-
$\textit{FastSVD-ML-ROM}$: A Reduced-Order Modeling Framework based on Machine Learning for Real-Time Applications
Authors:
G. I. Drakoulas,
T. V. Gortsas,
G. C. Bourantas,
V. N. Burganos,
D. Polyzos
Abstract:
Digital twins have emerged as a key technology for optimizing the performance of engineering products and systems. High-fidelity numerical simulations constitute the backbone of engineering design, providing an accurate insight into the performance of complex systems. However, large-scale, dynamic, non-linear models require significant computational resources and are prohibitive for real-time digi…
▽ More
Digital twins have emerged as a key technology for optimizing the performance of engineering products and systems. High-fidelity numerical simulations constitute the backbone of engineering design, providing an accurate insight into the performance of complex systems. However, large-scale, dynamic, non-linear models require significant computational resources and are prohibitive for real-time digital twin applications. To this end, reduced order models (ROMs) are employed, to approximate the high-fidelity solutions while accurately capturing the dominant aspects of the physical behavior. The present work proposes a new machine learning (ML) platform for the development of ROMs, to handle large-scale numerical problems dealing with transient nonlinear partial differential equations. Our framework, mentioned as $\textit{FastSVD-ML-ROM}$, utilizes $\textit{(i)}$ a singular value decomposition (SVD) update methodology, to compute a linear subspace of the multi-fidelity solutions during the simulation process, $\textit{(ii)}$ convolutional autoencoders for nonlinear dimensionality reduction, $\textit{(iii)}$ feed-forward neural networks to map the input parameters to the latent spaces, and $\textit{(iv)}$ long short-term memory networks to predict and forecast the dynamics of parametric solutions. The efficiency of the $\textit{FastSVD-ML-ROM}$ framework is demonstrated for a 2D linear convection-diffusion equation, the problem of fluid around a cylinder, and the 3D blood flow inside an arterial segment. The accuracy of the reconstructed results demonstrates the robustness and assesses the efficiency of the proposed approach.
△ Less
Submitted 24 July, 2022;
originally announced July 2022.
-
Generation of Patient-specific Structured Hexahedral Mesh of Aortic Aneurysm Wall
Authors:
Farah Alkhatib,
George C. Bourantas,
Adam Wittek,
Karol Miller
Abstract:
Abdominal Aortic Aneurysm (AAA) is an enlargement in the lower part of the main artery Aorta by 1.5 times its normal diameter. AAA can cause death if rupture occurs. Elective surgeries are recommended to prevent rupture based on geometrical measurements of AAA diameter and diameter growth rate. Reliability of these geometric parameters to predict the AAA rupture risk has been questioned, and biome…
▽ More
Abdominal Aortic Aneurysm (AAA) is an enlargement in the lower part of the main artery Aorta by 1.5 times its normal diameter. AAA can cause death if rupture occurs. Elective surgeries are recommended to prevent rupture based on geometrical measurements of AAA diameter and diameter growth rate. Reliability of these geometric parameters to predict the AAA rupture risk has been questioned, and biomechanical assessment has been proposed to distinguish between patients with high and low risk of rupture. Stress in aneurysm wall is the main variable of interest in such assessment. Most studies use finite element method to compute AAA stress. This requires discretising patient-specific geometry (aneurysm wall and intraluminal thrombus ILT) into finite elements/meshes. Tetrahedral elements are most commonly used as they can be generated in seemingly automated and effortless way. In practice, however, due to complex aneurysm geometry, the process tends to require time consuming mesh optimisation to ensure sufficiently high quality of tetrahedral elements. Furthermore, ensuring solution convergence requires large number of tetrahedral elements, which leads to long computation times. In this study, we focus on generation of hexahedral meshes as they are known to provide converged solution for smaller number of elements than tetrahedral meshes. Generation of hexahedral meshes for continua with complex/irregular geometry, such as aneurysms, requires analyst interaction. We propose a procedure for generating high quality patient-specific hexahedral discretisation of aneurysm wall using the algorithms available in commercial software package for mesh generation. For aneurysm cases, we demonstrate that the procedure facilitates patient-specific mesh generation within timeframe consistent with clinical workflow constraints while requiring only limited input from the analyst.
△ Less
Submitted 14 June, 2022; v1 submitted 13 June, 2022;
originally announced June 2022.
-
Recovery by discretization corrected particle strength exchange (DC PSE) operators
Authors:
Benjamin F. Zwick,
George C. Bourantas,
Farah Alkhatib,
Adam Wittek,
Karol Miller
Abstract:
A new recovery technique based on discretization corrected particle strength exchange (DC PSE) operators is developed in this paper. DC PSE is a collocation method that can be used to compute derivatives directly at nodal points, instead of by projection from Gauss points as is done in many finite element-based recovery techniques. The proposed method is truly meshless and does not require patches…
▽ More
A new recovery technique based on discretization corrected particle strength exchange (DC PSE) operators is developed in this paper. DC PSE is a collocation method that can be used to compute derivatives directly at nodal points, instead of by projection from Gauss points as is done in many finite element-based recovery techniques. The proposed method is truly meshless and does not require patches of elements to be defined, which makes it generally applicable to point clouds and arbitrary element topologies. Numerical examples show that the proposed method is accurate and robust.
△ Less
Submitted 15 April, 2022;
originally announced April 2022.
-
Intracranial hemodynamics simulations: An efficient and accurate immersed boundary scheme
Authors:
D. S. Lampropoulos,
G. C. Bourantas,
B. F. Zwick,
G. C. Kagadis,
A. Wittek,
K. Miller,
V. C. Loukopoulos
Abstract:
Computational fluid dynamics (CFD) studies have been increasingly used for blood flow simulations in intracranial aneurysms (ICAs). However, despite the continuous progress of body-fitted CFD solvers, generating a high quality mesh is still the bottleneck of the CFD simulation, and strongly affects the accuracy of the numerical solution. To overcome this challenge, which will allow preforming CFD…
▽ More
Computational fluid dynamics (CFD) studies have been increasingly used for blood flow simulations in intracranial aneurysms (ICAs). However, despite the continuous progress of body-fitted CFD solvers, generating a high quality mesh is still the bottleneck of the CFD simulation, and strongly affects the accuracy of the numerical solution. To overcome this challenge, which will allow preforming CFD simulations efficiently for a large number of aneurysm cases we use an Immersed Boundary (IB) method. The proposed scheme relies on Cartesian grids to solve the incompressible Navier-Stokes (N-S) equations, using a finite element solver, and Lagrangian points to discretize the immersed object. All grid generations are conducted through automated algorithms which require no user input. Consequently, we verify the proposed method by comparing our numerical findings (velocity values) with published experimental results. Finally, we test the ability of the scheme to efficiently handle hemodynamic simulations on complex geometries on a sample of four patient-specific intracranial aneurysms.
△ Less
Submitted 27 July, 2020;
originally announced July 2020.
-
Immersed boundary finite element method for blood flow simulation
Authors:
G. C. Bourantas,
D. L. Lampropoulos,
B. F. Zwick,
V. C. Loukopoulos,
A. Wittek,
K. Miller
Abstract:
We present an efficient and accurate immersed boundary (IB) finite element (FE) solver for numerically solving incompressible Navier--Stokes equations. Particular emphasis is given to internal flows with complex geometries (blood flow in the vasculature system). IB methods are computationally costly for internal flows, mainly due to the large percentage of grid points that lie outside the flow dom…
▽ More
We present an efficient and accurate immersed boundary (IB) finite element (FE) solver for numerically solving incompressible Navier--Stokes equations. Particular emphasis is given to internal flows with complex geometries (blood flow in the vasculature system). IB methods are computationally costly for internal flows, mainly due to the large percentage of grid points that lie outside the flow domain. In this study, we apply a local refinement strategy, along with a domain reduction approach in order to reduce the grid that covers the flow domain and increase the percentage of the grid nodes that fall inside the flow domain. The proposed method utilizes an efficient and accurate FE solver with the incremental pressure correction scheme (IPCS), along with the boundary condition enforced IB method to numerically solve the transient, incompressible Navier--Stokes flow equations. We verify the accuracy of the numerical method using the analytical solution for Poiseuille flow in a cylinder. We further examine the accuracy and applicability of the proposed method by considering flow within complex geometries, such as blood flow in aneurysmal vessels and the aorta, flow configurations which would otherwise be extremely difficult to solve by most IB methods. Our method offers high accuracy, as demonstrated by the verification examples, and high efficiency, as demonstrated through the solution of blood flow within complex geometry on an off-the-shelf laptop computer.
△ Less
Submitted 4 July, 2020;
originally announced July 2020.
-
Cell-based Maximum Entropy Approximants for Three Dimensional Domains: Application in Large Strain Elastodynamics using the Meshless Total Lagrangian Explicit Dynamics Method
Authors:
Konstantinos A. Mountris,
George C. Bourantas,
Daniel Millán,
Grand R. Joldes,
Karol Miller,
Esther Pueyo,
Adam Wittek
Abstract:
We present the Cell-based Maximum Entropy (CME) approximants in E3 space by constructing the smooth approximation distance function to polyhedral surfaces. CME is a meshfree approximation method combining the properties of the Maximum Entropy approximants and the compact support of element-based interpolants. The method is evaluated in problems of large strain elastodynamics for three-dimensional…
▽ More
We present the Cell-based Maximum Entropy (CME) approximants in E3 space by constructing the smooth approximation distance function to polyhedral surfaces. CME is a meshfree approximation method combining the properties of the Maximum Entropy approximants and the compact support of element-based interpolants. The method is evaluated in problems of large strain elastodynamics for three-dimensional (3D) continua using the well-established Meshless Total Lagrangian Explicit Dynamics (MTLED) method. The accuracy and efficiency of the method is assessed in several numerical examples in terms of computational time, accuracy in boundary conditions imposition, and strain energy density error. Due to the smoothness of CME basis functions, the numerical stability in explicit time integration is preserved for large time step. The challenging task of essential boundary conditions imposition in non-interpolating meshless methods (e.g., Moving Least Squares) is eliminated in CME due to the weak Kronecker-delta property. The essential boundary conditions are imposed directly, similar to the Finite Element Method. CME is proven a valuable alternative to other meshless and element-based methods for large-scale elastodynamics in 3D.
△ Less
Submitted 29 October, 2019; v1 submitted 13 May, 2019;
originally announced May 2019.
