at Pilot Nodes (b)
The pilot node at one end is fixed in all directions; the pilot node at the other end is rotated about the
Y axis by 1.2 radians.
Displacement
The vector sum of displacement (USUM) is plotted for the three models. The values are similar for all
three.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
776 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Computational Time
The following table compares simulation time and cumulative iterations for the three cases.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 777
Multi-Filar Coil Inside a Tube Modeled with Beam-to-Beam Contact
The above data shows that the beam-to-beam simulation (Case 3) is less computationally expensive
than the solid model.
Conclusion
The comparison of von Mises stress and computation time for these models illustrates that similar results
and reduced computation time can be achieved by using a simplified beam model and beam-to-beam
contact.
51.8. Recommendations
The following points are important for beam-to-beam contact modeling.
• Use pipe elements (PIPE289) to model polymer tube structures as line bodies. Currently, none of the hyper-
elastic material models can be modeled by beam elements (BEAM188, BEAM189).
• When internal beam contact is involved, set KEYOPT(3) = 2 for the contact elements to capture any internal
crossing and parallel beam-to-beam contact. In this example, the crossing contact type dominates between
the coil and the tube; the parallel contact type may potentially occur during New-Raphson iterations.
• Set KEYOPT(14) = 2 for the contact elements to allow each contact detection point to interact with multiple
target segments simultaneously.
• MultiFilarCoil.dat -- Input file for the simulation of a multi-filar coil inside a tube for all three models.
• MultiFilarCoil_case1.cdb -- Common database file for the multi-filar coil and tube model, Case 1.
• MultiFilarCoil_case2.cdb -- Common database file for the multi-filar coil and tube model, Case 2.
• MultiFilarCoil_case3.cdb -- Common database file for the multi-filar coil and tube model, Case 3.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
778 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 52: Sequential Construction of an Embankment on a Clay
Layer
This example problem demonstrates a method for modeling nonlinear consolidation using soil analysis,
where the load applied to the model is the self-weight.
52.1. Introduction
Embankments (or levees) are constructed to prevent flooding of land adjacent to seas, lakes, or rivers
and to support transportation by holding back water. Construction of relatively steep side-slope em-
bankments has become possible via staged construction.
Because soil and rock foundations typically behave nonlinearly with an elastoplastic stress-strain rela-
tionship, nonlinear material properties are necessary for the foundation. Several constitutive models
have been developed to assist with simulating the behavior of soil and rock, such as Mohr-Coulomb,
Drucker-Prager concrete, Cam-clay, and others.
Foundation settlement due to the weight of the embankment is an important consideration when de-
termining the stability of a structure. It is therefore useful to examine settlement behavior using finite
element models prior to construction.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 779
Sequential Construction of an Embankment on a Clay Layer
During the construction of each embankment layer, the clay layer undergoes consolidation. Each em-
bankment layer is 0.6 m thick. Loading for each step consists only of the weight of the embankment
and the clay layer's effective self-weight.
The material under the clay is impermeable, as are the contact regions between the clay and the em-
bankment, and between each embankment layer.
The embankment is assumed to exhibit linear elastic behavior. The clay layer uses an elastoplastic Cam-
clay material model.
52.3. Modeling
Because the self-weight of the embankment layer is applied gradually, this problem uses two approaches
to simulate the physical process:
• Case 1 – Divide each embankment layer into six sublayers (for a total of 18 sub-layers) using three contact
pairs. Twenty load steps are needed.
• Case 2 – Use three embankment layers and three contact pairs with self-weight and Young’s modulus as
functions of time and height. Five load steps are needed.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
780 of ANSYS, Inc. and its subsidiaries and affiliates.
Material and Contact Properties
The clay layer and embankment layers are meshed with 2-D four-node coupled pore-pressure-thermal
elements (CPT212). Contact pairs are meshed with 2-D two-node contact (CONTA172) and 2-D target
(TARGE169) elements.
The modified Cam-clay plasticity model, combined with porous elasticity, models the effect of voids on
the elastoplastic behavior of the clay layer.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 781
Sequential Construction of an Embankment on a Clay Layer
For Case 2 (p. 780), the specific weight and Young’s modulus are increased gradually over a load step,
achieved by making the quantities functions of location and time (TBFIELD).
Displacement of the embankment layers in the Y direction is restrained from motion in the Y direction
only when their corresponding elements are deactivated (EKILL). When those elements are reactivated
(EALIVE), the zero UY degree of freedom of the embankment layers is also deleted.
As the embankment layers are deactivated and activated, it is also necessary to deactivate and activate
the contact-pair elements associated with them.
The only load applied to the model is self-weight, simulated via soil analysis:
At the end of the geostatic analysis, the total deformation should be zero, confirming that equilibrium
is satisfied. All embankment layers and contact elements must be deactivated and displacement of the
nodes attached to those elements in Y direction must be constrained.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
782 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
The next eighteen steps for Case 1, and the next three steps for Case 2, are consolidation steps, each
step corresponding to the activation of a specific embankment layer.
Along with reactivating the appropriate elements, the displacement restriction in the Y direction for
the nodes attached to them is removed.
For Case 1, the command for reactivating contact pairs (EALIVE) appears only in steps 2, 8 and 14.
Excess pore pressure is the pore-pressure increase in excess of the initial state pore pressure when a
saturated soil is loaded. Because the excess pore-pressure evolution must be checked, that data is output
for postprocessing (ETABLE,SADD) .
Also shown (b) is the time-history evolution of the excess pore pressure of Point 2 (p. 780) (in the middle
of the clay layer under the center of the embankment). Excess pore pressure increases during the con-
struction of embankment layers, and dissipates gradually after the end of embankment construction.
Results from both cases (p. 780) indicate a similar prediction of settlement and pore-pressure evolution.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 783
Sequential Construction of an Embankment on a Clay Layer
As shown in the following figure, the pore pressure increases at a rapid rate from the beginning to the
end of construction (4.5 days):
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
784 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Figure 52.4: Evolution of Pore Pressure During 280 Days of Sequential Embankment Construction
The following figure shows the displacement in the Y direction during embankment construction:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 785
Sequential Construction of an Embankment on a Clay Layer
Figure 52.5: Settlement Process of the Model During 280 Days of Sequential Embankment
Construction
The inactive layers have no displacement, meaning that those layers have no influence on the settlement
of the clay layer during the steps in which they are inactive.
Also notice the gradual settlement of the clay layer. During the 1.5-day consolidation steps (p. 782) as-
sociated with activating the embankment layers, the clay layer undergoes a gradual settlement process
(where the portion of its top surface underneath the embankment descends gradually).
As shown in this figure, excess pore pressure increases during the construction of the embankment:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
786 of ANSYS, Inc. and its subsidiaries and affiliates.
Recommendations
Figure 52.6: Excess Pore-Pressure Evolution During 280 Days of Sequential Embankment
Construction
Excess pore pressure dissipates gradually after the construction of the embankment is completed at
4.5 days.
52.8. Recommendations
Consider the following hints and recommendations when performing a similar type of analysis:
• Set contact parameters properly. Ensure that contact pairs are always bonded. If not done correctly,
contact may be lost causing rigid-body motion to occur.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 787
Sequential Construction of an Embankment on a Clay Layer
• When using element birth and death (EALIVE and EKILL, respectively), enable large-displacement
geometry (NLGEOM,ON).
• Use the unsymmetrical Newton-Raphson solution method (NROPT,UNSYM) with the coupled pore-
pressure-thermal mechanical elements (CPT212 in this problem).
• Ignore the contact-element status warning that may appear at the start of the geostatic (p. 782) solution.
(Abrupt status changes are expected, as the contact elements are being deactivated and then reactiv-
ated.)
• SequentialEmbankmentCase2.dat – The Case 2 (p. 780) input file with three layers, and specific weight
and Young’s modulus as ramp functions of time and height.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
788 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 53: Acoustic Analysis of a MEMS Microphone
This example problem demonstrates how to analyze the response of a silicon micro-machined microphone
using acoustic elements and electrostatic-structural coupled-field elements.
• Linear perturbation
53.1. Introduction
Most digital devices, such as cellphones and tablets, include one or even several microphones. Micro-
electro-mechanical systems (MEMS) technology is very useful for designing these products due to their
miniature dimensions (millimeters).
A MEMS microphone adheres to the condenser principle. It consists of two silicon-based electrodes
separated by a thin air gap; one electrode is rigid (called the back plate), and the other is a membrane
that deflects under sound pressure. The air gap acts as a dielectric material between the electrodes,
and the capacitance varies as a function of the distance between the electrodes [1].
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 789
Acoustic Analysis of a MEMS Microphone
Figure 53.1: 3-D Model of the Condenser MEMS Microphone with One Symmetry Plane
This example illustrates how to analyze the response of a condenser MEMS microphone.
The microphone is composed of a sound port from which the pressure waves enter and reach the
membrane. The silicon moving membrane has a diameter of 0.6 mm and a thickness of 0.5 μm, and it
contains holes that allow ventilation of the pressure between the two sides of the microphone. The
membrane is separated from the rigid backplate by an air gap of 2.2 μm (dimensions taken from
Czarny (p. 795)). The backplate contains perforations that play a role in the pressure distribution on both
sides of the membrane and on the housing cavity, which is also part of the acoustic design.
53.3. Modeling
The 3-D model of the structure is created in ANSYS DesignModeler and meshed with solid elements.
The structural bodies use SOLID185 elements. The acoustic cavities (sound port, air gap, and housing
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
790 of ANSYS, Inc. and its subsidiaries and affiliates.
Material Properties
cavity) are modeled with FLUID30 elements. The air gap is meshed with a one element layer of SOLID226
electrostatic-structural elements (KEYOPT(1) = 1001) that use the elastic air option (KEYOPT(4) = 1).
The acoustic element size is chosen to ensure having at least 6 elements per wavelength for the highest
frequency of interest.
All bodies are grouped in a single part to obtain node connectivity at the body's interfaces.
1. The interaction between an acoustic pressure wave in a viscous fluid and a rigid wall is taken into account
for specific structures according to the low reduced frequency (LRF) approximation.
The LRF model is defined by the TB,AFDM command with the thin layer option:
TB,AFDM,MATID,,,THIN
TBDATA,1,2.2E-06
Property Air
Gap
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 791
Acoustic Analysis of a MEMS Microphone
Acoustic elements are ignored in this analysis, but the MORPH command is used to morph acoustic
elements so that they are properly shaped for the downstream linear perturbation analysis. Setting
StrOpt = YES on the MORPH command allows morphing in the model with structural elements:
MORPH,ON,,,,,,,,,YES
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
792 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
In this harmonic response analysis, the linear perturbation method is used to include prestress effects;
a DC bias voltage effect of 0.45 V is considered. An acoustic velocity is applied on the sound port inlet,
which leads to the vibration of the membrane:
CMSEL,S,AIRPORT ! Select the sound port inlet
ESEL,S,ENAME,,30
SF,ALL,SHLD,0.01, 0 ! Apply surface normal velocity
The linear perturbation method for full harmonic response analyses is described in Prestressed Harmonic
Analysis in the Structural Analysis Guide.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 793
Acoustic Analysis of a MEMS Microphone
of the applied pressure. Ideally, the sensitivity would be constant over the frequency range, which is
almost the case here:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
794 of ANSYS, Inc. and its subsidiaries and affiliates.
Input Files
53.8. Recommendations
When setting up an electrostatic-structural and acoustic calculation, consider the following hints and
recommendations:
• The acoustic mesh should be fine enough to properly capture the pressure waves.
• To account for deformations resulting from the static structural analysis, use the MORPH command in
the static analysis to morph the mesh in the acoustic fluid.
53.9. References
The following references are used in this example problem:
1. Dehé, A. "Silicon Microphone Development and Application." Sensors and Actuators A: Physical. Vol. 133,
No. 2. 283-287. February 2007.
2. Czarny J. "Conception, fabrication and characterization of a MEMS microphone". Ph.D. Thesis. National In-
stitute of Applied Sciences. Lyon, France. 2015.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 795
Acoustic Analysis of a MEMS Microphone
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
796 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 54: Reinforced Concrete Joint Analysis
This example problem demonstrates how to use a gradient-enhanced coupled damage-plasticity micro-
plane model to simulate a reinforced concrete beam-column joint.
54.1. Introduction
Reinforced concrete (RC) beam-column joints are critical to the global behavior of framed RC structures.
Insufficient reinforcing in joint areas is a primary reason for brittle failure under seismic loading. Joint-
failure simulation is necessary to understand and examine the effect of both concrete and steel reinforcing
on joint performance.
Modeling damage and softening typically cause numerical instabilities, convergence failure, and
pathological mesh sensitivity. The simulation presented here remedies those issues by using a material
that employs a nonlocal implicit gradient regularization.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 797
Reinforced Concrete Joint Analysis
54.3. Modeling
The 3-D model consists of concrete and reinforcing element types:
• The concrete uses the coupled pore-pressure-thermal mechanical solid element CPT215. Two extra degrees
of freedom per node, associated with implicit gradient regularization, are activated via a key option (KEY-
OPT,ITYPE,18,2).
• Reinforcing is simulated via the discrete reinforcing element REINF264 (generated via EREINF).
The solid elements and the reinforcing element are connected at the nodes, such that no special bond
interaction between concrete and steel is considered. Using symmetry, only half of the joint is modeled:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
798 of ANSYS, Inc. and its subsidiaries and affiliates.
Material Properties
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 799
Reinforced Concrete Joint Analysis
MP, EX, 1,
MP, NUXY, 1,
TB, MPLA, 1, , , DPC
TBDATA,1, , , , , ,
TBDATA,7, , , , ,
TB, MPLA, 1, , , NLOCAL
TBDATA, 1, ,
The rebar steel is modeled using von Mises plasticity with linear hardening (BISO material model) and
the following parameters:
• A load-controlled axial force of 94.8 kN is applied to the column and kept constant.
• A displacement of 80 mm is applied gradually (via two separate load steps) near the tip of the beam.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
800 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Automatic time-stepping is active, and the size of time steps is controlled (NSUBST).
In this case, controlling or modifying the convergence tolerance is unnecessary, but it can be done for
the nonlocal fields (CNVTOL,GFRS).
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 801
Reinforced Concrete Joint Analysis
The softening is caused by a strong evolution of plasticity and compression damage in the middle of
the joint:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
802 of ANSYS, Inc. and its subsidiaries and affiliates.
Results and Discussion
Figure 54.6: Tension and Compression Damage Profiles at the End of Simulation
Although the failure starts as a tensile cracking at the top left side of the beam, it changes at higher
deformation to shear banding in the common column-beam area.
Observe that plasticity tends to localize in a narrower zone than the damage, which is more diffused
and extends outside the plastic zone:
The plasticity and damage results are in agreement with the formulation where the gradient enhancement
is applied for the damage part only. Within this framework, the damage can be interpreted as the mi-
crocracking spread over a larger zone, while the plasticity represents the macrocrack emerging at the
center of the damage zone.
Considering the deformed shape and the plastic strain in the steel reinforcing, it is obvious that the
deformation in the rebars is concentrated in the region with the maximum tensile stress:
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 803
Reinforced Concrete Joint Analysis
Figure 54.8: Deformed Shape and Plastic Strain in the Steel Reinforcing
54.8. Recommendations
The damage parameters and the nonlocal parameter are critical in this type of analysis. To find appro-
priate parameter values, consider the following hints and recommendations:
• Experiment with values by checking the resulting size of the damage zone. Typically, the size should have
a reasonable ratio to the size of the entire structure.
• To gain a better sense of the concrete parameters, consider first running the problem without reinforcing.
• To allow sufficient resolution for the nonlocal interaction, set the element size to a value less than .
• Nonconvergence may indicate that is too small or that the damage parameters are too large.
54.9. References
The following references are used in this example problem:
1. Chalioris, Constantin E., Maria J. Favvata, and Chris G. Karayannis. "Reinforced Concrete Beam-Column joints
with Crossed Inclined Bars Under Cyclic Deformations." Earthquake Engineering and Structural Dynamics.
37.6 (2008): 881-897.
• joint.cdb -- Common database file for the reinforced concrete joint model, called by Joint.dat.
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
804 of ANSYS, Inc. and its subsidiaries and affiliates.
Input Files
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 805
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
806 of ANSYS, Inc. and its subsidiaries and affiliates.
calibrating a hyperelastic constitutive model, 227
Index Cam-clay material model, 779
Campbell diagram analysis, 205
Symbols camshaft, 105
3-D surface flaws, 237 cardiovascular stent simulation, 379
304L stainless steel plates , 423 centrifugal impeller blade, 177
ceramic materials, 269
Chaboche nonlinear kinematic hardening material,117
A Class IV flextensional underwater acoustic transducer,
absorption coefficient (acoustics), 631
501
acoustic analysis, 501, 789
co-rotational frame, 591
acoustic element morphing, 789
coherency functions, 513
boundary layer impedance (BLI) model, 631
cohesive zone model, 81
examples, 463
complex eigensolvers, 1
low reduced frequency model, 631
complex geometry: representing with beams, 353
noise mitigation, 631
complex material orthotropy, 129
normal surface velocity, 631
component mode synthesis (CMS), 283
surface port, 631
composite laminated T-joint, 553
acoustic waves: absorbing, 463, 501
composite material, 81, 353
adiabatic surfaces, 93
composite overwrapped pressure vessel (COPV), 129
aeroacoustics
computational fluid dynamics (CFD), 93
viscothermal resonator, 631
constitutive model parameters: selecting, 227
aircraft fuselage construction, 81
contact, 1, 15, 45, 63, 81, 105, 117, 177, 205, 227, 251,
aluminum liner, 129
269, 301, 379, 391, 403
anisotropic hyperelasticity with viscoelasticity, 489
contact friction, 423
anterior cruciate ligament (ACL) simulation, 489
containing vessel, 365
augmented Lagrangian formulation, 1, 15, 63, 177, 251
contour integral, 237
automatic time stepping, 105
convection loading, 93
axiharmonic meshing, 205
convergence difficulties: overcoming, 15, 63
axisymmetric and nonaxisymmetric parts: using both,
cooled turbine blade, 93
105
copper bar, 251
axisymmetric models: performance benefits, 205
coupled damage-plasticity microplane model, 797
coupled pore-pressure elements, 717
B coupled pore-pressure-thermal mechanical (CPT)
B-bar method, 63 technology, 141, 797
back pain, 141 coupled-field analysis, 423
beam element technology, 353 CPT elements (see coupled pore-pressure elements)
beam-to-beam contact, 769 crack-growth simulation example, 553
beam-to-surface contact, 769 creep, 535
bilinear isotropic hardening material, 403 creep response, 141
biomechanics, 141, 269, 379, 489 critical speeds: identifying, 205
biphasic model, 141 cubic interpolation, 353
Block Lanczos eigensolver, 177, 513 curve fitting, 45, 227
Block Lanczos solver, 117, 205, 301, 313, 353 curved-shell structures, 149, 447
BM3 piping system, 161 cyclic symmetry analysis, 177
bolt preload, 603 and linear perturbation, 177
bonded contact, 1, 81, 105, 177, 205, 301 example, 177
bonding, 423 cyclical thermal loading, 535
brake squeal, 1
buckling analysis, 313 D
damping, 1, 205
C debonding, 81
C*-integral calculation, 665
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 807
Index
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
808 of ANSYS, Inc. and its subsidiaries and affiliates.
linear elastic fracture mechanics, 237 operation, 447
linear elastic isotropic material, 1 nuclear island, 513
linear perturbation nuclear piping system, 117, 161
example, 1, 177 nuclear power plant, 513
linear perturbation method, 603, 789
linear static analysis, 237 O
load-limit analysis, 749 o-ring seal, 45
lumbar motion segment, 141 offsets, 129, 149, 313, 447
Ogden hyperelasticity, 15, 45
M orbit plot, 205
material properties as functions of time and height,779 orthotropic thermal expansion , 149, 447
MEMS microphone: acoustic analysis of, 789 output controls, 237
mesh distortion: repairing, 15, 63, 403 overconstraint, 283
mesh refinement, 15, 63, 403 overview, xxxvii
mesh-independent reinforcing, 749 overwrapped pressure vessel, 129
mesh: axiharmonic , 205
metal bar on rigid wall impact, 251 P
metal forming, 63 Parker seal, 15
migration model, 727 partial nonlinear perturbed modal analysis, 1
missing mass response, 161 perfectly matched layers (PML), 463
mistuning, 685 phase transformation, 617
mixed u-P formulation, 63, 379, 403 piezoelectric materials, 501
mixed-mode stress-intensity factors, 237 plastic heat generqation, 423
modal analysis, 117, 205, 301, 463, 513 plastic strain, 63
modal harmonic cyclic symmetry analysis, 177 polymer tube structure, 769
mode-superposition, 301 post-buckling analysis, 313
Mooney-Rivlin hyperelastic models, 227, 379 postprocessing
multi-filar coil, 769 sound pressure level (SPL), 463
multibody system, 283 velocity, 463
multiframe restart, 15, 63 power spectral density (PSD) analysis , 301, 513
multilinear kinematic hardening material, 117 pressure loading, 313
multipoint constraint (MPC), 1, 105, 177, 205, 283 prestress effects, 177
prestressed full harmonic analysis, 789
N prestressed modal analysis, 117, 603
Nelson-Vaugh rotor, 205 prestressed modal cyclic symmetry analysis, 177
Neo-Hookean material model, 391, 591 primitives, 403
nesting, 15 printed circuit board (PCB), 301
Newmark time-integration method, 251 pseudoelasticity, 617
nonaveraged results file format, 353
nonaxisymmetric and axisymmetric parts: using both, Q
105 QRDAMP eigensolver, 1
nonlinear buckling analysis, 313
nonlinear large-deformation analysis, 63 R
nonlinear perturbed modal analysis, 1 reduced-order CFD model, 93
nonlinear sparse solver, 15 reinforced concrete joint analysis, 797
nonlinear stabilization, 45, 81, 313, 379 reinforcing, 129, 365
nonlinear static analysis, 117, 129 reinforcing (mesh-independent), 749
nonlinear structural analysis, 269 reinforcing element technology, 129, 797
nonlinear transient analysis , 105 remeshing, 15, 63
nonlinear transient dynamic analysis, 251 residual stresses, 269
nozzle extension residual vectors, 301
fabrication, 149 response power spectral density (RPSD), 301
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 809
Index
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
810 of ANSYS, Inc. and its subsidiaries and affiliates.
wind turbine blade, 353
wire bonding ultrasonic transducer, 603
Z
zero period acceleration (ZPA), 161
ZPA frequency, 161
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. 811
Release 19.2 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
812 of ANSYS, Inc. and its subsidiaries and affiliates.