Professional Documents
Culture Documents
Ram Jet Applications
Ram Jet Applications
March 1982
AGARD-CP-307 -
Ramjets ad -Ramrockets
for Military ppications
NORTHATLAN IRATOGNSICEIO
V
A NOuG$AI AflI
== I~--A-
p!
- Continuously stimuLkting advances in the aerospace sciences relevant to strengthening the common defence
posture,
- Improving the co-operation among member nations in aerospace research and development;
- Pruviding scientific and technical advice and assistaice to the North Atlantic Military Committee in the field
of aerospace research and development;
- Rendering scientific and technical assistance, as requested, to other NATO bodies and to member nations in
connection with research and development problems in the aerospace field:
- Providing assistance to member nations for the purpose of increasing their scientific and technical potential:
- Recommendipg effeaive ways for the member nations to use their research and development capabilities for
the common benefit of the NATO community.
The highest authority wiIhin AGARD is the National Delegates Board consisting of officially appo"nted senior
representatives from each member nation. The mission of AGARD is carried out through the Panels which are
composed of experts appointed by the National Delegates. the Consultant and Exchange Programme and the Aerospace
Applications Studies Programme. The results of AGARD work are reported to the mmber
.- nations and the NATO
Authorities through the AGARD seriez of publications of whic this is one.
Participation in AGARD activities is by invitation only and is normally fimited to citizens of the NATO nations.
ADI?113j
NATIONAL TECHNICAL
INFORMATION SERVICE
Papers presented at the Propulsion1 and Enerigetics 58th Syrnoosiun. held at Church Hiouse,
London, United Kingd-m 0n126-29 October 198 1.
, ]~
7. Presented at
8.Author(s)IEditofts) 9.Date
Various March 1982
I10.Author'slEditofs Address -i1.Pages
Various 274
Ramjet engnes
Rocket enrines
Ducted rocket eneines
Missile propulsion
14.Abstract
The Conference Proceedings contain 32 papers presented at the AGARD Propulsion and
Energ tics Panel 58th Symposium on Ranjets id Ramrockes for Military Applications
which was held in London, United Kingdom, on 26-29 October 1981.
The Technical Evaluation Report is included at the beginning of the Proceedi ngs. Questions
and answers of the discussions follow e-ch paper.
The Symposium was arranged into six sessions: Survey Papers (3); Propulsion Systems (6);
Inlet Diffusers (5); Interference and Drag Reduction, Engine Testing (5): Combustion,
Fuels, Propellants (i !) and Integral Booster and Transition (2). A Round Table Diseuss'==
followed the sessions.
The aim of the Symposium was to provide a forum for discussions to research scientists and 4-
volume contain the Unclassified paperk the Classified papers aft pubushed in
-This a -
Supplement.
1t'C-
NOTICE
S_ _
m r
PROPULSION AND ENERGE-FICS PAINEL
CHAIRMAN: Professor flEjCoyin EP CHIMN M.Ie Profescur Ch.IEhim
D e answn of Aero,nautics andAstro~nautics Wflje Unimesitecit Bomw I
Mmsachiou insfitute of Tlechnoiov im Srmnnncr-
Catidoa. Mrnaachusetnr Mi'9-S Plcinlaan 2 - 1050 Brnusid
Be-L2iurn
PROGRAMME COMMITIEE
DipI.-Inz. B.Crispin tQ~airr=! Profcssor. R-Monti
McssschniuBa~ow~iinr
Goli stiuso i Asodaincs.Univest dli Studi
Unterahmmunbjc 1 -ch Rauun-falist - Ab: RT 31 PinaeTmslin 80 - 80i'
1__a\poli. Itay
0
Pos~fach go J!. RLXWmoMincie 890. Gennany
Md.'1 hdhten
=cbefM.Pilano
professor .E.Cdmkd C-0 b~- -de
ecirhsnTurbomacldncs.
Prufazr ofnt*-mana& Aplid Phywts ONRA
California Institute-aj eiry
of ~ 2-9 A--u d D.-=;o Lelerc
J.Dwilmn~ Im Ir T
Dr~flunhazW ~ hn='
olo-v
Narionsi Gas Turbine Eks~ NpFA~~cEsern
Pyestock. Famboran-asHan C (If40S.UK k.M ~- E- J ub
S Delft. Netherlands
PANEL XECVIE
AG.ARD-NT
7 ncr- Ancaile
9220 N Mily sur Seine
Francc
ACIISMLfl)GENflT
The- F.oma1-- and n-tk= Pane to ~.ex its tt
Ie UK aiolDerics
to AGAIRD form - to -olW its Ssth Stu rbT.1l-
and~ ~ r fortk~e
t= hj
PANEL OFFICERIS AND PROGRAMIME COMM~fITTEE
-OC!rN~DOUN SFATOREAOEIJ
A ACCELERATEUR INThGRE'
3EN
III - pii-s DIPPU.soR
COMEa? MA o O DDIM-ENTr r S -FG-LTI DVI E DtMIUSSLE
by ILO- ,icar
13-KTL
f aza
F
SERVICE EXPERIENCEMWTHTHREEGENERATIOSO Al
MlifE
ROLE OF LIUIDI PH.ASE D-ROFLT DECLOM)SIMN IN AN -ADVANCED
AURBRE=ATHIY NROPULSIO' SYS"5
by _ _ _ _ _ _ _ _ _ _ _
UWM fl -YI ORMM W01MNPOI
Ib
A FROM
t
Refrzence=
-I-
---
TECHNIAL EVALLATI REPORT
L. INTRODUION
rM Aw-J'
!-4 fint h
-- ~ S4w~i
=n1 -= --- C f=--t
I T ~ cnits ard t
- ~thn-t w;-. to=-=-O-i '~2I stfor Tr mwk
i==h ihffvmm
WUMMARY AMWCOWItN
Q;Qoa - r~
cWRhik.
- d ;ma
bran I~ S MW
mee eod
1h~~
&es 1~andiI1
TozMet- rS-nVI and 11was to (arkii "ren- survuy On moden t- itnraeket propulsion
teft~ inmsie"d r=J- _moiite -M zaa his'nr '- 'm-M
prapara n. t solid' znd liquid tuch eon ret- 1CIpropulsion-
n'~
a as discru nr*et
s- s-%ell typ ~pa
sion S'-'--x-- o th- n-ext sneraio mPapevs N -r cocie with -an - -' 5 A-results of a
freer! Hui - w -I'2 P-',gt lest programt Pn- 2 described
t"r propuls-on pel areaudn
cuwr se.- ~.a latie fuels for rcidys =tWe missies. dvanar -'_ hu n- sm-terr. over solid
p tj. cK MAEr discussed. Pape
,
Pae retMdtailed discussion or 4-ee-omt '
envclow si. ofit- 'b'oc-=- 'r W(si elt:
f'~
ME-- requizeents imposed o-
-1t:vw-u=-.,-onents. 'ricludp-' t'-= -nLh
of the lkuwd =m -MU P- f_ an
-bon
h transition .. n-I '- ro-t boost to.
6and --h i.mpoelaince of cost in deses' o == wazdressej. Paper-S
-=en missiles. Paper 6 eo'n.. rndw.iSei
smllran-
rvieCni
___ i sSan integral. nonlelcmsoli
prn hlnt-,oser whereas Pa-per A
tioag we-- i a hnw iud fuel integral rok.r'r = -ost issue-
x ssn.ynohlssin flopu- "zenrr.oe ae
du- yitcsn.- m = n impressive modec ws -~ WSt tes tor this progra
dleai " -e~r P-- iqueti Ln that it
integration uequkont .- Scieran t epne-_S for hvri-
-ilapictos
Reuls rme~ tuyonvaios-e we includd. N'-- 9pe a altlal study Of
~ zjdsusddsses="
rrw~~~~~~~~~~~ sn I -espect in p-rfoTn-- RCSts showed that i
adc~uo=
'a we. Thrut eruiii ~ ahimpotan
.
'd cmpnsns
Pa'_ IU _~g- " rnmewof val,.. -vubiel cottclued that V h-fikss pui.S~ets are
bcs' soldtla -Une ss thani 600 Ncwto=- .dn.'r-s'boz -r=-s'iW' cost.
S:m-='vM.M__-~ of domina"tbnr'=-
Session Ill
Pa--- ~-'pen presetied !'--to -Aie inlt diffser ;--vs-' cs
ofilet oonfia, ~ ~ Lsmar
emi-circWar I='~rv~n ubro
inct.r~nd m-I re v to missie body. - - fetnal
w, boundary
the penena aiit - icusd n eti
i_tasnand- floow'~rHci=
2cu 0n~n whe L'mco-Incnia
como
Combsti nuenr. A computer r'no w =Wte with herin i'snthe ine
at
Sstern "am- 11=-=r-e
data. Paper 13 wa n mn...-n Sdsntr -ape osur
son fiv- re'ferences listed. A -- i-- 'emrrrecl odel ar.-' co
== -upriesented. Paper 14 di -vr.- ci 'trus with
of theas o-efrmxf
avanet" o'i .nis and the effect
-M--' in rerards to the r-s a.fz Pper ISwas Pam'== it with
M tMWct- o__ fr rat.Jet powered r'usz.=cx. R' = -- 'nthtb c"ovton
methods5
USL-- 1otc capabil;iis or In-. rn -ru an~~
system, range- nmroiv-u mtaft- C01I
cpm&4oafto muft-=emcnts. can result W'aS
-eu- II e~ nn==vntozafl
Thi of divers t-a - "-
with a variety A--
Will, r__ w- missles the MCHio,,O. C-WUMCWa~ we-he is;-:X 'ea sn~==
nt= rem an
=-''-ow
~'n arabia
tu.truer She 12-Wers.
an aprai wa intraxno avn- f07 tr' !nftuc
and reads -t 4 L f th r-=
'i- Zrk-"'marv leves
poimn of bopmren -e
tor a- or.esoatta paper I
Wd
me----t- -- ---se and externa-l buinh here;its orepefr
n ..x4mrri
vunne e bwa 'it aMuc-- aiestr. Pr
hvdr~-
- ~rrwenciuted with CO. liHe N. used as the 'esIt w- rbmn it= tr-ecmnaso
1~ ~ pre''s.
.oal eatrelese at-P
1=r n ~ o roanret urop- pr 8der ied asm- "
-9 was an extensire r-erw the
'of
ite
-" o eattest faility
nr tev-
u kaile
seve
'~rs in
==--- Frupellant boosters,. P-- was
We, Z b-s d-.~'-c
As''tw
teai
RAF' Royal Nanr It wMS sr-n WEtC~ ,
am-.
mns 4ability to tha of t'-e SO wmret, in'stc of in n~st,~.ieofrn
c7=_-"bilily.
Suetch r-
___ led Tthe
"-MK - ' "woode
"- t''BadaiigS ~ ytm n
aLs-uodfor the hqui fin '==-s aneac
== r~~Tb
two u add-essd esunbitonm ue n w c ~prilata~
-Vill'
grains discussed, two were hydrocarbon fuiels and two wer, boron fuels. Paper 21 discussed dircct connect
'lle1-rich
combustion lests or a solid ruel rumJot consisting or a rearward facing step us its flumeholding device, an ignitor, a
combustion chamber containing the hydrocarbon solid fuel grain, and an afterburner for enhancement and completion
of the combustion. Regrvsion rates and combustion erficiency results were obtained over a range of flight conditions
for several hydrocarbon ruels, Including polyethylene, and ror dirferent port geometries. Paper 22 described the results
or water tunnel tests and two-dimensional windowed combustor tests with two opposing 450 side air inlets. These wcrc
used to provide qualitative Insight into the effects of important combustor parameters on the overall performance or
boron propellant gas generator ramjets (dueled rockets). It was found that the highest combustion efficlencies were
achieved at the lowest ruel injection momentum and at the lowest air injection momentum. Reasons were postulated for
these results. Paper 23 provided a comprehensive review of the boron combustor investigations conducted a. MBB for
ducted rockets. These Included boron reactions during the primary combustion (burning of the ruel-rich propellant
grin), as well as the behavior of boron during ram-combustion. Also discussed were special experimental techniques
applied during these investigations. Paper 24 described the development and characterization or a class of hydrocarbon
propellants ror ducted rockets. A third generation ruel bearing the reference number 1603 was shown to have a relatively
high heating value (8020 cal/g). excellent mechanical properties and a low exhaust signature. Paper 25 discussed the
development of ruel-rich cast double propellants for ramrocket applications. Paper 26 was not a combustion paper, but
rather discussed the use of high energy fuels, particularly hydrogen, in advanced mixed cycle airbreathing configurations
for military applications.
Paper 27 described a research effort involving the coupled effects of decomposition and evaporation of a typical
liquid fuel spray. Results indicated that liajid phase droplet decomposition at higher pressures may cause a significant
reduction of combustion efficiency in advanced alrbreathing propulsion systems. Paper 28 discussed the use or swirl in
ramjet combustors and showed that a properly designed contoured swirler can achieve higher combustion efficiencies,
along with significantly lower pressure losses, than a typical frameholder configuration. These same swirlers have also
been applied with success to the solid fuel ramjet. Paper 29 presented a detailed discussion of a modular model for the
analysis of sudden-expansion ramjet combustors. This model is based on a concept in which the recirculation zone,
treated as a stirred reactor, is coupled to a parabolic boundary layer formulation for the flow outside the recirculation
zone. Hydrocarbon oxidation kinetics and turbulent kinetic energy turbulence models are included. Paper 30 described
the theoretical approach for the calculation of axisymmetric and three-dimensional ducted flows suitable for ducted
rockets. A numerical computer code employing finite difference forms was developed from existing free-flow
computation techniques to calculate axisymmetric, non-recirculatory ducted flows. Future work would develop a
computer code to calculate three-dimensional ducted flows with due allowance for flow asymmetry and recirculation.
Paper 31 described a simple three-step model, empirically derived from a premixed propane-air baffle-stabilized ceramic
lined combustor, which can predict composition and temperature profiles in ramjet like combustors including the effects
of vitiated preheat and heat loss.
Session VI
The two papers which were presented in this session dealt with details associated with the transition process from
integral rocket boost to ramjet combustor operation. Paper 32 gave a historical accounting of the development of an
Integral Rocket-Ramjet booster for a propulsion technology flight demonstration program. Paper 33 discussed multiple-
inlet and transition development for another integral rocket-ramjet program.
The large participation (over 60 authors and 75 observers for "Ramjets and Ramrockets for Military Applications")
shows the importance and timely nature of the subject chosen by the AGARD Propulsion and Energetics Panel for its
58th Symposium. Comprehensive surveys presented on modern ramjet and ramrocket propulsion technology have shown
that neither the lack of technical progress nor the lack of attractive missions are the reasons for the hesitation in applying
new generation ramjets to operational military projects. Rather it was sugggested, that ramets need to improve their
"image" with military planners and missile designers.
Afuture A GARD meeting might be held, perhaps In conjunction with another panel, to address the overall missile
systems aspect and cost effectiveness of using ramjet technology. Propulsion related technology areas to be discussed
should Include:
- Missile guidance and control and Its effect on Inlets and propulsion performance: skid-to-turn vs. bank-to-turn
- Propulsion system observables: I.e., IR, RADAR signature and visible smoke
Miss iceeffectiveness
-
Poe -Awthots
g ~ ~~ L
Fiealdsp. fLim
g, Sto cb ice I* I
10 Ken liid
I
5 - Laruell4
1
6 1 eiN du
7l
RoNy. etaerds
~
8~ ~tcbig
~~~~Fgr 1atutBlfg
L~~~ ~~ Kramer.__ __ ___ ___ ___ __ ___ ___ _
LOW-COST COMBUSTOR FOR A SUPERSONIC TACTICAL MISSILE '7
T. D. Myers and A. P. Peters
Chemical Systems Division/United Technologies Corporation
P.O. Box 358
Sunnyvale, California 94086 U.S.A.
INTRODUCTION
In April 1979, a low-cost liquid fuel 'integral rocket ramjet (LFRJ/IRR) was success-
fully flown at the Navy Pacific Missile Range at Point Mugu, California. The low-cost
LFRJ/IRR propulsion system was a derivative of the earlier flight-demonstrated advanced
low volume ramjet (ALVRJ). The ALVRJ baseline propulsion system seen in Figure I had
performed perfectly in a series of five flight tests in which typical missile boost,
sustain, and flight maneuvers were commanded. A summary of the ALVRJ flight test results
is seen in Figure 2. These five flight tests were made over a two-year period, starting
with the first flight in November 1974. The propulsion system ground test development
program began in 1968 and continued for the next six years up until 1974.
After the key technical aspects of the LFRJ/IRR propulsion system had been demon-
strated, studies were conducted to identify the best mission application for the ALVRJ pro-
pulsion system. One of the most promising mission applications for the ALVRJ was for the
Supersonic Tactical Missile (STN) shown in Figure 3. When compared to an equivalent
weight- and volume-constrained rocket motor, the ALVRJ propulsion system could provide
over twice the ranoa !*.h a higher terminal velocity. Moreover, when compared to a compa-
rable weight- and volume-constrained turbine engine, the ALVRJ could provide higher veloc-
ity and a multiple shot capability, increasing the mission hit probability.
Cost studies revealed, however, that the ALVRJ unit cost production was approximately
five times higher than, desired for the STH. Therefore a design study was conducted to
determine where production cost savings could be made for the ALVRJ propulsion system
without a significant sacrifice In performance. The cost reduction design study was con-
ducted by the Vought Missile Systems Division and was supported by both Chemical Systems
Division (CSD) and the Naval Weapons Center (NWC). Vought was responsible for designing
the vehicle and integrating the propulsion system components; CSD was responsible for
developing the low-cost ramburner technology; and NYC was responsible for fabricating the
integral booster motor and providing inlet and combustor test facilities. The specific
cost goal was to reduce the cost of the experimental configuration from $200,000 by a fac-
tor of 4 to S with no performance penalty.
The approach for developing the low-cost ALVRJ propulsion system for the STM and
other tactical mission applications is summarized as follows. First, a series of propul-
sion system performance/cost vehicle systems trades were conducted. Key propulsion sys-
tem requirements that most strongly influenced cost were: (1) fuel flow rate accuracy,
(2) inlet pressure recovery, and (3) required combustor operational life. From these trade
studies specific low-cost ramjet component designs were selected including: (1) inlet,
(2) fuel controller, (3) rocket booster motor nozzle release mechanism, and (4) combustor/
thermal protection system (TPS). The low-cost combustor/TPS developwent effort is the pri-
mary subject of this paper. The low-cost inlet technology developed by Vought and the
low-cost fuel controller developed jointly by Vought and NWC are subjects for separate
papers. Although the basic low-cost combustor development- was primarily an In-house
research and development effort by CSD/Vought, some of the key component ground tests and
the flight test were conducted by NWC.
SUPERSONIC TACTICAL MISSILE (STM) LOW-COST RAMJET COMBUSTOR DEVELOPM4ENT
The ALVRJ ramjet combustor case was rolled and welded from Inconel 718. The inlet
entry ports and nozzle shell actuator mounts were machined from forgings. Thermal
protection was provided by a refractory zirconia layer and local hot spots protected
either by regenerative fuel cooling or air cooling. The high cost of Inconel 718 combined
with the extensive machining required for the local fuel and air cooling resulted in
a very expensive design. In addition, the fuel/air ratio was required to be held very
lean to avoid comust ion temperatures in excess of 1,922 0 K, which limited ':he maximum
thrust capability of the engine.
The ALVRJ TPS design was selected in 1968 and reflected the best state-of-the-art
technology available at that time. By 1977, however, separate industry development pro-
grams had produced a new ablative thermal protection system that proved to be a very
effective insulator for ramjet combustors. This carbon-filled elastomer material,
designated as DC 93-104, has excellent low thermal conductivity and is also resis-
tant to the oxidizer-rich ramjet combustor gases. By using DC 9S-104, CSD could select
the less costly case material, 17-4 ph, and also eliminate the expensive machinery opera-
tions associated with local fuel and air cooling. DC 93-104 has lower fabrication
costs than the zirconia liner. In addition, the combustion temperature limit needed for
the zirconia-protected ALVRJ combustor case could be removed with DC 93-104 and the full
thrust potential of the engine could be realized.
An illustration of the low-cost combustor is shown in Figure 4. Together with the
primary cost reductions associated with the use of 17-4 ph steel and DC 93-104, several
supplementary cost savings were realized by using castings in place of forgings for the
inlet entry sections. The tensile strength of the castings was increased by hot iso-
5- static pressing (HIP) to-a level normally attained' only-by using a more expensive lorging,
, These key cost reduction design features, which permitted the 4-5:1 produ:tion unit cost
projection, are summarized in Figure S.'
LOW-COST CbMBUSIOR DESIGN AND FABRICATION
The low-cost combustor assembly is shown in Figure 6. T1 ombustor assembly con-
sists of two major subassemblies, the forward combustor case a, 'he nozzle, which are mar-
ried in the final assembly by a circumferential SB weld.
Critical airframe interfaces are machined after iinal assembiy by Weans of Vought-
supplied locators and templates. Machining interficing surfaces -after assembly ensures
accurcte--alignment of the booster motor nozzle. The forward end of the combustor is a
dome formedto a shape approximating a 2:1 ellipsoidal dished sectioO with an 0.24/0.22-cm
wall thickness. A-boss, butt-welded to the center of the dome, so-'es multiple func--
tions: it acts as a centering hole for the TPS casting mandrel, a casting port for the DC
93-104 insulation, a centering hole for the booster grain mandrel, and a pressure port for
combustor pressure measurement. After the dome-has been joined to the inlet actuator
mounting pad/nozzle support ring, the subassembly' ,s aged to the H-775 condition before
the nozzle skin is installed to facilitate nondestructive testing (NDT) Inspection of the
critical fin support pad welds. Following qualilty control (QC) acceptance of the aged
subassembly the nonstructural nozzle skin is welded in place. Subsequent heat treatment
is not required.
The nozzle liner is molded from chopped squares of U.S. Polymeric FM 5020. The mate-
rial is bonded to the nozzle support structure with American Cyanamid HT;924 adhesive. A
silicone
across the0-ring-between-the
nozzle bond line. liner and the sealing
support is
ring ensures that
the gas d&ot not leak
Additional provided-by DC 93-104 nozzle
entrance fillet, which is installed at final combustor 'assembly, and a soft silastic mate-
rial that is installed between the aft-radial surface of the liner and the support rings.
Alignment surfaces of the snap-ring interface and the ramjet throat are machined
after combustor assembly to ensure that the required alignment tolerances ,are met.
Acceptdnce of the completed combustors was in accord with CSD Acceptance Specifica-
tion rT-2630-003. Weld efficiencies meeting or exceeding" design requirements were demon-
strated for all weld types. A weld qualification program to ensure sound welds was con-
ducted for the 13 weld or weld-type joints used in the fabrication of the low-cost combus-
tor. -Weld procedures were developed for each of-these joints in accordance with the stan-
dard CSD weld'qualification specification.
An extensive materials evaluation program was conducted in support of the low-cost
combustor d6sign. Vehicle weight restrictions required qualification of a material meet-
ing high strength requirements while still maintaining sufficient toughness and stress'cor-
rosion resistance.
Material for the low-cost combustor was initially specified by CSD as a 17-4 ph aged
to the H-811 condition; however, in response to Vought's weight reduction requirement, CSD
determined the best compromise between material 0 strength and toughness. The selected com-
buster design, based
z
on 17-4 ph and aged to 2 775 K condition, has an-ultimate stregth f
1,275 x 100 N/m for sheet, 1,241 x 10 N/M for forged materials, and 1,172 x 10 N/m?
for case material. Since resistance to stress corrosion cracking was very important,
stringent NDT and acceptance test procedures were used to ensure that components and fab-
ricated assemblies were free from defects.
Preliminary design values were based on review of available literature or data from
previous Vought programs. These values were revised as CSD material test data became
available. As a result of the materials evaluation program and selective procurement,
design stress values approximately 20% above MIL-HBK-S design allowables were approved for
use in the low-cost combustor design.
To achieve CSD cost objectives, the complex air inlets (Figure 7) were cast instead
of machined. The cast inlets reduced the inlet cost per combustor by approximately 7S% of
the cost of-machined inlets, rdpresenting a savings of about $12,000 per combustor. This
is the first -time CSD hid used castings at high stress levels as an integral part of a com-
bustor case pressure shell. In a continuing effort to reduce fabrication costs, the tech-
nology
entire gains
forwardfrom the section.
inlet inlet casting
This effort
design,were expanded
though to investigate-the
not completed in 'time for casting of the
this -initial
program, would reduce combustor costs further by eliminating the complex jigging and weld-
ing required to form-the forward assembly. Designs were prepared and a contract released-
for casting a single inlet section, formerly comprised of the four inlet castings, the-
I:
four roll-formed interconnecting sheet segments,' and the forged forward dome/forward'
attach y" ing section. This integral inlet sectioni sho'wn-in Figure 8,-met all inspec-
tibn-requirements established for -the single 'inlet castings used inthe low-cost -combusfbr
development-program (LCCDP)combustors. The integr'al 'inlet section was subjected to th-
same HIP process as the single inlet- castings. No defects du-to the HIP process were
dete6cted, and a' posttest dimensional inspection showed'the castihk- to be dimens-onally
stable during HIP.
-An extensive insulation-processing study was -_conducted- to develop procedures for
-ins -f -n . .. . .. " "
____ -i
- 1
AL.
- --..--------- = - ! -- - -
a{
. -
j ---w -- --- - :- 4-
41 -Z
aa El r. -- rz-
0 00 0 0 0 0 0 0 0
- -*. 0 C 0 U
ME =-- -- - =a - 0 - -d ZF
5.4 -4 9. -
E S."
4: Z - 2 0 0 -0 4 2
=0, 0 woo00 0 >.
tO~~ i U ~L L c li Li 0 i L
ao m4V -.
>. 0
l a --
Go= 4..
. a
u.
o-c-si
W..'" 0',- a-- V4> aw
i
u
a
W- 04.- u
cs
5d =000
M.0 M- ...
a0. = - s . c a
J
*>.S.4-=
~
0 '.. F50 I
ik0 omo Ao
4 u. 0 ud 54Z .4'-4 X.
-
0II M-OM
=j"rt-
MO :M M2 Cfl
a'- . v
A
MOV G- a , M0'=
0:::
- 0
EOa Q 0 to0 U L.
.1
3. 310a-l to a = 0 a a to~n.. .
-4r
0 -I-
00 U ~~t)J. yO 0 0 U C
V2 Zl -2 Li
H Z r-
C *0 0- 144
4m 4 1
ul.
as00 .C : 00 cc 400
C24 04
L) cm & .1 0 u 0 -
.4
-. 4 4l 1134 t- cc o
.00
A~ 0 00 M~ . .- 2Cs.
t
-A I!
A-1 _ _ _ __ _ _ _ _ _ __ _ _ __
___ _ _ _ _
and____________
-~Low altitude
crouise
Pitch
maneuver
IDive, yaw, and
popup maneuvers
Fourth flight Ffhfih
- ~ _ _ _ _ _ _ __ _ _ _ _ _ _ _7
Figure 3. Supersonic Tactical Missile
VI
__
METAL CASE
* AST LETs EXTENSHIOM ANmD ACTUAT1OR
*NOZZL E SHELL -MACHIKE FDRIING
* C931D04 INSIJATO (NO ACTtVE COOLING
REOIRRED
OD INCREASED TO RETAIN ALVRJ
ALVR' C"MUSTO A PROPELLANT BRAIN CONFIGURjATIONj
*INCONEL 1
*DOME- ANS CYLINDER MCHINE FORGING-
*CYLINDER -AND NOZZLE SIIEAR FORMED
-AND MACHINEDJ ____
- _Fa~I
-S
2:1 z -a--
-- - I
It-!
flu
ZIME
Of f4 Ib
9!4
so ~
is=
aL rata
Um ISW
_w _H_
In 0 T.ZZ
4--1
ELI ctrr Af
igm -W Ceorr\ze-Pp Tes niea Enduranc
Cas
-.R-Iy*~ gase
ID 093T104
Cobsh of pyoyi ae
kk pehdtnzn
Fitu-I
s 11
f- ma f;O so C9-0 2--n *
I7 -
~P d
IM
IO
7-13
II
-CD)
ce
cl; -D 0
T4-
CD
CDC
LU
cc)
1 1K
ImI
AL
7.14
~ ~88922 922
S478 *4I 700 Low-c70s
S367 Lo-ot478 478
000
Test time Test tima Test time
S92292
S700 700
a478 Low-cost 478 Lnw cost
I-
0 0
Test time Test time
Figuire 14. Low-Cost Combustor TPS Performance
IMMR
IM
STM Vehicle on A-7 Aircraft
S Booster ignition
Powered dive
Range
STM Flight Trajectory
A ~21 April 1979, Pt- Mugu Test Range
Iigure is. Supersonic Tactical
-missile
ENOINE SIZING AND INTEGRATION R,:QuIREMENTS
FOR HYPERSONIC AIRBRFATHING MISSILE APPLICATIONS
by
Paul J. Waitrup, FerikS. BilliC and Richard D. Stockbridge
The Johns Hopkins University
Applied Physics Laboratory
Johns Hopkins Road
Laurel, Maryland 20707
U.S.A.
SW MARY
A procedure that provides a rational means for selecting an Inlet/oZbustor con-
figuration for a hypersonic airbreathing missile is presented. The particular problem
that Is addressed is the design of the sustained engine of a two-stage missile that Is
constrained to be launched from a stowage volume that is either square or circular in
cross-section. Thle sustainer enbine accelerates from a low altitude separation at Mach
~4and climbs to high altitude for cruise at Mach 8. The results show that a missile
with an axisymmetrlc nose inlet provides a somewhat higher thrust capability and slightly
better fuel efficiency than a chin type Inlet. Aft entry Inlets are shown to have a
substantially lowter thrust potential and lower engine efficiency. A criterion for de-
termining the maximum contraction ratio of a fixed geometry Inlet is established and
applied to the exemplary missile designs. Combustor area ratio Is examined and found
to have a relatively zmall effect on engine performance for area ratios equal to or
larger than that required to obtain maximum thrust at the take-over Mach number.
NOMENCLATURE
A - area
- total vehicle drag roefficient
anose, leading edge, cowl wave additive, body friction, and
Dn,le,cw~add,f,tw,tf tail- drag coefficients
CN - normal force coefficient
CT a gross engine thrust coefficient [Eq. (A-!),
CT = net vehic~le thrust coefficient C,. CD
D = diameter
ER = fuel-air equlva-lence ratio
F strean. thrust
f s - stolchicmetric fuel-air ratio
L =length
M Mach number
P =pressure
Q= heat transfer
q,~ free stream dynamic pressure
r, R =radius
T thrust, temperature
U =velocity
=fuel f'low rate
WE =box width (Fig. 11,
x distance
Z altitude
a = argle-of-attack
(0-90)/2
y =ratio of specific heats
* = inlet smile angle
=90-0/2
~ME~~c =Inlet kinetic energy, exit nozzle and combustion efficiency
0ubs- io5 = station numbers (Pig. A-1) ApprovedJ f-or Presentation and
a = body publication at the NA7/AGAM/
c = cruise PEP 58th Symposfu- . Additional
= maximum circle in box distribution hlmted to U. S.
Ia s=Inlet cowl Government agencies oniy. Other WI
5max maximu= requests for this document must
n -nose be referred to the Johns Hopkins
ref - reference Vniversity/Applied Physics Lab,
t = walz.07
total Johns Hopkins R~oad. Laurel,
__ _ _ __ 'A___ __J_ _I
A.
INTRODUCTION
The design of a missile Involves the consideration of numerous factors that
include both the mission requirements and the constraints that are imposed due to
weight and volume limitations. It is the purpose of this paper to present a rational
procedure for guiding the design of a supersonic combustion ramjet (scramjet)-powered
missile constrained
chosen because a particular
it istothe cross-sectional
most efficient area. for
propulsion cycle scramiet has
Thehypersonic beenwithin
flight
the atmosphere (Ref. 1). To explain the techrique It is necessary to select a particularA
mission requirement, bu. the approach is not dependent on this selection.
The problem is posed as follows:
1) The missile is to be designed into a launcher that has a fixed cross-
sectional area, either square or round, but, as will be explained below,
the detailed analysis will be directed to the square cross-section;
2) The missile Is a rocket-boosted fixed geometry scramjet that has an end-
of-boost Mach number of 4 and a cruise Mach number of 8.
3) The scramJet is of the dual mode type, i.e., at high fligh Mach numbers
and low fuel/air equivalence ratios, the flow in the combustor is supersonic,
whereas at low Mach number and high equivalence ratio, a normal shock
precedes the combustion (see Ref. 2 for a detailed dlscusslon); and
Inlet capture area is either circular or a sector of a circle which is
h-he
co.monly referred to as a "chin" inlet and, by permitting the inlet cowl
} lIp to be located downstream of a forebody, an aft entry Inlet is also
represented.
The objective Is to develop a rational procedure for selecting the geometry of the
Inlet capture area, the design Mach number, ME ,the Inlt ccntractIon ratio, the
conbustor area ratio and the nozzle-to-linlet area ratio.
INLET C0NS7RAINTS
The problem begins as an exercise in geometry to determIne the range of "smile"
angie, :4,of chin and aft entry type Inlecs that can be considered. Initially the
crIteria for selection will be based on the maximum proJected frontal area of the
:nlet, later the effects of reduced inlet area will be -dIscussed. Figure 1 depicts
the packar=ng of a chin type inlet constrained to fit within either a square or
cIrcular cross-section. In each sketch the axis of the inlet Is located to give the
maxi Inlet area for a given 4. For the square, or box,.
constraint (Ref.3) the
Rxis Is ccentric with the bo axis for 3600 > I > 270' For 2700 > 4 > 900 the
axis moves along the diagonal and reaches the corner at € - . F r 4 Z 900 the
axis Is in the corner. Each sketch also shows tha-suilne of the circular body,
conceantrc w'-'h the inlet axis having the maxiu' ra For 3600 > 4 > 2700,
ra W_ 2, where WI s the length of the box sde < 210 ra W decreases
-
notoni to 00at I= 900. It would be per-i:Atsle to have a radius ra 'MAX
and still have the maximum inlet area. Moreover, r can Le gre-ter than r but
-only
b decreasIng the inlet area. aa:Ax
For a cIrcular constraint and 3600 > > 1 00, =ax'm= - I let area Is obtained
with she chln Inlet axis concentric with--the-axis of t-- circular "container." For
!cl!3 0 the Inlet axis moves progressively outward, reachIng the periphery at 4 =
.
9U For 2 < c9- the axis is on the periphery.
The ratio of the inlet area Ai to the area of the circumscribed circle Ae.
i.C., -W2/4 or ;r.2 for ra = r5 Is:
MAX
aBOX
no > 0r
4$
.J1
0 270r -3W 0 206= Bi I0 3
A.
i. A
Where * goo m
6
Wc > >=
180/1 +-
N-(9)
_Ai=
4<4
_ 84
-Even though Ai/Ae values-are shown-for the entire range of €, some designs would
not be reasonable in that it would be impractical to faIr the asymmetric inlet Into a
cIrcular-or near circular cross-section that would be -typical
0
of a missile. A reasonable
I---t would be-r./_ Z 0.3i which corresponds to 0-- i' for the box constraint.
Thus, for € < 140% smaller values of Ai/Ae would result. in this study this limit
was not enforced but as it turns out, It is a moot point because for other reasons
the chin Inlets of interest have € > 1400. Moreover, to lImit the scope of the
analysIs only the box constraint was studied in depth, which, as shown In Fig. 2,
gives the chin (or ift entry) inlet a~more favorable Inlet capture area relative to a
cIrcu ar inlet for < 2700 than would result with the circular constraint.
1.0
OR
6E
2 Circular Iauniier
X 02-
Fig. 2 Maximum inlet air capture area as s function of inlet smile angle and launcher type.
IMU-ET PERFOPRMANE -
-Me next step in the analysis is the determination of the inlet air canture
ratio, addItive drag az.d cowl drag as a function o- Inlet design Mach number M..,
= 0s Mach number M_, and ungle-of-attack. Here, to lirit the sty Is necessary
to be swewhat arbitrary in defining the geometry of the compression surface. Prev!cus
exerence has shown. that a surface ha-vng an InItial- co-pression anle- or 12.50
vle.o a good cc prI-se between nh_ atnd pressure luss. Consequently, as shown in
°
12.5 was used throughout. of the addl:Icnal compress.cn -an the inltC
was ass"_-ed to take n!ace downstream the last chaacterIstic passlng through th
4
cow- - n ts concal flow, compression weves would fall insid the cow-
li- regardless r he Mach number ant an;le-of-attack. For the aft Inlet, a 2.6:1,
3/-poer forebody S-pe was selected a-ad was Joined to a circular cy!inder. Here
--gi, twas necessary to be somewnat arbitrary in selecting 0.6 as te ratio of the
radus of the nose --- Inder to the radlus of the cowl ra/r. A smaller value would
n -ert suffcient packaging In the forbody, whereas a larger value would decrease
the a n-t of ca-tur area. Even 1- the cross-secticn of the aft In't 'a eircular,
I 60-ax-um
60, . -- air cp-ture ratio is less than one because the cowl lip
I. iside the law shock generated o- :.!,forebody. For example, If the aft inlet
was .aced in the mest forward positIon, as shown dotte-d Fig. 3, the shock generated
r
12e.50 surface intersects the bow shock at values of A = 0.3BS; 0. t9 and
-c T-- -
_ _ _ ~AL --
A/
(a~ Ahyrnnt~
r~12.e
S 0.4-
= ~ f I I A I I I I
C-rlwr=Ml-
nL4 -n v
ratio- -- l. - -Rslso
ul dceae-.n-.on4m!y - -e coi -a
ho re'
lhis s how~~~nelr I'ee .1n
19)I inFi. Us F
ar avilbl dat fro tests ntai of
hih -A (Rfs %heo
1 an ;lrm
calculbeens
aeo .v Sent forta sad er ei to og~qe a= 5 -. were.
effc dtency In txtesiv
otnffciw tv ded- Ianfl of-a-c
cenpr (For
zbho) Tre
I a tIns shown astth = A' 9
inFg6-nIn i.
ar dt aalal
f ess fhll-a: :nes Rea-4at h2.o
Mt=
I4
As-
5 I4
I M -A
OMAR-
Dram:6
%A~p.3
I II tI- i t z0
f -
ed
0 kr
4--~~ - - = -=
zeen~u-
-W of-c-=-
a-t.eav~ ar -C d
M=-
AA 1
the
'0ba WlC~n-.=
th e '=rm
d M wnm-
Zc t!, = rs
-e
s t - c - *= = - -
e4f- 74-~ 4
7- ' re -
=-- ---
=
--- s delaf- r- -4-
tNx =he~rin ex ztz tlc.f
metlo rat- :creca-I
is etuy--hez a~~=-
Pa en=-- =c4444
-- ==-=-- Cas se C0
.*-w
t
us fMla-i - mm -
Um-: a=----
-~~--~ ---
-am7
made
m- MM
-- . = ~
$- =
= - ~ ' =m-
* 'u
ms-=d h
n same-a-
mud4~- _ __34
- -Au
A2 ~ S J~..' i
- Os=
Ou--t . -
am - =
2
Fj I- =
fim me an 4_
aIg -b e-
ru-= : r s S I 'n e-X'
4 -K
I - -m
t Se
as' an
ssc ~ ~ -- - U: !E
-pe
I~
_ _ OE _ _ _ _ _ _ __ e-___
~~ne- a_ rw _______
_1a ar ,Z
Ilil
op perhaps MD S = 5 would be a reasonatle choIce. However, this would require the inlet to
operate significantly above its design Mach number, a factor that was not adequately
addressed in the selection of the correlation for IKE (Eq. :0). None of the data covered
tests of Inlets at M0 considerably above their respective MDES' Presumably, for the same
MI/M 0 , nKE would decrease for MDE S > M0 . Thus, MDES = 6 was deemed the praferred zhoice.
The choice of the larger A4 A2 was based on the slightly better average CTN at a = 0.
TN mAX
Figure 11 shows corresponding C. curves for chin and aft entry inlets all having a
-N
not atypical smile angle of 1600. Because of the prtvious arguments regarding inlet
Nozzle exit-to-inlet geometric area ratio, AJA i
09 1.126 1.393
I 1.492 1.333 ,1.0
M6 s 4
A4 /A-
=
O M, 4 Aref A5
I..W
06
S0.7
o 06
X O.r
03 i "
90 ISO 270 '60
Inlet smile anqle, 0 (deg)
Fig 12 Maximum net force coefficient at M 4 as e function of smuie
angle for nose, chin and
aft entry inlet designs with MDE
S
operation at H0 > MDE S) M < 6 were not included in the aft entry summary. Results for
other values of 4 exhibit similar trends. The previously stated arguments for the nose
entry used to select MD S n~d A4 /A2 are valid for the chin and aft inlets, so again Moms =
6 and A4 /A 2 = 4 were chosen.
With MDES and A4 /A2 specified, it is now possible to compare the efficiency of engines
having different types of inlets. To accomplish this, comparisons need to be made for
cases where all of the engines produce the same net force. Figure 12 helps to serve in
explaining the methodology used to 'ahe this compariuon. Here, maximum net force coeffi-
cients (i.e., M 4, ER = 1, a - 00) are shown as a function of 4 for engines all having
0
MDES 6, A/A, 4. The minima In the chin and aft entry curves are a direct consequence
of the geometrical constraints on inlet size shown in Fig. 2. To enable comparison of
engine efficiency at the same thrust level for all 4, CT c 0.628 and 0.382 values were
chosen. Thus, at the higher level, nose and all cin inlets can be compared and at the
lower level, all three types can be compared. To reduce the CT for 4 9 2090 engine
11MAX
designs, the engine inlet area is reduced holding A5 constant. For example, for the nose
inlet A5 /A4 is increased from 1.0 to 1.53 to reduce C from 0.89r to 0.628. For <
- - 3600, the reducti.,s in Ai are smaller since Cm are lower and the corresponding
NMAX
A /A,, shown on the top scale in Fig. 12, are larger. An alternative method would be to
5
reduce the engine ER, but this was not examined herein.
Z- 4
i -8
Figure 13 summarizes results for the chin and nose inlets at 1 O = 4, a = 0, ,0
5 ., Fuel flow is used as
= 00 and +50. the index of effio-eny. The a 00 curve is
for ER 11 and thus is representative of efficiency during acceleration. The 1 - 8, a
+50 orve shown in the middle figure is for C,. = 0 and thus is representative of cruise
efficiency. For either case the nose inlet engine is more efficient than an engine with a
chin inlet. if efficiency were the sole criteria f:r selection of A, then a large srile
angle would be the choice. However, as suggested by the sketches shown in Fig. 1, engines
having * greater tan about 210O would not benefit from Improved packaging efficiency
relative to a nose inlet that has often been ascribed in other studies. In any event, the
differences in efficiency between, say a 6 = 1800 chin and a nose inlet, are nactprofound,
amounting to gbcu: 3% at H0 4 and 91 at M0 = 8. Maximum accelerative capability at M
8 is also slightly better for the nose inlet, as shown in the figure at the bottom. Had
the basis for selecting the respective engine sizes been equal maximum thrust at N0 8,
there would be a greater difference in the relative efficiencies between the chin and nose
inlets but the chin would have greater thrust potential at lower Mn. The general conclu-
sion, however, is that on the basis of engine performance, the nose inlet would be pre-
ferred.
65 1,-! + + I I |
Ms 4
CT =0628@M,
0 4,a T MD0 <6
i G =• nA 4 "A2 =4
~~
pFR=1 ~~C'an leAet
6
7 .0
-
CI -Nose inlet
3 551 1 I I i I I| i
oNose iCetsi
j W-0
0 -8
Chint inlet
5- Y
Ine dg
VONsml,~
+ --. 41
- +- . . - . .. .-- -
+ _- _ t - t _
When the comparison is based on the lower value of C, at M 4~ a =00 so as to
include aft entry inlets, the arguments relative to the comparison of the chin an~d nose
configurations are similar. On a fuel efilency basis, the aft entry is poorer than the
chin !'or toe range of interest, as shown in Pig. 14i. The curves relating to accelerative
capability a'. 1.1 , shown in the bottom figure, exhibit trends for the chin and nose
inlet similar to thonle for the engines designed for higher thrust. For the aft entry at
a = 00, thrust levels are 60% to 70' of the corresponding chin inlet having the same 9
At first glance the curve for ;he aft inlet at a 50 appears to be anomalous. In fact,
the very large decreases in with increasing and the significant differences relative
to the chin inlet at the same 9 are directly attributable to the far greater sensitivity
In air capture of the aft inlet to change ir a. For these conditions with smile angles
less than about 22-0*the relative ai~r canture of the aft inlet increases more rapidly with
increasng a. For larger 0, the trend i'sreversed because the "shadowi~ng" effect or. he
leeward side Is more pronounced as the Inlet is moved aft. In general, an engine with an
aft inlet has less tnr .st potential and is less efficient than either a chin Inlet or nose
inlet engine.
4.0 082' 0 4a
;ER -l
inseinlt
~
Inletnie ' dg ~
sAl re
L ER I
At
A.1
The results of this stuly lead to the following~ conclusions and :,commendatiofl5;
1. The net performance of nose inlet missile designs Is somnewhat better than
comparable chi- inlet designs when subjezted to the same launcher cross-
sec' ional area constraint. Aft entry inlet designs have substantially less
thrust potential and lower efficiency than either the nose or ain designs;
6. The choice of an optimum smile angle is not clear. A lower bound can be estab-
lished when the zadone size is defined. The optimumt, however, would depend on
a study In which internal packaging efficie ncy is Included; such a study should
be made.
REFERENCES
1. Waltrup, P.J., Anderson, O.Y., and Stull, F.D., "Supersonic Combustion Ramjet
(Scramjat) Engine _evelopncnt in the United States," InvrIted paper presented at
th grnterationa Svmn:,.1um on Air Breathing Engines, MUnIch, Germany, March
19767U P Preprinit.
2. Billig, P.S. an~d Dup ' - .J '.If Shock haves and Heat Addition
r,' :h~sc
in ;he Design of uei-'scnh- Corbst__s 12tczAos05.u- (International) on Combustion,
Tht? Combustion z Ltzute P-tbrg ?., 19 , F . -_i25 5:
9. Wdaltrup, P.S., Billig, P.S., and Stockbridge, R.D., "A Procedure for Optimizing
the Design of Sca Engines,"
Propulsion Conference, .-
.e invited paper presented at the AIAA/SAE 14th Joint
as Vegas, NV., July 25-27, 1978, AIAA Preprint No. 78-1110;
t
Also Journal of Spacecraft and Rockets,Vol. 16, No. 3, Ma-me1979, pp. 163-172.
_,- - -
A. =-=
_ -
8-16
I~r
n
z --6R2
Command through !.r. 3.
mozat
(SEA-Epn;''.he,
_ wor warrs S.r'.AeaSstm
.0
U1---
31
Yn0~~C
AI Yt±t
I. is APpendix Contains
engine internal perform descripaoso h rcdrsue
e external aeodf~ oot~ ne n
by ~hessle Pcdregs
and
~ rius es~ns an hentfoc roue
alngw~hpc~~cexmlary sacesAl orthe wor~ng gratz and equtIo~ is are-pre-sened
for nose, chin _-ndart entry--
InetmsSIle QeI~~
- Iur
A- oesnte a schematic Illuttcn
re~~~jet~
(srja~nle along with the st'-tAo of a typIl'
into regIons of iao itrst-vz.-h -
conr ustnd oz le. In the s rarjet Int numbeats -Sa trc-!v :I~ I
(station i;s~dlrrused to e c ept t-- h yerson ic
lower sune-so= c apead 4 s e '4 rea -
szn
the Isolator aIr duct. An Utt-
shcyestem located at the sIr dutIA eurd rvp on
ec--
advere1 Interacting with entrance or the supe "soni cct0to f-~
4. r~nue
the Inlet,'Offlow ~a on
o =a n oblicue angle from the -. i Injected et-
then- burns in the wall Int -e air srren-at sta_-
nozzle (statlons( supersonMc to!r Combustr, z -e-'.--
me heart or the acra-fet
over a wide rane ofg.
en-gine~- ~ rwb=
2tos
P- Mlgh to c.at
and 30.5 km (100,4000 ftj or h... e between s.-_
the combust-on pocess. - .s 5
Fuel Is in~ected at statIonl Hee --- ~ete-
w
procerses Produce a OzrsIle (' and the combiation of- th '4I-e
-.- 8o ±-l
~ that X. te--'-A
i4 rs Fig A-e2 as
prpertles a-t statio- !2:
per
areatles at statzon Q b :n a 1Con~u t-c oaca--
ratio (AL A.J, the te de;e- -- - - - - - - -
tal- ressure a-csses due to n r - - -c
wall skI.. th
%rtwi arcnd ~, '"'-'4-ic
thav I teat trar=n '='d heal
f-- r ~-- ~ a
T-he strers- 2 -. =
Iron
a Machwave fo eAe AI
a c-xi~u= co--- . a~. -Cr
c o n b_ st o
Beceusat4.e C..U-nf-A- C__
Strength
auperson-ronl
cc!:z
a-' -'p- I
is. p_7cee 4 by...==-
-- n wtt-
. 4 . -
-- a.-
-o
E4.. --
!
--
iI
8-18
veracity of the analysis for predicting scran4pt engine performance has been
established.
VEHICLE D
D:4 N- CO-NSTRAINTS AND MISSION HZQUIRSMENTS
With the size of the missile thus ccnstrained and the flight regIme established,
the resainder of this appendix will present detailed descriptions of the inlet and
engine performance calculations and aerodynamics and drags used in this study as well
au sample calculations of the results presented in the main text.
INLET PEREOIMA CE
The Inlet a,- capture and additive drags were determined for 'he annur. hI
and aft entr-y nlets depicted in Fig. 3 of the -an text using a three-Imens-na
solution of' - hyperolic equations cf =tlon. The analysis is very slmii
. to
that reored - for example, Refs. 4 and , and codes of this type are now readily
available.
The results of these computations are given 'In Figs. A-3 through A-16. Free
strea M h n"r-ber- of 3, 4, 5, 6, 7 and Z were considered along with angles of attack
of 0 , +5 and +!A . Figures A-3 through A-1Z present Ao/A. and C as a-- n-o
0 Uadd
of inlet s=- ag-le (A), M 0 and angle of a c the nose/cA-. 'ni" h
=
Fig. 3 of t- a- text with M - 6 and . Figures A-ll through A-- Presentf
DES
the corresponding results for the aft entry Inlet depeted in Fig. 3(c) of - n--a
text. Here, however, only inlet design Maeh -. bers of 6, 7 and 8 were considered.
Note that the reference area is taken as the Inlet geometric area (A or
P l-
AA
1A.
(2) The loss In air capture at P. Ia-_ZS and a - +100 of the nose
inlets-is eli=Inated by using a chin Inlet with a smlle angle
of 2700 6r les; and
(3) For V. < YOF% , . > 0' and chin Inlet designs with * < 2700, the
air capture increases with Increasing and decreasing 14 over
the zero angle of attack values. For example, for Y.0 S - 6,
V0 - 4 and # - 1800, the air capture at a ±10O is 2Y% higher
.
than at a - O
A-heresults in Fig. A-19, on the other hand, show that the air capture ratio of the
aft entry Inlet ! always les than that of' the n or chin Inlet d-sIgns except fer
.Z a a 410 and here the aft entry inlets air capture is 65 to 12n
higher than for the chin.
The reason for these differences is that the bow
never intersects the owl !Ip, even at angle-of-attack,shock of the aft entry de.!. n
except for the above mentoned
cases. In additIon, ..a three-dimensional flew for e < 360 a lateral (non-axlal)
flow is established -hlch pernits a portion of the air to spill around the Inlet which
would not occur in an axly~etric ot two-dl=enslcnal flow. Consequently, the a~r
capture characteriattcs of the aft entry and, to so=; exte-t, the chin Inlet are lower
than would be expected for a comparable two-dImensaonal n..et des gn flylng at angle-
of-attack. Thus, whereas an aft entry inlet Is a vIable Inlet design for a more
nearly
engine two dimension
(Ref. 12), _'tflew on a than
-_3less body desirable
at angle-of-aztack,
air artujre such as the NASA for
characteristics hyperscnIc
o
hypersonlc missile. ThIs is especially so at low Vach numbers and lcw to mderate
angle-of-attack.
M;INE PERFO3RMAN=
stolchlometry, fe - 0.g7284
heat of formation = 20 kcal/ -mole
lower leading value - 9.97 kcal/gm
chemical formula - --0
molecular weight - 16.
specific gravity = 1.07;
(3) The air consists of 0.75525:2, d,23t 02 and 0.01325 A by weight;
(4) The cm -stor wall wetted-to-inlet are& ratio (A/A
2) Is 40.0;
(5) The inlet and combustor flows are In ther=oche=ical equillbr - 9
F (6) eat transfer through the walls of the combustor are neglI'gIble; and
(7) The nczsle exit stream thrust is based on an efficiency of 0.98 for
an expansion that is one third of the way between fromen and equilibrium.
With these assumptions, a =atrix of gross engine Internal thrust coeffIcients (I)
were computed for an altitude of 15.24 km (.00 ft.) for effective contraction
of three to the axlmu pe.Itted (Fig. 7 of' the main text), equIvalence ratios ofratios
0.25
to the maxlu= permItted for H. Z 1 In 0.25 Intervals, combustor area ratics (A/A )
2
of 2, 3 and 4 and effective exit nontle expansion ratios (A /Ao) of 1 to 5. 3ross
5
thrust coefficient in these cases is defined as
-[
C' Po O (As AO)] Cqeo) Ai
-A-
wJ
where F is the stream thrust. An altitude or 15.24 (50,000 ft.) w55 chosen for
simplicity. Although s=all variations in C. will occur at other
wiItudes * he
9
CT a given are a good approx=ation for C0..all altitudes of Interest.
The results are presented in Figs. A-19 through A-24 where C_ Is g ven as
AEROPYNAPICS ANDC.RAGS
T.he final parameters needed before a co-plete vehicle perfcrmane et Iae tan
be made, I.e., f-lnd!ng the net vehicle force coeff! lent (C. as a functIo- of..
x-
a and engine des!gn, are the nor-l force coefricIent per deree of ange-zr-atacK
(CN/e) and the zero lift external drag coefflcients. Fcr st-lIctty, a Inle :arve
for CN/o has been assa=ed which is representatve of' those cc-puted fzr ±ther hper-
sonic missile designs. The relatonsh!p is
a leading edge dIameter of 0.25 c= (0.1 ln.-. Trhe .cdy a. ta1 frlct!cn d70 Cc
effllents (CD,0 ) were computed ass=lag a -=-ot, ad a wa wh a
Df C tf
developed turbulent boundary layer. The tall wave drag coeflcen I'-
' as ctalned
°
assuming g two-d-mensIonal ftou at the free streamcnditnn 'Or-over a 15 angle wedge
with a 55 sweep angle, a total planfor- area of 256 C4 "a . - a -! d ... ns ..
surfaces wr" 5- -_
ratio of 5I. it should be noted that four c... -..-- - -
inlet vehicle designs assIng skd-to-turn aeuer, and three surfaces Trthe chin
and art inlet design ssu=ing back-to-turn maneuvers. The total zail eianfe- area
In both cases, however, Is the same.
The resulting drag cOefficients are shown In Fgs. A-S throug -_- as a
of free-stream Vach number. Al! are references to a 5C-c= (1.-- - e.
Note that De and C w are also functions
of H1 For :I- t, and are
given per unit body surface area and per unit projected frontal area r vely,
where the values of C are based on an AA. of 2- i- nt-duae less than a
42% error in CDew for 1LA5 /A, : 4. Base drags are not Included since It is ass-
that the engine exit nozzle rll1s the =axl-=- circle afforded by the launcher constraints.
With the inlet performance, engine perfoance and aercdyna-m cs and drags
established, it Is now possible to preset detalled exales of how the net force co-
efficlents (C-N - . - C, ) at N0 - 4 and 8 In the =an text were generated. S-ec fIcally.
the cases for which C, - 0.382 at N0 - 4, u - 0 and ER - 1 (F~g. 12 of the maIn text)
0
for the nose inlet and * - 1600 chin and aft entry Inlets will be used as examples.
Note that NDF S * 6 and At/A 2 ' -0 in all tree cases ano that an Iterative procedure
is required to arrive at the A5 /A. needed to produce a C - 0.352 at M- 48, a 0 and - - -
TH
Iz
- -= - -r - -
ER- 1. The exa ples, however, only Include the final coputations.
To Initlate these calculatlins requires char the Inlet geometric area contractIon
rat o, A,/A,, be chosen such t does not exceed the =axi-um contraction ratios,
(A 0 /A 1 , set forth In Fig. ?of -he maIn text. Since
= t/
AA)
0 max /(A/_Ao 4 ,
and A/A, var~es with both an a, bounds on both Fo and a need to be established before
A./A, can be selected. ne Mo As for -he veh cle designs considered are M0 4 to
Also, mss1les of this type generally follow a nearly zero l1ft trajectory dur-ng the
Initial stages of oi--cum fron = but requlre negative lift during the latter Portion
R of the cl4-cut In order to le-ve-off at the cruise altitude. This generally occurs at
t
Yos > . Once a t crulse a II"e a poritive a is required to maIntaIn I "g" of lIft.
Consequently, a - at - an ad a = +1' at M. Z 6 have been assumed as the rexions
of prInc- ;al Interest. HM.e-ro nOrally occur in terzInal maneuver at whIch time
engine ;erforz ance ,'ey
---
yre n-'portant.
With these -os ea ( 3) en used to determine A4 /A1 = f (VI, a) uslnm
Fig. 7 for (AC/A-) =a% and A- and .A-Il, respectively, far A/A. of the nose
chin and aft entry Inlets with H_ 6- -e minium of these values is then selected
°
as the maximum A iAI, (A/A)_ , f r each inlet type. For the nose and li 0 chin
nlets, the minimum cc*-,--s at G and (AA,) max ' 5.24/0.745 = 7.034. For
the 4 = 1600 aft entry Inlet, (4A,.)_-j = 8.32/0.756 = 10.529 which occurs at Mo a,
+10 0 .
-had a = +100 teen th - "-cti criteria at M = 4 and 5, then (A /A)=ax for the
nose would still be 7.0Lbut tha t for the €= 160 chin Inlet would be 5.413 which
occurs at V0 , a = C1a ---- t he
--- aft entry inlet would be 7.686 which occ-Us
at YO n 4, a = + a
u=m ry of tes results for all values of * is presented in
Fig. A-31, where AIM --- as a function ef inlet smile angle using the two
With A1/A. establoshed ' -sh design the remainder of the cozponents of the net
force coefficlent are calculated as sh-wn below:
A, =a---,z-=A..3:9
= 82.5= X.127.34fl .
cowj
= ti
(25. iL..i 6P
.).}
-
l = a (16.74) (.-) . (4.00 in.2)
= (33.
(3 7. -) U
4-7i
IF
U AYU
= ==A
MEr
tz -S - *-- f aa m
t A- _
Wa
u mr o-'s Z- v-I 31
mt-
SibA T"i_
F~- 2.2,yCM 0.75 :-
Qrz A
-=-
2 )I
A=K
MMI u
ff:33v- 3
* a-r~-
-
____ON
* .mw-
p U-52- 2-7c WD
AV- (.2zsf
Jm ~
Z = pm- -- f
mc A
lb
-30.8 (12.14 in.)
r -30.8 [.1-(.'059 1/2
*29C-.m(11.142 in.)
howevver, to Lc consistent with the inet flowtields generated in Pigs.A-11 and A-112,r,/r,. 1.66?. 7Ttrefore
r, 30.8/1.667
r.-18.148m. (7.27 in.)
A - *(18.148) 2 (2001360)
2
, 596.ocm (92.39 In. )
Ae (2)2 _ 1327 _ 596.0
2 2
- 40.L=s (628 in. )
1t + 35112
39,525
2 2
- 43,67cm (6675) in.
0
iezo ift drucoeffiients at M0 4 an a-06 referenced to A.are:
CDcd- . 107'0(6.28)
(1.617 - 0.0010
N~ote trata m, offset or -o.-ised for ttesft ertry' s~wndClin Figs. t-26 andA-27. WshIsabecause
tbC Ftach ntrber or the inlet's rnse just prior to entering tie second oemression field is zero to 0.2.
MOchnw-bers lower- thenYO
COn1 (0.0757) 192.39/304.342) - 0.0230
Cot - (3.53
x 10- ) (6675) - 0.0235
CDtw - 0.0092
CDf .0.0030
CD - drag coefficient
total -. 1t
__________________M m______I
8-27
add
=(1.248 x 10~ (6.28) =0.0008
C~t.=0.0088
C-Dt=0.0013
C~a =0.0228
CrD
At N8 w a OP,CD 0 butC- ~(r0168)/1.48 0.0114.
0D
Therefore,
CD=0.0928 - 0m228 - 0.0.125 + 0.0114 =0.0689
Again, the sawe procedure as before is used to obtain the Values of CT correspording to A -1.M8.
Hower, at - 0 4, ER -I anda =O, /A =.470. Thus A./A= (10-529) (0-47) 4.949,A 5/A0 3.149 and
CT 0.4972
CD -0.1145
CD - 0.0928
8 ~ R ~ 1 ~n MO.
For~~~ A/l 0.532, A./A=5.6011A 5 /A 0 =2.782anud
rr = 0.1730
cD - 0.0689
Im
8.2s
8-28
177
LI
LLI
--
8-29
[E z O2
0 0 - 0 0 0
~
C,, C,
'9
II s
bb2
E E
C; c i j'
II\ z
PPL33
6~pJ~!3jj*C
A11PO -JU
MW.
Ill
i
InC
0
'~~~'NifL'
_0
I- iiA
2c
Ix
\\104
I --
8424
co IL
8-33
C.,
L\\ L
-co- ta -T.C
-0
E=
IQ 0l 0, 0 0 C.0
Co f 0 V. 0 0 0 n 0 0 0
%~~~~~zld
L-4pW
'1.!JJO TJ
-34
L II
\ 'Kj
0 Q 0
-tA
At
'I At
'I~
/l
I~ t l 1 1cc,
XLm
~ii A,
0o
CAI
At/
I I
Al w
Cx
IIIIEI
NE" A1
LI
ct I
As
'4
Iii
tA
_ 411 li
A- U R.
ti --
Ii ____________z
___ ___co_
___ __ _ _4
fw__
4-j
rIL7
- - - - - -
\1 - -
AEli
Xif
EU ME
IfI
Ellit
II
Nii'
rii
-~ 0 0
Dill.
~ iir.CA
I= ~ liJA
ia w0'
c-li
I -
7-
III
k I
I
1/2
I
/ -
I _______ ______
-A.
8.40
I co
GO X)c
oa cc
h
F -1'E
I.E
Mf
IfI
acL)n el oio Ao
im /CI;aw
,w -~jO GN 0AJ_
cd Ili d
Io:
4j amunuxp
C
-. o
DESIGN CONSIDERATTON AND NALYTICAL COMTARISON _/
OP DIFFERENT TYPES OF RAMJETS AND RANROCKETS
by
Peter Benkmann+)and Peter A. Kramer++)
Institut for Raumfahrtantriebe
University of Stuttgart
Pfaffenwaldring 31
D 7000 Stuttgart 80
Germany
SUMARY
Several types of ramjet and ramrocket (or ducted rocket) propulsica systems with
hydrocarbon fuels for high sub- to supersonic missile application are theoretically
analyzed and compared. Design considerations are discussed with respect to their
performance impacts.
PREFACE
The results discussed below have been gained during an ongoing study under the
sponsorship of the Gernan Ministry of Defense in cooperation with the industry and the
Deutsche Forschungs- und Versuchsanstalt f~r Lu/t- und Raufahrt (DFVLR). The purpose
is to support ongoing experimental and design work with theCretical analysis.
The high level of the Specific Impulse realized by ranjet systems can sometimes
not be used because of the demand for higher thrust. Mostly the ramrocket can operate
at the higher thrust level but at lower Specific impulse than the remjet. The design
cf an airbreathing missile depends not only on 'he demands on Specific Impulse and
thrust, but also on the limitations oL combustion pressure and temperature and the
geometrical requirements. At a given Jituation of altitude and Mach number under the
iomtations a.-- requirements mentioned above che choice of . sufficient operdting
syste.n needs a detailed insight in the mutual dependance of the various parameters.
Some theoretical layout of missile propulsion systens ba3ed on their benaviour is
developed for some given criteria e.g. flight situation and geometry. Fig. I shows the
cross-section stations of the ramjet and the ramrocket (ducted rocket).
1. METHOD OF ANAlYSIS
Specific Components:
Inlet diffuzer performance according to AIA+)and
off-design correction according to Ref. 6
Combustion of all propellant combinations with
NASA SP-273 Program (Ref.7).
Fuel: Liquid Hydrocarbon, Oxidizer: Air, N 2 04 resp. HN0 3
Mixing and combustion chambers with variable area and pressure
Nozzle flow with defined freezing point at the nozzle throat
Nozzle expansion to ambient pressure or to maximum engine
cross-section area
The comparisons given in this paper mainly show the two most important performance
parameters:
1. The propellant Specific impulse Is as ratio of thrust to
total propellant mass flow, including oxygen when a rocket-
type combustion takes place.
2. The Thrust Density DF as the ratio of thrust to maximum
engine cross-section area.
Both parameters are equally applicable to all air breathing, rocket and combination
propulsion systems without any limitations, The Thrust Density D. is shown here rather
than the more usuai thrust coefficient F=F/O,5 o V 2A which is less useful at varying
and/or low speeds.
For a high altitude mi -i!e with a ramjet propulsion system the correlations of
fixed internal geometry anj flight Mach number at constant altitude !.pe been investigated
thenretically and experimentally at the DFVLR at Braunschweig/Trauen (6). The goal of
this study was to explore the whole range of physically existing solutions. Yet it
turned out that the simple theoretical model in use sofar was not sufficient for that
analysis. Therefore tne program described al-ove has been adapted, For constant altitude
and variable flight Mach nwmber nainly the internal gecmetry of the combuster and nozzle,
i.e. combuster exit/entrance and nozzle throat/ccmbuster exit, have been investigated
as parameters. Additionally a combustion temperature limit had to be considered. Fig. 2
shows as an example the Specific Inpulse versus flight Mach number with fixed geometry
and/or limitatiun of the tesoeratare at the exit of the combustion chaber. The var:atxon
of the flight Mach number results in a variation of fuel mass flow and Specific -mpulse.
Considerino the line for a fi-.ed cross-section area f3=o.6, the Specfic inpuise s
rapidly descending from the beginning at Ma=2.75 (point A) to the end at mac4.5 (point D)
with a slightly change of the bend of the curve. The temperatures are also varying and
the dash-pointed part of the curve (point B to C) requires temperatures higher than the
considered maximum of 2000 K. Another curve (dotted line) is plotted for a fixed
temperature and therefore a varying cross-section area f3 " Here a maximu of I, is found
Aircraft Industries Association (Refs. 4,S)
IA
at Ma=3.25. A third curve combines the limitations of the former two curves in the range
from about Ma=3.5(point B)to Ma=4.4(point C), where at the former curves either the
temperature was higher than the allowed maximum of 2000 X or the cross-section area at
the throat was less than the fixed value of f3=0.6. This combination of the two procedures
with its limitations requires a diminution of the density of the inlet airflow. This can
be realized by a supercritical operation of the inlet within a certain range which has
still to be defined.
This third curve (solid line) fits the dash-pointed part of the curve with f3 =0.6 and
variable temperature. For the curve from point A to D the behaviour of the total pressure
at the end of the inlet diffusor and of the temperature at the ends of the inlet and the
combustion chamber is shown in Fig. 3 and 4. At low Mach numbers, the critical cross-
section area of the inlet diffusor is less than the cross-section area at the throat.
Without any combustion one obtains thermal blocking at toe throat (point A at Fig. 2).
Lncreasing the flight Mach numLer decreases the critical cross-section area of the
inlet diffusor. So combustion cannot satisfy the fixed geometry requirement. This happens
at about Ma=4. The total pressure curve with loss in Fig. 3 shows a maximum at about
Ma=4. From Ma= 4 increasing the Mach number decreases the total pressure and increases
the critical cross-section area of the inlet diffusor. The fuel mass flow decreases,
temperatures decrease as in Fig. 4, passes the temperature limit (point C at Fig. 2),
and finally no more combustion or fuel mass flow can taxe place at Ma=4.6. Here the
critical cross section area becomes equal to that of the throat.
Specific Impulse Is and Thrust Density D. are the most important design-values of a
ramjet engine beside its specific weight (weight per unit of thrust). Modern missile
requirements tend towards high flight Mach numbers and long ranges. The pure ramJet does
not necessarily fullfill the thrust requirements for high speed missiles at low -1titudes.
The combination with the rocket cycle can but does not always improve thrust with a
penalty in I s . To define the op-imum propulsion system Df this f- % LIe whose spectrum
of the pure rocket over the ramiet-rockets to the pure ramjet has to be considered.
For example a propulsion system of this type for a cheap target drone shall be designed
for a sea level mission at Ma=2.5. Five different rocket gas generators have been
combined with the ramjet cycle and plotted over a wide range of air/gas - mixture ratios
in Fig. 5. In this case red fuming nitric acid has been used as the rocket gas-generator's
oxydizer. The rocket gas-generators have a range of mixtore ratios from stoichiometric
(curve 1) to extreme fuel-rich (curve 5). This leads to a ramjet afterburner from pure
mixing without secondary combustion to nearly all-secondary combustion like in a pure
ramjet.
The other three cases are located in between the extreme ones as true combination
systems (ducted rockets). Fig. 5 shows the characteristic values I and D F over the
secondary (ramjet) mixture ratio as the independent variable and the primary (rocket gas
generator) mixtue ratio as parameter. Additionally the primary and secondary combustion
temperatures have been introduced as boundary conditions.
These four diagrams car be used as a rapid graphical design method for ducted
rockets. For the example discussed here an estimated drag coefficient for the given
constant flight situation and the requirement to launch the missile out of an existing
launch tube with a given cross-section area leads to a minimally required
5
D =1 .9 x 1O N/m (Diagram I).
min
All curves below Du do not fullfill the thrust requirement. The projection of the
q limiting points of tie curves I to 5 into Diagram II cuts off all IS curve pieces with
insufficient thrust -dotted lines). In this case here the ramjet-like curve 5 gives the
best values in Is wi'h sufficient DF . Projecting the limiting points into the temperature
diagrams IT! and IV onnects the 1s and DF crves with the related combustion temperatures.
A chosen temperature limitation in the secondary combustor (ramiet) of 1000 K cuts off
most of the solid li es in Diagram III. Only curve 1 (stoichiometric rocket gasgenerator)
and curve 3 are left A further temo
perature ismitation of 2200 K an the primary combustor
(rocket gas generato -' finally leaves only curvo 3 in Diaqr Iv with an a</f = 2.47 ratio 't4--t
fullfills all requirL ants. A ratio of a/f=28 leads to an optimal Is value. This gives
a relatively well def ned set of prelLinary design values for that engine. Yet in
practice one would in,:rease the limiting temperature in this case here in the secondary
conbustor to 1100 K i: order to make use of the twofold I of curve I (ramjet-like syst)
compared to curve 3 (-ucted rocket). This requirement is an additional result of that
graphical short analy;is.
Similar to the case of the chapter before a pure ramjet engine shall be designed for
several different fEight situations and for a pregiven minimal Thrust Density D.
tmain.
on-C-sign behaviout is plotted in Pau. 6 a 7 with the according minimum thrust require-
i ments.-
The curves of the Specific Impulse for the four situations are overlapping. Only at
hioh excess air the curves are spreading, where two opposite trends exist: higher altitude
increases the Specific impulse values, but lower Mach n-mbers decrease them. So with the
given high Mach numbers at low altitudes no specific trend of the Specific Impulse values
can be predicted. In addition at high excess air a zaximum of the curve may exist. Yet
the Specific impulse at lower excess air, where the thrust requirements are fulifilled,
is much lower.
As in the chapter oefore the basic design consideration has been made in the diagram
for DF (Fig. 6) rather than in the diagram for Is (Fig. 7). High flight Mach numbers and
low altitudes increase the DF values. In addition to these flight situations excess air
leads to a rapid increase of D. Flight situations with lower thrust values need much
more fuel for the same increase of D,. At the D, line the flight siuation with the
'min
highest altitude and the lowest Mach namber requires stoichiometric combustion. At higher
altitude and lower Mach numbers the DF requirements cannot be satisfied at all.
min
The examples discussed before of possible ramjet and ramrocket resp. ducted rocket
designs show tnat the behaviour of the Specific Impulse and of the Thrust Density cannot
be predicted by experience. This is due to the coupling of all effective parameters like
pressure and mixture ratio in the primary rocket gas generator, cross-section area ratio
and mixture rati in the secondary ranjet combustor. Thus, decreasing D, values are not
coupled with increasing is values and vice versa. Generally, ramrocket rosp. ducted
rocket propulsion does not necessarily deliver higher DF and lower I values than pure
ramjet propulsion. For a flight situation like Mach=0.8 at sea level the four parameters
just mentioned have been varied.
A =
- Pressure in the rocket gas generator. This parameter has only a small influence
on is and D. An increase of the pressure increases I. and D, slightly. The
gradient is less at higher pressure. But, at this subsonic flight situation, a
high pressure is essential for supersonic nozzle exhaust. The pure ramjet of
course cannot run with !%;prsonic exhaust at this subsonic flight situation.
- Mixture ratio in the rocket gas generator. increasing mixture ratio decreases
DFlslightly. I has a flat max,um at ecuivalence ratio=2. This relations are
shown in Fig. 8 for fixed values of the other parameters. The =axim.u Is occurs
lecause of the optimum share at this case of chemical energy and kinetic energy
from the rocket gas generator. The ramjet combustlon chamber cross-section atea
ratio is f.=4 and air/fuel ratio is a/f=1.
- Crcss-section area ratio of the ramjet combustion cha.ber f,. A s=all value of f,
permits only little energy supply, e.g. values uf f, = are only aoplicahbe for
those ramrocket (ducted rocket) propulsion systems with only little secondary
co.bustion. On the other hand higher values of f- are only applicable with
sufficient energy sunlv. In all considered variations of the parameters there is
a =aximum for I and D, at f=4.
5 r
- Air/fuel mixture rat!: aL the ra m et combustor, shown in Fig. 9 and 10. For Pure
ramjets an increase of the air/fuel mixture tat-o increases I s . Under certain
circmstances a maximum of is can occur. DF decreases slightly. For ramrocket
(ducted rocket) proulsion systems an increase of the air/fuel mixture ratio
increases I-s.These values start at a higher level than the value of the pure
ramjet, but are q4ite lower rt higher air/fuel mixture ratios. The values for D,
are rather high at low air/fuel mixture ratios. They decrease rapidly with
increasing aIr/fuel mixture rLtlos down to the lower leve! of the pure ra=iet.
An additional variation of altItude and flight Mach n'ber can considerably change these
interactions. Also it can be seen, that even at this extreme low flight situation at
Ma=O.8 for the use of a ramjet propulsion system, there are cases at which the operation
of a pure ram-jet can lead to better 1 s and D. values than competitive ra-rockets. However
supersonic exhaust at subsonic flight can generally not oe obtained witn the pure ram-jet,
but partly with the ra:.rocket.
6. OFF-DESIGN BEWAvIOUR
First I and D_ at on-design conditions and variable altitude are shown in Fi'. The
relations are similar to those discussed in chapter 4 and shown in Fig. 6 and 7. Howeve-r
the flioht Mach number is constant here. The parameter is the aIr/fuel mixture ratio. As
can be expectcd the 1s values are higher and the D_ values are lower for the higher air/
fuel mixture ratios. Therefore the maximum is values might not he useful .,ecause of the
related low D_ values. The gradient of the D, values is ext -me at low air/fuel mixture
ratios. At lcw ar/fuel mixture ratios the Is values are nearly constant. Fig. 11 shows
purelv on-design-values for Is and D_. whrch generally decrease at off-desln condition.
The choice of the design-point has to consider this. Next Fig. i2 and 13 show the off-
design behaviour for at two different design points with an individual interral geometry
each:
- Design point at sea level , Ilch. Mach number Man3, Mach number at the comtUstion
chamber entrance =0.207, cross-section area ratio of the combustion chamber is
f'=.1544 and cross-section area ratio of throat/combustion chamber exit f?=O.8.
The relatively high flIght Mach number of 3 for the design point leads to sufficient
g
values at lower off-desi n Mach numbers. The lower fuel supply in this case also
lowers D The Ts values even run through a maximum at about Mach=Z. At the left
end of the curve, the values for TS and D, might still be applicable.
l I
° - Design point at sea leve-, flight Mach number Ma=2, Mach number at the combustion
/ !
chamber entrance = 0.415, cross-section area ratio of the co.mbusti-n chamber f,=i, and
cross-section area ratio of throat/combustion rhamber exic f 3=0.8. The low on-design
Mach number of 2 lowers the fuel supply, Is and D F rapidly at off-deiten condittons, No
maxi"um exists for I s . V, is at an cxtrme low Iml±. The possible fuel supply is mucn less
than in the first case. A des.gn point at the disadvantageous low flight Mach n-uber of
Ma=2 leads to a corresponding disadvantageous internal geometry which limits the fuel
supply at lower off-design Mach numbers. Not discussed here are cases in which the off-
design Mach n'hmbers are higher than the on-design Mach number- These cases are presently
under investigation and will be presented later. The results can be expected to be
similar to these of chapter 2 where a fixed geometry with variable mixture-ratios and
flight Mach n' ")ers have been investigated.
7. CONCLUSION
a. RlEFERENCES
0.S.
00
jk R
4 .
z,. 0
* -OU
I ~~ ~
G-~ 0X1
Lk C.Z 0 -
UU
U UU j~r-
U~~2 UUoU U '3
0 t
49.
x 0
U- In-
K: IS
1Z r _
.
F9 .
ALL
q.--
LLE 'o
C.)) i
II t0 aI
-4.
I
.t~
zlz
t::
-
- -
0-
*10 C -J
/ ".7 t.
1 * .0..
II
I / .7'
~
V
- 7'. - -
I
7,.
__________________I
'--0 2
o o 2- x 0 to
I 20
N.RAIM-f
3 O 5
ea leve- altitUde. -a O
X, 221
751 \\f±1.2J
1> /f2 C
altitude 1km!
2
OX10 I
k11
200X 0io to
18001 9 P
Ds 9
16001 88
1400, 7 7
1200
1000
56
5 /5FAt T_
O
6
3
60
40 2 2
200 1 M
o 05s 1.0 U 2.0 2.5 3.0
2000 10 as0*
1800 9 is OPR
1600 8 .0.4
14007
A, ma
A 1000 5
800 4 a2
- ~600 3
400 2 0.1
200 1
0 05 1.a3 0
0 0s 1,5 ZO 2.5 3.
z_ t77
-E~
294
DISCUSSION
R.Marguet
Quel est le donjaine d'utilisation pour lea cycles que vous envisagez?
Author'iRenly
Thie applications arc mainly for target drones and low altitude,
high Mach number ramiets.
L . .-.... r- . -
IO-i
SUMMARY
A review ia Presented of predicted flight performances,
static test-stand data, of ourrent technology based on well established
with alternative forms of propalsion suitable valveless pulsejets and comparisons are made
that valveless poeJ eta- have some performance for small, subsonic, vehicles. It is shown
for short-range missions requiring low thrust advantages over other forms of propulsion
lbf), The advantages of valveless pulsejets levels of less than about 600 Newtons (= 130
In terms of simplicity when compared with
other propulsion systems are also shown to be
substantial.
Possible ways of improving pulseJet performance
involve means for obtaining I-mproved Precombustion are discussed. The systems cobsidered
ly-firing pulsed combustors and rre sophisticated charge compression using twin alternate-
units incorporating dynamic pressure.
-exchangers.
1. INTRODUCTION
Whilst the writer is of the opinion
tion, in the field of flight pronulsion, that the best and most generally uz: ul
for valveless pulsed combustors lies in applica-
gain combustion system: for otherwise conventirnal pressure-
be a potential for rather restrictive, and selective, gas turbine engines, there appears tc
tops, or more specifically pulse 2ts, are used applications in which pulsed combus-
ers. The prime motivation for incorporat ng in their traditional role as thrust produc-
to generate an increase of stagnatin pressure, a pressure-gain combustor in a gas turbine 15
to providing the temperature rise asseciated across che combustion system, in addition
for which the pressure-gain combustor is with the conventional steady flow combustor
substituted. it appears that Reynst was
first to suggest this applicallor of valveless the
has shown that a 6% stagnation pressure rise pulsed combustors (1). Recent wurk (2,3)
ratio of 2.5:1, under laboratory conditions, can be achieved at a stagnation temperat-u-e
using a small [2.875 in (73 nun)combustlon
zone Internil diameter] valveless pulsed combustor.
Thero is a promise of greater pressure
gains from larger units.
The use of a pulsejet
which is easier to implementas than
a thr-st producer is in some respects a type
a gas of applIcation
Jet does not have to be integruted with turbine pressure-gain combustor because
turbo-machinery. However, the desirability the pulse-
'otherwise of pulsejets as a propdlsion device or
characteristics compared with those of alternativeclearly depends upon their performance
portions of a highly rated valveless pulseJet propulsion systems. The general pro-
developed by the French company SNECMA and
further improved by Lockwood (4) are shown in
Fi5 . 1 (a). The device consists of a short
inlet duct leading Into a short, enlarged diameter,
end to a long divergent tailpipe. A shorter combustion zone connected at its far
rultiple inlet version
writer's direction appears in Fig- r (b). The essential philosophy developed under the
of the multiple inlet
concept Is to reduce, in approximate proportion
required for the ming of air, fuel and residualto the reduction of length, the time
plished by reducing the representatve path products of combustion. This is accom-
length, proportional to the inlet diameter,
which particles of air, fuel and residual products
This concept Is described in more detail elsewhere must traverse during the mixing process.
(3,5).
An inherent problem when working with nulsed combustors,
consequently obtaining a detailed understanding is that of predicting, and
internally. For highly rated puisejets potentially of, the non-steady flow events occurring
is virtually essential to use a -metho4-of-characteristicssuitable for propulsion appllcation It
Hoever application of the e-.ehto .ulseJets type theoretical treatment.
primarily, with the diffic..y of modlllng the presents a number of problems associated,
combustion process in a satisfactory manner.
The peformance prediction technique and other
aspects of pulsejet design, operation
and performance will now be considered.
2. THEORETICAL TREATMENT
The method-ofcharacterstics
for continuity, energy and -cmentu= involves, in effect, soving simultaneously the equations
in a time dependent flow field.
ferential equations are of the hyperbolic type. The appropriate dlf-
Details of the procedure can be found in
many texts on compressible fluid flcw.
The earliest solution obtained at The University
of Calgary, due to Narzouk (6), was
the result of a --nua1 calcuaatio o a cold- flow simulation of a very simple configura-
p tion featuring an inlet and outlet of uniform
cross-sectional area. The solution, in the
A
--S_ - _ . . . . - - . . . . . . . . . . . . - - -f - . .
IA
form of a wave diagram, was used to check a computerized representation of the cold flow
model. The numerical technique employed for handling the forward (time) stepping procedure
waOs;,hwas
known as the Hartree backward difference technique. A description of the cold
flow model, including the boundary conditions used, has been given by Marzouk (6).
The next step, in order to advance beyond a cold-flow model, was to model the combus-
tion process Itself and combine such a model with a hot-flow method-of-characteristics
representatlon of the tailpipe and inlet flows. Attempts were made to represent the com-
bustion process using various overall reaction rate type models but these did not yield
realistic rates of pressure rise corresponding to those measured experimentally. Accord-
Ingly, it was finally d6ided to represent the combustion process by an empirioal heat-
release rate model the heat-release rate varying during the combustion process. A usable
model of this type was derived by M. Rehman (7) and later incorporated In a hot-flow type
method-of-characteristics computer programme by Cronjo (8). In Cronje's model the heat
released during the entire combustion process was made proportional to the mass of air
Inhaled Into the combustion zone during the introduction process. A typical result, due
to Cronje, for a configuration with a tapered inlet and tailpipe is shown in Fig. 2. The
pressure-time trace of Fig. 2 agrees very closely both in magnitude and duration with
experimentally obtained traces.
it can be said that further wc(rk is needed in order to model, correctly, the combus-
tion process. A successful and generally applicable model of the combustion process
should, ideally, then permit valveless pulsed combustors to be designed, and their geometry
optimized, by use of numerical procedures only. Ievertheless, at this stage it appears,
on the basis f Cronje'a theoretical work (8), that the SNECMA/lookwood configuration Js
an optimum one as attempts to evolve a configuration having a superior performance, by
(6,8).
modifying the parameters defining the geometry of the numerical model were unsuccessful
From the plot, on Fig. 2, of dimensionless velocity (local gas velocity divided by
the acoustic velocity at ambient conditions) for the inlet flow it is apparent that a
substantial backflow occurs, as indicated by the negative value of U, from the open end
of the inlet. This, of course, gives rise to a transient reverse thrust unless this back-
flow is rectified, or redirected in a rearwards direction.
3. FLOW RECTIFICATION
There are at least three possible ways of handling the backflow from the inlet of a
valveless pulsed combustor. One possibility is to use an Intake flow passage offering a
high resistance to backflow but a low resistance to Inflow. Devices of this type, usually
known as fluidic diodes, have the disadvantae of introducing strong irreverslbillties
into the flow system. Another, more attractive, possibility is to employ a rearward facing,
inlet (or 1,'lets). On the basis of flow visualisation evidence (2,9) this concept can
result in poor mixing in the combustion zone when the inlets are attached to the same
end wall of the combustion zone as the ;ailpipe. When this latte r probler is avoided a
very good performance can be obtained as evidenced by the SNECP.A Ecrevisso (Crayfish)
pulsejet unit although induction problems can arise if a suitable engine cowling is not
used (10).
The third solution is to employ a flow rectifier, a "heavy current" flutdic device,
which redirects rearwards backflow 'rom a forward facing inlet whilst offering minimum *
interference to inflow. Two types of rectifier are shown; one is illustrated in Pig. 3,
the other in Fig. 4. The earlier device, knodn as an auxiliary passagu flow rect-fier,
shown in Fig. 3, relied *ipon a combination of Coanda effect and a Sao curtain to deflect
backflow (indicated by the arrows in the upper portion of Fig. 3) inco a return bend.
During induction flow passes more or less directly into the pulsed comLustor (or pulseJ-t)
intake as indicated in the lower portion of Fig. 3. The concept works, so far as can be
established, perfectly in the sense that there Is no backflow from the air irle . Simpl-
fied single outlet asymmetric versions of this device were also develped which were
equally effective in inhibiting backflow from the air inlet.
Unfortunately the presence of a rectifier of the auxiliary p=asage type tErves to
reduce substantially the thrust producing performan.e of highly rated pulsed sombustorSl
to which it is fitted. It appears that this problem is due to the itimate aoupling,
within the rectifier, of t"e inflow and outflow pa ;ages. 4
Figure 4 shows the most successful of the rectifier concepts tried. This is, in
essence, a venturi type non-steady flow thrust augmente' of the kind pioneered by SN9CMA
and Lockwood (11) but bent around into a U shape so that tha backflow entrrins tiirfrom
the inlet into the convergent-divergent return bend (12,1J. The virtues of th~s arrange-
ment are that the return bend portion of the devtce is uncoupled f-om the pulsec 6on.bustor
inlet and the additional air entrained actually results I a small thrust augmeration "
the thrust increase more than offsets the losses due to turning the 'low 4nrough I10*."
As with the earlier device there was no detectable backflow from th: inlP1 ports -
iigure 5 shows the performances obtained by regulating the fuel inflow rate of a Z -
propane fuelled 2.875 In (73 mm) diameter SNECYA/Lockwo-cd type 1.ulsejet tested statically
with two vezsions, an early and also a more advanced form, of the auxiliary passage flow
reitifier shown in Fig. 3. Figure 5 caso shows results obtaineJ with the thrust augmenting
type rectifier of Fig. 4. The resu. for the "no rectifier" case was achieved by simply
adding the opposing thrusts, as measured using thrust plates, due to the tailpipe flow and --- -
7F
mA SZ
-rPfYX
As
- ---
u
zadkflow from the inlet. It should be noted that all the results shown In Fig. 5 were
obeined at a test size located 3600 ft (1096 is)above sea level; the test results were
not corrected to compensate for altitude.
duct or the combustion zone of pusejeti from either a single In.jector or a series of
~ressurised A system-of
fuel tank isthis type Implies
employed. that either
T1h0 letter a pressure-pump
arrangement has obvious isadvantages
available from
or a
the viewpoint of simplicity. The only other auxiliary equipmen t needed is an fgnition
system and an air Jet to initiate flow throih the combustor. Since both the Ignition
systefa and the air jet are used only at start-up they do not have, to be carried in the
flight -'ehicle.
-Figure 6 chows a relatively sophisticated alterabtive fuel system, suitable for
gasoline or kerosene type fuels, featuring a carburetor and nence not requiring a pres-
sutibet fuel tank. The system shown was actually designed for a high. intensity hand
helId space-heater (14). The foel iS pumped from the tank to the carburetor by means of a
pressure'-pulse actuated diaphragm pump. The pressure pulses are derived from the puiseJet
itself- The remaining Items are not required once the engine has started and do not,
theafoKre, have to be carried onboard the flight vehicle. A developed version of thimi
system illustrated In Fig. 6 has now been opasatonal for severa years in connection
with the space-heater application. The pulseJet unit of the srace-heater is of the single
Inlet SNECMiA/Lokwood type and has a combustion zone dIameter of 3" (76 mm).
6. FAPF0flRNNE EXPECYAT10.ISS
BasedA 'eta displayed In Figs. 7 and 8 augmented by adaltional Inform tion zttained
f6a desg - S,
e the characteristics ar e nre,;ened in Table i of a family of single
Inlet SNEI? -Ad type -lsejets equipped WI-th flow rectIf=ers of the kind !hown In
Fig. T .o
zure a-- location of the shroud duct referred to In Table 1 ire
Illustrated
in Pig. 9.
o-be 2 resents co:' espondlng. Inforation to that of Table 3 but for shor-ler four-
Inlet type puiseJets. It is notewcrthy that She a pGifto weigh, of the four-inlet type
actuVly exced hat of The single inlet vailety due to the lower specIfic thrust obtain.-
j bl. tis probable that more than four Inlets will be opt'Ia for units larger than 3
In (7h mm) combustion zone diameter. Experi.-enL.. shows that a1 Inlets Is the maxim-um prac-
tical number for the 3 'n (76 P-en)dlaA-ser rachIne. Hypothestsang with the number of in-
lets results !n the information presented In Table 3. The latter table should be inter-
j p -tdas cbpjctre so aahtt guided by experilment.
- The thrust-to-weIght ratio quoted in the three tables are ra conservatIve
- and are representative of values which should be achievable with falrx:' long life units
which may, foe example, be suitable for, say, recoverable, and reusable, target droes.
For some applications the great length of the single inlet concep- rm7 actuall, be sdv&n-- -
f-
6
A
II
TABLE I
SINGLE-INLT TYPE PSEJETS
-
TAB 2
TABELS 3
DIMENSONS
jMAX. S.F.C. AT
I THRUST: WEIGHT
RATIO (ESTIMATED) MAX THRUST MAX THRUS
f
OMBUSTOR OVERALL S.L.S. MAX NMi. - /TI
INT. DIA D ILENGTH THRUST THRUST S.'F.
& NO. OF
INLETSO WITH NO lbf/in
in NO. OF in l mb SHROUD SHROUD
2
(mr) INLETS (r) (N) (kgiN'ii) (k9/N.h) DUCT DUCT (kn/m ) (lm)
3
(76) 4(1062)
42.0 9.5
(42-3) 2.2
(0.217) 1.8
(0-184) 2.0 2.8 1.34
(9.23) .71
1(39.9)
tageous if, for example, the pulsejet tailpipe 13 also used as a tail support structure In-
a very simple vehicle. The shroud duet should be useful in minimlsing side view radia--on
from the hot section of the engine. it may also be possible to achieve a useful Jeasure
of ramjet assistance, and drag reduction, under flight conditions.
. . .. -o --- -_<-
7. COMPARISON WITH ALTERNATIVE PROPULSION PLANT
The single-inlet pulsejets of Table 1 have been compared with conventional gas-turbine
turbojet and turbofan units. Data for the gas turbines were obtained from the literature
(18). Turbojets and turbofans intended purely for lift purposes were omitted from the com-
parison although derivatives of these types intended for a thrust boosting function were
included.
Figure 10 presents plots of thrust per unit frontal area versus sea level (static)
thrust for static engine operation. It can be seen from this figure that pulsejets are
only competitive for very low thrusts which are generally applicable to unmanned vehicles
such as drones of various types. The corresponding specific fuel consumption relationship
under static conditions is presented in Fig. 11. Here p'"!sejets are not really competitive
although it will later be shown that forward flight can result in a significant reduction
in the specific fuel consumption of carefully installed pulsejets. Nevertheless, the im-
plication of Fig. 11, other considerations aside, is that if a pulsejet is selected pre-
ferentially it is only likely to be an acceptable unit for short range missions.
Figure 12 shows that the uninstalled thrust-to-weight ratio of valveless pulsejets
appears to be quite competitive at very low thrust levels. It may be possible, with a
more sophisticated, but probably more expensive, design approach than that selected by the
writer to move the maxima of curves 2 and 3 further to the right.
An important characteristic of valveless pulsejets which is not apparent from any of
three previous curves is simplicity and, as a consequence of this a potentially high re-
liability. Whilst highly rated valved pulsejets have earned a reputation for unreliability,
due primarily to short flapper valve life, valveless pulsejets appear, on the basis of
fairly extensive running experience, to be inherently reliable since, provided a pressur-
ised fuel supply is used, there are no moving parts operating at engine frequency.
The writer, and other workers, have gathered fairly convincing, but somewhat clrcum-
stantlal evidence which shows that, at the condition of the maximum fuel flow extinction
point, highly rated valveless pulsejets appear to be operating with a mean fuel:air ratio
in the combustion zone of only about 50% of the stolchlometric value. If a way could be -
found to approach, more closely, a stoichlometric mixture strength a substantIal increase a
in thrust should be obtainable for a specified frontal area and structural welght.
Another means by which an improved performance should be attainable relates to in- 4K-
creasing the normally rather minimal pre-combustion charge compression. A possible approach -=.A -
- i1-
-7-
is to employ two valveless pulsejets each arranged to supercharge the other mutually. Such
an arrangement is currently undergoing testing. Antiphase operation has been achieved
successfully, with an accompanying reduction of noise, however, mutual supercharge has not
yet been obtained. Figure 15 Is a diagrammatic illustration of the twin- 'opId
6nit to-
gether with a schematic of a simplified wave diagram showing the gas-dynamic principies
involved.
Proceeding to a yet higher degree of complexity it may, In some circumstances, be worth
advancing to a propulsion unit, also utillisng non-steady compresslble flow, based on the
dynamic pressure-exchanger concept. Here a cellular rotor, with open cnded passages, runs
between two closely mounted end plates as shown in Fig. 16 (a). By providing such a ma-
chine with suitably arranged inlet and outlet ducts a Jet propulsion type unit, with either
combustion within the cells or in an external combustor, can be contrived. if such a
machine were to be developed it would probably be of use only at low thrust levels below
the range of normal, reasonably efficient, turbo machines. A pressure-exchanger thrust
producer could almost certainly be arranged to h.ve a specific fuel consumption signifi-
cantly superior to that of a pulsejet. A low thrust per unit frontal area would prolvably
render pressure-exchanger Jet thrusters non-competitive with turbo-machinery for thrust
levels for which turbojets and turbofans are normally used. Figure 16 (b) shows, diagram-
matically, the construction of a typical pressure-exchanger. Further background material
on dynamic pressure-exchangers, explaining the principles involved, is available t19).
10. CONCLUSIONS
Enperimentally obtained static performances have been presented of representative
current technology valveless pulse~ets and the performances of these units have been com-
pared with those of alternative propulsors. it is concluded that current technology valve-
less pulsejets are best suited, in the subsonic regime, to applications where low initial
cost, simplicity and reliability are of dominant importance.
11. REFERENCES
1. Thring, M.W., editor, "Pulsating Combustion, the Collected Works of F.H. Reynst",
Pergamon Press Ltd., London, 1961, Chapter I.
2. Kentfield. J.A.C.. Rehman. M. and Marzouk, E.S., "A Simple Pressure Gain Combustor
for Gas Turbines", American Society of Mechanical Engineers, Journal of Engineering
for Power, Vol. 99, No. 2, April 1977, pp. 153-158.
3. Kentfield, :.A.C., Rehman, A. and Cronje, J., "The Performance of Pressure-GaIn COm-
bustors Without Moving Parts", American Institute of Aeronautics and Astronautics,
Journal of Energy, Vol. 4, No. 2, March-April 1980, pp. 56-63.
4. Lockwood, R.M., "Summary Report on Investigation on Miniature Valveless Pulse.ets",
Report No. ARD-307, Hiller Aircraft Company, Palo Alto, Calif., 1962.
7. Rehman, M., "A Study of a Multiple-Inlet Valveless Pulsed Combustor", Ph.D. Disserta-
tion, University of Calgary, Calgary, Alberta, Canada, 1976.
8. Cronje, J.S., "An Experimental and Theoretical Study, Including Frictional and heat
Transfer Effects, of Pulsed Pressure-Gain Combustion", Ph.D. Dissertatlon, University
of Calgary, Calgary, Alberta, Canada, 1979.
9. Kentfield, J.A.C., " he Flow Rectifier, A Fluidic Al ernative to the Pulsed Combustor
Non-Return Valve", Paper No. 18, Proceedings (D.J. Brown, Editor), First international
Symposium on Pulsating Combustion, University of Sheffield, Department of Chemical
Engineering and Fuel Technology, 1971.
10. Marchal, R. and Servanty, P., "Note sur le Dgveloprement des Pulso-Reacteurs Sans
Clapetsa, CR. ATMA, 1963, pp. 611-633.
11. Lockwood, R.M., "Interim Summary Report on Investigation of the Process of Energy
Transfer From an Intermittent Jet to Secondary Fluid in an EJector-Type 1hrust Aug-
menter", Report No. ARD-286, Hiller Aircraft Company, ?alo Alto, Calif., 1961.
A=~
A.I
12. Xentfield, J.A.C., Marzouk, E.S. and Pehn-ans M., "An Alrbreathing Pressure-Gain
Combustor Without Moving Parts", Paper No. 1317,15th International Symposium on /-
Combustion, Tokyo, Japan, Combustion Institute, 1974i (only extended abstract pub-
lished).
13. Marzouk, E.S. and Kentfield, 3.A.C., "Pressure-Gain Combustion, A Means of improv~ng
the Efficiency of Thermal Plant", Proceedings 9th Intersociety Energy Conversicn
En.gineering Conference,- (ptub1Ished by the A.S.M.E.), San Francisco, Aug. 1974, pp.
1125-1131.
1h. Kentfield, J.A.C., "A New Light-Weight Warm-Air Blower for Rapidly Pre-Heating Cold-
Soaked Equipment", Paper presented at A.S.M.E. Winter Annual Meeting, Atlanta,
Georgia, Nov. 1977, A.S.M.E. Paper No. 77-WA/HT-20.
15t Kentfield, J.A.C., "Dulsed Combustion: Proising Areas for Further Research", Paper
Presented at the Technical Meeting of the Combustion Institute, Canadian Section,
Banff, Alberta, Canada, May 1977.
16. Rchman, A., "A Theoretical and Experimental Study of' the Effect of Size on Liquid
Fuelled Carburetted Valveless Pulsed Combustors", M.Sc. Dissertation, University Of
Calgary, Calgary, Alberta, Canada, 1978.
17. Taylor, J.W.R., "June's All the World's Aircraft 197b-~75, Jane's Yearbooks, London,
p. 771.
18. Sawyer, J.W., editor, "Sawyer's Gas Turbine Engineering Handbook", Gas Turbine Pub-
cations Inc., Stamford, Connecticut, U.S.A., 1972.
19. Azoury, P.H., "An Introduction to the Dynamic Pressure Exchanger", Prooeedin&s,
institution of Mechanical Engineers, Vol_.180, Part 1, No. 18, pp. 1j51L'RO.
12. ACKNOWLEDGEMIENTS
The auth~or is grateful for firnancial support, for much of the wAork reported here,
from the Natural Sciences and Enginesring Research Council of Canada. The assistance was
in the form of Operating Grant A7928.
rig. ~~ ~
ywticFo ~ ~3 ~ 'o
eco~r
theAuilar PssgeTye
IM T
elf TWE*IM4
WUDED
'
Fig. tSyn_-etric Flow Rectifler of 0 n
the Th-rust Agetn p.o a 4 0tla
nnWIOMUtMo -SPA CXM-S=an n"
-0hre FlwCe
LO -n
r2o Is
Co~ustur inlet z
ar eser-
r-t t
0L
Tr~t
- O= - --
on fl
LI. 11
tnelll Cflsuet--
Sz-a~~r
- Leve Sarve
Coapre3_-s
Pure Jets wth azx:copea
ntt-.r~
Best t--.---,-......-.
Valveleasz pulse let -,
LO /
+eI
ZAA
t
u, nr.-W- Rl~at
--
rutt-Wta % G = ersus
TIM~r.Gas
0f
turbine enve -*=e curve
1
2-SN VaACIes 9SeJ-_ without
shrouA A-to
~ r 1aeFig.13 ~t
ce o at 7-jghtValveless
Lfluece ;=bn~e r sh rou_ d= nul-le- with
Onthe Pertor-! '
e of'a _ost
-,Ipped vi th a Flow :Recifier
t0 T ype i!nn
eV=
-1-7As
10-10
a 3.0
x 8L
0 * ..
i D,
0.48u
z- I
vxr-L
am =0 amTV..
(9a(
DISCUSSION
U.SobersE
(a) Was the design n'odel computerized?
(b) Was the model used to design the unit that was flight tested?
Authoris Reply
(a) Yms the design model, which is bastd on the method-of-char-cteristics as.-pplied to one (space) dimensional
time dependent compressible flow, was computerized. Bricf details of the procedure, with mention of some
of the restrictionS, are givent in Section 2 of the printed paper.
(b) The flight vehicle referred to in the paper (Section 8) was not due to the author. It was. in fact, a French
prototype vehicle, using a valveless pulsejet. built sorme years ago by SNECINA. This work is dec=ibed briefly~
in Referenrce 10 of the paper.
Author's Reply
The application uppermost from the feasibility iewpoint is that of the propulsion of sinrll. relatively low cost.
t;-zget drones etc. For such applications the noise problem is not likely to be regarded as serious. If the device were
to be used for active military purposes it may even be cuse-vable that the loud noise could be advantageous in
helping to frustrate. whilst overlying enemy positions. orl communications etc. Certainly any activity requirh-.;
stealth is not a profitable field of applications for pulse-jer powered -4ides.
It
0=
MULTIPLE CTAXES
FOR ?ACOCEZTS
by
E.-0. Krohn and X. Tri!sch
Deutsche Forschunga- und Versuca stalt
f.r Lut- und Raumfahrt e.V. (D-E.)
Postfach 9C 60 58
5"000 Kdln 90, Germany
SUMMARY
SYMBOLS S"JES.PTr'S
A area free-stream condition
d diameter 2 exit station of single inlets
h diverter height 3 mixing station
M Pchnumber 4 nozzle --hroat station
& mass flow p Pitot conditicn
p pressure P-.S root-mean-square value
Re Reynolds number t total condition
x distance from body vertex
a angle of attack
6 boundary layer thickness
1 ratio of specific heats .
0 angle or roll (def. Fig 2)
1. IffLWDUC':ON.
Modern air-aunched ramjet-powers-d issiles have n -nterated rocket-ra--et, for
exa=ple, the very advanced "low vol,me ranzet (ALhJ )'. -Thebooster, necessary to accelerale
the missile to the racket's mini-m operating speed, Is installed within the ramjet com-us-
tion chamber in form of a solid propellant grain. After burnout of the booster air for
ranjet operation is supplied through several (usually four) side-mounted inlets. iins
or the individual airrlows in the co-on chamber causes sudden changes in the performunce
of the inlets under certain ccnditions, especially -duringflight =aneuverm. This may be
accompanied by choked flow, reversed flow, and if;et bu-zz. The interference effects of
cultiple-inlet systems have been studied in close connetion with ra=Jet missiles uner
development by MS.
3. COUPLING OF INLETS
How does a system perform in which several inlets are connected with each other over
a common combustion chamber? As long as the chamber pressure is so low that the inlet in
the most unfavourable position still works in a supercritical mode, no problems will
arise. This condition can - assuming a constant nozzle area - be practically achieved in
two ways:
- The engine is designed for optimum inlet operation - meaning near the critical point -
at cruise flight. In this case the fuel flow has to be reduced during flight maneuvers
(liquid fuel) in brder to maintain supercritical operation in all inlets. This, on the
other hand means a loss of power just at the moment of increased thrust demand. This
procedure is therefore only applicable for missiles with moderate maneuverability
requirements.
- In the majority of cases the engine is designed for the critical maneuver case, keeping,
in general, the fuel flow rate constant (solid fuel). Consequently, the inlets are over-
sized, resulting in increased external drag and higher supercritical normal-shock losses
at cruise flight.
These considerations have raised the demand to optimize the inlets for cruising, and
to allow subcritical operation during maneuvers. Initial wind tunnel tests with a 4-inlet
system (Figure 3) revealed abrupt changes in pressure recovery and mass flow ratio in the
subcritical regime even at zero angle of attack. In addition, a hysteresis was observed
when reopening the nozzle. Analysis showed that throttling below critical operation brought
only one inlet into the subcritical regime while the other three became more supercritical.
With a total mass flow rate of 65% the summation of the individual mass flows requires
reversed flow for the subcritical inlet with a rate of 35% of its supercritical mass flow.
The consequences of throttling this inlet system into subcritical operation can be summa-
rized as:
- Breakdown of mass flow by 30% results in loss of thrust and possible overheating of the
combustion chamber
- Breakdown of pressure recovery increases the internal drag, ausing loss of thrust.
- Reversed flow in one inlet causes steep increase in external drag.
- Buzz in the reversed-flow inlet transfers into the combustion chamber and possibly causes
flame-out or mechanical destruction because of alternating overload.
For the investigation of multiple-inlet systems in the supersonic 60-cm DFVLR wind
tunnel a model has been used (Figure D), which enabled separate measurements of mass flow
ratio and pressure recovery in the individual inlets by meane of a quartered and calibrated
Venturi-nozzle, and total measurements by a hydraulically actuated choked nozzle. Asymmetric
inlet operation has already been observed with supersonic fighter aircraft having a twin-
intake system and a single engine. This has been investigated by BEKE (Ref. 2) and MARTIN
(Ref. 3) and related to missile propulsion by JELL (Ref. 4). The conditions leading to
sudden changes in inlet performance are discussed with the help of Figure 5. Two assumptions
are made:
- parallel entry of the flows into the mixing chamber. This is realized in the wind tunnel
model,and
- constant static pressure in the entry plane. This assumption is so far realistic (Ref. 3)
but errors are to be expected for strong reversed flows in single inlets.
Mass flow ratio and pressure recovery are related by the following linear expressions:
%3 Pt3 A4 1 2(c~)M= 2(~ - P+_
~ ec~l I
3 p3
P3 f(M, M3 ) for static pressure
That means: Operacing points for constant nozzle areas are possible only on straight lines
through zero. If for example "b" is the critical operation point, a single inlet could
Isystem
also operate subcritically at "c" or "d" with the same nozzle area. If the system is sup-
plied by two inlets, one operating supercritically and the other subcritically, then the
operating point "c" is only possible with a combination of points "a"land "le".
Accordingly, the following combination is obtained for a four-inlet system: three inlets
operate at point "a" and one at point "f". Theoretically, a solution with two inlets each
operating at "a" and "e" would be possible, but in pratice one inlet reaches the critical
point first-either because of fabrication tolerances or because of disturbances in the I
entering flow. In this manner, the maximum slope of t'e subcritical branches of single
inlets -n a multiple-inlet system is obtained. Exceeding this slope causes the described
unsteady performance.
N X
,, ,==
- Two methods of solution have been investigated. In the first method, all possible combi-
nations of the single inlet operating points were calculated. Then the solutions were J-7
arranged in order of the nozzle area, non-possible combinations eliminated, and one se-
lected in case of several equivalent solutions. This method gave good results only for
simple cases. With an increased number of points, the computing time and the necessary
storage capacity became too large.
In the second method, the operating points are examined stepwise on all single-inlet
performance lines. CUly those soitions are allowed which give a steady decrease of the
nozzle area A4 when stepping through the regime from supercritical to subcritical opera-
tion, or a steady increase when going the other direction. The method will be explained
by an example with two inlets. The static performance lines of the two inlets (Figure 6)
are given by five points each (a.. .e, A.. .E) connected by straight lines. Four character-
istic operating regimes can be distinguished:
- one supercritical with constant mass flow,
- one subcritical with falling pressure recovery, the slope is different for both inlets,
- one subcritical with slightly increasing pressure recovery, and
- one region witnout mass flow, representing the reversed flow regime in this simple
example.
In the diagram showing the calculated performance of the system (Figure 7) the appro-
priate operating regions of the single inlets are marked, the first number denoting inlet I
and the second inlet II. Starting with both inlets in superuritical mode, inlet TI reaches
its critical point first at "b". Since the slope of its subcritical branch is flatter than
the limit, no unsteadiness is encountered and the inlet runs through regions 2, 3 and 4
while inlet I remains supercritical. After exceeding the critical point "B" of Inlet I,
the performance line falls continuously with inlet I operating along branch "2" and inlet II
along branch "4". Reaching point "d" inlet II switches to branch "2" and inlet I to branch
"4". This event is accompanied by irregular system performance and by a sudden change in
external drag and longitudinal moments. After that the system performs smoothly again. A
reopening of the nozzle causes no unsteadiness at this point and the performance follows
a different path, so causing a hysteresis. At the end of line "31", inlet II operates in
supercritical mode and inlet I at point "c". From here inlet I "jumps" into supercritical
operation raising pressure and mass flow of the system by 30%. Irregularity can be avoided
by increasing the Machnumber at station "2", because the slope of the subcritical branch
of the static performance line is reduced by this means. This method is, in general, not
practicable because of increased mixing losses.
A second example shows the attempt to calculate the case shown first (Figure 3). The
measured performances of the four single inlets have been simplified for the calculation
by straight line segments (Figure 8). The inlets showed consideraole differences in their
critical operating point which can be attributed to high fabrication tolerances. A common
feature of all inlets was the absence of stable subcritical operation. Prter exceeding the
critical point, the inlets ran into immediate buzz, and dropped in pressure recovery and
mass flow by almost 40%. With the available missile model the performance lines could only
be measured down to "zero" mass flow. The plotted reversed flow branch has been estimated
from different test runs with large-scale inlets. in reversed flow, below a certain flow
rate, the throat acts as a choked nozzle, and thus the operating points lie on a straight
line through "zero". The lowocritical pressure recoveries are produced by model-dependent
high subsonic losses (two 90 corners,Venturi nozzle,Carnot losses). Subcritical pressure
recoveries below normal shock values are possible during buzz, since the operating mode in
a buzz cycle varies between subcritical and highly supercritical operation with accordingly
large shock losses. The result of the calculation (Figure 9) is not in complete agreement
with the test results (Figure 3), but the tendencies are iufficiencly predicted. The saw-
tooth-like breaks at mass flow ratios 0.6 and 0.3 as well as the strong hysteresis between
= 0.6 and 1.0 are in good agreement. A closer analysis of the calculated data shows
that first inlet II and then inlet III exceed thc critical point and immediately go into
reversed flow, and thus initiate the abrupt changes in performance. Inlets I and IV remain
supercritical. This result agrees with the tests.
In the last example, the angle-of-attack performance of a 4-inlet system at Ma = 2.5
is investigated. The measured performances of a single inlet mounted on the side of a
missile in various roll positions is shown in Figure 10. The inlet station was 7 diameters
behind the body apex. The inlets were nearly free of buzz and had, in almost all cases, a
steady transition from supercritical to subcritical operation. Evidently the most problem-
aSic position is that under 45 roll. An acceptable performance is obtained only up to about
4 . At higher angles pressure recovery and mass flow exhibit a rapid decrease because of the
vortex influence. In te "x"-configuration two inlets in 45 -roll position are combined
with two inlets in 135 -roll position which show a moderate angle-of attack sensitivity
of less than 10%. A comparison between the computed (Figure 11) and the meaLua:d (Figure 12)
performance of the total system shows good agreement. There is a surprisingly large region
(shaded area) with at least one inlet with reversed flow. The c.lculation terminated here.
since no data with reversed flow were available for the single inlets. A comparison with
the single inlet performances shows clearly that the pressure recovery of the total system
is determinded by the "worst" inlet. To improve the performance of this system the entering
flow field had to be changed for the upper inlets. This would be possible by shifting the
inlet plane forward, by influencing the vortex position with strakes, or by positioning the 04
inlets between wings. In the "+"-configuration one inlet each in top and bottom position
are combined with two side intakes (Figure 18). The performance data of the top inlet varies
I within a bandwidth of less than 10% up to 12 angle-of-attack, because the vortices by their
rotation maintain a relatively good flow. The bottom inlet exhibits a considerable increase
in pressure recovery and mass flow ratio. The side inlets show a strong angle-of-attack
Sensitivity because of the increased local flow angularity, and they are therefore decisive
j- 3-.
A.
12.4
for the system performance. The calculated (Figure 13) and measured (Figure 14 perfor-
mances are again in good agreement. In spite of the fact that there is reversed flow
(through the side intakes) below 50% flow rate, the performance decrease at angle-of-attack
is, however, considerably less than for the "x"-configuration. In addition the "+"-config-
uration has the potential for substantial improvements of the performance by use of inlets
less sensitive to flow anf~.lsrity.
5. CONCLUDING REMARKS
Based on the results of this study, these conclusions zan be made:
- The calculation of the performance of a multiple-inlet system from single-inlet data is
in good agreement with measurements. Since the inlet in the most critical position governs
the system performance, only its data need to be known exactly.
- Gubcritical operation can be permitted to a certain extent when the inlets are free of
buzz, which can be obtained by boundary layer bleed. Rapid changes in external drag and
longitudinal moments must be taken into consideration.
- Inlet systems in "x"-configuration can operate properly only at small angles of attack
up to the point at which vortex flow enters the upper inlets.
- Systems with "+"-configuration of inlets have a much better performance and can operate
up to high angles of attack, if side intakes with low flow-angle sensitivity are used.
II
REFERENCES:
1. THIESCH, K. Tnfluence of the Plowfield ArGind an inclisied 0give-cylinder
Body upon Supersonic Aft-intakes, in:"Contributions to
Gasdynamics and Aerodynamics", pp. 61-78
ESA-TT 518 (1979)
2. BEKE, A. Criteria for initial Flow Reversal in Symetrical Twin-Intak
Air-induction Systems operating at Supersonic Speeds,
NACA RM E55LO02a (1955)
I7t
M5
I4
\ N4
.x/dI8
x/d =12-
Fi.pto-
rsr
dVto ruda nlndbd
2.0, Re i8.61ij
.9
.8
0 45 90 135 180
W4f 0 -
-0
I
-2 .31~
' '
ASS___ ATI
-i
fu-nrsse
F. 3 mesue pefraneo
2.5,*-*-=-
00 -
A A.
.
.0 .1 .2 .3 4 . .6 .7 .8 .9 1.0 1.1
=~S-FLOW PA 10 -
Fig. 5
* ~.0 I
IN4E FF1
d 2
.-
Fig. 7 Otwi-nlet
Performanceof sysem _ -MSFtM~ ~
11C
.2~
1 1 ~
* *I-
WlSS-LO P~
ATIO -
P g- 9 Caculaed tour
erfor-i-ce of nl~e t s uplr-
12.10
av. 1
0 .1 .2 . 6 7 . 9 - .
i. 0 Pefrac .- of n Thyma~e
2-.4/ N.
aIo
-i Fft
S
w 77 1 1
Fig.
12 Cax>jz
DI_
1
C ______
ca EQ 1
~ i. ~ a u mn. e ~ - n . u - r m
= - C
- - - - -L
vs,
_ _
_
01
a11
-11 0
M- I
.0. -1 3 . 5 . 7 . 9 . .
I U41 =TI -
za- a
ftg-2 ofa*wiltss-f 14
121
12.3
41~ - .7 - 9 16
I 7F
'I'll''
12.14
DISCUSSION
E.Sttffker
What was the basic difference between the intake system demonstrated
first with the generally negative pressure
recovery slope below the critical operation point into the subcritical region
and the improved intake system?
Author's Reply
The intakes used in the first 4-inlet configuration (Figures 3 and 8) had
no boundary layer bleed, whereas the intakes
with improved performance in the later example did have bleed. The effect
of bleed upon the inlet performance is
shown in Figure 15.
t2
5 1
- - ~
c~m-
One-,-imensional Non Linear Considerations on Supersonic Diffuser Buzz /3-
Hermann-L. Weinreich
MBB Raumfahrt, Postfach 80 11 69, 8000 MUnchen 80
SUMMARY
This paper will deal with the fundamental aspects of supersonic intake buzz. A description of the basis
phenomenon characteristics and a delimitation to other unsteady propulsion processes will be presented.
A simplified intake propulsion stability analysis shows the possible influence of diffuser exit Mach number
on the stabi ' Iboundary in comparison to other well-known theories. Non linear examplc calculations using
smooth intakc !-essure recovery curves can illustrate some experimentally ubserved buzz characteristics.
The limits of ie theoretical approach will be discussed,and the tendencies will be compa:red with exper-
imental work.
LIST OF SYMBOLS
1. EINLEITUNG
The first experimnntal investigations on multi-shock intakes for supersonic ram propulsion were performed
by Ludwieg and Oswatit~chf 1 2. They used rotational intake models, where one or two oblique shocks were
generated upstream the terminal shock with the aid of a centre body. The terminal shock stands in front of
the intake cowl during subcritical operation. The capture mass flow therefore depends on Lhe back p.-essure
generated by the consumer (Fig. 1). During supercritical operation, the area of the swallowed stream tube
does not depend on back pressure; in this case, the terminal shock stands within the intake diffuser.
During the tests of the multi-shock intake modes, Oswatltsch found during supercritical operation an
increase of the total pressure p. exceeding the Pitot pressure with a closing throttle (Fig. 2). During
subcritical operation, a sudden unexpected pressure breakdow, in connection with a loud buzzing noise,
was detected [ 1 .7.In spite of steady incident flow, periodic movements of the shock system, and
associated pressure and mass flow pulsations, could be observed in nearly the entire subcritical regime
[ 1 .7.The phenomenon was called buzz, due to the characteristic noise.
2. REMARKS ON THE EXPERIMENTAL EVIDENCE
The following statements on intake buzz are based on published papers, which have been collected during
our own theoretical work on the problem [61, 637.
Even in the case of unJisturbed axial supersonic indicent flow, buzz has been observed during subcritlcao
operation in many ty, s of intakes:
o Pitot intake [ 3, 4, 5, 52 1
o external compression intake
- axisymmetric intake [ 1, 12 - 23, 28, 29, 32 - 34, 37 - 41, 46 - 48]
- half axisymmetric intake [59, quarter axisymmetric intake [ E2
- rectangular (two dimensional) intake [26, 27, 42 - 45, 53 - 3..,57)7
- inverted rectangular intake [58); external ramp intake [51.7; scoop intake [8]
o mixed compression intake [ 24, 30, 50.7
o internal compression intake [ 6 M
The operation
In this case, with
buzz the
couldangle of attack,
possibly alreadyoroccur
with free stream distortions,
at subsonic incident flowmay change the
conditions buzz An
[ 51.7. boundaries.
increasei
in stable subcritical operation range, or a damping of buzz pressure pulsations, can be caused by shortening
the tube between intake and nozzle [ 14, 15., or by the use of a turbo engine instcad of a nozzle.
R
.. .
Among the many observations conducted on the intake buzz, the descriptions given by Trimpi [ 14, 15 ],
Dailey [18] and Stewart [38, 39, 40 ] show the greatest accuracy.
3.1 Simplified description of a buzz cycle using experimental observations 15 ] and theoretical consider-
ations ( 61 - 63 .7
Start of cycle:,
The terminal shock, stnnding in front of the intake cowl, will be forced upstream by upstream running
compression waves. The associated mass flow reduction is propagated in form of a pressure lowering rare-
faction wave into the intake. The causes for the movement of the shock system, the decrease in massflow,
and the internal pressure breakdown may be as follows:
the entrance of a vortex sheet (with a different total pressure on both sides) emanating from the shock
intersection,under the cowl lip, called Ferri instability, was observed many times in connection with
buzz [ 13 - 15, 25, 39 - 44, 45 - 48, 60].
- a shock induced boundary layer separation on the compression surface, which was termed Dailex instabil-
ity, [ 14 - 20, 33 - 41, 46 - 48, 60 ].
- a local total pressure dip inside the free stream captui. tube, increasing relatively with throttle
ratio [ 51, 60].
- a certain declin. of intake total pressure, with increasing throttle ratio (several general stab'l:ty
theories give different limits for the corresponding slope of the intake characteristics) [ 2, 7, 6,
21, 22, 35, 61].
The upstream shock movement continues until either, for relatively short intakes, a new highly subcritical
quasisteady equilibriumstate (producing eventual backflow) will be Fced, or for longer intakes, the
shock will collide, during the upstream movement, with rarefaction waves returning TFrbii-Thenozzle, and
generating a reduction in back pressure, in order that the shock can return into the intake.
The exhaust phase will be terminated even in the case of relatively short intakes, when the 3bove-mentioned
rarefaction wave, which starts the exhaust process, will return in the intake after reflection on the
consumer. Supercritical pressure rise in the backfilling phase:
Due to the reduced back pressure, the terminal shock returns and enters again the intake diffuser (super-
critical operation). The associated rise in mass flow propagates in form of a compression wave downstream
to the consumer. There the wave will be reflected back to the intake, whereupon the terminal shock will be
forced upstream. If the shock reaches now the instable position, a new buzz cycle will again be initiated.
In such a case, we have a period, which is longer than the first mode period of a one sided closed organ
pipe of equivalent length, only by the reaction time of the inlet and consumer.
;R} W
Three other unsteady propulsion phenomena
mentioned. which partly come up in connection with buzz
will only be briefly
a. Sometimes a weak pressure pulsation connected
with small motions of the terminal shock can
during subcritical phase of the buzz cycle be observed
This phenomenon could be interpreted as an [ 15, 18, 40]1, or before buzz is initiated [a4, 59]7.
interaction jetween '.he befurrated part of
shock and a local separation bubble [ 15.7, the terminal
or as an interaction between terminal shock
and extarnal
sonic line.
b. Unsteady total pressure profiles at the
intake operation. The unstable interactionintake exitplane are also observed for highly svprc
of terminal shock and diffuser boundary layer :al
app.. 75'be
the cause for the distortions.
c. During the cooperation between the supersonic
intake and the turbo engine, it is also possible,
buzz can provoke compressor stall [647, tnat
and compressor stall can cause buzz.
4. UNSOLVED PROBLEMS AND A POSSIBILITY OF
A NEW THEORETICAL APPROACH
During earlier work [ 61, 637, fifteen different
theoretical explanations and descriptions
have been discussed. These can be divided of intake buz.
into two main groups:,
o Theories which assume that the intake
alone
The theories presented by Ferri, Nucci [13 determines the stability behaviour:
7 and Dailey [187 belong to this group.
for the subcriticai drop in efficiency presented While the causes
investigations, the explanation, that an in these papers have been confirmed Dy many
other
due to vortex sheet ingestion [ 13]7, or intake buzz cycle is initiated by choking of the intake throat,
boundary layer separation [ 18), has not
been readily accepted
[15 - 18, 25, 61, 63].
Some other stability theories:
- are correct only in cases, where the shock
system breaks down at the stability boundary
reaction by the throttle (Pearce without
* seem to be oversimplified and not [35 ], Trimpi [ 15.7),
generally valid (Oswatitsch and Teipel [2],
Hall [51,
Billig [65 J)
- are valid oniy for special cases of Ferri instability (Fisher,
- remain unexplained (Beastall [ 49]). Neale, Brooks [41])
o Theories which assume that the stability behav'our is determined
by interactions between intake and
throttle:
- For relatively short ntakes, the theory
by Mirels [ 7 7 appears to be accurate.
- The famous Helmholtz-Resonator theory worked
out by Sterbentz, Eward and Davids [21, 22
the existence of internal pressure waves 7 neglects
and was therefore not accepted [ 15, 61 -
63.]). A similar
-
theory was given by Fraiser [8].
The theory presented by Leynaert [ 36] could be useful in a case
slow acting intake and a short fast acting of a cooperation between a long
- The initial theory worked out by consumer.
Stoolman [ 3. was later completed by Kowalewicz
Hsu [9, 10]7, Endwards [ 11 ] and extended [ 5], Chang and
by Hagashima et al [ 12 ]. The primary critisism
this theory is concerning the connection of
the velocity of the terminal shock movementbetween disturbances (4 ,aM) in the entrance area
[ 61, 63]1. According U [7, 61, 63], with
ances depend primarily on the location of the disturb-
the 'arminal shock - the shock velocity can
additional effects of the order (diameter cause only
- / L).
The different theories given by Hermann ["24
] and Fisher [ 42 ] exhibit experimental
or theoretical
contradictions.
The study of the existing literature shows the following
unsolved questions:
Q1 Is it possible to describe the different
buzz phenomena (Ferri, Dailey, Buffet [39
due to total pressure dip inside the capture
stream tube) with the help of only one -simplified
41 7. instability
global
theory?
Q2 Will buzz be caused by non-linear intake behaviour
alone, or arrise from the growing of weak inter-
actions between intake and consumer?
Thereforethe theoretical approach should have the
following aims:
o An attempt should be made to include the different
observed phenomena within a unified theory, which
should be physically correct at least for a theoretical
o Within the scope of question Q2, it should limiting case.
be examined to what extent the consumer has an influence
- the stability boundary, and on
- the pulsation cycle.
5. FORMATION OF A SIMPLIFIED THEORETICAL MODEL
The experience shows that buzz begins when the
limit by the connected consumer, which intake mass flow is t-irottled below a certain stability
Using a turbo engine instead of a tube can be a tube with throttle, a ram
with throttle nozzle, an increased combustor,or a turbo engine.
damping, can result [28, 29, 50.7. A ram combustor stability, due to additional
can
intake nozzle interaction, demending on fuel properties, produce additional damping or excitation for the
combustor geometry, or combustion behaviour.
- Simplification 1:
The first step in the investigation of the intake
consumer stability behaviour should therefore be
to interactions between irake and nozzle alone. limited
In spite of simplification 1, the stability behaviour
by Fig. 5) will be determined by at least 24 differentof an appropriate intake nozzle module (illustrated
parameters [ 617. Most of these influential para-
- T
.- #
_----z-
I
meters describe the geometry of the intake. Therefore a general calculation of the unsteady intake flow,
in regard to
- the shock boundary interactions (including separation effects),
and vortex sheet and shear layers
without the use of doubtful approximations, seems hardly possible. On the other hand, the principle
features of intake buzz are maintained with lncreasingduct length between intake and nozzle. This was
demonstrated experimentally by Trimpi £ 14]7. Therefore it seems to be apparent, that it is necessary
to simplify the theoretical model in a radical manner, with the aid of
The pressure amplitude of the strongest internal waves during strong buzzing does not exceed 30 percent.
Therefore the internal wave processes can be approximately treated as isentropic:
Simplification 4:
The increase of entropy due to internal shock waves, which may be generated by converging compression waves
is proportional to the third power of pressure amplitude and may be neglected.
In contrast, the entropy of the quasisteady intake exit air varies with intake exit total pressure, and that
changes of entropy should be taken into account. Any change of intake total pressure will therefore, q
according to Fig. 7, generate an entropy discontinuity with separate regions of different entropy in the
internal flow. The entropy discontinuities are special path lines and represent perturbations of density,
temperature, Mach number,ad ina constant pressure region. Under the previously mentioned simplifications
2 and 3, the spatial extend of the simple compression waves generated by intake - or nozzle reactions are
negligible small compared with L. Furthermore, the diverging rarefaction waves should also be depicted in
a summation manner in the x,t plane according to Fig. 8.
Under the previously mentioned "simplification 4" (neglection of entropy increase due to the steepening of
internal compression waves), all state variables in the constant state regions of the eshed network can be
calculated step by step using the bimple wave theory. Using the state characteristics for unsteady one
dimentional flow [ 66 - 69_], all state changes, due to simple pressure waves,or entropy discontinuities
respectively, can be calculated in an easy manner.
With the abbreviations (4) " P T.
LI?
and using the state characteristics, in conjunction with the gasdynamic relations, the following relations
for simple pressure waves may be obtained:
V57_
6 -4--
I-
xf7 Z7
9;r77r , r I.
f'._.7M -y.;
x' --M 7r 7 i)/-
or as a function of mass flow perturbating respectively:
_;
Sign convention-1
The upper sign always refers to changes of state due to upstream traveling waves, while the lower sign
denotes changes due to waves traveling downstream.
The differential equations (5)can be integrated in closed form using the static pressure as an independent
variable. This was one reason why the unusual po, p plane was used in earlier work [61 - 637, whereas
in this study a graphical presentation of the internal changes In state will be given in the po, A plane.
Using j - 9 to denote the intensity of a weak entropy discontinuity, the following relations can be ob-
tained: -1 X
-2 7 (7)
With the aid of eqs. (5)- (7), all changes in state from one constant state region to the neighbouring
can be calculated if the intensity and the type of the enclosed wave is known.
7. SOLUTION OF BASIC PROBLEMS
For the calculation of the state variables inside the meshed network, the magnitude of the individual waves
is required. Therefore the changes of wave magnitude, due to fundamental unit processes, were determined
using eqs.(5) - (7)with the respective boundary conditions:
o The intersection of simple pressure waves will not lead to a change of pressure amplitude within the
scope of 3 theory of small amplitude.
o The intensity of an entropy discontinuity will not be changed by the intersection with a simple
pressure wave. The amplitude of the incident pressure wave remains constant within the scope of an
linear perturbation theory. The amplitude of the reflected pressure wave Is proportional to the pro-
d&tof the amplitude of both incident waves and can be neglected withing the scope of a linear per-
turbat.on theory.
0 In the case of a subsonic channel exit (characterized by M and MN ,see Fig. 9), the exit boundary
condition for simple wave reflection is the conservation of the ambient pressure P,,at the exit. For
small perturbetions one can find [ 61,62]:
/1" [(1-rVIY
Z h 4 H -r. ( -n)t,
II,,,(Ii.n)tU -
VfV/(B (8)
- the open end with MN = M: The resulting mass flow amplitude ratio (I-M)/(I+M) means, that the
pressure pertubation vanishes: rjj , -?tz
- the critical (or supercritical) nozzle (N = ):
compared with the closed end, the loss inamplitude
According to eq. (8), this result is valid for nozzle and diffuser type exits respectively. While an
entropy
includeddiscontinuity
in eq. (10): (special path line) can never strike a closed end, the following special cases are
- the change of gas density at constant pressure can pass an open exit with M MMN, without causing
any wave reflection.
j - the critical (or supercritical) nozzle will result in a pressure wave reflection of less than half
the mass flow distortion :
-~ .1 -l
W_.
/=
o In the case of simple wave reflection on an intake exit (Fig. 10), two boundary conditions had to be
satisfied: Firstly, the pressure and mass flow perturbations of incident and reflected waves must be
compatible with the intake characteristic. Secondly, the total pressure change due to the waves had
to be in accordance with the total pressure recovery characteristic. If the intake characteristic
curve is continuous in the considered operating point, the slope of the total pressure characteristic
is given by C,
and the amplitude of the reflected waves can be obtained,after some manipulations,
3 , _ g r, (13)
,r17
The entropy discontirtiv 43 vanishes in the case of constant total pressure recovery (0= 0),
while the mass flow artpji.ude of the reflected wave 9132equals 5#21. In this case, the intake is very
similar to the op.n end vf an organ pipe.
The linear intake response, defined by eqs. (12) and (13), can also be obtained graphically by using the
state characteristics (5) - (7). Therefore the unknown strength of the pressure wave 32 and the entropy
discontinuity 43 had to be iteratively brought in agreement with the intake characteristic in the two
different state planes.
Using eq (11), a semigraphical construction of the intake response in only one state plane (according to
Fig. 11) is possible:
Starting from the indisturbed intake operating point 1, the mass flow may be reduced by*?2 1 by means of a
small compression wave. Depending on the slope of the intake characteristic G, the intake will respond
in a different manner:
- In the case of supercritical operation without bleed ( = - com), the incident mass flow reduction had
to be compensated by the refelcted waves.
- In the case ofIg= 0, no entropy discontinuity will be generated. The reflected pressure wave will double
the incident mass flow reduction.
- The magnitude of the reflected pressure wave will increase with growing U until an unlimited reaction
will result within the scipe of a linear perturbation theory, when the slope reaches the value of
The value 6" denotes the greatest possible slope of the intake characteristic, because at this operation
point, the intke would react strongly, even by small perturbations *ithout feed back from the consumer.
9. WEAK INTERACTIONS BETWEEN INTAKE AND SUPERCRITICAL NOZZLE
If friction between intake and nozzle can be neglected, then the intersection point of intake and nozzle
characteristics represents the respective operating point of the intake. The question, whether such an
operation point can represent stable interaction between intake and nozzle, can only be answered by the
perturbation theory when the intake characteristic isdifferentiable inthe neighbourhood of the operating
point. This is not self-evident, because the boundary layer separation or vortex sheet ingestion would
normally produce a discontinuity in the characteristic curve.
Under the assumption of supercritical nozzle pressure ratio, the nozzle characteristic will be a line with
a slope of one in the V. plane. First of all, it should be examined whether stable interaction of intake
and nozzle ceases at the top of the total pressure curve, as mentioned in [ 2]. Within the scope of the
perturbation theory, the corresponding total pressure characteristic appears in form of a line of constant
total pressure intheleft side of Fig. 12. Assuming stable operation at time to, the nozzle discharge may
be immediately reduced by921. The resulting shift of the nozzle characteristic will cause a small compression
wave to travel upstream. The total pressure rise 7o2, coupled with the mass flow reduction Y2, depends on
the internal flow Mach number in accordance with eq. 6.
The compression wave T1 will lead to an excessive total pressure at the intake exit. Therefore the intake
will respond with a rarefaction wave 3, decreasing the mass flow, and pressure, down to the new intake
operation point 3. No entropy discontinuitywill be generated because the total intake pressure recovery
did not change within the intake reaction (3'2 3). The mass flow 3 is now below the required nozzle through-
flow. Consequently, the nozzle will react on the perturbation 32, and accelerate the gas with the aid of
a rarefaction wave 0'. Therefore, the intake will respond with an attenuated compression wave 54.
The pressure wave decays while oscillating between intake and nozzle and illustrates stable Interaction of
intake and nozzle at the peak of total pressure recovery. On the right side of Fig. 12, the analogous inter-
ference is treated for the case of positive slope of the intake characteristic. To obtain a clear picture of
the main process, the pths and consecutive nozzle reactions of all the entropy discontinuity generated by
intake reactions have been neglected. Within the scope of that simplification, a reduced attenuation of the
wave cycle can be seen. T
-_ . .. ..-.
51 E ;-
i2
Looking for the boundary of stable interaction (using eqs. (6). (7), (9), (11), (12), (13)). it can be seen
that in general, an undefinite number of weak secondary pressure waves are generated at the nozzle by entropy
discontinuity reactions [ 61, 62]. For an undetermined number of special cases, it can be shown however 13'7
[61, 62] that all the entrcpy discontfnities have a damping orexcitingeffect on the formation of the
main cycle. Fig. 13 shows the two most simple cases.
After some manipulations, the following boundary, for stable interaction of intake and supercritical nozzle,
is obtained [ 61, 623:
g J7 da~ping effect on
If~ (16)
I 1"3t,-1-JP- - ]
rlzJ{-qflJ -I exiting) entrop. disc.
The value of q depends on the mean effect of all entropy discontinuities on the formation of the main
cycle.
10. COMPARISON OF DIFFERENT STABILITY THEORIES
Fig. 14 shows the influence of the mean internal flow Mach number H on
- the limit of absolute intake instability slope 0 Al
- the limit of intake nozzle interaction instability slope 0': for the critical cases of maximum
attenuation q = 1, indifferent effectioness, and maximum ei(tation by the entropy discontinuities.
Any change from the above assumptions (differentiable intake characteristic, length of intae and nozzle
negligible compared to total length L, end for interaction stability boundary'a supercritical nozzle
pressure ratio) car, result in quite different stability boundaries:
+ Increasing the relative intake length, will cause a growing zone of wave dispersion at the intake and
result in additional attenuation of the main cycle [ 61, 61 .7. Therefore, even iith an intake greater
than OAI, the intake flow may be stable.
+ The boundary of stable interaction can be changed if the consumer characteristic does not agree with
the assumed super:ritical nozzle.
+ A discontinuity within the intake characteristic, which immediately moves the slope from stable to
unstable values, may not be observed because pulsations can start just at the discontinuity. Therefore,
a measured slope of the intake characteristic at the origin of iistabilitydces not reed to coincidewith the
interaction instability limit XIS.
The preceding stability statements can be compared with other well known theories under the above-mentioned
simpllfications (1) and (2):
o The result obtained by Mirels [ 7] is equivalent to eq. (6) under the additional assumption of small Pach
number M, the entropy discontinuities being able to be neglected (q - 0) / 63 1.
o The statement of Pearce [357 does not take the interactions with the consumer into account. The result
would be comparable with eq (15).
The widelyused Helmholtz resonator theory, proposed by Sterbentz, Eward, DaviJs [21 , 22_, has been
criticized extensively [ 14, 15.7. The resulting instability slope limit is proportional to the relative
length of the intake. In the case of a negligible relative intake length, the theory would state a stabil-
ity limit of O - 0, which was also st~ted by Oswatitsch and Teipel [ 2]. The Helmholtz resonator theory
can not explain why sometimes(for higher internal Mach numbers ar.1relatively long intakes) stable intake
characteristics with large slopes can be observed.
The methods for the evaluation of small interactions between intake and nozzle, which was presented in
chap. 7 to 10, car. easily he extended for the investigation of strong nonlinear interaction processes. It
can be seen that the intake characteristic is the source of principal nonlinearity. Using simplification 4
(chap. 4), the change in state, due to arbitrary simple wave, can be depicted in the pc, p plane in closed
form [61, 62],,or can be integrated step by step in the po,m plane with respect to the parameter M.
In the case of a semigraphical construction (Fig. 15) of the intake response, the relation between the
mass flow ratio of the entropy discontinuity, and the intensity of both pressure waves, becLes, according lip
to eq. (14), r7l -
t.(423 p~ (17)
Using eq. (17), the effect caused by the entropy discontinuity can be outlined In relation to the intensity
of the reflected pressure wave R. The example, illustrated in Fig. 15, shows the nonlinear intake response
due to a mass flow reducing compression wave after a suO'n reduction of the nozzle exit. The change in mass
flow and total pressure caused by the entropy discontinuity is very small.compared to the intensity of the
pressure waves. Therefore, the investigation of the principle nonlinear interactions between inta;e and nozzle
can be simplified by the omission of all entropy discontinuities. Such simplifications of typical nonlinear
interactions are presented in the following figures:
Fig. 16 shows an isolated buzz cycle as the result of a sudden nozzle discharge throttling up to a value
near to stability boundary. The strong rarefaction wave coming from the intake will he reflected 1rom
the nozzle, Due to the reduced back pressure, the terminal shock will return from the position in front
4 of the cowl, and try to reach a supercritical state of equilibrium. The respective operating point 5
was attained by a simple downstream running pressure wave (and an associated entropy discontinuity)
j
starting from the disturbed operating condition 4.
Within the scope of the simplifications I and 2, it is not necessary to pay particular attention to the ->
St
mm
different possible intermediate states during the unsteady change of intake operation. In thesequence, the
j3 . intensity of the internal pressure waves running back and forth is attenuated because the new inter-
section of inlet and nozzle characteristic still represent a stable operation.
Fig. 17 shows a series of simple limiting cycles for various throttle ratings. They have a period correspond-
ing to an organ pipe and are typical for intakes with a wide stable subcritical operating range. With
closing throttle the total pressure amplitude grows, starting from a considerable high initial level,and
goes through a maximum, the pulsation dies away when a second stable subcritical erating
range is reached. In the illustrated example, the intake enters the reversed flow regime. The in-
creasing intake total pressure with reversed flow has been observed experimentally [ 59 J and corres-
pondends with the transition of a nozzle characteristic.
Fig. 18 shows limiting cycles for an intake with a small stable subcritical operating range and a large
amount of subsonic diffusion (small value of M). A throttle ratio just below the stability limit will
produce a staged supercritical pressure rise and an appropriate elongated periodic cycle. With closing
throttle, the number of supercritlcal interactions during the fill-up phase of the cycle is reduced
since the amplitude of the pressure waves traveling down and back grow progressively. Finally, a
limiting cycle with an organ pipe oscillation period will be reached.
An additional consideration of the effects caused by entropy discontinuities and internal wall friction can
lead to the following results [61, 62 ):
o Due to the impingement of enZropy discontinuities on the nozzle, a large number of weak secondary
pressur'= waves will be generated. If such a secondary wave, during buzz cycle, strikes the inlet,
operating just near the stability limit, it can provoce the next cycle. Therefore, the cycle period can
vary slightly. Even without regard of the Intake and nozzle delay, the period does not need to be a
multiple of half the organ pipe period.
o P :ocal internal friction zone causes attenuation and stabilizing effects wnen it is concentrated at the
int%-: exit. Total pressure losses, due to Instrumentation at the intake exit, can therefore produce a
too optimistic stability behaviour.
12. LIMITATIONS OF THE THEORETICAL MODEL ANDCOMPARISON OF THEORETICAL TRENDS WITH EXPERIIMNIAL WORK
The principal simplification of the theoretical model, L <<' L indicates a significant limitation:
Calculations of the transient behaviour, made for an intake with a long throat [ 61 j.showed that many
interactions between the external shock system, and the internal diffuser can be required to obtain a new
operating point after distortion. Remembering the high Mach number level inside the throat region and
lheresulting low propagation speed for the upstream traveling waves, a long reaction time is produced.
Even in the case of - very long internal duct, a stabilizing and attenuating effect, caused by wave
dispersion at the intake exit, can be expected. In comparison to the demonstrated example calculations,
a smaller oscillation amplitude may result.
In contrast to the simple theoretical model discussed, an intake of medium relative length will not
reach a quasistationary operating point during a buzz cycle.
In the case of similar length-of-intake and connected consumer, the dynamic behaviourof the intake may
be slow compared with the consumer. In this case, a totally different theory has to take the dynamic
behaviour of the intake flow into account while, in accordance with [ 36.7, a quasistationary consumer
characteristic may be assumed.
In spite of the strong limitations, resulting from the principal assumptions, the presented siEple model
can explain in closed form different observed features of diffuser buzz using the associated intake charact-
eristics:
The final question, as to how far the consumer influences the onset of buzz pulsations, may be answered
by the following statements: I - I
L~z~ L
~U
If the intake characteristic is differentiable in the neighbourhood of the stability boundary, then the
consumer will have an influence on the stability behaviour. The stability boundary will be situated at
Such a throttle ratio.,where exciting by the intake is compensated by the damping effect of the constmer. 13
Whenthe stability boundary is crossed, a rapid buildup process of buzz pulsations may result from a Warked
curvature of the intake characteristic. So it may be possible that the stability boundar" is determined by
small linear intake consumer interactions, even in the case vhere the first discernible cycle shows the
entire buzz pressure amlitude.
REFERENCES
17 K. Oswatitsch, Der Druckwiedergewinn bei CGeschossen mit R'jckstoantrieb bei hohen Oberschallge-
schwindigkeiten, Forsch. u. Entwickl. des Heereswaffenamtes, Bericht-Ur. 1005 (1944), Nachdrtuck:
DVL-Bericht-Mr. 49 1957/MACA TN 1140 1947
C-2]1 K. Osatitsch, I Teipel, Die Pulsationen von Sto8diffusoren, ZAIMPVol, IXb. 1958, S. 462-477
(r4] J. Nicolas, Etude exp~rinentale des coulements pulsatolres dans une prise d'air supersonique,
La Recherche Aeronautique No. 58, S. 23-28
r5. A. Iolewicz, Stability analysis of a normal-shock diffusor, Archivn Plechaniki Stosowanej 5,16,1964
f8,7 H.R. raiser, Supersonic inlet dynamics, J. of Aero/Space Sci. Vol. 27, S. 429-436, June 1960
L9 1 C.C. Chang. C.T. Hsu. Solution to 5toolmans external diffusion equation for instability of a normal
shock inlet diffuser, Jet Propulsion Vol 28, No. 7, S. 457-460, July 1958
Z'10_7 C.C. Cang. C.T. Hsu, Aerodynamic instability of supersonic inlet diffusers A.R.S. Journal, Vnl. 30,
No. 6, S. 468-475, May 1960
i-11.7 R.H. Edwards, Solution to the linearized quasi-stationary external diffuser equations, A.R.$.
Journal Vol 29 No. I S. 54, January 1959
f127 T. Nagashima, T. Obokata, T. Asanu=a, Experiment of supersonic air intake buzz, Institute of Space
and Aero. Sci., University of Tokyo, Report No. 481, May 1972
17137 A. Ferri, L.M. Nucci, The origin of aerodyamic instability of supersonic inlets at subcritical
conditions, MACA IN LSD K30, 1951
[14_7 R.L. Trimpi, An analysis of buzzing in supersonic ram jets by a modified one-dirensional nonstatlon-
ary wave theory, MACA TM 3695, July 1956
15] 1 R.L. Trimpi, A theory for stability and buzz pulsatior. a4plitude in ramjets and an experimental
investigation including scale effects, NACA Report 1265, 1956
[-16-7 R.L. Trimpi, Comments on supersonic diffuser instability, J. Aero.Sci. Vol. 23 No. 6 S. 611-612, 6.56
L 17_7 R. L. Tri=pi, Reply to further comments on supersonic diffuser instability, J.Aero.Sci. Vol. 24
1No.11 S. 845-846, Nov.1957
r18_ C.L. Dailey, Supersonic diffuter instability, J.Aero.Sci. Vol. 22, No. 11, S. 733-749, Nov. 1955
(-197 C.L. Dailey, Further comments on supersonic diffuser instability, J. Aero.Sci. Vol. 24, No. 1,
S. 70-71, Jan 1957
1(21.7 1.H. Sterbentz, J.C. Evvard, Criterions for prediction and control of ramjet flow pulsations
N.ACATI 3506, Aug. 1055
f223 12.1. Sterbentz, J. Davids, aplitude of supersonic diffuser flow pulsations, ACATM3572, Oct. 1955
!231 J. F. Conno-s, Effect on ramjet pressure pulsations en supersonic diffuser performance
MACA 101
E50 H122,
Nov. 1950
£24] R. Herm-nn, Supersonic inlet diffusers and introduction to internal aerodynamics, Minneapolis-
Honeywell egulator Co. 1956
f 25 ] D. D. Wyatt, A review of supersonic air intake problems, AGARDograph No. 27, Pergamon Press S.21-47; '58
( 26. L.2. Gertsa. N.A. Beheim, Performance at Mach rJ=!ers 3.07, 1.89 and 0 of inlets designed for
inlet engine matching up to Mach 3, MACA RM ES 813, may 1953
- R=[ -.
-A.
f 27] .7.- Beheim, L.W. Gertsma, Performance of variable two-dimensional inlet designed for engine inlet
1340D matching. I-Performance at design Mach nw ;ber of 3.07, MCA RM E56 1123, Nov. i956
L[28]J N.A. Beheim, G.U. EnalertEffects of a J-34 turbojet engine on supersonic diffuser performance,
MACA RM £55 12s, Jan. 1956
.29H.T. Mslal, Comparison of effect of a turbojet engine and thre cold flow configurations on the
stability of a full-scale supersonic inlet, KACP, RM1£56 K23, Jan 1957
['30-7 D.!. Motycka, J.8. Murphy.Experimental investigation of inlet-to-inlet shock interference, SAE
National Aeronautic Meeting, Washington, D.C., April 12-15 1965
['313 B. Crispin. Einlaufdiffusoren fur den Oberschalflug. DVL-Bericht Nr. 14, April 1961
Z[32.7 W. Trosdorff, Versuche an eirnem fertigungsgudnstigen Mehrstobdiffusor bei Oberschallgeschwirdigkeit
DYL-Bericht Nr. 44, Sept, 1957
1 33] W. Tromsdorff. Thtersuchungen an Triebwerkseinl~ufen, DL-Forschungsbericht 66-34, Mai 1966
[34_3 E. Sbffker, Untersuchungen an einem OberschaltuiplaB-Diffusor mit stabilisierendem Ring, ZfE 8,
r
Hleft 2, S. 33-44, Febr. 1960
135_3 R.B. Pearce, Causes and control of powerplant surge, Aviation Week, Vol. 52, No. 3 S21-25, Jan. 16,*5D
* [363 J. Leynaert, Pompage dans les entrebs d'air supersoniques, OflERA TP866 ou 1 Aeronauti&,"e et I Astro-
nautique, N22, 1970
L 3731 D.G. Stewart, Proposals for a furthe, programme of intake instability research, ARI.Techn.emo 170,
April 1957
[32 0. G. Stewart, Design of conical intakes for irnstability investigations, Tehn.Memo. 179,
TR.
Jan. 1958
[-39)Y 0.6. Stewart, Supersonic intake instability - the effect of design geometry on intakes of 25* cone
semi-angie at MN 1.83, AR!.ReportM.E.10 helbourne, Oct. 1962
f[4037 0.6. Stewart, Supersonic intake instability - further investicatiuns of intakes of 25' cone semi-
angle at Mach numabers up to 2.14 with and without boundary layer bbed. ARt.Report N.E. 112.
Melbourne, Ja-n, 1964
(r41,.7 0.6, Stewart, S.A. Fisher, Instability and turbulence in supersonic air intakes - the effects of
free stream turbulence, Reynolds number and surface roughness on some axisymmetric intakes at
M = 1.8 and 2.2, AR!.Report 14.E. 129 Melbourne, Nov. 1970
[*42,7 S.A. Fisher, on the mechanisn of supersonic intake instability with some thooghts for further work,
AR!.Note M.E. 313. Melbourne, March 1970
[C43-7 S.A. Fisher, Tests at Mach 1.915 on an air intake proposed for Concorde, NGTE Re'or- No. POO0, 4/67
['44-7 S.A. Fisher, N.C. Neale, A.J. Brooks, On the subcritical stability of variable ramN intakes at
Mach nwroers around 2, NGTE Report No. 1.311, Febr. 1970
t-45_7 N.C. Neale, P.S. Lamb, Tests with a two-dimensional intake having all external compression and a
design Mach number of 2, ARC, Current Papers No. 937, 1967
[r4637 E.!. Goldsmith, C.F. Griggs, The estimation of shock pressure recovery and external drag of conical
centre-body intakes at supersonic speeds, ARC, RM No. 3035, 1959
f(473 C.F. Griggs, An investigation of two methods of suppressing shock oscillation ahead of conical
centre body intakes, RAE, Techn.Note Aero. 2551, March 1958
nae tsproi pes
( 48-7 C.F. Griggs, E.!.. Goldsmith, Shock oscillation ahead of centre-body i..ksa ueroi pes
RAE, Report Aero. 2477, Sept. 1952
[49_3 0. Beastall, Flow instability of centre-body diffusers at supersonic speeds, ARC,R&H No. 2933, 11/53
[ 503 R.G. Schweikardt, R.P. Grippe, A supersoric intake control system for the external compression mode,
J. of Aircraft, Vol. 8, No. 12, S.995-1000, Dec, 1971
[r5137 G.1.Hall, A criterion for prediction of airfrare integration effects on inlet stability with applic-
ation to advanced fighter aircraft, AGARD-CP-150 Paper 2, Sept 1974
['523Y J.E. HaAins, YF-16 inlet design and performance, J. of Aircraft, Vol 13, No. 6, !_.436-441, June 1976
[53_3 1.1.Retti, W.G.E. Lewis, Design and development of an air intake for a supersonic transport aircraft
J. Of Aircraft Vol. S. No. 6, S. 613-521, Nov.-Dec. 1968
[5437 W.R. Haage-son, L.M. Rendall, Inlet development for the B-11 strategic bomber, AIMA Paper No. 74-1064,
Oct. 1974
[r55_7 W.F. Imfold, Development programme for the F-IS inlet, J.Of Aircraft Vol 13, No. 4, S.286-291, 4/76
L 563J F.M. Burcham, D.1. Hughes, J.K. Holzman, Steady-state and dynamic pressurp phenomena in the propul-
sion system of an F- Ill Airplane, NASA TH D-7328, July 1973
(r57_3 J. Peikert, Regelung von verstellbaren OberschallufteinNufen, insbesondere zweidimensionaler. Schrg-
sto5diffusoren ftir Gasturblnenstrahltremerke zum Antrieb von Hochleistungsflugzeugen, Deutsches
Patentamt, Auslegeschrift 25 45 019, 20.03.1980
[-58)7 6. Laruell, Personliche Hitteilung Uber das Stabilitltsverhalten der invertierten RechteckeinIhufe,
ONERA, MATRA, Rolls-Royce, BMe, ME-Intake Meeting Dez. 1979
r593_ E.-O. Krohn, Pers~aliche Mitteilung Uber das Stabilitltsverhalttn der halbrotationssymetrschen
Einlauf-Oiffusorpn. Diskussion: DFVLR-IKdln, institut fur Experimentelle Str1nxngsmechanik/MBB 1979
£ 60-3 C.R. Dawson Supersonic diffuser instability, a summary and a bibliography of current literature,
Aircraft Research Association, Bedford, Wind tunnel note, No. 47, June 1962
[ 61)_ H.-L. WeinrgChi, Findimensionale Betracbtumqen
zun Stabilit~ts. -haltel von Einla-ufdiffusmr-n fu;r
Clbermchall-Flugartriebe, Dissertaticn, TH-Oarmstadt,
19791-i
f 62)-7 .-L. leinneich, Eir-e (In'tersuchung
instztion~rer eindi_-nsicnale- Str~mungsvo-ange.
bericht. Fachgebiez Flugantriebe, MI 1978 instit~-u-
[ 63]- H.-L. Weinreich. Vergiteichende LiteraturdurChSiCht
zin Them. des Djffuso rrmmns in-titutsbercjt
Fach lbet Fiugantriebe. TID 1978 .
f 64)7 C.S. Jell, Air intake aerodynamics
and operational and installation effects on
performance..AGAp-vv Lecture Series No. 98 m-issile powerplant
Supplement, 12.-*6.3.:g79
f 657 L. 0. Stillig, Automatic detection an
suppression of inlet buzz in instrumntation
propulsion, Proc. of the Symposium, MIT-press, for airbreathinoa
S.453-460, 1974
[ 66) R. Sauer. Hichtstationre Probe!3 der
Gasdynamik, Springer Verlag 1966
[ 67]7 A.H. Shiapiro. The dynamics and thermodynamics
of compressible fluid flom, The RonaldPress
[ 68.7 E. Becker, Gasdynamik, B.G. Teubner, Stuttgart 1954
1966
[ 69)- R. Courant. K.O. Friedrichs, Supersonic
Flow and Shock $(aves. !nterscience Publishers,
flew Yurk 1-148
pressure
_ 8P.
an intake recovery
without for 2 EZ
internal contraction __
Detail: &aWtes
shrCe
Fig. 3 1!Fig.
~ j.."""~ Yortex sheet from shock 4 Shock induced biiundary
triggers the Ferri layer separation which
instability
instability
Fi.5
Soeeometric parameters influencing
A buzz in the case of a simple axiswetnicr
Wduel
Fig. 6
Simgle elongated model for___ -- - __
stability analysis
I N.
'E7 IS&
4-~L Fi.,l
t Agglomaration of a
diverging rarefaction
PFig.7av 17
Propagation of
pressure waves and
6entropy discontin-
uitles after a Fig. 9
distortion of Reflection of'a pressure wave
nozzle througliflow from an arbilrary open end of a tujbe
AA
13=.. I ReflectiOn Of a Pre-Ssre *-v 37, and an en-t.-' , i %yy
.2after distortion of th~eintake qperasm. pIq ni s
Fig. 11
N
intake
response 'fo d"-fereft
slopes of the iptake cheracter~st~c 22
2 35
x 4, , 01irt, 225
7r
t~ F7OS
ig. 12
'ten;;iation Of a =ass
O~O 2
( 1 +a
25 flowdisturbance after
sude. dischtrge
3 3
Shol I
-
/ x~ x, 'mozzle
'h4
______
UZ X
Fig. 1A
unstable /10 ~ Two
IK a
most sivple special cases:
1: ax. attenuation
Stable
0 .
f I NK
-o - G n--
of Zzne
P-0-
tttuwti
7
--
32
Example of an ink
phase
hsnai stable sacritia oeazo
cM a~ r Onat
0o.an ga tyoe OSci lai o witrn So-tn,, f h
. mh
o--ni.
e rt r D
-- .e s _
th-G-e
t --
13-14
DISCUSSION
C.Limage
In recent experiments in the USA, it has been observed that the inlet instauility has been driven by the ramjet
combustion instability. Have you considered thi. type of inteaction?
Author's Reply
The interference between inlet instability and combustion has been known for many years and have been reported,
for instance, by Triinpi 4 ," s. Daily2" and Sterbentz 22'2 .
There also exists some theoretical approach for this problem (Mirels', Trimpil4 IS).
Because the buzz phenomenon is not generally altered by the combustion process, it was decided to avoid the
complication of calculating the wave dispersion and reflection by an internal combustion zone.
It seens to be possible that the theoretical method presented in References 61, 62 which include effects of internal
area change and local friction could be extended for the consideration of the effects due to local heat addition.
H.Wittenberg:
In the last section of your paper, you have compared the results of your theory with experimental work in a qualita-
tive way. Have you also made comparisons with experimental data quantitatively to investigate if your method can
be used to predict the stability behavior of inlets?
Author's Reply
No, we are now just trying to make such comparisons with experimental work performed at the DFVLR by
Mr Krohn. In my theoretical work which was performed at the technical university, I had no opportunity to make
experimental measurements. I used for calculations one experimental pressure recovery characteristic which I found
;n the literature.
Mi
r- n # -z y- e I
,A..
RANGE MAXIMIZATION METHOD FOR RAMJET POWERED MISSILES
WITH FLIGHT PATH CONSTRAINTS
by
Ulrich M. Schttle
Institut f r Raumfahrtantriebe
University of Stuttgart
Pfaffenwaldring 31
D 7000 Stuttgart 80
Germany
SUMMARY
1. YNTRODUCTION
Ramjets are currently being considered as the primary propulsion candidate for a
number of future atmospheric missiles as they offer the combined advantage of high speci-
fic impulse, high speed, continuous thrust and compactness (1). The integral rocket
ramjet concept in its solid fuel ramjet version or solid dicted rocket version is of
particular interest. The present paper addresses the problem of range maximization of an
4 integral rocket ramjet missile and reports on application work which was performed in
order to establish the performance capabilities, throttle requirements and parameter
sensitivities of a given missile. The vehicle design and propulsion characteristics as
well as the mission requirements were defined in previous industry studies and were
prescribed (2). The mission of interest is schematically shown in Figure 1. We assume at
the outset that the structure of the optimum trajectories cosist of climb, cruise,
descent and sea-skimming segments. Not shown in the sketch is the rocket boost phase
which will accelerate the missile to ramjet ignition velocities- Only the rzmjet powered
flight will be considered. Mission objectives require trajectories below a prescribed
maximum flight altitude in order to delay radar detection and a final weaving flight
phase at sea-skimming altitudes. This type of maneuver will maximize the survivability
of the missile against shipboard defense weapons. Ramjet trajectory problems have not
received much attention in the control literature yet. This may be attributed, in part,
to the fact that these works are rarely published beyond industrial reports for security
reasons. Open-loop optimal trajectories for minimum fuel-to-climb , minimum time-to-climb
and minimum time-to-intercept problems for an air-to-air Liquid fuel ramjet-propelled
missile are considered in (3) A separate important issue is the on-board implementation
of the optimal control solution. This requires that the solution be found in a closed-
loop form. The focus of the work and (5) is mainly concerned with the derivation of
such optimal control laws using singular perturbation theory.
Our goal is to obtain the optimal open-loop control program. The control problem is
formulated by constructing a cost functional which penalizes violations of the inequality
4 Z 2
- -- '
Iconstraints. An efficient method to optimize complex atmospheric trajectories is to trans-
form the continuous optimal control problem to one of parameter optimization. This
technique has seen wide application to a great number of optimal control problems of
diffinc o l et, 3 (7)iu of vaitin3eho s
varying complexity, to (13) and in our example proved to be less complex anc.
difficult to solve than calculus of variation methods.
The remainder of this paper is organized as follows. In section 2 the e.r'imization prob-
lem is stated. Then, in section 3, the numerical solutio, technJque is der ribed and in
section 4 the missile simulation model is defined. The final section 5 presents some
numerical results. The first examples concern level flight at sea-skimming altitude.
Fnally results for low-high-low flight profiles are obtained.
2. PROBLEM STATEMENT
t [ f ) ,1 ] 0(t
O (3)
In equations (1) to (5) x(t) represents the n-dimensional state vector, u(t) denotes the
r-dimensional control vector, n represents a design parameter vector of dimension m,
q)is a q-dimensional vector function and 0 , L are scalar functions. The subscripts O"
and 't" refer to the initial and final values of the independent variable time 't',
respectively. The initial time to corresponding to ramjet ignition is specified while tf
is free. Final time tf is determined from the specified final mass which is used as
stopping condition in the numerical algorithm.
* contains the specified values for terminal velocity vf, flight path angle Tf,
altitude hf and possibly lateral displacement yf. The inequality cons,raints (4) and (5)
result from structural limits, mission and propulsion requirements which bound ;he
altitude-Mach number operational region and the controls. Specifically, it is required
that recovery temperature Tr (t)<Tmax , Mach number M(t)>Mc(h,1), flight altitude
O<h(t)<ha, propellant flow rate umi n < u I < j(t) < , < with throttle ratio
2
1 / mx and the resulting angle of attack r(t)=VCt (t)+8(t)
2 _ max stay within
pregiven limits.
, Optimal control problems with constraints on the state and/or control are often
solved by converting them to a sequence of unconstrained suboroblems by means of penalty
functions. These are added to the primary indicator of the system performance which in
2 our case is the downrange coordinate x(tf). Thus, the functions 0 and L of equation ')
are augmented to include
:3 -%
",
!L
2 w 2
((tf),?_
f = - x(tY) + = i. (6)
and 2
L S R(7)
i=1
where = i 0 if hi < O (8)
0 if hi L 0
Ki > 0 if Si (9)
0
O 1-f S i 0
P£' Qi (Ci) and Ri (Ki) represent coefficients penalizing violations of the different
constraint conditions (3) to (5). The functional inequality constraints (7) are evaluated
by integratinq the constraint violations over the Irterval [to, tf] along with the
numerical integration of eguation (2). A sequence of ortimal control problems with in-
creasing penalty coerficients must be solved. In the limit, minimizing the augmented
alty function maximizes the range performance and drives the constraints into satis-
.action. The derivation of necessary conditions for the optimal control problem stated
bove by means of the calculus of variations is straightforward and well documented in l
literature (6,7), and will not be repeated here. Some aspects of numerical solution
techniques applied to the ramjet trajectory problem will be discussed in the next section.
3. SOLUTION METHODS
The most widely used iterative computational algorithms to determine optimal controls
represent variations of either the gradient method or Newton's method in either function
or Euclidean spaces and are local in nature.
Function Space Conjugate Gradient Method
in the beginning we sought the solution by use of a function space conjugate gradient
(FCG) method (6,7). We used this technique suc~essfully to obtain optimal controls for
rocket ascent trajectory problems to near earth orbits. However, when applied to the
ramjet trajectory probiems the FCG method failed due to excessive computing time. This
failure may be attributed, in part, to 1. the highly nonlinear characteristics of the
missile and the ramjet propulsion system and 2. to the high number of equality and
inequality constraints involved in our problem, which were handled by the method of
penalty functions. During the iterative process the trajectory tended to leave the
bounded altitude-Mach number region, for which valid ramet propulion data were available.
Hence, comparatively high penalty coefficients had to be chosen already at the beginning C
in order to force the trajectory to stay within the limitations of the propulsion model. 4
Adding penalty functions to the performance index containing periodically increased
constants may change the original optimization problem to a large extent by creating
narrow valleys in the state space res'tlting in extremely slow convergence. Therefore, the
= FCG method has been abandoned.3
Parameter Optimization Methois
The appearance of efficient and roh,-t algorithms to solve nonlinear prograosming
problems (NLP) during the iast decade has led to the replacement of the more complex
calculus of varie&ion methods by techniques which dicretize the problem in some way. Such
an approach reducos the criginal continuous optimal control problem to one of finite-
dimensional parameter optimization. i.e. NLP. The approximation of tho original problem
may parameterize the states as well as the controls ( 13) The methods employed in the
remainder of this stu -y: however, parameterize only the controls. Parameter optimization
I __ _ _ _ _ _ _ _ _
methods and unresolved problems connected to it are surveyed by (8).
/ ' T An obvious direct parameterization method is to represent the control vector function
11 (t) by an explicit function of a parameter vector p c Em and thus control the trajectory
by means of equation (2). Although it is natural to treat the controls as the independent
variables and the states as the dependent variables in the systems equations (2) it is not
essential to do so. In problems involving terminal state constraints and/or state
inequality constraints it may be more convenient to treat some of the states as
independent or nominal variables and to determine some or all control variables in terms
of these independent variables through the state equations.
In our method of approach some control variables ui (t) are approximated by specifying
ui(t) = Ti(2i, t) i = 1,2,..., r1 < r (10)
where Ti are known functions of the mi-dimensional parameter vector p,. The remaining
x-ri control variables Ur1+1, Ur1+2,.., ur will be uniquely determined in terms of a
selected set of independent or nominal state variables. Similar to (10) a specific para-
meterization for these nominal state variables is adopted
xj(t) = T(12j, t) j = 1,2,..., r-ri (11)
where each T. is a known function of the mj-dimensional parameter vector p.j.The form of
the parameterization (11) chosen should be consistent with the terminal conditions and the
state inequality constraints of the problem. If the j state variables are enforced
according equation (11), the control variables ur1+1, url+2,.., ur are uniquely deter-
mined through the state equation (2). Unfortunately, inmost cases these control variables
cannot be determined explicitely in a straight-forward manner without having to integrate
the differential state equaticn (2) numerically. To circumvent this difficulty we consider
the variables x. of (11) as nominal values for a pitch controller, whica determines the
controls u.(t) in such a way as to drive the current state variables to the nominal values
of equation (11). Thus, the controls url+1 ,..., ur are defined by a control law
uk(t = Tk(0 , xk(t), Q = Tk(pk, t) k = 1,2,..., r-r| (12)
where Tk represents a specific relation of the nominal values xk(t) and the mk-dimensional
vector Rk' which combines the parametervector jj of eq. (11) and pi. The control models
utilized in different flight path segments are described in the next section.
T LetT n, T TI
= mI + m2 +... + mrr + m'7 and define the m-dimensional parameter-vector p -by
P1I 2 I ... r . For given initial values x (t0 ) the dependent variables x can
then be solved in terms of p by numerical integration of the differential state equation
(2) in the interval It , tf. The cost functional J(u,7) reduces to a function z(p) and
thus yields an ordinary minimization problem. This will be solved by a variable metric
penalty function method employing either the Davidon-Fletcher-Powell (DFP) algorithm or
the Broyden-Fletcher-Goldfarb-Shanno (BFGS) formula (10, 14). iowever, with respect to the
boundary conditions (3) either penalty functions or projections may be emplc-,ed. We found
for our control problem that the DFP or BFGS alorithm coupled with a ;Tadient projection
technique for handling terminal state constraints offers computational advantages. Most
trajectory optimization examples of section 5 were solved by an accelerated gradient
projection scheme. In this case the penalty terms associated with the condition (3) are
dropped from equation (6) and the parameter optimization problem may be stated as follows
minimize z(P)
subject to the (13)
boundary condition J(P) 0
Note, that the differential constraints (2) do not enter into formulation (13) as they are
1 _._ _-_ _
+~
A I ,,,__,,,,_,,,,,,,_
where ak represents the step length taken along the search direction sk and the super-
script indicates the iteration cycle. In the restoration phase the task to be accomplished
by the parameter update scheme is to meet the desired end conditions within prescribed
tolerances. The parameter set is updated using the displacement
with a nominal value ai = 1, which may be reduced within 0.05 < a < 1 to enlarge the
convergence region. In equation (15) Hi is a diagonal weighting matrix which attempts
to normalize the effects of the various parameter and will be updated as described below.
The notation q indicates gi = (I). The matrix i contains the partial derivatives of
the constraints with respect to the parameters and q is the vector of constraint errors
for the ith-iterate. The minimization step is accomplished along
s Hk Fk (16)
using cubic interpolationto !ucate the scalar parameter ck which yields the greatest cost
decrease. This search is terminated prior to passing the minimum at a k if either the
parameter changes or the constraint errors exceed some prescribed bounds. In this case a
new restoration cycle is initiated without modifying the matrix H. Otherwise H is updated
by use of quasi-Newton methods. These include the famous Davidon-Fletcher-Powell (DFP)
and Broyden-Fletcher-Goldfarb-Shanno (BFGS) formulas (see for instance Ref. 14):
Hk +1 = ¥kHk + kk(AFk)T
k~k -p , -Hk Fk --
i ]~. -p
Apk (62k)T
k k k
( - k(A)kkT LF (An )T ( o
z lr
1f
_z = z + - z(p) 3
gi gi- g(n) (21)
apj 6pj " pj 6P.
_ = = -- - =g -g (22)
LZ z~ ag gi
g2 6pj(
apj 2 6pj J apj
+
where the notation z+= z(pV,..., pj + 6 pj,..., pm) and z- = z(pl...., p.- 6pj .. p )
is used. In the latter case a scheme suggested by Hull and Williamson (12)is employed to
verify the magnitude of perturbation. However, the use of the central-difference formulas
is avoided whenever possible to reduce the computational expenditure. In most examples
the use of forward differences was appropriate. Another way of generating the partial (11)
(1e)
derivatives is by the quadratures of the error of the optimality conditions . In
gradient generation by forward differences compare favorable with error-quadrature
gradients.
4. SIMULATION MODEL
Dynamic Model
The flight of the missile is described by a set of seven differential equations
written in a relative velocity coordinate system. This system has its orgin at the
current position of missile's e.g. relative to an Earth-fixed nonrotating xyz-axes system
and its axes are aligned as shown in Fig. 2. The equations of motion are as follows:
[F.cos a-cos 8 - W!
-=[ - g9sin y
= I- [F-sin a + A] - -0cosy
-.cos y Cos X
v-cos y . sin X
h = V •sin y
The associated elements of the state vector x = [v, y,x,m, x,y,h]T are the relative
velocity magnitude v, the flight path angle y, the relative azimuth X, the mass and the
coordinate of the missile point mass. g0 is the surface gravitational acceleration. The
missile is controlled by the control vector u(t) = [JU(t), a(t), B(t)J whose components
are propellant flow rate ii(t), angle of attack a(t) and sideslip angle B(t)- Structural
limits, mission and propulsion requirements impose four operational constraints the
violations of which are numerically integrated along (23)-
I
) 2
_ Tmax
-(Tr(t) if Tr(t) > Tmax otherwise = 0
0 f -
- I
The ae.'odynamic and propulsive forces are obtained from the relationships
L CL aM).q.S
D CD( e,m,h).q.S
(25)
Q = C0 ( aM)-q.S
F = Cp(M,h,u)-q.S
2
where q denotes the dynamic pressure q = v and S is the reference area. Thrust F is
considered to act along the vehicle body axis. The acre-thermodynamic heating load is
described by the recovery temperature T r which is given by the adiabatic equation
T 1 2
r = T.(I + - .M ) (26)
where T is the ambient temperature and K is the specific heat ratio. The atmospheric proper-
ties, i.e. density o, temperature T, speed of sound a to define Mach number M = v/a,
are derived from the 1962 US Standard Atmosphere model.
The aerodynamic data are given as drag polars and the dependence of the aerodynamic
coefficients on angle of attack a and sideslip 8 is assumed linear at each Mach number,
thus
CiL CLa(M)-a
2
. C (M)-'I+c'h) + K(M)-C (27)
cO - CQe(M)-
Since the missile is onsidered to be sy-metric we assume C0(M) = CL (M). The aero-
dynamic coefficients Ca (M), CDO(M) and the induced drag constant %(M) are represented
as tabular functions of M and are determined in dependence of the current Mach number
by means of cubic spline interpolation. The thrust coefficient CF(4,h,0) is characterized
by a strong sensitivity to the flight conditions M, h and to the propellant flow rate u.
It is also given in the form of tabular data. CF is modelled for different values of
= const. by bicubic spline functions and the dependence on ji(t) is determined by cubic
interpolation between the several planes CFCM,h)I = const
The lower operational Lmit for the ramjet propulsion considered is given by the critical
-
Mach number Mc = Mc (h,) ,-Wic is al.* given in form of tabular data and .modelled utilizing
bicubic spline functions. with respect to the fuel flow rate u the propulsion
characteristics require :.(t) to obey umin i (t) < I*,<
-m with tne additional
restriction of Ua/ Ul f 4-ax for the throttle ratio t.
Control Model
i The trajectory is determined by three control variables: the mass flow rate u(t), which
implies thrust magnitude, the angle of attack a(t) and the sideslip angle (t).To = -
The time interval it , tf can be partitioned and the para.meter of p can specify the
magnitude of the control variables at secified points in time. Simple linear inter-
polation or cubic spline interpolation (l1. can then be used to def ne the control
variables at intermediate points in time. Alternatively, the parameters can be regarded
as coefficients of some mathematical model, which may be either a function of time or a
function of the current state. Both methods are employed in our control model.
in_ - _
-ML
=t
The angle of attack a(t)-parameterization assumes constant flight path y segments and
/57: cruise'phases at constant altitudes h, i.e. a flight profile as shown in Figure 1. The
desired values of constant y or h within path segments are considered as the nominal
values for a simple pitch controller which defines U-control in a closed-loop form.
(t)
M - i(t)
a(t) as (t) astc + (28)
Sh(t) - hn -.(t)
a(t) ast(t) - t- +Ut) (29)
C
hj
In equations (28),(29) denote Yni and hnj the desired values of flight path angle and
flight altitude, C,, C, Cn represent weighting coefficients and a is a preset constant.
Furthermore, as denotes the steady-state angle of attack
0m(t).go
cos (30)
s= (F(t) + CLgq-S)
resulting from the second state equation (23) by setting 4 = 0 and sin a a. Note, that
T
ni = 0 for co;:stant altitude flight phases. The bracket terms in equation (28) and (29)
represent linear combinations of the error function and error rates of change. The
a-control derived from (28) or (29) is restricted to la(t)I < -ax wth : a specified
control bound.
At least four flight path segments are assumed: climb, constant altitude cruise, descent
and constant altitude sea-ski--ing phase. The climb and descent chase may each be
partioned in additional flight segments. However, only two constant altitude flight
segments will be considered. In the constant altitude cruise leg the desired nominal
altitude hn is set equal to the specified maximal altitude hmax and in the sea-ski4ir
Phase the nominal value h is chosen to correspond to the specified terminal altitue hr.
As wi)' become obvious from the next section, h represents the opti-a
a cruise alti,-ae
for our trajectory problom.
"
The nominal values Yni as well as the weighting coefficients C_.. C.., C.. for the piC '
controller (28) or (29) of the different flight path se=_ents i are collected ifl tne
parameter vector p to be cpti'ized. Prover selection of the weighting constants will
result in an optimal transient from the current states y(t) or h(t) to the desired.
nominal values.
The a-control model chosen :epresents an attemot to ell--inate the explicit appearance
of the final states h. and y. in the boundary condition (3). Both are satisfied directly
by the control (29). Furthermore, satisfaction of the path ineq-ality cnstraints
h(t) < h ax and h(t) > 0 is achieved easily without the need of nigh penzlty coefficients
in the associated penalty functions. Hence, the numerical difficulties mentioned in
section 3 regard-.ag the use of penalty functions are kept to a minimu=..
The sideslip angle is S(t) = 0 for all but the last path segment. In the final sea-
ski;-m-ing flight leg lateral aneuvering of the missile is w.com-plished bv a ban-bang
type -control. The magnitude of S(t) is numericallv determined from the third state
differential equation (23) for a specified lateral acceleration. A control constraint is
imposed:
8(t s
8(ti r
< /S - a2 (t) (31)
where a(t) results from ecation (29). Hence, high lateral acceleration requiraments
resulting in higher -angles will not be realizable.
The optimization technique deter=ines the controls u(t) and O(t) in the sense of the
_i3T4
-A.
A..
above stated approximation. with this control model the resulting trajectory may not be
optimal for the problemo originally stated but shotzlG be very neatly so for most problems
of interest.
S. IMMERICAL MMAPLES
This section presents two examples llustrating the potential performance im-prove-
menits that can res--lt from lte =npienentitilon of optimalI controls. -he first example
considers a missile flight mission. at se-trigaltl.- s ind demtostratez the
improvements gained by ot~~v throttle carsabiliries. The second eyxample correspondAs
to a lohg-l n ission crof lie. For all cxam-ples to -1a presented In this section tie
following rem-arks held: all trajectory calculations were performed on a USIVAC 110t
coputer us±.ng dcble preczicn arithuetic. ll smlts flights ujerde initiated a: the
same condition, which: corresponds to ram;ericnitlcn. and 131! had to meet the same
boundary condItilons v_ , 4 and !,- -.he trajectory calculations were terminated at
depletion of a specified amoun-t of fuel.
AS Mentioned above tine fzIrst missi1on to be con-zIdered is one of Ileve flight at l1w
altItuwde and Includes a fInal-3ain flight pnth sen _nt of Prescribed length. The sea-
t
skis n;g alitd.h.ich is chosen. ec:al to the temialnndiltion h:,. Is inamtiedb
-siti-CUntrols determ-ied from equation (293 the coefficients of Which h--- been optims'-,
in a separate calculation. ThS. only the fuel f.*o- control wit., is to he ottizsred.
since the sidsip andy~ S Is deatermIned byI te requ-rement of a speciXfi-c lateral aiccel-1
ratiam. ~~ ~ dli. ~ ~
,*1i'adv*0'AreTebreryc~ ~ ~
O satis5fied ditl by s-control
(29). The ratncvr.4dltlor-s to be met are the termlrnal velocity v., the reqit-red
ditnecf weavinq flI -u_~ anid flight pat iecuaity Nosrits14.mte.
c
conditIon on the iateral disolacenen-t is ipediceytlis qn~l~aa
sioescase bets- - a-d xz cuaad he satiste (withi etclwerrl
utiliing a Control paraeter consistIng of luszt two naramerers PS ie
coestent fiuel flow rate ;: and the r-tltr tinev s of wearin fli7ht. flowerer, since
there are two end condit-lofs and cmlv two con'-'oll ca-ame-se there Is =n parame-tr left
for optilnltics_ an- th=s the rolmredutCes I-- One of tarcting. 7-r soeCl-_fzcas1on o
T, and- 1, c-nloely defines the zwo parame-s- metceoh ncr o_csr are
r Imescly
deter-mid iny a es.
-"nzn- i-c cyce (Section 3;. As this spcial case
P2 is of ConepatclItrsmceslsae presentedIs inFseres 3 to, S.
Figure 3 shows tLe effect of parameters :5 and tW7 on Weam~ fltscit distance for a
specifIe final v-unc-ty r, and e11fen:ent ltalacceflerazr a,.
&_.. . x
re e. the
cbolcew Of fuel fl-wiwt, 0.8 - Coma toectrer w2ztL 06 US ield a nav-ny
distan-ce of 2.!1mfwhile, seetinq the odto Tm
, !or a aamvrer ac-le-atioc, cf a. 2.
It lshofin zirc 3 that for high raceriiaet_eents a, M 3 and
only shourt WeArn ;%tAnce_ can. b.e ac&hieve vie= with he'. I reer setting. lartheranre.
high ateral accrolerat ion a - A may not be maintalzned de-te o.er'-tie of tiesaini
flight _--e to the assciated rapid elctdcr Asean d;e to nhem=,me sidell
anilea acorin F.qat'on "i.in F jartbe IlI time t, Is a-so clottecd: the
difeenesofwhchfo svral a lratl- a,. ale too -l as to be shc%=n in -he
grah.
The above disusiondosnt cesder the costraint Tr T maon the n-.2teatr
ALS
Ifand will be decelerated in the interval tw t Ctf of mancuvcrlng flight to the
prescribied terui"a condition. The thrust associated with the specific fuel flow rate
ui(ti - const. required to =,Let both end cond.-tions v, and ix Is not-sufficient to
cancel the increased drag during this flight rnase. Thus, in o-rder to seet mncreasing
ZX--requlroents the missile has to be accelerated to higher velocities In the inte-
Val t 't cct - f the termin-al conid~tlon v, is :0 be arnieved. -.his Is 3cccopll.shve
by utilizing higher fuel flow rates :;. Hawe-er. this Lay result In a violation of the
operational constraxnt T_ C7 (i.e. V Li a= is the case for the etasples fx, - 2 and
Ixv-4 shown in Figure 4. Hence. a constant rowemr Setting Is not appropiate for
wediu to high Ztar - distance requirezents and siaryfor higher rmaneuvering
acceleration-s a-.
he Stared ~e
th4 cit-4 _--d lee svay 1-c parltio*ed tact hero
sevweral ~ ofh desizred vomtazr "osz~ P-tn _lea . taNis ".te at -- r three-
ci i&-nsnec
.tre derntsfle~t ~hes wzd 4st -a sa~~'1
c*aom pas ha L-.
ei-es opotossed It-as . 2-arterira*N.-M fue'l f lo. -;It: andanl
.. f attachaetrl Ilkrver. I= vmrd- to ra ce the tat :enal, esjend * tare =wnry
zt0$ .. Z"sec~ tze dtat=s.om of the, pa-incter ce P was eg e to toi I-
in
c -misf -sqIrm-tz. zz .. sere vxaples -- e simliflmq- asa,,amof
l
1-rdl q_ 1.21 asS (Iiere1
z- cftr -- the corrvtn ptmrdvts f
pa-rmiWas ~ mo and tans imcnded frtm the Parasrer veto
. mrical
cm~arisafw::ts eas accrorac-. :*-s ' a 0.4 Percent of the s-I e e
for
estmla'onof zzsterest.
flqure; 7.us~cazt.--te histories- of .. gh at~e hmn. velocity nt) and
~~ w..t.. 'MI*S trak,-ect---f to Iore~ se r-s~e s
Irr
2rI-na f1iphT distance sllateral sc:eratles AL, 2. regarding bln ooaa
t
cruise occurs at the na=xinal allowed altitude. There is5 a =:all violation of the ititude
constraint hnxat the begimnnInq of the cruise secrent. Tids violation could be cancelled
at the expenditure of additional coputing effort, however, it was Considered to be
tolerable. The vel-ci4ty Um history shows a Continuu Inraeoriotecuiecae
indicating that wit teretcodio nseAdy state is ahievable. The ragcnitude
of vct) undergoes rapid chancmes during the descent and -al sk -- n Pase- Its n---&
fro= Figure Z that f-Iflew rater i.e-. thrust, is increase alreaAv durim the descent
phase. Thus, veloci11ty as Increased during the final oorro odescent in= order to Cancel
the deceleration of thIe w-eaving, flight set-ment and to flee-t the_ boundary condi tionv,
Th.e recovery temperature increases accordinalv and hit T_
Max ni o asotie
Onceagai,ths vilaton -Could he further decreased bq acctzonal
& connotation but was
considered accentable. -h rorresnocng Cotros a-reiern
the fuel flow control. vnic- As pareteriz&d bv ee'- t ln se---s. it is
noted that fuel flow isse-ilvreduce_ duA.. rsIs -li- -i- contrast to C:e
classical steady State r~f~~_-.e analyrsis an to -.e 0 n renar
throttle iS founId to bm----us--= sznula tmd-Z -j ex-pl- t~
utilization of maxi4__fZe resua-ts in a ef 1nl.%tcnsraint
-'ltin
rne
trajectory cpti=misaton: nrcties for a ramjet pawe-red missile is soi-&wt
a oarar-ete-cp t-"--cie =etho&- The trajectory coflsistS of o1 i distincwt phases (climb,
cruise, descent, wezmvirn flight at sea-ski-incz altitudes)'. Penalty functions are
cmlved --n four state variable i-neiaali- contraints. - crical sniulato resul-ts are
Crsn:or Z range of d__' fercnt -'iss'lo- -eaairen-ents- The results deonstrate tecaps-
b~i::of
hearoedre o oic ~mPlex. realistic trajectory cp)timeio roblems- it
isshown- t:hitdeean enr 103-s cet cc ostrtle cabi itU ofth
rfl-et pro-ul-iS4= system- resulin m inproventsz of aboumt 5 to "I ziercent ~e
t-- mre ccntro. --naddit4ancontinuous th rottl... - -
~-V=
='-=-'t - e'a- - ataims_
- m-- *
;=-7102 -- d Contro C-'f~reo
-. E: at- s.Tx !a
m oAt cZ la -m So I
em A o ' . i s * m s t
t ~--=
~-~ ~ . ad--as Vcvxsus Eror '=-aacre cradienlts ' _
I/g
sc. r
x-cs
I: '5 -.
- - -I L
r
X (tf) x(f 6 i
X AxX
xv
yvv
Fi~u.R 2
;ltufmITIONi OF CONTROL VARIAB CS A!qD FORCE=S
sea skim flight t'ftw,
h l1in 10
control parameterp-t
4 Vf specified
>a( -1 2
08 -
( 3
WVw
2-
2- 0.6
\2 _
0 '
06 08 10
PROPELLANT CONSUMPTION,,. (nd]
'09
S05
~
0 020. ~ ~ ~ .5 V t
or,,
tr . tV
05 1 coto prm
101
':0.9at 2
08
0 025 05 0.75 10
TIME I (nme]
10
-. An x
C _-F
06 _ _
0o~
4 177 ,II 04
0
020
TIME I [ndi
FIGURE 6: PROPELLANT FLOW RATE Ys TIME FOR THREE EXAMPLES
OF LEVEL FLIGHT
10 11
08 -
10 W.a
-
06 >
I-S
I0 ILI
02 100 06
operational conditions,
hra 0.8
-C
1i0
&
, 0.8 a---
1j -x2
tL06 4
z j(t) 0 O
2.0 Zi 02L
1- 04.--. -02
.j
IL Q
-10
0 __ _
0 _
Q2 0.4 06 Ole 1t0
TIME t (nd]
1
8:CONTROL
FIGURE ASFUNCTION
VARIABLES OF Tiff EXAII
FORFLIGHT IE OF FIGURE
7
X.f~7 ~
~W
A..1
C -
co0 C >
-~
000J
I -
0/0
/ E
/ j 0
/ I z
- __ _ ; 0
a C4
ItI
95- -
0 aD 0
15-18
* 10f
x 08
P5
ZI AX,,.2 2
z
0
0 1 2 3
LATERAL MANEUVERING ACCELERATION a,(nd.]
10-
2 a -2 h -110
THOTL RATI
KIUR AOOHRHEP F~tttC
h2 .0'aF1~RTL
O tWH -O
ILITPiItE WT RIEA LTTD SlOE
__
- ~ -
ly-
z~~~-
06~
15419
DISCUSSION
H.Wittenberg
Thec optioium control for maximum range, which you have shown in your examples, seems rather complicateti to
found by optimiza~ion theory. Have you
perfor in actual flight. T1his is often the case for soltutions. which ame
co'iqpared your optimum range wi~h the ranges% hich are obtained by simpler contrcl functions. which approximate
tlhe 'deal ones?
Author's Reply
Some examples of trajectory simulation results utilhzing simple controls are _-iven i-i the written paper 'Fhe
simplified cc-strols result in a range performance decrease of abo--t 5 to 10 percent when roinp',red to it., optimal
solution. However, the niain advantage, of the more complex control programs isseen in the fact that these controls
cdfnbe adjusted such as to avoid violations of the operational constraints.
The problems of on board implemoentation of the contro: solution has not been considered,
T.13ell1inger
Have you considered using your method with fe-Aer constraints. parttularly with regard to the alttude lim.its which
seem a little low? If this were higher, wvould the throttle limits still be crncounzered? If suich were tlte- ease. I wendcr
if you would encounter a miore continuous control for both the angle of incidence and for the throttle?
Author's Reply
Both the altitude constraint and the throttle limits imposed oin our problem are due to mission requirermnts and/or
due to the characteristics of the propulsion system. We have considered higher altitude trajectories: however, the
altitude had to be resticted to the range of available propulsion data, which is limited with respc:t to height.
Also. for a wide range of trijectcery optimization problems, we have neglected the constiaints on throttle ratio,
Allowing higher altitudes and ncegleIcting-- the profellant flow constraints would result in th. saine structure of the
flight profile due to she control model chosen and hence svotid not charge the characteristics of both the angle of
attack nod the throttle control.
Author's Reply
Well. I think it doesn't make much difference whether I use trajectories for several maximum altitudes in.the range
between 10. 15 or 20 kilomctres. It doesn't make any difference to the optimization effort. Theireason why vwe
constrained ourselves, to this range is that we %ere considlering dutmcd rockcet ramjets with restricted throttle
capabilities.
JET-AFTERBODY INTERFERENCE ON MISSILES IN SUPERSONIC FLOW x) 14-I
by
BERNHARD
WAGNER
DORNIER GMBH
Postfach 1420
799 Friedrichshafen
West-Germany
SUMMARY:
For the axisymmetric base flow problcm in presence of an exhaust jet a study is carried out taking into
account essential physical parameters. Using the method of characteristics ror the iniscid flow regions
and Korst's treatment for the turbulent shear layers, the influence of the approaching boundary layers
is introduced by the concepts of equivalent bleed and origin shift, an4 the ONERA angular criterion is
used for the recompression process. With respect to the interference€ in the confluence problem of inter-
nal and external shear layers special attention i3 given to the morentua terms in the ONERA criterion and
a m-dification is performed in order to achieve agreement of boti'reattachi'ent pressures at the confluence
point. The investigations include separation from conical boattails by applying a simple separation
criterion and extend to small angles of attack by computing the inviscid flow with the aid of a linearized
cbaracteristics method. Furthermore, for the axisymetric case a first result is provided from numerical
solutions of the full Navier-Stokes equations which are presently calculated in order to gain deeper in-
sight into the flow structure.
1. INTRODUCTION
The importarce of the base flow problem with respect to performance of missiles is obvious since a great
part of tne total drag might result from the afterbody if its design is not optimal. Therefore, the after-
body contains frequently a boattail for drag reduction (see e.g. /1/). But the increasing pressure on the
surface of an axisyimmetric afterbody when the diameter diminishes leads easily to separation. Once the
flow is separated the stability is affected because the separation position moves asymmetric in consequence
of a disturbance in the angle of attack. Additionally, in presence of an exhaust jet the separation re-
giuns may be largely extended due to high internal pressuru resulting in loss of control if the control
surfaces are covered by the separated flow.
1 the past some authors developed prediction methods for -nis complicated flow problem in the axisyme-
tric case by use of a component model (see e.g. the survey paper /2/), based on the approach of a simple
downstream facing step. The supersonic backstep problem can be calculated in the following way. For a
trial value of the bxse pressure p , which is assumed to be constant in the stagnating separated regions,
the inviscid boundaries of the bas region are determined by the method of characteristics. Then a viscous
shear layer is imposed on each boundary and the streamline discriminating between the external mass flow
and the mass being reversed to the base region can be found. Finally a closure condition must be formula-
ted for the reattaciment crocess using the flow conditions on the discriminating streamline. By iterative
improvement of the base pressure this condition is fulfilled and the solution is found. The first success-
ful closure was accomplished in the theories of Korst /3/, Chapman /4/, and Kirk /5/. They assumed that
the f'w on the discriminating streamline stagnates at the reattachment point in an isentropic stagnation
process and that the corresponding stagnation pressure is equal to the static pressure far downstream.
While Korst predicted the base pressure satisfactorily for the case of turbulent shear layers without any
account of an initial boundary layer at separation, Kirk has already introduced an approximation for this
feature. Detailed analysis of experiments showed the quantitative succe:s of the first approach of Korst
having been due to the fact that two effects cancelled each other approximately, nairely the neglect of the
anproaching boundary layer and an incomplete pressure recovery at reattaciment /6/, /7/. This knowledge
led to the construction of angular reat.achment criteria which produce good results in connection with
initial boundary layers.
The more complicated problem of a flow over an axisymmetric afterbody containing a centered propulsive jet
has been treated by Addy /8/, who applied the ideas of Korst's restricted theory on the confluence of the
two shear layers. But for good agreement with experiments it was necessary to modify the original recom-
pression criterion by an empirical recompression coefficient depending on the ratio of nozzle radius to
afterbody radius only. Further investigations for boattailed afterbodies led to modifications of this re-
compression coefficient introducirg further parameters for such cases /9/. Since such a procedure cannot
account for all physically important influences, especially not for a Reynolds number dependency, it led
sometimes to still unsatisfactory flow predictions. Hence, it was decided at Dornier to introduce the
boundary layer corrections in combination with the ONERA reattachment criterion into that method and to
study the sensitivity of results to changes in uncertain empirical parameters or in differences concer-
ning the treatment of the approaching boundary layer /10/. While this procedure corresponds in essence to
the ONERA calculations for cylindrical afterbodles /11/, the addition of a first approximate account has
been made for the interference between both shear layers during recompression. The present paper continues
the previous investigations by additional discussions which are stimulated by an excellent survey paper
given recently /2/.
__ tz=
_AL
If the external flow exhibits separation already on the boattail surface in consequence of high ratios of
jet pressure to external pressure, a simple separation criterion has been included into the method /9/
extracted from many experiments on upstream facing steps /12/. Further evaluations of experiments indica-
ted the importance of incipient separation effects /13/ and thus led to a modified version of tne separa-
tion criterion which is presently used.
Only a few investigations are known which deal with angle of attack effects for afterbodies containing a
propulsive jet (e.g. /14/, /15/). Restricted to small angles of attack for avoiding lee side vortices t0e
separation line nas been calculated in ref. 15 by neglecting the angle of attack influence on the invis-
cid external flow. Then, the variation of the separation position in circumferential direction followed
solely from changes in the separation criterion due to corresponding changes in the boundary layer thick-
ness known from the experiments /15/. In the present study the angle of attack influence on the external
inviscid flow is added by linearizing the angle of attack effect with respect to the non-linear axisymnee-
tric solution. This approach uses the well known linearized method of characteristics (see e.g. /16/) and
is justified by restriction to sufficiently small incidence excluding the shedding of lee side vortices.
Knowing the inviscid pressure and Mach number distribution and the three-dimensional boundary layer deve-
lopment, the base pressure and the separation point can be calculated for each meridian plare separately.
Since a base pressure calculation with the aid of the component model requires a lot of empirical para-
meters, which contain more or less uncertainties, the results are somewhat uncertain, too, if no further
detailed experiments become available. Hence, the progress in numerical solutions of the Navier-Stokes
equations suggests the use of these methods (e.g. /17/) in the present context in order to get further
insight into the details of the flow structure. Although such numerical solutions may include some un-
certainties, too, due to deficiencies in the knowledge about the turbulence structure in separated regions,
this way is at present pursued at Dornier for the axisymmnetric case and a first result is included in th
paper.
2. COMPONENT APPROACH
Assuming homentropic flow and prescribing trial values for the base pressure p and the separation location
the external inviscid 'low in tne axisysetric case is computed with the -,doY the method of characteris-
tics for potential supersonic flow. Hence the separation shock (Fig. 1) is represented by an isentropic
compression and lip shock effects cannot occur, which were supposed to cause considerable effects /18/ and
could only be calculated by methods of characteristics for rotational flow. Correspondingly, also the
inviscid interral jet flow is calculated, assuming it to be uniformly conical if not known otherwise. At
the implngee't point of both inviscid boundaries of the separated base region a system of two oblique
shocks is presumed for determining static pressure and initial slipline direction behind the confluence
point.
For cases with angle of attack additional characteristic equations have to be solved for the disturbances
of Mach number and flow direction due to the incidence. Fortunately, the partial differential equation for
the disturbance potential is linear and it exhibits the same characteristics in the physical space as the
quasi-linear equation for the underlying axisyimetric case. Therefore, the numerical procedure is straight-
forward once the axisymmetric solution has been found in an iterative way. If the basc pressure varies
in circumferential direction, the parts of the inviscid flow fields which are influ-nced by p must be
calculated separately for each meridian plane, the axisymmetric solution implicated. In appligations of
linearized characteristics methods discontinuities appear in the solutions .hen those characteristics are
crossed which emanate from contour points of discontinuous curvature or discontinuous slope. Fortunately,
the influence of these discontinuities concentrates on local zones while the global effects remain
negligible. More details can be found in Ref. /19/.
The method imposes two-dimensional shear layers on the three-dimensional inviscid boundaries assuming an
asymptotic error function profile for the velocity distribution in the shear region. The local breadth
of this layer depends on the distance of its origin and on the spread rate parameter, presently used in
the form of Korst and Tripp /20/ as a pure function of Mach number M
d 12 + 2.758 M . (1)
Since the experimental values for o-exhibit large scatter, the influence of that scatter has been studied
in /10/. Because the pressure is presumed to be constant and identicol with the base pressure across the
shear layer, the density profile can be derived from the velocity distribution under the assumption of
Sbeing
unity turbulent Prandtl number. This correlation follows from Crocco's relationship where non-isoenergetic
cases are admitted with the stagnation temperatures in the external flow and in the separated base region
different /21/. Hence the principal treatment of hot exhaust jets is provided in the method /8/, but
presently no use is made of this facility since the poor experimental information on those problems affects
the reliability of corresponding predictions. The location of the velocity profile with respect to the
inviscid boundary is found by fulfilling the momentum conservation law while the mass conservation fixes
the position of the discriminating streamline already mentioned.
If an compression
approaching boundary is present it undergoes a rapid expansion at the separation point or a
rapid in case oflayer
a shock. This event can be calculated with sufficient accuracy by a simple I
2'
.....
..
-A.
for-mula of Nash /22/ for the change of the S-ouentum thickness
with the subscript a denoting the inviscid flow outside of the shear layer. Differences between eq.(2)
and other treatments can be shown to have negligible influence or the oase oressure. Introeuc-g the
initial disturbance by this layer and adding a bleed stream into the base region, the momentum and =ass
conservation laws yield that the position of the discriminating streamline and hence also the velocity on
it depend only on a general bleed coefficient.
q LI 9aia 9al 3
where m iz the mass flow and i the momentum of the bleed stream. The length L consists of the length of
the mixing layer from separation to inpingement and, in addition, of a virtual upstream displacement of the
shear layer origin due to the finite thickness of the initial layer. In that way eq. (3) contains the
classical concepts of 'equivalent bleed" and 'origin shift" vhere the origin shift is presently carried
out by the experimentally well proven procedure of ref. /23/. The present forwulation of C (3) is sllght-
ly different to the original one of the ONERA (see /2/ or /24/) in order to make the recoression crite-
rion formulas independent of e (see '10/).
Many efforts have bean undertaken to model the recompression process properly 12/. From an engineering
point of view the well established OWERA angular criterion is very convenient /6/ because it gives simpie
correlations and contains comprehensive empirical informa3tion. It is based on two experimentally evident
facts for simple backsteps. Firstl), the recompression deflection angle for the restricted case of no
bleed and no boundary layer T cannot be predicted theoretically but must be taken from experiments. For
isoenergetic cases this liniting angle is a pure function of the outer flow Mach number shown in Fig. 2.
It is very close to the corresponding angle calculated from Page's criterion /7/ -n the important Mach
number range (24-Ma 6 4), but differs essentially from Korst's restricted case. The limiting angle is well
approximated by the hyperbola given also in fig. 2 /24/. Secondly, the changes of the actual recomprossion
angle i as function of the general bleed coefficient C (3) are well predicted by the theory of Korst which
gives an analytical exoression for the gadient of iydpending again on the Mach" n-mber alone for isoerer-
getic cases. Therefore, from the first two terms of a Taylor series expansion
CqO
the general bleed coeficient C can be calculated which states by eq. (3) a balancing condition between
approaching boundary layer and 9ase bleed. For a simple backstep the base pressure pa is now iteratively
improved until eq. (3) is fulfilled while for the two layer confluence problem eq. (3) gives an expression
for the flux leaving each shear layer between the discriminating streamiine and the streamline which sepa-
rates the original external mass flux from the entrained mass, the so-calleo J-streamline.
Whereas this procedure signifies an independency of the reattachment pressure development in both shear
layers except with the final pressures being equal, the treatment has been modified in Ref. /I0/with
respect to the expected interference of both layers /10/ In reality already the pressures at the confluen-
ce point must be equal and that comon pressure can be affected by the downstream flow conditions only in
a very restricted manner /25/. Hence, instead of prescribing the reattachment by the inviscid flow direc-
tion behind the trailing shock system, the reattachment pressures - identical with the stagnation pressures
on the discriminating streamlines - are required to be eqtil for both layers implying a curvature of the
slipline between reattachment and final direction. The resulting reattachmeent direction accounts in a first
approximate form for the interference between both reattachnent processes in the confluence problem. This
procedure is called 'modified ONERA criterion" in the following.
The Sum of the entrained mass fluxes leaving both shear layers az the confluence point in addition to the
original external and internal mass streams mst be balanced by an eventually present base bleed
m +m m. (5)
e 1
Also if no real base bleed is present, eq. (5) permits in general a mass stream through the separated
region from one shear layerAo the other. The determination of m or m from eq. (3) requires therefore
an estimation of the corresponding momentum term i. While this tm hal been neglected in /10/, it seems
to be of relevance in the corresponding O ERA calculations /2/ and its effects are therefore investigated
isome detail by the present study.
If both shear layers are treated independently the mass m and the momentum i can be taken from the outflow
-of each layer. This way is denoted by "momentum terms without interference" in the results presented.
I
- _ _ _ _ _ _ _ _
- ~
_ _ _ _ _
i'
However, in the important case of no base bleed also the bleed terms in eq. i3) for one of the shear
layers can b understood as a result of the negative bleed due to the outflowing -ass in the other layer.
/4 1 That method is presently characterized by 'interference eomentEt-E.ms' since it models an interference
influence between both layers in addition to the mass flow interference (S).
Finally, in all cases the solution can -e found by improving the base pressure until eq. 15 is fulfilled.
In tae non-isoenergetic case which means different stagnation temeratures in the external and internal
inviscid flow the solution could be fourd by using an additional iteration loop in order to fulf-ll the
energy balance for the base region. Although Addy's method provides this facili , no attempt has been
made to compute those flows since the ONEPAcriterion has nut been established for such cases ano only
few corresponding experiments are known.
where the subscript 1 denctes quantit-es ismnediately upstream of separation and p means a plateau pressure
developed downstream in the separated zone or the final pressure immediately upstream of the step depending
on the value of the constant C (0.365 or 0.5 respectively'. But eq. (6) did not lead to generally good
agreement with experiments and when recalculating C fro= experiments, a strong variation of this constant
was found ;f the separation point moved within the rearward part of the afterbody /13/ as shown in Fig. 3.
Since the values of C could approximately be reduced to a single curve if they are plotted against the
distance from the base plane divided by the boundary layer thickness, these variations could be assigned
to systematic incipient separation effects and could be approximated by the relation
V(7)
where s is the distance from the base, & the boundary layer thickness ard C. and,- two constants which
approximate the experimental findings best for the values 0.5 and I respectively (Fig. 3).
Since MI and pl in eq. (6) for an arbitrary position of the separation point are known fr- the inviscid
solution, eqs. (61) d (7) relate the base pressure uniquely to the separation position. Hence, in tie
solution procedure the iterative change of the hase pressure could easily be replaced -by an iteration of
the separation point without affecting the rest of the solution pracess.
Figures 4 and 5 contain comparisons of the theoretical base pressure plotted versus the ratio of nozzle
exit Pi to external pressure pa with FFA-experiments /131 where the data of the nozzle boundary are un-
known. The case of Fig. 4 was used in /10/ for co-parison with all variants o' the co-ponent model dis-
cussed in that paper. The best result was obtained by applying the modified ONERA recospression criterion
in rombination with the other components as indicated above. This result is represented by the ;olid line
in Fia. 4. Changing over to the usual ONERA-criterion but including interference momentum tenS (dashed
line) leads to the deterioration of the predictions by a lrge &wunt in tlhedirections of the original
Korst criterion of isentropic recompression up to the final static pressure. How-e-er, introducing the
momentumn
terms without regard to the interference (dashed-dotted line), being identica! with the ONEPA
results given in /2/, the results achieved ere for small pressure ratios similarly good as with the modi-
fied ONERA criter*on while convergence failures occured for hig;4pressure ratios just as in the Addy
method which considerably ,nderpredicts the bese pressure. The converoence failures cnuld be shown to be
due to the fact that the necessary flow deflection in the external inviscid flow at impir Jment exceeds the
maximu, possible deflection angle in an obliqu shoak. Perce, this proble is probably related to the
experimental observation of embedded normal shocks /15/ ,tsultii1 in a flow oattern similar to the well-
known Mach reflection.
Fig. 5 shows a further FFA-case having a higher external Mach numer. Now only the mwmentuz terin without
interference are used and the conclusions are analogous to those for fig. 4. In addition, a result is
presented for the modified ONERA-criterion including momentum terms whicn leads to a further improvement
of the predictions. However, for high pressure ratios essential discrepancies still remain which can possi-
bly be attributed to lip shock effects /18/ being not calculable with the present method because of the
restriction to homentropic flow in the inviscid regions.
Fig. 6 depicts the comparions with an ONERA expc: iment for two different Mach numberii in the nozzle
exit which has a relatively small diameter /2!. Since tne data of the nozzle bounreary layers are known,
they are introduced in the calculations. The ONEF.-criterion including momentum terms without account for
interference leads similarly as the ONERAtheory /2/ to good agreement with the experimental values up to
tne high press-ire ratios of the case Mi = 2 while the modified OMRA shows slightly higher deviations and
the intererence -momentum terms cause large discrepancies, rig. 7 a~ain presents a comparison witI'an
ONERAexperiment which agreed excellently with the ONERA theoretical prediction. /11/, v.e ONERA theory
almost coinciding with the Addy-curve in Fig. 7. The oresent method overpredicts the bNse pressure some-
what, al3o if momentum terms without interference are used in the normal OXERA ceiterion. In order to give
IR-
I ________________________________________R__
Ih
an impre.siCn how the very thin nozzle boundary layer, which is regected in most investigations, affects
the base pressure, corresp -AiPgresults with and uithout internal bovm-ry layer are gi-en in Fig. 7.
Although the internal bo-undar" layer is about an order of magnitude thi.mer than the external one, it in-
fluences considerably the base pressurc.(The complete experimenm"'l -t for this configuration are kn n
from private connunicat-';0-I
Summarizirg these results it may be stated that the mmentum term in the reccrression criter'on are of
obvious imortance. &Atbiecause of doubts en the correct way to introcd-ce them a further uncertainty
arises in the method in addition to those already discussed in /10/. Further-ore, the nozzle boundary
layer appears to be always of reasorable influence also if it is extre-iy tiin. hence, in order to remove
uncertaintics from the prediction methods further thoroughly perform--' experiments are urgently needed
whi;h give detailed insight in the structure of the flow ficld for the two layer confluence problem. With
regards to these aspects also r'-erical solutions of the full tine averaged .iavier-Stokes equations can b;
very helpful.
according to Fig. 8 a good prediction of the separation point is achieved for all boattail angles.of the
FFA-axperiments /13/ i-i the rear part of the afterbody using the modifications of the separation criterion
discussed above. But disagreement takes place in fo-ard positions which is caused by the sensitivity of
the separation criterion to smll base pressure variations and by differences between theoretical and ex-
perimental pressure -nd Mach mn--er distributions on the afterbody due to viscous effects at the body-
boattail junction. ne results of Fig. 8 are achieved using the eirical recompressior coefficients for
conical boattails 19/ which give -specially gd results for the FFA experiments. A change to the more
physical.y based approach described in section 2 Krsens the agreement between theory and experiment, again
revealing the sensitivity of the separation point position to snll changes in the base pressure.
Fig. 9 shows the corresponding results for the coeffictet of the total afterbody drag C, plotted versus
boattail angle fe with the ratio of internal to external static pressure as parameter. The agreement
with the experiments /13/ is cood for small boattail angles ad low pressure ratios where no sepa.'aticn
or only incipient separation is present while the predictions become worse if extended separation regions
occur.
In conclusion these results for boattail separation also establish the need for further detailed experi-
ments andior Havier-Stokes calculations in order to explore the sensitivities showm and to improve the
reliability of the predictions.
First, for verification of the results from the linearized characteristics m.thod a calculated boattail
pressure distribution is cr-pared with a corresponding exoeriment /14/ in Fig. 10. In three meridian planes
which can be identified easily with the aid of the princip.l sketch given in this figure, the differences
between the pressure coefficient of the angle of attack case and the corrzspording axisymmetric case is
depicted versus the distance from the body-boattail junction. The results show clearly the principal ad-
vantage of linearized characteristics method ("present method', compared with thc usual linearization for
supersonic flow ('linearized theory'), which means a linearization based on the undisturbed parallel flow.
The present method represents much better the tendency of the measured pressure_ than the value: fr
linearized theory which are constants in each meridian plane. Nvertheless, in the rear part of the after-
body the dgreement becomes obviously poor because of boundary layer separation. But one must not be con-
cerned on the following comparisons with FFA experiments from this point of view since the corresponding
afterbodies extend only to one caliber behind the body-boattail junction. Powever, other discrepancies
appeared in these measurements as already pointed out in Ref. /13/.
The boundary layer data needed for the separation prediction are taken from the experiments at angle of
attack /15/ but these data are available only immediately upstream of the afterbody. Fig. 11 shows the
calculated pressure distributions on an annular base of a conical afterbody comared with FFA-experiments
/I5/for different ratios of jet pressure to external pressure. In general the experimental trend is
approximated satisfactorily by the theory. For cases without separation on the boattail surface
(pi/P -I) the theoretical base pressure distribution compares well with the experimental pressurc aistri-
bution at the end of the afterbodyand the agreelent between theory and experiment is better by use of the
empirical recopression coefficients /8/ than by use of the more physically based approach ("fundamental
influence parameters'). This inaccuracy of the last ..entioned and above discussed approach is probably due
to the unknown changes of the boundary layer thickness Over the boattail surface. For higher pressure
ratios only the recompression coefficients are used because of the worse separation prediction by use of
the more physically based approach as mentioned above. The predicted base p-essure distribution exhibits
little sensitivity to changes in the separation criterion as can be seen by comparison of the calculations
using the real local pressures and Pach numbers with those based on the underlying axisymetric data.
In contrast to the base pressure behavior the position of the separation line is considerably sensitive to
such changes in the separation criterion (Fig. 12), especially in the forwar' part of the afterbody.
Fig. 13 shows a comparison for the inclination of the plane going through the outermost leeside and wind-
ward side points with the measured inclirtion and with the simplified calculations ty neglecting the de-
viations from the axisymetric case in the external inviscid flow 1IS/. Obviously, the present method gives
worse agreement for the higher pressure ratios but for a final judgement further details on the experl-
mental separation line are needed.
aM
_7
In conclusion. for the angle of attack case tne reliabilIity of tepredii tions suffers evern ore than in
zhe axisy etric case iron the lack of detai led experimerwa information. avier-Stokes solutions couOl not
,e. be rrovided because of the high coanputational effort needed for fully tree-dii.ensional calculations.
5. CONCLUSIOeiS
The component approach in tr.e present forma allows to account for initial boundary laoyers and leads witt tie
aid of the modified IEPA angular .riterion to reasonable predictions of base pressures in spite of a few
;,remeters and components remaaning still somewhat uncertain. 'he separation point on boattail surfaces
can be satisfactorily determined for the incipient separation range. For small angles of attack the calcu-
*ated circumerential behavior of the base pressure is in agreement with experiments while the calculated
separation lines sem to be less accurate. in order to improve the prediction method, additionl detailed
experiments are needed with respect to the confluence of two sheer layers, to the separation on boattails.
and to the angle of attack effects. Corresponding numerical solutions of the foil lNaiier-Stokes equations
can, be very helpful, too, although then other uncertainties may arise from turbulence modelling in sepe-
rated regions.
6, REFERENCES
Q'~WIT.RA Optimum afterbody considerations for supersonic flow with a centered propulsive
nozzle including the effects of the flow separation.
FFATH Atl-863, 1971
/2/ DCLERY, J Ecoulements de culot.
SIRIEIX, M In:AGARD Lecture Series No. 98 "Missile Aerodynazmcs*, 19-9
131 KORSI, H.H. A theory for base pressure in transonic and supersonic flow. .ou:-nal of Arplied
Mechanics, Vol. 23, pp. 593-600, 1956
/4/ CHAPMAN, 0.11. Investigatior on. separated flows ir-
supersonic and subsonic streams with emphasis
KUENN. D.M. on tine effect of transition.
LARSON, H.K. 'IACA Rep. 1356, 1958
151 KIRK, F.11. An approximate theory of base pressure in two-dimensional flow at superi.onlc speeds.
.'tAE-_TI
Aero 2377, 1959
/1 CARRIERE,P. Facteurs d' nfluence do recollement d'un 6coulee'ent siupersonique.
SIRIEIX,N. Proc. 10th nt. Cong. of Applied hiechanics, Stresa, Italy, 19603
/7Y PAGE, R.N.i Reattachmient of two-dimensional supersonic turbulent flows.
KESSLER, T.J. ASK-EPaper 67-FE 20, 1967
i.'LL, W.6 Jr.
181 ADI. A.1. -AStudy of fle-w separztion in the base region jnd its effect during pomwered
KOAST, h. fght
RIaiTE R.A. 1973
6GROC-124,
/9/ R.A.
WHI1TE, An irgproved axperirrntal-theoretical nase P-essure correlatio-n for conical- and
A.L.
ADDOY. cylindrical at ter-odies with centered propulsive nozzle.
- rAUELJ. Proc. 10th lmt. Sm si=on Space Technology end Scien-e, Tokyo, Japan,1973
U01 ILASEr, B. Su;ersonic base flow problem in preserxe of an exhaust jet.
,., 4A--3ournal , Vol. 18. pp. 876-882, 198W
1 SQiMIX, M. techercnes exp~rinnles tcndazentailes Sur les Ecouieeents sebaris at ap-lica
f'rqY, J. tos
'I krn3A, 3. COIERATP .10. 620,I 1967
/121 V'XOSKC. 6.E. Turoulent ooundtry-layer separation in frt of a forberd facing Step.
Ma-ora 5, op. !7t-li55 '96,
U-
-- -- LiR e _ _
F!G.~ ~ ~~ 1:C311*!t-_aMU
IX,
J A
3,~e St r_ __
A. M _________
ONEA (9 Mbifi 0NK /9~.
MMUM-(ML4*AHp
1-57.3d cs0
Tes rotigrato.
.4l
i -0 ii=3_ _
/,y v-
6f ~
ISIR
*lG
tq 1 4,1
___ 0.3_ o
amI- ft;edtq...- rr ____ ~ ~
t
G. - eatti.te Asl
ir"
4:BsjrsueRst 2!
letIm~ -to
TII
at rr
l-. .. - . . .. -- .... 4 DA
Y 0
2o -o mm
- I 04R~ |
930 •4
1.1 '4.0 " 03
l/.0 0 -
e 9oN* - -9 11111,
1 9OMPA
I 13090VIA
1 H
9439994093 09993
I msI
l 003 39
g .0
I 9! Oel
.° 0'
_, -- - 333900390
ll1 (0(99l I/Jfb) I, ',aj
_",, = . - (I(N
,1!__?. , 02"
Vj/".MI
0 MI/
,,. ,_
"I-OUR
.! -0MI-
1""
1"1M'1:
- ....
- Test Configuration .. .I.9 ,A ~.'-.z,,0J
3:.0-' - -J - -
: or :0'M'
I 4 0i/p-
Mis{5
'
Axisynaaetrk ver Conica
31X1IET oFlowo
~ .. 1 l
"-T Y , F,6-,9"
.... " ...
.
11SEPARATION
. ( o 1
20. pessue
IG.1:
ditriutins
Bae
angl
de t
of
tta
~ -4-AXISMK:RI MACH
NUMEIER
7 SEPARAION1
14~I CRITERICII i
OFULL
OCAL MCH
K14ER
INSEPARAPiWi
CRllERION
/ I
-SARA71IINE olg. I \1
0
2
I1/R .De: 8
~ I f 0 &Exrient AgrelI/Wtte0'
x Calculation
-
76 6 ;5o 01 9 97
M.0 a_2.o.0;r... u/
7550D-
0 --- 0 4
0.0 0.0 6.0
7z __
16.11
DISCUSSION
Do you have an idea as to whether your mrthod can be used with a hot jet efflux, auid do yott know of or plan to
carry out any new experiments using a hot jet efflux?
Author's Reply
We do not intend to carry out such experiments, but we appreciate if research establishments would undertake
datailed measurement; on th::t subject. In principle, the method iscapable to treat hot gas effects similarly as
Addy's method. But at present. the ON13RA recompression criterion accounts only for isoenergetic cases (equal
stageration temperatures in the jet and in the external stream). Hence, we need fsurthuer experimental information
in order to make the predictions reliable for the case of hot~ets.
.J.Berard
L'auteur peut-it preciser si, Ason avis, le mod~le sueculot qu'il a d6velopp Iest capable de traiter le cas des jets
rdactifs? en parliculier, dans le cas de-, -noteurs-fusde 5 propergols solides, les gaL de combustion sont ruducteurs, et
la pr6diction de la post-combustion eat tin probldme important.
Author's Reply
I think that in principal, it could be introduced, but we don't know details of the hot gas effect on the recompres-
sion process.
z7
'WI
BASE AND EXTERNAL BURNING FOR PROPULSION i--I
by
Warren C. Strahle
Regents' Professor
and
3ames E. Hubbartt
Professor
School of Aero-pace Engineering
Georgia Institute of Technology
Atlanta, GA 30332
USA
SUMMARY
Experimental studies are reported upon concerning base and external burning behind an axisymmetric bluff base
body in Mach 3 flight. Pure hydrogen and hydrogen diluted with C0 2 , He and N2 were the fuels. A variety of base
configurations were used and external burning used radial j'.t injection with both subsonic and supersonic jets. Some wind
tunne! interference problems arose which limited consideration of the results to base pressure elevation to free stream
Static pressure. A maximum performance condition was identified as a line in Isp - base pressure rise space which was
achievable with all configurations and combinations of base and external burning. The major determinant of ISP
performance is the total heat dump rate into the wake, and only minor differences in performance result from
configuration and fuel type details.
INTRODUCTION
There are severdl conceivable missions for bodies in supersonic flight where drag reduction would be beneficial.
One method of achieving such drag modification is by external or base burning or a combination thereof. Various spurts of
attention have bee-t paid to these techniques, with the latest concerted effort having taken place recently.
I
In external burning the idea is to inject a fuel peripheraily from a body in order to place it into the more-or-less
inviscid stream external to the near wake. The compression waves from the combustion in the supersonic flow would then
impinge on the largely subsonic near wake and elevate the base pressure. In pure base burning the injection take- place
into the subsonic near wake and combustion elevates the base pressure by increasing the stagnation temperature of the
dividing streamline. This in turn destroys the capability of the fluid to sustain a reattach,.,ent pressure rise and the base
pressure tends toward the freestream static pressure. In pure external burning there is, in principle, no such constraint on
the base pressure and levels above free stream static pressure could be expected.
(l)
Analysis has revealed that with pure external burning the specific impulse performance is quite poor for large
(2) (3)
base pressure rises. This has been confirmed experimentaly. However, analys;s and experimentation(2 )have
determined that a) for low drag reduction pure base burning is -fficient and b) the combination of base burning and
external burning may be attractive for moderate drag reduction.
Through a sequence of programs, starting with cold flow manipulation of the near wake and progressing through
burning tests, this laboratory has gathered substantial data on external/base burning phenomena. The purpose here is to
draw this work together and attempt any generalizations which may be made.
TEST FACILITY
The blowdown-type test facility was designed to simulate the base flow for an axisymmetric body at Mach 3 with a
bluff base. The Reynolds number, based on the base diameter is 3 x 10'. The test section is shown in Fig. I. The hollow
cylindrical model is supported in the ducting upstream of the nozzle where the Mach number is 0.07. This virtually
eliminates support effects. Gases for base injection and instrumentation leads are ducted into the model through the four
support struts. Base pressure is evaluated as the average of five pressures measured on the base plane. The tunnel flow is
not heated and the stagnation temperature drifts downward from about 483K to 433K during a typical run. Several
pressures are measured on the model and tunnel surfaces to ascertain that flow conditions are repeated from tet to test.
•-A.
A
A computer based data acquisition ystem controls testing and data retreival.
i7 ~-
From time-to-time a traversing probe is inserted into the wake to measure pitot and static pressures wnd stagnation
temperature. This is especiaily crucial when wind tunnel interference effects are to be studied. In this regard, noting Fig.
1, any reflected wave frurn the base lip will reflect from the tunnel wall and rentersect the wakc at 9.3 base radii
downstream of the base. If the wake it not fully supersonic by this station, it is possible (aid likely) that the bas- pressure
will be altered by this interference eliect.
The base configuratio.s studied are shown in Figs. 2 and 3. Wih peripheral injection for pxteiial burning both
supersonic and subsonic jets were used. Moreover, bases with and without a circumferenflal channel (for flameholding)
were used. For pure base burning, a simple porous base plate was used. Cne other base plane, not shown, injected the
hyerogen rafially through a circumferential slot, since this was a configuration tested in Ref. (3) at Mach 2.1*.
For foreign gas dilution of tle hyorog n the facility is equipped to mix CO 2 , He or N2 with the H2 upstream of the
injection positions. This dilution was used to investigate the effect of injectant molecular weight and effective fuel
heating value (hRp f
RESULTS
The specific impulse (ISP ) is defined as the change ir. base pressure with and without ir jection times the base area
divided by the injectant flow rate. That is, it is the drag ciange divided by fuel flow rate. The base pressure, Pb ' and the
free stream static pressure, p,, are determinants of base drag. If FbZ p, base drag has been eliminated.
For pure base burning, cold H2 injection, radial slot injection and subsonic radial jet injection the results are
summarized in Fig. 4. These results are all for pure H2 . Except for the radial jet injection, a unique ISP vs injection rate
appears on Fig. 4. The injection parameter, I, is equal to the injectant flow rate divided by the .ree streari wind tunnel
air mass flux through an equivalent base area. The slightly poorer performance shown by the radial jet injection is
probably due to some axisymmetries, which ar known to reduce performance.(6)
Perhaps a more illuminating way to present the data is on a plot of Isp vs base force due to burning, as shown in
Fig. 5. Here the maximum performanc. lne of Fig. 4 tecomes a stralght line which has been placed in Fig. 5. On Fig. 3
have been shown only data with supersonic jet injection combined v'ith ba-e burning or for pure base burnit,g. There is
significant scatter on this figure; the lower performance points were visually identificd with poor combustion wit%the
radial jets. But the major point is the following, If _vecything is working well, it doesn't matter if pure base burning or a
combination of external burniag and base buning are used; the perfor-rance can achieve the maximum performance line if
base ;)rce changes are less than or equal to the base drag. The efficiency i4 also impressive, 83% base drag reduction can
I.. .meved at an p of 4000 sec. This is comparable to ramjet eff:ciencies at these conditions and is achieved without
i* let, combustion chamber or nozzle.
Three points lie to the right of the maximum performance line on Fig. 5, and they do so when attempting strong
radial injection to increase the base force level (note the scale break on FiL,, 5). H;gh force levels were achieved here but
probing of the wako indicates wind tunnel interference;, the wake still contained subsonic rrtions of flow at 9.5 radii
downstream of the base. Since basc pressure on these runs was above free stream st-tic, the wave emanating from the
base lip is a compression waye and undaubtedly increased the base pressure upon reflection and intersection will :be
wake. Cons.-quently, these points must he regarded as suspect. In this regard, Ref. (2) hid a free jet boundar, as the wave
reflection surface and the opposite effect may have occurred, causing an underestimate of performance.
For the case of base pressure les- than free stream static pressure, the wive emanating from the base lip is an
expansion wave, and, if interference occurs, it is argued that performance is degraded. Hence, the results here are
believed conservative at low base force levels.
Turning now to the effect of diluents, Fig. 6 presents the results at force levels correspoding to base drag
reduction. It is found that by and large. if ISP is scaled by the ia"erse e hRp , the maximum performance line again
represents all data. Them' is some scatter, especla~ly at the mir..drag reduiction j oint, but there is a remarkable collapse
of the data. With dilution the molecular weight, density, flame temperature r.-n turbulent mixing rates all must change.
There is also a visible change in flame location. Nevertheless, !he prima-Y deter-,'-ant of performance appears to be the
IK
Ml
__ - c-~-
A7.
1-
total heat dump rate into the walv:.
Tempetature traverses on the wake centerline are shown Pi Fig. 7. it has always been found up to 3 radii
downstream of the base that the mixture must iie air-rich. Interestingly, the temperature differences wi~th dilution cannot
be accounted for by the diluti-y effect alone. The mixing rates must be altered to account for thz differences in the
temperature with the C-12 Ciuton on Fig. 7.
Measuring temoeature and p-tot-static pressures on the wake centerline allows a Mach number estimate to be
made, as shown in Fig. S. Recal, U~ the Mach number does not exceed utity by 9.5 radii down~stream interference is
probably present, but the results should be conservative. On the basis of Fig. 8 it appears that the results presented are
interference-free if the base drag reduction is less than 80% and hRPI : 10,000 Btu/ib.
CONCLUDMIG REMARKS
When operating at base force levels below those required to totally eliminate base drag, it appears that either base
burning or combined external/base burning may be made to operate along a unique line in Is - base force space. This will
ozcur for any fuel type it I sp is scaled by the inverse of h .P At high base forre levels the issue of performance and
force level achievable is c:ouded by wind tunnel interference eliects with all available data. Presuming these performance
levels are of interest for some mnissionts, it is imperative to design future high foice level experiments in an interference-
free environment.
REFERENCES
I. Mehita, G. K. and Strahie, W. C.,"Analysis of axially symrnetric external burning propulsion for bluff-base bodies,"
AIAA Paper No 78-26, 1978.
2. Schadow, K. C. and Chiez:a, D. 3., "Experimental investigation of combined base injection and external burning,"
8AIA 3ourna!, 16, 1978, pp. 1084-1087.
3.chez, . A, Blli, F. S. and Favin. S., "Approximate analysis o1 base drag reduction by base and/or external
burring for axisynrnetric supersonic oodles," AIAA Paper No. 80-1258, 1980.
4. Nenle, 0. H., flubbartt, 3. E. and Strahle, W. C., "Mach 3 hydrogen external/base burning," AIAA Paper No. So-
0280, 1980.
5. Townerd, L. H. end Reid, 3., 'Somne effect.s of stable combustion in wakes formed in a supersonic stream,"
Supesonic Flows, Chemical Processes and Radiative Transfer (eds. Olfe and Zahhay), Mac Millian, New York, 19641.
6. Neale, 0. H., H-ubbartz, 3. E., Strahle, W. C.. and Wilson, W. W., "Effects of External Compression on an
Axisymmetric Turbulent Near Wake," AIAA Journal, 16, 1978, pp. 940-947.
ACKNOWLEDGEMENTS
This work was supported by the Air Force Office of Scientific Research under Contract No. FA962CL73-C.C003.
Dr. Leonard H. Caveny is the Programn Monitor.
8Z
17.4
ACCESS PORTS(8)______
SETIN zA AC
C35.1-utN
310O
SUO 21C.6
SECTION
CO10-1BL IGNITEMSTIN
RADIAL
INJECTIO
BASE INJECTION
E~fl.
JET ORIFICES (Six)
-pw
17-6
12000
100000- COLDHYDROGEN
INJECTION (BASE)
6000
4000'
2000-
0
5 10 15
--- OV..
-ZERO BLEED:.--------
.4 /ZLp;;inbase
L.radaj injection base
5 10 15
Ix Z
Fig-r 4. Specific Impulse and Rase Pressure as a Function
of Injecant Flow Rate.
13 CHANNEL FLAMEHOLDER
a STEP FLAN EfOLDER BASEj BASE
7
S0
DR H1RUST
D BASE INJ3ECTION
DD
DILUENT
0 NONE
IDN2
6 He
v, CO 2)
hRP,
500-
*
-7
17-8
CO2 DILUENT
0. .3000
hp
=1 52,000
200 10,300
w
CC
6 7180901
CO2 DILUENT
0.5- .52.000
DISCUSSION
Commeent by F.Stull
The slides shown do not match the preprint.
Authes Reply
While some of the slides do match, I have shown some later results M:an those in the preprint.
P.Choudhniry
1: seems that with "pre-burnin-4", the peak temperatur, will move toward the body. In view of thc superior drag
reduction with "pre-burning". isn't the spatial rate of heat release rather than thc total heat released more important
for your correlatian? Perhaps this will correlate both prc and after t-uming better.
Author's Reply
I am quite sure that is what ishapper~ing. We are getting combustion of hot gas injection, but it is altering the
structure of the wake so that heat s-ease occurs earlier.
R.Malutl
(a) Avez-votts fa-t des essais avec des combustibles solidecs appropri~s. compte tenu des r6sultats avec
(112 + N2 +-02). Combustibles sp6ciaux?
(b) Pensez-vous obtenir des pressienis culot supiricures i la pression statiquc e'tcmce. en utilisant une conibinaison
d'injection lat~rale + centrale (avec des bonnes impulsion sp~EctiquesPl
Author's Reply
(a) The reason that we diluted the hydrogen with the vauious dilutants was to simulate a solid propellant exhaust
with !ower lieatino value than pure hydrogen. We have been unable to find a solid propellant to put in the
facility that will bumn at the low' pressure canditions without developing, an actual rocket chamber to fire-. So
the answer is no. we have not done any !oud -ropcflant work. I would like to,
(b)I'The second question concerns lateral injection. Ir the supersonic case. we got very good performance. but the
problem is the wind tunnel interfcnt-rcc. I ain pos:'ive that our wind tunnel had broken down when we got
that impressive performance with both :atcrai injection :'nd base injection.
When we inject laterally subsonically. we get exactly the same performance that we got from the bwse injection
and could not raise the pressure above frectream siitic. But. wvehave been ib-v freestream static pressre-
with the supersonic lateral injec~lon.
P.Eraer
You mentioned an t 5000 sec for that system. which has a lower pressure and a lower tetrperature in the
combustor th~an a ranie. Did you compare with the rrnict? H-ow did you define 4
Author's Reply
Ipis defined as the change in base fore~ level divided by total flow rate. At Maech 3 -withthe diluted ftiel on the
same basis that we computed the numbs wev get bct ter lv's than those of a rarniet. The trouble, of cruise, is that
we cant -et ntt thrust. TT t rren thaut the I~ is better than sranjet is t'iat there isno inlet, which is a drag surface
in -. aMjet.
!-6
SERVICE EXPERIENCE WMh THREE GENERATIONS OF RAVJETS Jo--
BY: C F FLETCHER An) D R LANE
ROLLS-ROYCE LIMITED, PO BOX 3, FILTOX, BRI3TOL BS12 7QE
S U.ARY
IN-TROXJCTICOq
2 Thor Ranjet in
wFigure Bloodhivnd Mc
A}
m ~ Figure I Bloodhouand Nk2
2 .. ~ 'srCv
-RE OF BLOOO~ND AND.-SEA DART RMUET PWOV"LISICC Lixs
Bloodh d~
Daring the !ae '940's was
necessary to irovide the range, ecognised 'cty.n'ea.crl wc-d be
that
speed. a':- -derajetcapab'prolsion
flexibil..y for t.e surface-to-a-ir defence -- ssiles be n con template at that
time. Fol.onn-o, early s ccessafu fliht s of ramjet pellel tes vehic'es.
project studies by The t. Bristol Aeroplane Cmpany and Ferreati le to m
evolution of t.he 8l0e-.u-.d land base- anti-aircraft missie a- t'he II"r series
i
of ranjets to propel it. T wee the first produution ramjets i.n Lv -e to :e
offered for the propuliac' 'Supersonic missiles. The Thor engine was desionedj
as a pod ;-nit for exsr--! - ting, with this type of installation the
ramiet reevsa on - f.-n fl-ow of. air_ and the'ot tiPe And exhaust nozzle can
be adequately cooled v the e--ternal airflow. The twrist and steer control mtethod
adopted for the -isstil. sa relativey dt icicdence Ivels o- the e.
Develoonent work star-ed- on bhe Thor 100 series ezs imes, for Bloodhud Wi
1952 with w-VType Test nh .57. Some 300 to 400 missiles were fired &ring
- - -
amm . *-:- 5-
miJue a r t oem - is
}r
i;-
ice3rSa Dart
jiur Sea Dar 3.
The Odin ranjet for he 'Sea Dart missile, -rior 3, was evolved from the require-
meats for a lightweight, cnc moav
edi rann'- strfce-:-air missile -a'ich was
capable of beino verti'cally stoted in-, and rapidly 'unw from, snips. The
ce=.fign-'t-or selected involved the raPe beino built knto tbe main aifra -, as
sho n e The com ical ce-trebo4 supers.nic intake wh-ich conains
equipmen t 4n he nose of' e missile and, unlike Bloodhound, the
';ssile is atesian co.rolled and icidence levels can be relatively hich.
Th stteam -' 'ef t ebo- the subson- a-- 's transferred o a entral
clir-ca' hCtich iS surrounded by missile -umpien and the - nular fuel
tanks. Unlike r m Bloodhn, the buri-d installatiorn of the ramiet combstor
and exhaust mozle within the missile rear airframe made an intenal Air
arrn-in
-ecesca.,.
S iI
Development work started on this engine in 1963 and Type Test was achieved in
1966. The number of development :lights associated with this project was about
180 with 6 of these being nominated for early ramjet propulsion proving. Again
the availability of the High Altitude rest Plant was fundasental in reducg the
.,umber of propulsion firings %hich were required. The Sea Dart missile entered
service in 1975 and has been chosen by the Royal Navy as the nain defence system
for Type 42 destroyers and anti-submarine cruisers, and is also used by the
Argentine Navy.
The Cain engine comprises an air intake, transfer duct, combustor and exhaust
nozzle as well as fuel system components. Combustion is initiated by a single
pyrotechnic igniter. '.he orincipal items inkthe fuel system are a Fuel Air Ratio
Control to meter fuel at the correct rate and a Thrust Control Unit to control
missile speed in accordance with the measurement of air stagnation temperature.
Unlike Bloodhound, the fuel tanks are factory filled and sealed off from the
ram3et fuel system components by a stol- valve until the boost motor has fired and
the missile has left th9 ship. This is termed a 'dry' fuel system.
A chart showing the major milestone dates for the Thor and Odin engines, spanning
almost 30 years, is given on Figure 5.
P T
START T 1TENTRY- APVRXIIA E PROPLSIOSYSTEM
TESTlNO-
SERVICE
LEVELOF4WNT
STATUS
F ld., ROF
FIRINGS
KLIEAILITY
The results of these are summarised on Figure 6 and it is importaut to note that
they relate to Service practice firings with standard Service mai: tained missiles
of varying ?k s with either telemetry or with live warheads andtfuses. They have
covered all areas of the flight envelope. There have been no propulsion system
failures out rf nearly 10C Bioodhcznd Mk 2
0
firings, including one with a complete
missile cold soak temperature cown to -20 C. In addition# about 60 production
Sea Dart missiles have been fired from ships with no propulsion problems. The
100% achieved reliabilities to date of these propulsion systems during Service
firings exceeoed those specified. It should be noted also that an eq.kally
high reliability standard was achieved in the later stages of R and D firings.
Missile requirements dictated that the Thor engine should be .apable of operating
up to high altitudes and at speeds between Mach 2 and Mach 2.6 with a maxinum
flight time in excess of 2 minutes. Furthermore the engine mvst be capablr., with
high reliability, of light-up and maximum thrust develop nent witnin about 4
seconda of launch after long periods of exposure and, since this missile is land-
based, the u.nprotected environment in which it may be deployed ranges from arctic
to tropical conditions with very high humidity levels often being involved. In
addition, the engine and its fuel system components are subject to high loads
during boost acceleration and during manoeuvres in free flight and from vibration
associated with flight and ground transportation. Light-up and0 operation in
ambient tempezatures associated with a land environment, ie -40 C to +520C, at Sea
E Level, are required. These arduous operational requirements are sunmarised on
Figure 7.
L8
S, N
TYPE BASED,
I-AND SURFACE-TO-AIR I ENGINE
PROTECTIN
COVER
EJECTOi
SPEED
RANGE MALH 2 TO2.6 2 L16HT-UP
ENGINE RELIABILITY
ALTITUDE SEALEVEL
TOH1IGO
3 SIMUJLATEDTESTS
LAUNCI
FLIGHTTIME IN EXCESS
OF2 M1;'UTES
PEIOS
EXPOSURE GROUND
EXAUSSTIVE ATSIMUJLATED
TESTING CONDITIONS
FLIGHT
(FORLONG
PEID) WORLDWIDE
5 INERITY INVESTIGAIIOS
MECHANICAL
AIRTEMPERATURE
RANGE
ATSL -40%CTO452%
6 DROP
TESTS
TOSIMUJLATE LOADS
SHOCK
MAXHUMIDITY 90 TO95%AT35%
7 VIBRATION OF CMP"UERS
TESTING
MAXLONGITUDINAL
ACCELERATION IN EXCESS
OF 306
d) Exhaustive ground testing at simulated flight conditions over the whole of the
operational envelope of the missile. This includes combustion stability
investigations as well as thrust and fuel consumption determination over the
e_:tremes of ambient temperature and at representative incidence levels.
- - -~~~~ fWAK
STEADY
TRANeICMACH
LIMITS
ERIC
4TURTESTS
ESTASJSA1
,..~---.
III"AV
RC
7
FLOW
BOOST
Me AIR M StMACE-TO-tAS
91, MM,. SAETO-r si m te ct
ATMSALMTO PEIUMMW16 -
AIR1 *EIII*AT
kWI ATSt. -WCTO1
.30%
T~T e SHIPS
P5AZIE AltF w113 1 11, ,
II~l WIERAT RE
SIi Tt51. MULE I tAIL STAND
fM etlli SPILt
t Alt
t E19MA~
ZI
PAX LNGITDINAL SELING ll& MOM' "
ACtEMTION IN0= OFS SPILL SFMD
DIITL A
ILSIIMtME
l01 IN&
ACEEATIl UPTO
X6 Figure 12 Diagram of Test Cell
T"iiTAIOIi LAN.s. All - High Altitude Test Plant
SPECI.MSIWT 113
FMt SPIUAGE
Although the installed engine is normally stored within the controlled tempera-
ture and humidity environment of the shp's magazine, it is required to withstand
the environment which would result from an air conditioning breakdown as well as
the vibration levels as3ociated with flight anl transport in the ship or by air
ur on land. The very nigh boost accele-ation and that associated with missile
manoeuvre imposes high component loads which had to be taken into account in
design. Additionally, during the short time the missile is on the launcher, the
engine can be subjected to considerable water ingestion and salt spray. Light-up
and0 operation in ambient temperatures associated with a marine environment, (ie
-18 C to +300C at sea level) are required, Because of the potential fire hazard
on board ship as well as the possible contamination of eluctre-ic equipment,
particularly in the buried engine Sea D-Art configuration, leaks from the engine
fuel system could not be tolerated under any circumstances. These requirements
are summarised on Figure 11.
RII
N LIE6114-P
RELIABILITY
EGN
2 SI11LATED
L.W TESTS ALTITUDE EBUSTI|N
LIMITl AD 4
3 ATTI.ILATED FLI-IT CONDITIONS
WIAUSTIIGRO TESTING .i +
4 I ECIUANIC, INVESTIGATOtS
INTEGRITY I W
|
I
5 FIRMKGS
TETHERED )R V
AT VARIOUS
ANGLESOF
6 RAINEROSION
TESTS
ONTICAIRINTAMI / IhCIDECE
7 TOSIULATEHOK LOADS
DR TESTS 1AAA"R--
FACH
FL16HT
Figure 14 Odin Ramjet
8 TESTING
VIBRATION OFMO:NJTS Ground Test Fmvelope
9 OPERATION FUEL
WITHDIRTY Figure L3 Odin in Sea Dart
Development Testing
angles. For Odin, major items in the development test programme are summarised
4o-L. on Figure 13 and include:
a) Engine light-up reliability investigations which were similar to those
carried out on Thor.
b) Launch ano light-up demonstration on a real time transient facility at
extreme ambient temperatures.
c) Fuel control priming investigations using neutron radiography techniques.
d) Exhaustive ground testing at simulated flight conditions over the whole of
the operational envelope of the missile and at incidence, as depicted on
Figure 14. These tests included an investigation into the stability
characteristics of the combustor, particularly at extreme altitudes, and
thrust and fuel consumption measurements..
e) Mechanical integrity tests on simulated 'worst case' trajectories.
f) Tethered firings in which the complete missile and engine are 'flown' in a free
jet facility to determine the vibration levels imposed on missile equipment by
the engine. A test rig for this experiment is shown on Figure 15.
g) Rain erosion tests on the air intake.
h) Drop tests to simulate the shock loads which can result from depth charge
action.
i) Vibration testing of components
j) Operation with dirty fuel
During production, manufacturing quality of the Odin is assured, as it was for Thor,
by Production Quality Tests on engines selected at random at annual intervals.
These tests, which are carried out on the High Altitude Test Plant facility,
comprise:
- An engine thrust check at a specified flight condition
- A demonstration of the engine altitude capability at a specified flight speed
- A mechanical endurance test to a schedule which embraces the most arduous missile
trajectories
6DEPAT~IOS
S
tlI B 3
ISERVICE
ENTRY
INTO CIJYI CA'sRLU
8.1 General
The aim has always been that of achieving standards nf maintenance equivalent to
those of gun ammunitionie no testing. In the days of Bloodhound Mkl however it
was necessary to Lstablish the serviceability of the fuel systemat quarterly
intervals,by running the turbopump in order to check the characteristics of the
fuel flow controls and,in addition,careful priming of the fuel control was
"equired to ensure a reliable light-up. Bloodhound Mk2 progressed to annual
tests of the fuel system,and turbopump ruaning was eliminated because the flol:
characteristics of the engine could be checked from an external fuel supply. In
Sea Dart the fuel tanks are factory filled and only two components in the engine
fuel system need be tested. Currently this occurs at 5 year intervals but
clearance for a longer period is being sought. It is clear therefore that the
reduction in maintenance complexity and the progressive incr.-ase in time between
in-service tests over three generations (see Figure 16) gives confidence that no
tests will be required in future generations of ramjet.
z5
82 Thor Hainte.~angce
CONTO
FWNG EXIESFUELAIR PATIO
~R I
mar_ .r1 -"
Prgu18 raigoOdn'ate nC
_W say
IM I M I
'P0
Figue Mk 17Tbo- inBloohoun
Maneac Feaure
rtTIPLLI 'XS AE
FiueI rwnofOi anear
sa
extreme temperature conditions. The Fuel Air Ratio Control and Thrust Control
units require relatively simple calibration testing every 5 years at a shore
depot and removal of these is via the access panel on the missileas shown on
Figure 18. The decision to test these controls was taken in the absence of firm
evidence of the effect on them of storage in the ship's magazine. Evidence is
gradually being accumulated which is aimed at permitting the time between tests
to be extended to 10 years. Eventually they may. in agreement with the Royal
Navy, be dropped altogether. The reliability demonstrated with the twin pyro-
technic igniters of Thor and early Odin showed that a single igniter unit with a
percussion initiating device would be satisfactory for Sea Dart in service. This
unit requires no testing and currently has a life of 3 to 6 years, depending upon
the environment.
The manpower requirement for the maintenance of Odin is about I man per 200
missiles. Using the same definition as for the Thor engine, analysis has shown
that the Odin defect rate is I per 30 missile years. This improvement is a
direct result of the lessons learned from Bloodhound and the generally more
favourable environmental conditions enjoyed by Sea Dart.
I - MI UYSTUGE4EPEP1EU
AtrIN! WI IJlAO l PTAL
1sjFISSJUE
t P
iX EYiDE
AYXXX mAW,
M~.tiE
UAtL .=ITiMe
L;4Wr kAr-Y2B
2 IE OF -,'ATECu18i
CfVSR15)/,S
PW4HW
CI9"M
I NISH
tEtIPIE XJVVAEsCS 2 WDIUt K.'tsasAZ VEt95
s-AM IS
S -=-2.=[t
iA
51018515 PARTS
ACLEWJTI-N
FFSTCIS
.toI!At SEX AS75
ONC OFA 'D2T'RaL SnSICS j GEEUNXS - -AW
FstAE
Figure 20
'A' :LSI
For the Future
AN; C
The design task is to obtain instant readiness with trouble free operation and
the correct performance levels after long periods of storage and ground handling.
As a result of the experience gained from the Bloodhound and Sea Dart engine
programmes.there has been a steady progression in the understanding if the key
design areas and the precautions which are necessary to achieve high reliability
and the elimination of in-service testing. These are summarised on Figure 19 and
comprise:
a) The use of stainless steel where possible and protective dipped coatings
where necessary to elitinate corrosion during long term service life in all
environmental conditions.
b) The avoidance of electrolytic action between mating parts by careful matcning
of materials.
c) The adoption of a 'sealed' fuel system in which engine co=monents remain dry
until launch.
e) The use of engine blanks while the missile is on the launcher to reduce
environmental effects.
9) Moving parts in the fuel system positioned to reduce the effect of worst
acceleration forces.
h) Utilise boost acceleration and fuel control position relative to the fuel
tanks to assist the fuel system priming process.
i) Design for ground transportation/handling protection to prevent, for example,
sensing probe damage and skin denting.
Fron the foregoing, it should be clear that the design and development of reliable
ranjets depends critically on the early and precise statement of the missile lono
term storage and deplo-ment envirormental conditions.
10 FOR TI UTR 20&9
2 THIS
A~IE(,W1 C~fAy gtIEDu r.
ztriLiMstMSJ
ofr SSO
CCS16TEDMM TRP-W THM-
tF5ATIIS
J, Ihi ,
5 THI Rk-t!T Lr4YXUAtt=
F r, -1~jaa
F igur 21Coclsio
WJET_
: :7,r,
EPERIMENTAL INVESTIGATION 0OF A HYDROCARBON SOLID FUEL RAM4JET
by
Dirk Meinkdhn and JUrgen W. Bergmann
Deutsche Forschungs- und Versuchsanstalt fUr Luft- und Raumfahrt e.V. (DFVLR)
Institut fUr Chemische Antriebe und Verfahrenstechnik
D-7101 Hardthausen-Lampoldshausen (Germany)
Summary
A solid fuel ramjet motor was investigated in a connected-pipe air supply system under
simulated in-flight operating condition By taking into account diffuser losses based
on empirical efficiencies, a set of states of the air upstream of the ram3et combustor
was theoretically derived in order to simulate certain in-flight conditions of a ramjet-
powered missile. The experimental investigation of internal ballistics, combustion effi-
ciency and overall combustor performance covered a range of Mach-numbers between 1.5 and
3.0 for altitudes between sea-level and 6000 m. Polyethylene was chosen as hydrocarbon
fuel in most of the tests. It was found that a suitable fuel regression can be estab-
lished for varying performance requirements resulting from a particular rZnjet mission,
even by providing for trajectory dependent combustion efficiencies.
I. Introduction
Within the family of airbreathing engines, the solid fuel ramjet represents a design of
utmost simplicity due to the lack of any moving parts (like valves, pumps, etc.). It con-
sists basically of an air intake system, a thrust producing exhaust nozzle, and a corbus-
tcr comprising the flameholding device, the ignitor, the combustion chamber containing
the solid fuel, and the afterburn chamber for cnhancement and completion of the combus-
tion.
For all its apparent simplicity, the ramjet engine brings about particular difficulties
in its operation, since the combustion processes strongly depend on the flight conditions.
The operation of a solid fuel ramjet, in particular, depends on the interaction between
the air intake system and the combustor, because the fuel regression as well as the mixing
and burning of the propellants are governed by the physico-chemical laws cf hybrid com-
bustion.
It was our objective to investigate, as a first step, the combustion behavior in a model
ramjet combustor for varied flight data (speed and altitude). The flight data were taken
from certain points of a flight trajectory of a ramjet-propelled missile, which resulted
from assumptions as to thrust and drag under the conditions of a preselected ramjet
mission.
In order to work under realistic conditions for the upstream boundary of the combustor,
the influence of a typical air intake system on the flight data was calculated for us by
the Institut fOr Raumfahrtantriebe of Stuttgart University.
g 7;
- X
!h
air temperature by passing the air flw through a heat reservoir consisting of two elec-
hg-.. tric heat exchanger units of 7 kW each. After passage of a two-way valve, the air flow
then was directed into the actual ramjet combustor. On entering the combustor, the
state of the air flow is defined by a set of three parameters, namely the iir mass flow,
the total temperature, and the total pressure of the air. For a simulation of the se-
lected conditions of the flight trajectory, these parameters were matched to the flight
data which were modified by the calculated influence of the air intake system. It was,
therefore, possible to simulate the ramjet operations in a connected-pipe system and the
application of real diffusers was obviated.
The ramjet motor itself consisted of the flameholder, the ignitor, and a fuel block with
a cylindrical combustion channel, an afterburner, and an exhaust nozzle. It was rposslble
to test ramjet combustors up to mach 2.5 at sea-level air conditions.
The aim of the investigation was
- to derive the fuel regression rate in its dependence
or the air mass flux and the combustion chamber pressure
for polyethylene, which was selected as reference fuel
because of its clean combustion;
- to determine the combustion efficiencies;
- to investigate the effect of the combustion chamber
geometry and the combustion time;
- to investigate different hydrocarbon fuels.
The hydrocarbon fuels under investigation - other than polyethylene (PE) - were poly-
methylmethacrylate cPHMA) and hydroxyterminated polybutadiene (HTPB). Depending on the
initial state of the fuel (fluid, powder), the fuel blocks were prepared by casting or
by compressing.
5. Test Results
5.1 Regression Rates and Combustion Efficierncy for Cast Polyethylene
For these tests,*the geometry of the fuel block was Ycept unchanged.- The total temperarure
of the incoming air was the sae for all -rests t280 X). The combustion time was kept at
60 s, if not otherwise explained.
Fig. 5 shows the influence ol the combustion pressure and the air =ass flow on the PE re-
gression rate. One sees that for chamber pressures between 2 and 1C bar the pressure ef-
fect on fuel regression is not rmuch lower than. the one of the air flow. The influance of
the air mass flux Aa On the regression rate is nearly the saze for different chamber
pressures, as Fig. 6 reveals. Using double'-logarithmic znsordinates to display i as a
function of % I the average slope can be found to lie arcund 0-36. Further analysis as
to the pressure dependence finally brings abo-ut a po-wer la-c* approximation for the PE re-
gression rate of the equaticn
0 3
(11 r = 0.0414 - PC . a0.36
Fig- 7 shows how the approximation curve fits the expermental data.
"'he combustion efficiencies zeasured in these tests are displayed in Fig. Z_ the diagrazz
reveals. - as ezxpected - the strong Influence of the chamber pressure on the efficiencyof
the combcustion. Decrasing efficie!nces for raising air mass flows - as can be extrapolated
frm the- graph - aro, due to the fuel lean OIF ratios used in these tests.
In Figures 9 and 10, the cozmbustion Channel regression is depicted for varied combustiom
time ta and for varied air mass flow.
5.2 Varied llvdrocarbon Fuel
in Table 1, the effect of varying hy-rocarbo- fuel is displayed. In these tests. fuel
blocks of one and the same geonzetry w~ere burned under the sa;me set of conditions.
uel Compostion PC rr
ZYA
S 214
In-Flight Altitde Pc On I
Mach- er
-WN [barl! iFS -114
1.7 J 610 1 .8
3.38
175 16.4
16.2
014
0_.,35
,96.8
1 9i.8
II
AIt
o. ___ _ _ -
i , " I !I -: I |
I i I : ; I
_L_ i i ; ! _I
t i _- j I " -
I i i i ! I i -L
I -s.I-13 + + -- - "
It II " " i i
I + I '
I'=C
- I i _ - -=
01 i i+ __sp burning
.i-I -;
dragcompensagon
by thnsust
0.5
I I
__ _ __ ____ ____ ___
I _ _ __
o __
0.3
__ __ __
0,6 _
_____
Fi-2: Air Massilow resp. Combustion
Pressure Requirement-
' i
8M2, bar
I1= Ia
I~ I Alri
7
IN,
2 x'
-n
MA5 MY
KY Fuelbtock J'CY
__ __ _ _ ._ _i '- .,
02 -- i-
0.23 4_%±8/
mkz525 5 9/s_ I
_I I I ii
0L S
c-] h-31015 5 ,
++++|+ +
0 1 c 3 1. 5 £ 7 8
I a
PC -----
H4-5: Regression Rate as a Rmetion of Cori6SUiOU Pressure gs [barl
0,25
mm
0,20 --- - --
--- 4 ~,5 --
Pc3,5 bar
8 9 10 15 20 M, [91cmI. s] 30
0,25
mm
s TAir 28 0 K
0J20
r 0044 ~, 3. ~0,36
01I5
0,12,
2 3 4 p ,3 . j366 7
TIE--
1)~~r8 - io740±28 9/'s
9 7- -m 0:,525±1S /s
95 rma1415 -15 9/s -
94
- -,3019s
92---
91---t-
90--
89 ------ .
0 1 2 3 4 5 6 7 8 9 10
Pc-~ [bar]
Fig. 8 Combustion Efficiency as a Function
of Combustion PressurE
mm
--- 15s
30- - _____
150 100 150 mm 200 '
-Il'F
50 _ _____
____-00,_10.70
__ ___
575
mm
301_ _ _ __ _ _
Fig. 10: Combustion Channel Regression for Varied Air Massf low
0,3
ms IC* efficiency
=0,97
\Combustion
Stoichometric burning
0,2-
0 0,3 0,6
H/H max -
DISCUSSION
P.Kort ing
Did you observe during your experiments oscillatory combustion, and if so. what are the rest conditions (chaml-er
prussurc, air flow) at which oscillatory combustion occurred?
Author's Reply
Oscillatory combustion hds been observed, however, no unstable low frequency oscillations of large amplities.
P.Boszko
Did you have any problems obtaining ignition of the polyethylene charge? Did you use a conv.,rtionl Pyrotechnic
composition?
Author's Reply
There were no problems at all in igniting the propellant. We used a conveirtional pyrotechnic device and therefore
had no real igniion problem. The problems arose in initiating tile igniting devilce. This is don. CiLctrically and some-
times the wir-es became disconniccte,'
P.Cazin
Pouvez-vous nous pr6-aiser si le taux de rdgression du combustible vanie entre l'avant et l'arri~se de la chambre do
combustion.
Author's Reply
The local fuel regression is a function of air flo..
rate and combution time. An example of the time dependence
of local regression is given in Figure 9. An exam, le of the dep .,dence of local regression on air flow rate is giver
in Figure 10 of the paper. These two figures were not shown in my presentation of the paper.
B.Petit
Quels sont les nombres de Mach des dcoulements aux diffirents endroits do I.- c'ambre de combustion (entrie.
accroche-flamme, avant et arri~e du chargement et en fin de clianibe'.
Author's Reply
Inlet: NM <.2; Flameholder: M < .7; Combustor Chamber: M < .5; Afterburner Chamber: M < .3.
T.Myers
What was the LID ratio of your secondary mixing chamber downstream of the fuel grain?
Author's Reply
Nearly one.
F.Cuilicc
What were the amplitudes and frequencies of the osciMations?
ClEcary
What is the minimum pressure you experiened in the combustion chamber?
Author's Reply
Bet-t.'m approximately'-2 and I5 bar. The lowest pressure was !.8 bar.
_.L A.
TsORN COMBUJSTION
L', DiUTIED
ROCKETS
by
Klaus C. Schadow
Naval Weapons Center
Chini Lake, California 93555, USA
SUIL'IRY
l-editmensional, windowed cor=bustor and water tun.nel tests were performed to study the qualitative
effect of inlet and combustor parameter on mixing and overall performance for solid boren propellant gas
generator ra.jers wit), t,o opposing 45' sia: air inlets. Highest cozbustion efficiencies wure achieved at
lowest air injection momentu (resulting in minim-m mixing rates at the fuel/air intersection point) and
:ow, t fuel Injection momentum (resulting in highest gas-phase combustion temperatures duriag boron particle
ignition because of fuel-rich plume ignition at the fuel injector). Te decrease ef the combution effi-
ciency with increasing air inJeccion momentum may, in part, explain t! e observed combustion efficiency
decrease with Jecreasing ramj-t cortbustor pressure and increasing air-to-fuel ratio, which ct consteat cir
inlet gertry resulted in increasing air injection momentum. Tests with mltiple fuel injectors showed
Optic.
hi'at plum ignltion at the fuel injector was more difficult to achieve at decreasing fuel mass flow
per fuel injector. Toe conclusions of this program are qualitative becaist of lack of detailed data,
however, they provide insight into the importance of the combustor aerodyn.,mics for achieving efficient
boron combustion.
A Area
a/f Air-to-fuel ratio
c* Characteristle velocity
I Le.gth
AV Me"ontum
o Mass flew
P Pressure
V Ve'ocity
W dv.
l'f
a Angle
P. Combustion fficiency
AP Press ire di.fareuce
Subscript
a Air
a,w Air simulated by water
CP Cuomusclnn; gas generator
CS Combusticn; ranjor conbustor
f Fuel
f,w Pel simuleted by I:.er
t, Throat; primary actor
t,P Throat; gas genaator
5 Throat; 4-amjetcombustor
Ac* Difference In charaeteristc velocity
INTRODUCTION
Efficient bare particle combustion in so-id propeiiket gas generator romjets (Fig. 1; (du.ted rocketz)
depends on a iuamer of ceupled paysical add chemical rocesses. These include turbulent MleAig between -ps
air and the ga: senerator plume, which contains boron particles and unreacted gaensou tuels, ignt.ionicom-
busttoc of the gaseous fwls, and ignition/combustion nf the boron part,ciew.
he imporar,-e c the gaseous fuel c' "stion (gas-phase coxbustion) for sron particle ignition/com-
bustion was studied by $cha-'v (Ref. l-'' r coaxial mixing flow field;. It was shown that e.ly as len;
as the gascouq fuels of thu gas genera ide to react with the air before excessive mixing occurred,
the resutring gas-plase combution cc, ;-ereabove 1950k, and therefore hlh enough, to free the
boron particles from thi inhiviring oxi yer.
These texts indfcate that for acnleving nigh boron c- bustlon effici<ncy, in Senaral. ntc requirements
are demarded f.7on the aerodynamics of .e combustor, G' to establish flow regions with near stohebso-tric
gaseous fael/air oxtnre ratios and (2) to initias ga-phase conbL.ticn in them miting rezions. roe
importance ot chess reairemen ior cenbustors -j.h coixiat mixing hag been well established and w-ll be
I
further illustrated in the folowing paragraph. In ths paper, the inr.ight into coaxial eit:ing will be
used to.interpret experiments whic. were performed to Sguina qualltaclre -2nder.stand.g fnr non-coaxial
"il '; in ecmbustor wit, side air inlets. Again. enrhesis 'ill Le pla-ed or the effect of the coustor
aet dynamics on gas-phase ignition and o rail perfecrance.
For coaxial mixing, near stoichiometric mixture ratios :ith highest gas-phase combustion temperatures
always exist in the extreme fore-end of the mixing region Fis. 2. ref. 2). Tnerefore, for achieving h ?h
boron mombustion efficiency, it is nece,.sary to achieve gas-phase ignition at the gas generator nozzle
(fuel injector). This -as demonstrated in test, with solid boron propellants at low renjet combustor
pressure, Fr example, with supersonic gas generator exhaust velocity (fuel injection velocity) gas-pbre
ignition occurred several injector norezle diameters downstream of the fuel injectir after excess air rtrtt
had taken place and the resulting gas-phase combustion temperatures ware below the boron ignition tenersture a -
... 2
-A..
£ COAXIAL
A] R -- -- --
2 FUEL
s
0 SIDE DUMP
FU EL
IN
TOR
CC
-7
Figure 1. Gas Generator R jet.
0
T< 2400 K 1"5°K NWC/BYU
FUEL
(< 1950K). As a result, the boron combustion efficiency was low. However, at subsonic fuel injection
velocity, gas-phase ignition occurred at the fuel injector in the extreme fore-end of the mixing region
where highest gas-phase combustion temperatures exist-.. As a result, boron combustion efficiency was high
(Ref. 4). It must be emphasized that gas-phase ignition at the fuel injector has to be accomplished
without increasing the mixing rate. If gas-phase ignition is achieved at the fuel injector with mixing or
flameholding aids, the mixing rates may be too high and therefore flow regions with near stoichiometric
mixture ratio. might not exist and gas-phase combustion temperatures would be again tocolow (Ref. 4).
In general, coaxial mixing is most favorable to provide initially high gas-phase combustion tempera-
tures for t.ieboron particle ignition process because the air gradually mixes with the fuel. These favora-
bl- temperature conditions can be achieved independently of the fuel injection momentum and aix injection
momentum. On the other hand, coaxial mixing requires long combustor length and is not practical for many
applications such as combustors with side mointed air inletv.
For combustors with side air inlets, mixing at the fuel/ir intersection point (Fig. 1), in addition
to pl,.t ignition at the fuel injector, is critical. Therefore, in addition to the fuel injeetion velocity,
whch is ricical for gas-phase ignition, the air and fuel momentum become important parameters. These
parameters are critical for f:el penetration into the airstream and mixing rates at the fuel/sir inter-
section point. Test conditions may exist, at which mixing rates and cooling rates at the fuel/air inter-
section point are too high, thus resulting in reduced fuel reaction rate.
To gain qualitative insight into non-coaxial mixing, experiments were performed at the Naval Weapons
Center (NWC) using a two-dimensional (2D) windowed combustion chamber with two opposing side air inlets.
The qualitative effects of combustor parameter such as fuel momentm (dv)f, air fuel injection velocity
air
nvf),momentum (fv) , and air inlet geometry on mixing/combustion and overall performance waq studied
by correlating mixing data from water tunnel tests with flame characteisties and combustion efficiency
datL from combustor tests. Experiments were made with n.n-metallized and boron-laden solid propellants.
The majority of the water tunnel tests and the combustion tests using non-zeallized propellants are pub-
lished in Ref. 5. Highest combustion efficiencies in tests with an air injection angle of a - 90' were
4 c..ieved at lowest experimental (irr) or v (to achieve plume ignition at the fuel injector)and lowest
(to achieve good penetration o[ the reacting plume into the airstream).
In this paper, the 2D combustor tests with two opposing 45* side air inlets and solid boron pro-
pellants, as well as additional water tunnel rests, will be discussed. Specific objectives of the teat
program described in this paper were;
(1) to determine combusticn efficiency as a function of independently varied vf, (Ev)f, (iiv), and
fuel injector numbers and as a function of air-to-fuel ratio and ramjet combustor oressure, w'hich included
the dependent variable of (mV)a;
-A.
22-3
(2) to describe flame characteristics in the windoaed ramjet comb~ustor as function of rhe sane para-
mter, In particular to qualitatively esseethe effect of positloo of ploso ignition. which oay occur at
the fuel Injector or nay be delayed to the fuel/air interorotion point. on overall performance;
(3) to describe water flow. nixing characteristics In particular at the "fuel"/"Sir" intersection point
as function of "fuel" and "air" Injection comentum and "air" injection angle; and
(4) to provide qualitative Insight into the effect of vf. viv)I and (CV) on nixing and overall per-
forsancc by correlating the results fron water tunnel and windowed coaustor test..
The variation of the foel injector number in the combustor tests wan performed at constant overall
propellant (fool) ason flow. 11. Therefort. with increasing Injector number. ite 6E per injector was
decreased. Eopcriaocot% in Rcl. 4 showed
6 that optixun plane ignition at the fuol injcctar was more difficult
to achieve with diecccsIng os. his ,-szaling effect. --fifith was related to the interaction between ganeoun
fuels and boron particles doring plunoc igojenu process, Is not yet folly understood.
The 20 design of the water tonnot and conh,.stor wan c'iosco to simplify the flow field and to mak, the
mixlngltosbootion more accessihle for experimental ob~,rvaticw. It =ust be realized, that the 21) flow
field does not fully represent a 30 hoaw In an asisymtrIc combustor with two opposIng side air inlets.
EXPERIMENTAL
SET-E1P
Fig. 3 s'n'~ a scileotic Iraoing of tihe laboratory coabuntor with nolid propellant gas generator and
windowed ramjet conhostor with two opposing side al, inlets. In.the gas generator with a 45% weight boren-
laden solid propellant, a two-throat arrangenent was used to vary vg at connta.it :i. Sonic and subsonicvI
were studied. The numsb"of fuel injectors was varied fro:: I to - and 3. Arrangeornt of ther foel injectors
say be neon from Fig. 3, In the ranjet combostor, easy Variation of the following air Inlet geometries was
achieved.- axial position. L, ; length, L ; width, Ii and angle, a.With the geooetric air Inlet changes,
the air inlet Hath numer varied bet-.on 8.05 < v .68.
During a typical test, the air nuss flow. A as varied between about 2 and 4 lb/s with = constant
at ahoot0.25 lb/s. Loaluation of the tests 4asap rforoed at the lowest ind highest air-to-f~ef ratio,
8/f - 6 /if of ahoot I and 20 corresponding to ar' equivalenrOe ratio, -.. of about 1 and 0.4. The raa]Ot
combust-rressure. P CS, wa varird octueO 25 and 92 psa by ".rying the ranjet conbustor throat diane;er
D With the changes of i , Ps and the air 4ilt area, A , (:IV) varied between 124 and 2910 lb ft/1s.
Ike fuel injectorpressureratio~ P /P no u - about1 -ad14.9, (iv),oetween Z and 1250 lb f t/s,
and the oentue ratio (-.v) 14). etes.16 sod 3.8. T1he (iAc)is thA product of the total 4 and v
of ec Inlet, (hv), the product t ooth -., Sod V, of one injector.' The detailed test renditions tor eacA
tent nay he seen fron Table I, welch givei a isuoary of the combustor tests.
Experiowntai rethods consisted of gas generator and ramjet conhustor pressure measurenents, I-, and
P tnsdhigh, speed color psotography of the flame In the ramjet combustor. For eacn test. t',e colus.ion
efficiency, -,Sc2 (referred to -,In the follutiog) ',as calculated. which 4as based on the P difference or
characteristic selocnry difference, 3c*, with and without combustion. because of t"e hcavy~ardware and
relatively low El signiforant heat losses and therefore low talculated '~even for efficient combustion
oust hr expected.
(' \W.
st 0, -2.45 IN1
30I IN (SCI) .11
45% WT B0O0ON
3-5 INI o 125 IN AIR INLET AREA Z52<A,4<LS HIT
401 IN x US IN AIR INLET LENGTHS 0.3L<A IN
Figure3. 20 RunjetCocbostor.
Fig. 4 shows a schematic drawin i f the water flow tunnel. Aerated water was used to simulate the gas
pentrator exhaust products ("fuel") and clear water to .1lnasitethe air. The coetric "alt" Inlet vari-
ablesare shown In Fig. i and are summaarized for each water tunnel tent in Table 11. "Air" and "fuel"
iontco. (&v) nd
I a~w
6V,.w' were varied by varying the pressure dif ferente at the "air" inlet and "fol"
to -
22.4
F.-- = ec- a
PCP 2b _.2._ UN
()a
Test no. e [ s
If$ l 2 _"__CS
Lot~~ ~ ~sPi n-
Is 1 6.2 756 138 0.18 0.052 .5*.
1 8.3 4.9 813 242 0.30 0.052" 81.1
2& b 3.
2 15
2
3
.9,2 9472470260.0.8
11.3
17.0
77.2
7.5
:.3
929
551
!51
153
0.16
0.28
0.C5
0.058
54.5
49.9
b 3 17.0
9 5.4 862 j 244 0.28 0.059 76.6
4a 1 10.1 3.1 795 130 0.16 0.048 62.2
b 1 15.0 2.2 714 205 0.29 0.050 89.7
5. 2 10.7 2.7 674 125 0.19 0.048 59.4
b 2 15.0 2.1 633 216 0.34 0.052 86.6
68 3 .5.3 2.2 409 240 0.34 0.05A 48.4
b 3 R7 1.8 501 240 0.48 0.056 82.0
78 1 33.9 1 156 342 0.91 0.054 52.4
b 1 20.1 % 127 .6 1.92 0.059 76.5
3271 7.
m 22 10.3
10.4 1 123 125 1.02 0.047 61.2
b 2 14.6 110 224 2.03 0.052 90.6
9a 3 13.0 :1 58 140 2.43 I 0.054 52.6
b 3 17.9 -1 52 241 I 4.63 0.058 78.1
10 3 0.7 3.2 687 594 0.87 0.24 44.4
14.9 2.4 94 985 1.42 0.25 67.1
12.6 I 1,6 592 3.02 0.23 48.4
b 1 17.4 I 188 1070 5.76 0.26 68.0
M 2 5 3.1 802 515 0.64 0.20 54.0
b 2 12.8 2.4 836 859 1.02 0._.1 83.6
13 2 18.0 "1 228 726 3.19 0.17 28.1
14 2 19.1 14.6 971 778 0.80 0.18 25.9
15 2 16.0 6.7 995 778 0.78 0.18 26.2
16 2 13.2 8.2 1220 2980 2.44 0.71 25.0
17 2 1.5.6 -1 315 2870 9.10 0.68 26.3
18. 2 6.9 4.2 1230 2410 0.20 0.092 63 .1
b 2 9.8 3.12 1250 * 408 0.33 0.099 92.4
19. 17.4 108 319 2 97 0.119 '9.9
20.3 1I 127 508 4:01 0.2 78.1
20a14.0 6.1 68 88 0 44 0.113 49.9
14.7 6.3 1027I, 4,17 0:49 0..19 77.5
w - -AIR-
INLET
A..
WATER . _
__
m _____________________
m , nm 3mm
TAB'E~ Water Tunnel Tests.
a~jaw a . A, 1pO .
L I
DEC 201 f 20V~ ~I
lN 1 4 .2' V -
09
ININ I PI w f
24 9. .030.24 10 o1822
2W 45
603W. 5.
3~ 90 3.9 12.329.7 4
4. o
4-9.10.017 5!0. 0 1 19
241 0.0451 201,
_gz
3.65
60 .4 2.3 .4 3.82
The water flow characteristics
shown In Fig. Sa. Th- were recorded with high
inf~rapcso sPeed and still Photograph)-.
Fig.
b. In Thspethe niiayn h ae flow field 'ere ergphastzed One example is
flowscharrhcteircs in~ sketches as shown in
will be described using
To ac hi e v e p r op e r such sketches.
f lo w s i u l a t i o n , s
moantu ratioe (6vusi4 the i i l ar i
waE chieed ' hawe to be aintainedi e s o f g o metry , Re yn o ld s Numnb er i l f C anLt r i
b
number similarity wasusig eat geomsetric dimensions between water tunnel and comu.Co/a a nar-ofe c -t.-i u el t
satisfacor with Re - (inlet and combustor) for
buster. The filr"-t.o--"fuel" 0.5 x 10 in the water both set-ups. The Reynolds
ratio In the water tunnel tunnel and Re -1.5 x1055
was varied between: In the comn-
whi.h
wa 8( (" a* a vI 39.4
whih
i saisfctoy
ws areeentwith the ia/A range
In the combustor with
A w n f aete"air"
Inlet and "ulInetrareas.
wa tnnewws
te The "ar-o-'u
vaidbetween omentum ratio In the
(5a)
197
is ~ 095
I (5b)
TEST RESULTS
Windowed Combustor Tests (Combustion Efficiency and Flame Characteristics)
FiS. 6 shows n for two P le.vels. Generally, higher r:was achieved at about 10 psia than 25 ia.
It should be noted that with Se reasing P,. (iv) increased froo about 2-5 lb ft/s to 750 lb ft/s or v
from 0.055 Ma to 0.16 Fa. T1hetests in Flu. 6 included a wide variation of (w) f which effect on nlwilla
be shown later.
Fig. 7 shows n for twoma/f levels. Generally, higher n. was achieved at about a/f11'- toan a/f-IF.
It should be noted that with the increase in a/f, (&.)
aincreased because of constant air inlet geometry.
Fig. 5 shows r as function of (iv) for wide ranges of P-, a/f. and (&V)_. All tests wert made with
5
one fuel inje::or. Itnyb.ee.ta generally doc-eased' with Increasing tf-v).
Fig. 9 shows n as ftnctito of (*v) for all the tests conducted in this program. It may also be Zetai
that ngenerally decreased with icesg(v)adespite the wide variation of PCS I.(v).PP/
and fuel injector number. S
6 2
Figs. 10 and 11 show rias function of (L-v)and P /P for nearly constant ( v) -of 130 lb ft/s .I
may be 3ean that except for Test 2a, n.was nearf7 indepeade6t of an hs4 t 5p .6v Itfo
P rznge. The color movies showed that for all the tests except Tes[ 1 iig~ IIad12 ln ±nt
occ urred at the fuel injector and good plume penetration into the airstream was aciiievedw In Test 2a,
however, plume ignition did not occur at the fuel Injector, but was delayed to the fuel/air Intersection
point. This delayed plume ignition may account for the lower niin Test 2a t:.en Tests Ia and 4a at the same
~.
(fv) The delay in plume Igniticn in.Test 2a may be t.icresult of the combination cf high (--v)
f or P&F.JPC
and low af per one fuel injector.
10
110 ~ ~<2TISo6)'-FT/S 2
13. -OIS
LZ
4-1
Z'1
FUEL IMUCTCR
4<
.<90 MSA 2
M0 < f&Af < 1000 LS-FT/S
J42
0.4! ma*
0 *10.
-
I 10b
00 116
*
(16). FT-LB/SZ 2.
2
L00 <FT/S <IZOLBFTS
171
0
2a
igr 9-*~<
(Cmbne
Teets).~rp.
fj
MT/S
e-y S i neto
2
JBFT
Icbsl
20 0 60.0 00 120 10
2
LBS
<FT/S MLaFT
45 < . < o pm
0 1 FEL IMCTO
A.I
1. 0
2 FUELINJCTOR
2FUEL
0 81T L-TORS
0.21-l L26LI
which
prbal refect the; effec of a tanl
Fig.
i shws te ef eo fuel ic-utiornune
E foreny t s Fre l In evls whch orepnd o he
laffset
'% fe ijectOR. Itr P 2cplayb that
seen fo Figbo 12 Thdr.5 werase
notn signiintl
(*v).Teclr=issoe occurred at the fuel
ignition hfpluzie faot6 Injector.
era it T~Pt thu l jow
e he
OEnbi , a
edofoevetor the fuel/i point
ndicesdwe interection ih gwos ivto
incrase G-, fr iiepnta 3.
oint eth. ain Tssaivdn2 a l of1tbis pertsinjetor an 3 igs 14-005bsprijco. as Ir-
so ' ealnd a inict that, IF
at the sP plume
.optinongua
wsaheeignition at the fuel i njector ws
Fi. 3rhwse asev t por i
cresngo (m 4.v, or.ndg terefoly decreased . Thhnceasn=-orens ora2to,
3wfeich ~ probably
a eetopo-v ulair
duryrfecste ) ing.
F...,
14 showse
snt ahstonffect
of Injco
(fueonl~ everl, dicresds
gubrfrtreFI thincresn
. I
1 1009
i J
j I
nl
-0 2 8 93 1
FUEL IMCIRN0E II
The 060) cr P 0 ,/P had no effect on - as long as plume Ignition occurred at the fuel injector. TheI
rdecreased with increasfing (jv), when plume Ignition was delayed dow.r.strea= to the fuel/air Intersection
point at low 6for low Pim.
The color novics indicated that giund fuel penetration into the air was achieved Independent of (--v)-
it appears that the 4531air injection angle was favorable to zaintain good fuel1 penetration vizh decreasing
(Aiv),, which nay also be seen from the following water tunnel test results. f'w
Water Tunnel-Testsj
Figs. 15 and 16 show the effect of (iv) I fv on the utming characteristics, In particular, atf
the "fuel"/"air" intersection points. At (v fiv, -0.5 (Fig. 15). at which highest -,was achieved
in the combustor tests, mi'xing rates were vsby lower than at (--v)
a /(&-) -.1.82 (Fi-g. 16). The
increase in nixing rates in the water tunel tests with Incre"asing I() -ay explain the- observed
n-decrease in the combustor tests wi.-hIncreasing (mv).
The effect of a on "fuel" Denetration Into the air for (me EIC ) -- 4 is shown in FIg. 17. It
nay
be seen, that at a 9O,/ sdinificant avon nt of "f:l' as reflected the"air" stream rackinto
the ples region. FrSA"herc, it flowed downst-re a on both sides of the air let along the t nel wall.
These flow patterns indicate poor "fue" p enetration into t heaie siren at a 9dpe At the sate
=t= ersatio. "fuel" penetration was icroved at a - 60bo probably
o uil%44 further ihroved at a
- 4? which was not studied in the watr tunnel.
AtrTnnlTs
.~. -A.
1,7 ES,1
12,7 TEST .
Efficient boron particle combustion depends on a number of interrelated combustor parameters. In the
I following,-primarily.the
urcherrx re. the effct qutlitativeeffectof
0£ will, be
er(Av)
_de Pne~rn (6v)considered
-and-v (or P /Pp9nts
which ) on out-oa
ryw ill i,
bescaling-,effect
discussed.
on n. From the available test results, it is clear that only a qualftative-assessment of the effect-of
each of the-parameters on n can be made because of lack of detailed flow data. For the interpretation of
the test results, the insight, which was previously gained into coaxial mixing, was helpful.
For the-combustor with opposing side air inlets, (iv) appears to be the most critical parameter. The
tests clearly showed that n decreased with increasing (iv)a. The water tunnel tests indicated that the
,-decrease was caused by increasing mixing rates at the fuel/air intersection point which probably resulted
in excess mixing or cooling in the reaction zone. Excess mixing is, in particular, critical for boron-
laden fuels because the high mixing rates may prevent flow regions with high gas-phase combustion tempera-
tures to be established which are required to free the boron particles from the oxide layer.
The demonstrated strong effect of (irv)on r,may, in part, explain the n-decrease with increasing a/f
and decreasing P which at constant air inlet geometry resulted 6 in increasing (6V) a. Although the n-
de.pendence on a/Fand P,, cannot be explained by the effect of ( v) only, it seems clear that highest n at
1eo PC and-iigh a/f e'be achieved at-iinimum (6) ( a.on
A strong effect also of (Ev)f on n was expected because it is critical for both fuel penetration into
the air and plume ignition. However, the (ixv)effect was less than anticipated because good fuel penetra-
tion was achievedoeven at the lowest (mv) , ang optimum plume ignition at the fuel injector was achieved
over a wide ruig of test conditions as hdlscusned in further-detail In the following.
Good-fuel:penetration for al the tests was probably achieved-due to a - 45*. Water tunnel tests
indicated that at-this a- , even at lowest- (&v), , "fuel!' penetrated Intoathe "air" which was not the
case for a
either; thtrefore
6
If
= 90". Inate combustion tests, no'visible effect of (1v) on fuel penetration was noticed
It must be assumed that any significant effect of (6v)f on n was mainly related to the
effect of ( v)f or v. on plumbe ignition.
At low (iv) or subsonic v , plum ignition occurred at the fuel injector and highest n was achieved
at all (VV)a or f levels. WAL plume ignition at the injector, probably highest gas-phase combustion
temperatures were achieved as demonstrated for coaxial mixing-_in Ref. 4. With Increasing (&v)¢, the n
remained constant when plume ignition was maintained at the optimum position at the fuel injeefor. This
was the case for 45 psia < P < 65 psia with 6f > 0.25 lb/s per fuel injector. The n decreased with
increasing (im)-, -when the increase in (Av) resulted in delayed plume ignition at fuel/sir intersection
point. With thls ignition delay, probably he plume combustion -temperatures before fuel/air intersection
wore lower-and-the combustion during fuel/air intersection was more adversely effected by the high mixing
rates. Decreasing n with increasing (6v)f wa observed at PCs-25 psia, And at 45 psia < PCS < 65 ps!a in
tests with more than one fuel injector.
The observed plume ignition delay at low Pc and high (iv)f (or high vf) is consistent with the coaxial
mixing tests. It had been shown that the ignition delay, which is probably caused by long reaction rates
and insufficient mixing on the mlecular scale, resulted in low gas-phase combustion temperatures and low
no
The observed n-decresse at high (fiv) and two or three fuel injectors point to the a, scaling effect
in plume combustion discussed earlier. T e results of this program and of Ref. 4 ndicatA that plume
ignition was more difficult to achieve at decreasing ff and could therefore be more adversely effected by
increasing (&v) or vf. This 6 scaling effect, which in Ref. 4 was related to gaseous fuel/particle
interaction durIng plums ignition, has to be considered when more than one fuel injector will be used. The
rn1scaling effect is critical because the gain in improved mixing with multiple injectors may be offset by
less than optimum plume ignition when i is decreased below a critical value..
CONCLUSION
For achieving high n in combustors with two 45* opposing side air inlets, the position of plume ignition
and mixing rates at the ful/air Intersection point were critical. Highest n was achieved at lowest (iv)f
(resulting in-plume Ignition at the fuel injector and therefore in highest gaseous fuel combustion tempera-
tures) and lowest Obv)a (resulting in minimum mixing rates at the fuei/air intersection point). The strong
effect of (&v) on n, may in part, explain the n-decrease with increasing a/f and decreasing P , which at 4
constant air iAlet geometry resulted in increasing (iv) . Teats with multiple fuel injector s'% ed that
optimum plume ignition at the fuel inpccor was more ditficult to achieve at decreasing Uf per in, tor.
The conclusions are qualitative because of lack of derailed flow data, however, they provide insight
into the importance of the combustor aerodynamics for achieving efficient boron combustion.
RE2EUN2CES
, K. Schadow. "Boron Combustion Characeristics in Ducted Rockets," C-aomution Science Tech J., Vo. 6.,
o Hay 1972, pp. 107-117.
- S.2. V. Abbott, L. D. Smoot, and K. Schadow. "Direct Mixing and Combustio Efficiency Measurements in
Duted, Particle-Laden Jets," Amer rnsf Aeronuti's and Atrarautice J, Vol 12, 1o. 3, March 1974,
pp. 275-282, AlA, New York.
-I K-
~1 _ _ __ _ _ _ _ __ _ _ _ _
_ z-- 74-~-
1 74
4. K. Schadow. "Fuel-Rich, Particle-Laden Plume Coabustion," Am-w Ir Aernrtics rd Act rzot,. rs .,
t/0Z 13, kd. 12, December 1975. - AW, New York.
-5. K' C. Selsadov aridiD. 3. Chiize. "ater,-Ttine" and Windowed Oimbustor as Tools for Ducted Rocket
Develop nt," pre sented at-the 1981 JAMAF-Proptlsion meating, New Orleans, Louisiana, May 1981,
(publication IJNCLASSIFIED.)
DISCUSSiON
R.Lo
Wouldn't you consider average temperture in the mixing zone, boron particle size, and geometric considerations as
more important parameters influencing combustion efficiency than air injection momentum?
Author's Reply
To achieve high boron combustion efficiency, high temperature zones downstream of the fuel injectors are essential
to free the boron particles from their boron oxide layer. Air injection momentum (air injection velocity) is one of
the major parameters determining the magnitude of the temperatures, and therefore, one of the major parameters
affecting boron combustion efficiency. (Note: The following paper by Mr Besser also points out the importance of
the air injection velocity.)
You showed us a slide where the ratio of the air moi.,entum to fuel momentum varied from .5 to 2.5, and yet your
flame pictures showed an optimum in-between these values.
Author's Reply-
The flame photographs had no relationship to the test data I showed. They were only used as an example. These
flame pictures were done with a non-metallized propellant that had no relationship to the combustion efficiency
results in this presentation.
LNadaud
Quel est le pourcentage de bore et le pourcentage d'oxydant utilisds dans le propergol de votre g6ntrateur de bore.
Author's Reply
The detailed information I cannot give you, but the boron percentage was 46% by weight.
R.Monti
I think that all your data points out that the main parameter is the stay time of the boron particle at high
temperature and so that would explain most of this data. The particle size distribution is also very important. Did
you make any experiments of different particle size?
Author's Reply
The original particle size in the propellant is sub-micron. Of course, we don't know what agglomeration takes place
in the gas generator and so we don't have detailed data on the effect of particle size.
3--
- W-9
---
Mk -- - -
ER _RETMSEAU PODTWO;
?ROPER4GIL SOLIDE POUR STATOl-FUSEE
JI,. BERARm
.- _G.-D0RlATH_- c. PERIJI -;-DIRECTION TEMNIQJE AUTOPROPULSION
INTRODUCTION
Depuis le ddbut des annges 1970, Is Socigtfi Nationale des Poudres et Explosifs (SlIPE) en
collaboration avec l'Office Natiooal d'E~udes et de Recherches Afirospatiales (ONERA). sous contrats
de Ia Direction Technique des Enjine (DMTE), a cherchfi A ddvzalopper des propergols aolides dont In
combustion produisait des gas rfdcteurs. Entre autres utilisations possibles, ceux-ci pouvaicat servir
de cosbustible poor ls propulsion de statorlacteurs me prodoisant pas de funfes. Nous nous proposona
dame le-cadri do cet article-de dfitailler Ia ifabrication de-ce type do propergola e de pr~ciser leora
performances.
Eneffet. dans on premier temps-Is SNPEa2 mis so point des-forcularious 1 base Woun oxydant
zinfral etdin liant hydrocarbone-. Celles-ci de-coulaiet de rocherches exfcotfes dams les laborstoires de
lONERA (3). Ce type de produits qui constitoait Ia premi~re gindiration, a sobi avee succeis des esais
en vol -de missile I carborant solide (1) (vair-tableao 1);
Poor Is troisiime gfofration dont lea propriftfs do'n repr~sentant vent itro dicrixes, Ilefform
a portf plutot- our I'augmentation do la-chaleur de combuatiom dana l'ir,- sans modification de dersiti
(voir tableau 1). Poor sugmenter-clle-ci, an a'est rspidouent rendu compre qua loutilisation des oxydants
mingreux classiques dans Rea-propergols iEtait insuffisamto. Ausai divers oxyisants organiques ont-ftE
alors testis. Ceox-ci donnent en gftfral des performances sopfricures car .1l'Stat put, uls bralent plus
facilemant qua lea oxydants- mingraux. 11 faot dooc; des quantit~s plus faibles de ces prodoirs Snergitiqtes
pour-ontretenir I* pyrolyse do propergol.
Sur cette base, a 9t9 mise au point one odrie do propergols a-viresse do-combstion et
exposant da pressalom diffirents. Pour lVona .os compositions lea tmieox connues, Is gain de performance
par rapport sux preziares ginfiractos, coest-l-dire la chaitur do comustion dana l'oxygfmo acteint 9 1.
Uoe telle formulation a one chaloor de cobstion mesurie dams I'oxygme do 8020E caig at
3
one masse volw -qme de 1060 kg/n .
- -~~ = Par -xtaoudiaire de-theruocaules platiee-platino rhodi& an pent avoir accis, a-is tempfrature -
-Aesurt doe gas. Ce'le-ci i's tablit A 920.KmiV1ro porle propergol I63
PRE
-7-
Par rapport sux propergola classiquss gnanrm des'gss nidocraura, qul sonr coastituls dWon
ozydant-ninirsl (ganiralement In psrthlorsaro d'incnita -et d'un liant hydrocarboni, llutilisarion d'up
ozydant organique 3 canacrire explosif renatbligsroiro-umeo caractirisarion fine do is sicuriti dur-ii-
sation.
Le tableau 3 fournir Ios valeurs do jugesent habitueloes: dams cc, domaiac. to sensibiliri ama-
ada par lutiliarion do cait oxytiars organiquos maet voisina do colic des propergois 3 ozydsnts =xiraux.
Ils soar donc clausfia-ca 1-3.b (OX?).
&Los quslitis d'adbirencc, do Iz cositioc 1603 &ur des-liners at inhibirours do type nBC'r.
PEMl ot PVC oar in ialuts an traction or an peageo. Ues prapriitC-s iftsnijaezi inrUi sariafaisanres
ananrou ls inrsdotpo PM. Par axmple, pour is- liter I'Ml, Is khrg
Is rupture car do
5,4 bars. Cotta valer ar- ri proche di is chari3s uhr do -sa coqosizion IW03-ui eat do 7 bars.-
routes les ruptures des sseages ianl6O3 cat lieu 44£
ges faoa-cchisiro d1ana it proporzal.
co qui confirm Is qualiti do co collar, at Ilaptivude doeces propergols 31wariier ats blocs noulfis-callia
(tableau 3).
--r-ft sidev=-
-- II-
44
do diazzire 3 4 enttfiis d'air. ittnese-einjlanr tint vitesso-de--Vo1-do Hscir2 or tine altitude -do
13 ~
it par tine socigth aiericaine ARC en chanbra do 132 mm.
bans -Ye premier tins, It bloc gin~rateur f-unctionne 3 tine pression zoyanno do 32 bars. at
-- dibite 0,280 Wgs pendnat 47 a. to datit d'air oat do 7,2 !ceje. La richease est do 0,38 (courhes do tr
onfgur a'5).La discbre satoriactour a parfsiii nrzfoncriou6 Ia porte do prosaio roate I tr
los -mancheai !air at 13 fits do In cb re do combustion cam do 14 2, Ye reudoocur do corzhuaritn dfif jo
cOMsE autla rapprt ontre Io di-bit do sazbrGiC et Ie do~bir do gat injecti est do 0,91.
La douiiaow ivaluation a'ost diroulic danas It -cadre -d'un conrrat -do lAFAPL. La SIU'E a lir6
3 ARC 10-chargeta spout fivahtiin -an bang ~et-en
ebmio statotiactotur do 152 =do diamitro. Cinq
ossais cut frE rfalisis avto succis dana cete ojr cunfiguration;- Lsredntaocohsin sont
ribdis (2).
19. Discr~itiond
dijt-dti stator~acmour
Povi los proporgola do la troisiice ginirarion, los produira do combuation des gas do pyrolyse
avoc Pair soar essenriollonent Se"t aux pressiona at tonyfranures habituollos. En effor, tIna soar
conarittiis principalorPar d'azoe er d'oygcne. or d'uzrje de-giazcamtoniqmde or d'oau. En raison do
l'efficacitido In cotniarion dos gas-riducrotirs,-tris pea:-d-hydriearlureafirbrlIs sortent do la chacire
do combusrion, at usx sour trts diEs. Aussi is transparenc do jar ear, mile, trz-3bonne, supfioiuro
* par exczpla an ~milleurs
proergols double base sans fmfi. En effet, dams cc dernier can, Ies ndditifs
balismiques (sets do cur., do plumb) et !a prisence d'inhibiteur riduisen-t ligirtent In transparence.
Sous cc rapport, lea formnulations a0 cc type scant Me s niues awc for~lntiona ati bore gui scat
trig indiscrts.
42.
*
U0LAME 3ELA VITESSE 2C(IMSTIM
La cowiruot dot Yes performaices carntR dicriras ti-deacon nWest que Ilunc des pnssibihitis
ccd riglage des pr.yegolu a.a
iisi gAnrAtion. En fainS-at varier dvrs iisxflerres, daysvarimites
peant We atteintes.
L'utili Iiatio-n d'one qualtS -plus grossef pour I *'nydnrnodif to p-anfandicoar Yen- qualitis
cinitiquos do tmirWU (figure 8). La -imesnc do cotusmion 3 50 bars chute do 62 Z. La sennibiliri do
In vitesse do cmurionI Ia pression devian-t beaucoap plus foible- to rn do risido ent higirn
sqiinitur A cohui-4o Is composirion 1603. 11 ear campris matre 13 et IS 2- L'urilization do vnr-ris
Wirniques doe'xdn n entnhine pas do nodification do- in vitesse do combstnra.
quo I type ;a foscoalation permit cae variation do vitesse do cebustion Von facteur 3 =iz..
__C_ __ 3
-24:1
3C0SCLUSI
Lintfirer de-o coaoizion 160 rildadanv sa flrrs- eboleur de cccuation dona i'osyg ne par
units de nse sea bonnes prop-riM.ts =Ecaniqu-s. sea poaaibilitra do 0odultion imparanres at a
diacration. 511e -a toutefola 'me &ame doei- sid cucbstiic relativenrn per-iftendue. 'me-e naitg
faible, un -tans-adc
t residu ilani. ces-pointa -pv it i6astituiiri hindicaps ja~t certainesaoplicailana.
Eie -fjaitiU bo-~aloain 1 e-aiculier, par diminution ck-taux de rasidu. la-ccposiion
ains anliodoenra-dan-undiveloppetent -expioratoiro act-ellenent. en -cours.
-REFEREUMCES
lire 1601tin
GG 7320 IWO00
2Zma gfinration I=0 6920!z
_7 zfi
h Au7
24-5
IT~~iraruxe
I Viteseift"C~ mkcantqueszLgt
*er m E jc er S=
Surtl
mia, 3. 2.5 6 z6
705 36
50
3.9125
-40
0.5
5
1 2i
20.9
.194
477 Ma
r- '01
2',.86
So 31-1 354.
Ios
f___A -635.7
S
35.8 156
32;
62222
65
15161i 159
1663
j2of
1300 1287.*
1
1 30r
aalinicus 220teen
'Cviu peefsxsesible
Aptitude la
Is Ematin oc r~sei
-(Cire cap Test I ransis) ;(silO Cm" WS) =
~
I0 1o
"WHObiit S Is ClfeIPat d'izflm=tines~
Watae cr5117v - an
-t diaatre critique
-
24-6
f PT - -(as
PlC -3.3
1.6
MCA diec 65
jC A jfuaziet
EN~ f I.MWWHM
7 (1)
000 I .
0 -- i
4 1 - I
rV~lrIOJBE icm 2
Is~~ (TI DEiLVrnVSIO! SPECMflEWE 44
1500T-
1000 EK '
10 Z
5 10 15 ~ A
M0
FIWEE 3
COME &IESS f log I/Rat
R - Vitessm, de tractiona sr
at facto= de rnlation
0-5C
* 0-1
0 .5
-10 0.7
20 j
40--
4. --
Mflrrt- swomL-E
IAm A
flTESSE
I ST~ ~ ~ ~GS ~IA miw C~vs~c.M26)
I I I-
-i-IF
FICUR-
LA GRAMEUH~ ME L OXflMM
Um LA VflZSS E UGS Ex EW)CS
VCI
am,- t e s WS
_ I
3 5.- wI
M2-- P -tz
I m1 =T m OBSU yM
0.3 -
4$ 3
24-10
3i 4- 1_
0, 9
0' 1
0.3
Concffnii yamiWwffaph wbidt shavwed rmombustorenuc i ort pr phBemt.as ;=t Aszsdehvin ninjet
iksition of ahma 5o5 sonds w=s 'rr-- rn you crnnment on this!
SU A
:Problems arising from high e.nergyfuals,_particulhrly hydrogen, impose remarkable changes ininterface
components, geometry and control, of current ramjets and r-arockets design.
Starting from typical missions scenarios, advanced config-trations are established for military~applics-
ti
i .s Variable geometry nozzls and-combustion chambers are considered in the light of their-combined of-
feet with the intake, to give good performance and flexibility.
A numerical eamle of overall design is given for very high flight Mach numbers considering the problem
arising famn enginoeairffaie integration and cryogenic fuel storage. Particular attention is paid to fea-
tures effecting installation of such engines in aircraft.
The idea of using oxygen and hydrogen as propellant is nothing new. It- was put forwa:d for the first
time in 1 903, by a Russian mathematics, Ziolkowsky, and c-nsidered by Oberth in Cermany almost at the
It ia-surprising that the development of this propellant has been " , starting only at the end of
1 958 whin the Pratt andWhitney engine RL 10 was being perficte .ay is attributable to the ex-
treme properties of liquid-hydrogen, its very tow boiling point ad cezy small specific mash, but now the
great advance-in-space resear6h tes rendered high energy propellants ,a6re interesting.
Liquid hydrogen/liquid oxigen engine - Nowadays, it is a ra,,able high- performance rocket engine-capa-
ble of various thrust levels. Ignited on the ground prior to launch, the cluster of three main engines
operates in parallel with the solid rocket boosters during the initial ascenL of the "Space Shuttle Trans-
portation Systems". Each engine operates at a weight LO/LH 2 mixture ratio of 6 to 1 to produce a sea level
thrust of 1 668 kMand a-vacuum thrust of 2 091 kN (25% mare than on t'.e ground). The en-Ines canbe throt-
tled over a thrust range of 65 to 109, which-provides for a high thrust level-during li..-off and the
initial ascent phase but allows thrust to be reduced to limit acceleration to 3 g' s during the final as-
cent phase.
Each engine has fuel and oxidizer preburners that provide hydrogen- rich not gases at approximately 1 030
K. These gases drive thd uel and oxidizer high- piessura turbopumps. The preburner gases pass through
turbines and are directed through a-hot- gas manifold to the main combustion chamber together with liquid
oxygen and burn at approximately 3 590 K. The main chamber operating pressure at rated power level is ap-
proximately 20 700 kPa. The ratio of the area at the aft end of the combustion chamber to the throat area
is 5 to 1. From this area, the nozzle-assembly isdesigned for a 77.5 to I chamber expansion ratio for
thrust efficiency at high altitudes.
The propellant feed system, figure 1, includes low and high pressure turbopumps for the liquid hydrogen
fuel and liquid oiygen oxidizer. Each-low- pressure fuel turbopump is an axial- flow pump driven by a two-
stage turbine, It raises the pressure-of the fluid being applied to the high-pressure fuel pump to preV-
ent cavitation, normally operating at-a speed of 14 000 rpm for increasing the pump pressure from 207 to
I 6O kPa. The turbine is drivefi by gase'ous hydrogen at a n6minal in-let pressure of 29 434 kPa.
The highm pressure fuel turbopump is a thr'ee stage cent ifugal pump driven directly by aat-stage tur-
bine. The latter, in turn, is driven by hot gas supplied by the fuel preburner. This turbopump per-ates at
a nomiAlsp..ed-of 35-000 rpm-nresn the-pressure from I 213 to-42 817 kPa. The nominal t'urieainlet
prsueadtemperature Are 35:605 k a d 961 -K, respectively.
The
presure
lw xidiz r ropump-is an axial flow pump hti rvnb i tg ubn n
- powered by oxidizer propeilit, to maintain s'fficientoinilet pressure to the high pressure 6xidizer pup-
to-prevent cavitation. Turbine-drivaluid-at-30 944,kp is prvided-frm'the ligh pressur-oxidizer. -
-: !:-:
-- tl -pu~ppressu-efrom-_690 .to:2-881=_kp,- _ .. --=:
pump _dscharge. The low pressure 'oxiie pup ominally operatesfat a spe -of 15 panid-increas
Is
thihe spesreoiie t 2 881otisperate- at- 'nominal -speed of-29-.057 -rpm with a turbinceinlat.--
pressure and- teomperature 3IV046ea and81, resetivel y ,The pf-hy resure increase iS 48_- -ror2
pump from 30392
-The
_p t 5 642 kk- -pressure-i.cresses -
31--937
_to ka- .
The dsign cf engines and lsunching-rockets -are- greatly- influenced by the very by4 temperature ,and -sma0l_
seificass o-liquid-hydrogn The mitr ratiloi ofspecial- "morace
becasterisn rciL
consain is selection, its value- meat be determied by n optimization ctdi.Tetrusiniinc
of it can best be aprcae ytefiguere _2j w h ie compaison between- tank sizes -for-tretns~-'
containing- the same propelln _mas ut for three dif7en:mxue ratios.
The ffct of ,osustion-pressure is of course dependigo ete rsue Ldo upfdsttmi
ii of hetwocotitens
_
nd xpcia~yth
~
smllmolcuar~.s
__--
o hyroen
_ _
-=---~-'s - =jM
- La
use- but in both cases the use-of-higher pressures-brings, a- parformance advantage. The main reason for the
pertrac- atg occurrin _k o ii combustion prestore liea in the ability to make use of nozzics
of~ high expansion ratio. -
A comparison of the theoreticaI maximum perfermance-utnder the customary reference conditions (conbus-
tion-pressure 6 890 kPa, expansion-ratio-68, ambient prieurtIO1l,kM) gives.
L02/Mi.kspecific impulse, 391 a; combustion tempairature. 3 000 K; weight mixture ratio, 4.0; bulk dea-
3
airy,-2,746 kIA/m ;
W/karosene: specific impulse, 300 s; combustion temperature, 3 700 K; weight mixture ratio, Z.6; bulk
LO
3
den~iry, 10,003 kNlm ;
K2'~0
4IMOO, specific-impulse, 292 a; combustion temperature,-3 300 K; weight mixture ratio 1.3; bulk
3
density, 11,964 kN/rn .
The small-bulk density of the L0 1 /1Hf2 propellant renders necessary considerable tank volumes and stru.ctur-
a1 loads which are still further increase byte strong thermal insulation required.
Liquid oxygen is 16 times denser than liquid hydrogen,-hence the bulk density of the propellant varies
conideibfyaccodin tothemixture ratio; and op.limi-zation of thir -atio requires careful study for each
missile design.
Dependence of mean bulk density and vacuum specific impulse on mixture ratio is shown in figure 3, as coam-
parisou of liquid hydrogen-with kerosene as rocket fuel, burned with liquid oxygen in each case.
Combustion pressure and mixture ratio depend on cobustion chamber cooling. TheL-best oxygen/fuel ratio
regardless of engine cooling or vehicle design in respect to specific impulse is shown infigure 4; the
optimum is-somewhere around 4 to-1.
Inacalculating the tank volume, and the propellant quantities, corrections ha-e to be made for the
following factors, 5% of the tank volume for the vapour apace above the contents; 1%of the initial con-
tents for the residual liquids remaining at the end of burning period, residual propellant vapour at the
end of the burning period; propellant lost evporation. Densities of propellaiftiqianvporn
the saturated conditions are shown in the following table; -the final columns indicates the percentage re-.
sidual vapour for tat:s initially filled '~* to 1 mass mixtur4 ratio stredlqudaaprsreo
100-kP-.
A.quid bydrogen/air~engine -- The- proipulon system selectee-for thla-discussiai consists of-turbine on-
- -= ines-for-acceleration to Mach 3 toz4-range and convertible scramjets for the--mission scenario. -The trt-
bine-enginies-for the lower part of the boost speed range is based on the advanced-state of- turbojet ,devel-
including the SST en"-natcahuology Mach 3.5-and 4.0.capability with hydrogen
opmcnt~for--tlie Mach 3 fli?.ht
fuel is highly -desirable. The-scraasjet-i's the'~only airg- breathing angina capahle of operation 'from Mach
3;5 to the proposed mission flight sod -Technology evaluations at-present ierfrta h-cine
pmises grat potential, .but that-the-feasibility of prartical operation and performance -remainsa-t-n be
- V' -jThe "Idual-soda" MAS-- Ai eerhHproi eerh Engine -(HRE) is a-imill axisyarseri6, variable-
geomtryenjiie esined to perfor'w fro-m ach,-to-$. In h ulmd siil b ~aino-t~
activa, u-inijector stagas and th-ofstouoedoetem olre duicis -_ s rh igt i ie
a
Th ke feturs o th cocejt Are severl
numer s dcresi~ gs -f ue
area jmutoj-jh~ 4tsyt ! aarotbsrm-odynmiu gemer -requireents _for eihr -suborser-
sonic-combuarion.-
-M- Q:_
_ M
One key to sustained hypersonic flight is an efficient airbreathin& hydrogen - fusled ongins, The integrated
scrmajet engine/airframe has been identified for military applications in the Mach 4 to 12 speed range@ The
hypersonie variable geometry propulsion system consists of multiple engine modules each with a forebody
precompression inlet end plug nozzle exhausting. The axisymmsetric multiple inlet scramjot modules effi-
ciently capture forabody shock waves and compressed airflow. The plug nozzles strve to increase expansion
area and allow the central body, figure 7, to be nearly stream aligned at the design Mach number for maxi-
mum installed thrust performance. The engines ale designed to be integrated fully with a hypersonic air-
frame, accounting properly for all forces acting on the integrated vehicle.
The propulsion system requirements of the low range boost vehicle bring out the interest for the air
augmented rocket. Suitable for high thrust and average specific tmpulse, such engine might be used at
flight speed less than Mach 4.0, with variable geometry.
The air augmented rocket, for the propulsion of boost vehicles, is known as an engine which improves the
propulsion efficiency of the rocket at low flight speeds, and which requires a smaller weight of manufac-
tured structure than other airbreathing engines.
Simple engines with fixed geometry and variable airflow have realized at best about one third of the po-
tential gain in specific impulse due to air augmentation. The failure to gain more is very largely due to
the difficulty of arranging a fixed geometry which allows for expansion of the exhaust stream, after mix-
ing is completed. With a fixed outer cowl, the specific impulse can be improved by arranging for the air
intake diffuser to vary intarnally. A beneficial variation in air duct area might be arranged with compar-
atively simple v.3ving parts operating in the cooler part of the engine. This has a double effect in improv-
ing performance; reduction in diffuser outlet area at high flight Mach number assists in intake matching.
SYSTEM CONCEPT
A satisfactory launch speed might rapidly be reached at low altitude by a turbo- ramjet- rocket multi-
cycle missile.
Such kind of hybrid engine is enough flexible to meet a large variety of cruise conditions with speed
quite higher than Mach 4. Like turbo-ramjet, this propulsion system, figure 7, is arranged with a lot of
inlet and exhaust variable geometries.
A hydrogen-oxygen rocket combustion chamber is used to produce high pressure and temperature steam,
for the count down and takeoff phase, during which the degree of super heat is controlled by the injection
of suitable quantities of liquid water.
The steam flows directly to the nozzle assembly driving a high pressure turbine in a first stage of expan-
sion, while useful energy for propulsion is realized in a central- body variable geometry plug nozzle.
The steam turbine is giving power to a cluster of multi- stage atmospheric air compressors accomodated
into propulsive ducts around the turbo- rocket central body, for a sequence of operations going from cold
to hot jets for take-off and acceleration phase.
The LH2 /LO2 rocket and LH2 /air turbojet propulsion is followed by variable geometry LH 2 /air ramjet and
scramjet operation.
Application of current compressor technology to meet the higher core pressure ratios desired in ad-
vanced engines would result in a large number of stages and blades. To minimize weight, volume and sthges,
high tip speed and high blade loading for high stage pressure ratio are required. It is nuw possible to
design compressors for an overall pressure ratio of 20 to 1 and only eight stages. To achieve this goal,
the pressure ratio as high as 1.8 in the first stage is decreasing to 1.25 in the last one because of the
increasing temperature. The first stage rotor has a design speed of 450 m/s. The hub- tip ratio ranges
from 0.7 at the inlet to 0.95 at the outlet. A compression efficiency of 87 percent is obtainable.
A major objective of fan or compressor for a possible application as in figure 7 is to obtain high pres-
sure ratio per stage with high efficiency. At 540 m/s tip speed, a stage pressure ratio of 2.2 has been
already obtained with an acceptable efficiency of 82 percent, Ref. 1. We are confident that a higher level
of efficiency in this high- speed range can be achieved by further improvement in blade design. With a
two- stage fan, an overall pressure ratio of 2.8 and a peak efficiency at design speed of 85.5, were ob-
tained.
Similarly, a higher core turbine inlet temperature is required for the application as in figure 7.
Elevated temperatures and pressures for a short operation do not require, however, sophisticated cooling
schemes to protect the blades.
Gaseous hydrogen GH2 and oxygen C02 are directly sobtained by water electrolysis at the launch place.
3 3 3
The most part of electrolytic cells are producing 0.2 Nm of CH 2 /kWh (0.0555 Nm /MJ) ane 0.1 Nm of GO2 /kWh
3 3 3
(0.0277 Nm /WJ), i.e., respectively, with specific gravities 0.881 N/Nm and 14.013 N/Nm , 0.04896 N/NJ
and 0.38925 N/NJ. The weight mixture ratio for the combustion is becoming 0.38925/0.04896 - 7.95, corre-
sponding, figure 3, about to the stoichiometric ratio, if all the electrolysis products are used.
The energy need for a modern electrolysis process may be at present time 14.65 NJ for producing 1 N
of GH2 and 7.95 N of G02.Liquefaction (L) may require 3.65 NJ per I N of LH2 and 0.19 NJ per 7.95 N of
L0 2 . Therefore, the total energy need for production and liquefaction, at atmospheric pressure, of I N of
LN 2 and 7.95 N of L02 is about 18.49 NJ, corresponding to 18.49/8.95 - 2.066 KJ per N of stoLchiometrLc
mixture LH2 - L02 .
The combustion temperature at a pressure (20 700 kP) of the order of the main LH2 /LO2 engines of the
"Space Shuttle" may be '3 500 K in the stoichiomstric mixture ratio. The combustion product is of course wa-
ter steam.
Such combustion temperature is obviously tnacceptable for the turbine blades. Considering the short dura-
tion of the LH 2 /L02 engine operatin, a temperature of 2 000 K may be chosen. Pulverized water injection
in the combustion chamber is used for decruasing temperature from 3 500 to 2 000 K.
In order to know approximately the water fraction 0 to be injected through the experimental values of
the specific impulse given in figure 4, the following procediire is applied.
The initial entalpy i, (for: combustion pressure, p, - 20 700 kPa; specific heat ratio, n - 1,25; mixture
ratio, 02/H2- 7.95; dissipation coefficient, v w 0,95), respuctively, for sea level (P2 = 101,3 kPa), and
5 000 m altitude (P2 - 53.9 kPa) at specific impulse (figure 4) 350 and 370 s, are
I w
./g A
.I (1 - (p2 1 )(nl)In/g (1)
The gradually injected water in the combustion chamber is increasing its entalpy from i8 -0.0043 KY/N
(10 kcal/kg) to is - 0.612 KJ/ (1 435 kcal/kg, for To - 2 000 K and p, - 20 700 kPa).
Approximately, disregarding heat exchange dissipation effects, to be considered in the expansion efficien-
cy, the fraction of the GH 2 /GO2 combustion products a, the injected water 8, and the hydrogen y - a/8.95,
it can be written, Ref. 3
a - i ( /i
(2)
a *B-l
from which
B - (i+ - i)l(i
1 is ) (3)
With a step by step procedure, from the ignition to the end of the acceleration phase, an optimized
turbo/rocket- cold-hot turbo/fan propulsion system may be designed, to be followed by the variable geome-
try ramjet and scramjet operation.
The procedure is starting with the evaluation of the weight of L 2 , L02 and water,for the acceleration
phase.
Establishing a mean value of the dissipation coefficient U (taking into account of an approximated equiva-
lence to the real hybrid engine used) and considering i I - is in ta. 1, we get Isp and w from eq. 1, from
which the propellant weight flow C corresponding to the thrust T
C - g T/w (4)
to be used step by step. The propellant weight is obviously depending upon the acceleration phase dura-
tion.
In operation, hydrogen, oxygen and water, are supplied from turbopumps to the combustion chamber where
steam is formed. The steam flows directly to the turbine and thence to the plup nozzle.
The expected overall efficiency of this hybrid boost engine is higher than rocket, because of quite less
mean gas exhaust speed. The propellent consumption is reduced by the high propulsion efficiency of the tur-
bofan operation.
The success of such boost engine combination is depending upon the real possibility of a large fraction of
water injection into the combustion chamber, to have a high work tranfer from the turbile to the compres-
sors.
A numerical approach may be obtained considering acceleration hybrid turbo-rocket as in figure 7with
-0.6 (taking into account che compressor work absorption for the engine equivalence), i1 - is - 0.612
NJ/N. and T - 100 000 N. We have from eq. I
and from eq. 4, for a 5 000 m ascent trajectory at a mean velocity H - 2.0 , a mean weight flow during the
time At - 5 000/2 • 340 - 7,35 a
C - 503 N/s
So, with a propellant weight C • at - 3 697 N, we get, applying eqs. 2 and 3, storage weights and volumes
3
8av * 3 697 - (water) - 961 N - 961/g - 1 000 - 0.098 m
3
2 736 • 7.95/8.95 - (oxygen) - 2 431 u - 2 431/11 199.2 - 0.217 m
i.e., propellant and water for the hybrid acceleration engine require the volume equivalent to a circular
cylinder 0.60 m and 2.67 a in diameter and length.
The advantages of the hydrogen fueled airbreathing ramjet and scrajet, for a flexible cruise flight
to the target, are deriving from the higher energy per mass unit of on-board propellant. In fact, weight
stoichiomtrLc mixture ratio for air/hydrogen is 7.95 - 100/23.14 - 34.356. Air/hydrogen weight mixture
ratio is ranging from 5 to 115, this meaning that a hydrogen fueled engine is very much flexible for speed
variation during the cruise flight.
Compared to the acceleration phase, the on-board fuel volume for cruise is requiring quite less vol-
ume.
Water electrolysis, by nuclear or solar energy, may be the primary candidate for gaseous hydrogen and
oxygen production and derivery through pipelines to the missile launch place, where liquefaction plants
are existing. Temperatures as low s 20 K (LK2 ) and 90 K (10 2 ) require vacuum- jacketed stainless steel
pipe lines for transport from liquefaction plants to the missile launch site. The need to kip pipe lengths
an short as possible is further emphasized in that each length is triplicated. Some of the liquid hydrogen
changes to gaseous state during the fueling operation, and at any other time .hen the temperature and pres-
sure exceed certain conditions. For reason of safety ad fuel economics, the gaseous hydrogen must be col-
lected and returned to the liquefaction plant through a second line. The third line provides redundancy and
can be used either to deliver LK2 or return
Flexible liquefaction plants produce LH2 and LO2 as much as it is used and to refill storage ves-
sels.
Gaseous hydrogen and oxygen are piped from the water electrolyzers to or near the launch site, where
electrical power is provided as needed for liquefaction.
The ground- to- missile connection is following a basic fueling concept: truck with dual vacuum jack-
eted flexible hoses manually connected between the hydrant pint and the single connection to the missile
tanks.
Based on the largest demonstrated hydrogen liquefier capacity (6.2 I/s, 534 kW/day), it has been yet
determined that a single plant having a capacity of 25.75 I/s is possible.
The-t- chnoogyrequired-to-build -reliable fuel- coold- structures -for the- hostile environnt wthin
{the-
- engineibas -been- devalopedi; -
Liquid- hydropjn -is stored -in uwell insulated -tanks, whose pressure- is scheduled to minimize boil -of
lossii. ten scl-aa of- fiidrezen are npotkedl b it that &s~ ene. jetak
-throu -tatyti an&L-ure
the converted to water vapor.
Several aluminua ilois and- stainless steels are applied in lightweight structure for cryogenic liquids
at extrenoly-low temeratures, Ref. 2.
REISREgCS
ill A.1- Harris and D.H. Sul*- "High Loading 1 800 Etlsec Tip Speed. Transonic Cooror Fan Stag" -
121 D. Dim, "Perfor-mance Characteristics of Turbo- Rockers and Tuort- Rmnjets Using High Enargy Fuel",
AGM Conference Proceedings No. 205, Pais, 6-9 September 1 976.
131 D. Dini and P. Psarudekis, "ttilizzazione direta dei prodotti dells elettrolisi dell'acqua per turbo-
geveratore ad alto rendinto", 3* Conresso naxzionale ATI, Viareggio (Italy), 5-9 ctober 1 981.
141 D. Dini and R. L zeretti. "Engine Intake Control Design for Advanced Fighter Aircraft". A.ARD Confer-
ence Proceedings No. 274, Cologne, Germany, 1-2 October 1 979.
- -i ... .
--- Fig. 1 ~Shuttle main engne . 2 for differn Lxure rto.-
lerant~- mi:ato
M
SAX
-
..........
-42.
000
'Lv" ftt-s-..
3D 40
3t"0 ~5j~~
3200s ~n
Pi
. = - DP
va i de c of
a seii ee 1)leo
bu k d si y
ituei
jFi ; 4 - Te es O y e / ue ra o
CW553mo-Wr___ epc t pcfc n
40'
(. -- -
4 P I5 £ i4
_pejjflUjs
jfpr 4b 4
4jvrct
_ I _0 _ _ _
AL0 kes~
DISMISSION
iRio
1 think-Piter Kramer of the University of Stutgan should comment on the overall idea of this type of turborocket
Ii amulet. I would like to make a comment and ask a question concerning the injection of water into the hydroge and
oxygensockq combustor. We are doing precisily this at my institute in DFVLR which we called a hydroge-
oxygen steam generator for push button demand electrical por generation. The problem of getting high droplet
evapoaton efciency become more difficult as you try to water
c into very hi;h temperattie systems as you
propo ri. Any droplets remaining would damage the turborocket machinery which is in the wake ofthe steam
-x- aw My did yeu sug~cst 2000 K in your srudic? Were you proposing to use ceramic blades or what?
I Author's Reply
This idea came out of what yon suggested as astati-- steam plant. A 20000 K temperature we can now
p6tentially achieve with transpiration cooling of the blades for asbort period of time. The problem is to in ect
water in a small volume. The problem is fora stationary power plant, but we may apply it to a rocket too.
Kramer
± You wanted to reach Mach 3.7 within 7 seconds? Whnat acceleration did you have?
Authors Reply
I don'i know the number, but normally in designing a missile, the acceleration phase gives an altitude frow 0 to
10:000 metres and takes about 40 seconds. In order to --t there in 7 seconds. the acceleration would be too igh.
Comment by P.Kramer
The turbomachinery Will not survive.
Author's Reply
I took a temperature of 2000' K. This was used as numerical example to put forth the idea. The 7 seconds is too
ht h an acceleration.
.Kramer
Did you compare it with t.e rocket booster instead ofa tiwborocket booster? You know we are doing similar
studies for Space Applications, and we found that the pure rocket booster balances out the weight disadvantam-e of
the turborocket, so we came out with the same payload without the turborocket, but with the pure simple rocket.
The turborocket saves fuel, but doem't impove payad. That was our result.
Author's Reply
We need to study this more completely, but the main purpose of this kind of propulsion system is to improve the
propellant efficiency. From a payload co.dideration, you are probably right.
i,
....... = _
1
-- - --
=._ _ - -r_= -- _
- -- . . . . . -• + - -= . ... _ - _ -+ : -L -
ABS7RAG
i has teen observed that wh-en a fuel Ar--oplet evacorates In a gh -Mle-L" ft+ c=n-
nent ;yp-cal6 an advanced ran-je;. the usual lin-ear D12-t relt0si da-e od
Inere is a strong evidence that at higher- nressues, equ _ 10
sble tar the sudden break n the relatonshp.ducts of dec on A
tarr- residue which Is practlcaliy nonvolatIle and burns only ve-y slowly.
The present paer deals with the coupled effect of deacc-C=s t--n a- -nd
a typical ' c-' y
he result this study s.ow tha drop!et decop ion at hizrer pren-re
ou,--e
and co rclal' fuel blends has a sl-ifl-cant Influence on "he 'us-_. f
Carbon build-up on the walls of the combustlon chamberend ss"I_ fa--"-
1plutants and varslculate matter In the exhaust.
ImnODJ CTION
in order to deter.-1ne the effect of droplet deco sjtIon and res~due t =n-
the overall perfornmance- of a combustor the h~story, of a heexa-'ane sray' withs rsic
d-on size distributIon has been chosen as an exanple. -7=-- as - d o sprayed
in a 60Ow, 4 at=. air strea travellIng with a velocity of 50 M/s S -b---flow IS
assused to be steady and one-dl=-nslonal with constan therc ta--el
pnopr _2 . --
distribution of drop size typIcal of an arblast at-'-ze- was sue fb V-o th da
shown n Reference (4). The percent of the total population oacu- eah
with a given =-ean subgroup diameter is lsted in Tsble . exadecan sp-ay waS ar r
to follow this distribution.
TAKE I
DISTR WION OF MEN SU-ROUP DIA.aMETE
Percent MRoan Din= ter of Percenit _Me-- X=--=
Population Sub--ou, (=icron) Populatlon Sub--up (-=ero)
f: 1
!5 20 17
A 27 40
-
Co,.e phe, n.a -nf ev_ -"tlon/deo.posl.tl -man --m-er of ca t- s
of th"-e _h-
were analyzed neglecting interpartioular interact on beween- Iff-r nt su-h-a- . in
other words, collIsion between droplets, shatterIng and thIr
tzterlcoalescene we-+ 13M ed
coaecec
..e folowing equations apply for the evpatlondecomp-sltlon -fa-- le- n- a
A one--len.- l gas stream:
s-momentum: r
---
. - u
du -r
dt PL Dj-
---- unpored by an AoR G'a t7 33 5h
-~A-4
I : + _ _ _
__ =+r_
---- _ow _
27-2
8; + 03 _RC.-7r -
- soid- A- rots~u x e~ u
=Z - 3d
a fuel
"I yd
C d
I- U~c orc-
ng jatee
follnowcief '-i-4 1~ -& use fo SOe e--vato onth
the A-~
--ar -~ a ' ~ =
i~ - 8' ~ t a T _- - - fe a"_a
~-d=_sn--_-s et ba
- !
lico d IteO-S-U -l s - ie ---
"so
su grv n r var' =3z= obsv a '~aba1r - e -o-s a-
an "'o
a zh-s =
---:-,4du
- e t =: m
tine te resdu ' bJ""='- ereit gain
s-loly h e ngcreu-1- -n do- -
""p ns'b~ oe (1) t'r - ug o1 ii*"''a
M-e codtos ie ye-nrres~
u si n g~~r
O d ar F n m - ut a -c= - - c u _l _-h
-- W 1 n a r i a y al et a
( R mf l b r - mocf ae = c aI ex a e e-a e w r a- ob -- a i . d f
TT00 IL!E . TJ
I- 80
us~ 60-
l
10
- 80EVAPO-RATION
o 60 CEASES AFTER
a 40t
20 2 HILLISEC
i0 X O L04 x=
- 40
20020 40 KG80-100 020 40 6080 100 0 20 40 6- 0 1OD
-DIAMTER (MICRon)
1003 4 x2
80EAOATn
CEASES ATE
us 409.C
20
T h*_
tb ~~ tl l 11-ra -ea
twon
IUZ A valum o' F =~=-
1- em .e- r w
con c:- Ps -nat~ te c-vosv the sje..oz
llit k*'A-
ow
-= Stfdle of =yoya-
torh r1 -~= th 5-14-41
fo D v=*am=
-lct~ld phase d~~mO'A
- * o ro-esoe
- ~ i to
ted ettl-- o_ -ze- e~ G-
4
A
cont so Je±.ce-
r-4-an --
con'r= ml
acotr rorthetot o t-=o sldue
cc=eare z-Eifl?)
stwe - r rsur
d~needr-rgy. rk-= . Y fea..
yo so~ th %ete f _.__e_--
_Mthe WRM
cae as Of 5(-zn
d -- -d a te u.. 'em
lnze-l _-_ -~ te~piratur~n
r -ae b leqn.Iri ACPer-4 oftelsrt~rt
wor'r Sugest
l~qu~d
flie
0.8 o
Ph_ m 0.6f
case-
of3-
ez- 3=_ e- - -W t- U
rae 0.2 Cm- f _
0 __
1 __ _ __ _ __ _ __ _ __ _ __ _ __ _ __ _ __ _ __ _
0 -e .2
liquibdn.TP
La -- 0ifa O ~ ~ ts
___BL- y;lglc e
- 27-S
1
5~ -era'
testl=-te a
been =-d assualmn=E-U5t' -- rba
cP
- =, ha fo -SxS ------
Is Iu Vutf t-- p-
rmr
M tuj
0-tine-lFfounr
x)
I L
GO 1f)~&-O
ltW"36jrV801
2i0&~r
xI
so4
6:11
X 80
60 -
ii 40
20
000460 80100 0 2040 60 80100 000000
DIAMETER (aICRON)
10 J 2 x =14 x1
I- 80
S60
S 40
~- 20
0
-0204060 80100 0 2040 60 80100 0 20 406080100
DIAMETER (MICRON)
==~-~ al -a- a
-lo=w - 5
~
wi - ~ s de
t4.d-p o ' O are jus ~
a CrlU A-n
na- 4
-u -s a- 'n- a= a' a t
k1cu
-e of in ttim
w
a 100 n rc pa as0 u- s r yr ~ mn n v - o mt l A
or80- i s ci n7w -*- e
0=uvw-- m e
rAA_-no
Z ~ ~ ~VPRAINCAE
~ -e A~6
II
5 6 7E-SE
1 3 4
0p 1 1-
nan~
assgr nw
For tho-omee of F10~ cm3/ghole-sec, the coupl1ing between evaporation and decomposi-
tnisrhewa. Teetre spray evapor -ates nearly compl-et-ely before residues-are
formed.Frtis case,-an i, - combustion-erric-iency of Neal %O ih be- exected.
Figure 7 ii-a plot of the pos e ideal combustion effioiency based upon the amount of
fuel. notualayf~evaj~orated. it is-plotted froml-Fig. 6 'and shows the oombustion efcec
as a fuhctio6Uthe-cbtracteristio length available for evanci~n We ecomposi-
tion rate Is faster tha.. tb0 evapr~r aesintecse of ol an increase i
the characteristtc length would not chnnge the burner performance significantly. The
maximum -possib-le combustion efficiency as a function of the frequency factor is shown
in Fig B.- It fdcii -jattentioA on_ the- ipop -tahnce of'liquid - phase decomposition in.a
burner with~nh-i'estrictlon6ii -the allowable characteristia length.
= ~ 100
Ii-F
2 2
-U-
U.
040
2
-20
(00
00 1
T.-
U
- i
S 20
(0
44
- ~~~--=e-~- A,
M- A
REFERENCES
5. Carnahan, B., Luther. H.A., and Wilkes, J., "Applied Numerical Methods," John
Wiley and Sons, 1969.
6. Obert, E.F., "Internal Combustion Engines and Air Pollution," Intext Educational
Publishers, New York, 1968.
7. Mansour, M., "The Vaporization Behavior of a Fuel Drop on a Hot Surface," Ph.D.
Dissertation, University of Southern California, Los Angeles, June 1977.
8. Gerstein, M. and Choudhury, P.Roy, "Liquid Phase Decomposition: A Possible
Problem with Fuels in High Pressure Systems," Proceedings of the 1980 Heat
Transfer and Fluid Mechanics Institute, Stanford UniversIty Press, pp. 79-91.
DISCUSSION
R.Monti
I think that the "sudden freezing" kind of model introduced by the authors is valuable and very proper as a first
approach to the probim. However, I think that a more accurate model should take into account the thickness of
the layer of decomposed liquid and relate it to the evaporation rate.
Author's Reply
The question raised by Professor Monti is very valid. Our mudel assumes that the entire droplet undergoes
decomposition as a unit. There is no evidence that such a so called "sudden freezing" actually occurs. Preliminary
observation shows that the droplet gradually becomes darker as the evaporation progresses. This might be due to a
"surface effect" as Professor Monti suggests, or it might be due to uniformly distributed. -cleiof insoluble residue
species suspended in the liquid phase. At the present time. we are attempting to interrupt the evaporation process to
determine the chemical composition as well as the nature of the products of decomposition. We are hopirg that by
the end of this academic year, we will be able to resolve whether a thin layer of decomposition product or a large
number ofdispersed insoluble nuclei exist. Only then a better and more appropriate model can be proposed.
The purpose of this paper was to point out the potential probl~m areas associated with droplet decomposition
particularly for synthetic fuel and blended fuels. We thank Pr6fissor Monti for his kind words.
E.Wame
(a) Could the authors explain the apparent conflict between Figure I and Figure 2 in relation to the time to reach
the cessation of evaporation. For example, 2 ms in Figure 1 and 0.4 ms in Figure 2 (2 cm at 50 m/s). 4
(b) What work was carried out to study heat transfer to the droplet sihce this is a vital component of the
evaporation scene?
Author's Reply
(a) Actually, there isn't a conflict. The particles have zero initial velocity and they are accelerated to the gas
velocity after a long period of time depending upon their diameter.' After say 2 ms,they have not attained the
gas velocity except, perhaps, the droplets with diameters under 5 microns. Also, the evaporation/decomposition
rate is diameter dependent, The time and locations shown in Figures 1 and 2 appear to be consistent. Perhaps
o during my talk, i did not stress enough upon the momentum equation showing Stokes' drag and the velocity
difference between the particles and gas.
(b) Detailed study of heat transfer to the droplet-was carried out by Farhangi, a graduate student of USC. (Please
see Reference in back of the paper.) In the present-paper, the experimentally determined evaporation cofistant
p- icludes all the effect duie to heat transfer. In ordei ot to confuse the central issue of droplet decomposi.
tion, we have used the experimentally detemuined lg rather than the effect of conduction, convection and
radiation heat transfer on the droplet. Thus, implicitly our study includes th -effect of heat transfer. i---_ - -
7_ -_ _:-
Ti
T- -
PR&ME
Cq4BU5TO -#9.qgLINGF) DEELPMN P PedRsK
fP. -garsha,-R. -D; delkian,-S. V.chi6tbocha*an aJ.- e~sn
-Combustion-Sciencee and. Advance&-aTechnology Department
21133 -Victo -Bouloevard-
Canog
Prk, -Calffnia 9130-
-, Utaoq
1. INTR~ODUCTION
Control of-flame stabilization-and-flame propagation in a turbulent flow represents
a key element in combustion chambi design. The placement and geometry of iual injectors,
flameholders', and air distribution-ports are basic design paameters that govern the per-
formance of a particular-combustori. Therefore it is desirable to compute combustion
chamber flowfields in order tol-understand the--phenomena that occur in existing combustors
and to predict the perfoimiance of new- combustor-condepts. Modeling can also be used to
analyze the behavior of flameholderi and fuel-injectors7-under-a variety of conditions.
The insight gained through the use.-of these "unit" -analyses can -be of substantial use in
thlnning of-a combustor test prga adi -h iteriietation of combustor and com-
bustor component test data. However, the computation of -a~gefioilized combustor flowfield
is a formidable task, involving a ~number of compl ex~coupled.-physaical and chemical proces-
s es, which can include turbulent, recirculating tflow, possibly-with swirl, finite-rate
chemical kinetics, and droplet evaporation and-combustion. -Thie obvious difficulties in-
clude not only the multitude of co-upled-mechanisema but'also-involv4e the typical disparjty
in characteristic length and-tim scale's. ih combustion chamber flows. Despite these prob-
lems, considerable progress has been made--in iecent-years-in- the development of calcula-
tion methods, using techniques for the solution of the elliptic governing equations and
simpler yet physically perceptive"modular" modeling techniques.
A "unified* model of a combustion chamber .flowfield in general requires the numerical
solution of the elliptic form of the governing equations: Most practical combustion cham-
ber flows involve large regions of recirculation in which axial diffusion is important.
-A c- nsiderable amount of research effort has been put into the development of numerical
techniques for these problems, and successful comparison of calculation with experiment
for recirculating flows with large heat release has been reported by liutchinson, et al.
[11 and by Abou Ellail, et al. [2], for example. Significantly, in both cases a careful
adaptation of 'the- numerical model to the ipecific-experimental configuration was reported
to-hefrequiied, and in.both of these papers it was noted that the details of the computa-
tilen iequired careful handling to obtain the accuracy demonstrated. Furthermore, Abou
Ellail, 6t al.' 421 note that it isnoi possible to. provide sufficient resolution in a de-
tailed .combuisto~r flowfield-compu~ttation to adequately describe pronesses, such as fuel in-
jectibn; in wli-ich-mixing initially occurs on-a scale much smaller than that of-the overall
combustion chamber. 'Thus;, while the-development of numrical models capable of providing
directso6lution of the'equati6na governing specific combuition cfiimber flowfields contin-
ues, a 'need exists for the development of physically perceptive yet mathematically simpler
models. This rejuirement arises-from the need to-provid6 a model-which 'allows reasonably
rapid, computation of a number of different, complex combustor ga&daetries, and the need- to
develop models for those processes, such as fuel injection, which 6cctir on scales smaller4
tiau can -be adequately resolved in a detailed-ovt-rll-flowfiele computeto.Tedvlp
ment of-apprbximate methods - modular models --is~-a response zo the reqfuirementi; just out-
The basic interest in the application of a#proximate techniquesi-6 to avoid the com-
plexities inherent in a direct calculation of a complex flowfleld by making suitable
assumptions that allow the flow to be computed using simpler approaches. Clearly the
simplest possible procedure is to assume that the flowfield. is effectively one-dimensional
thus avoiding any necessity for definition or calculation of vel-ocity or species profile
effects. A somewhat more sophisticated approach is to assume rhat-the~combustor flowfield
can be -btoken'down into- separate zones, each of which can ~be calculaied individually in,
some-detil,
'aud thn'~cupin~togthI in soe i:6in to obtain an 45verall comufatiiiil
-analjo-f the -combuit&r-flbi . Suhpproachei arm -termed modular models, exasipeso&f
whimh h-av4e begh-iotdyRhetle al. -(3] an:wteba t2 4); theifedniula- i!W i
di~i iwh-il-hsperiQie--similar in that the overall combust'r flowfield-,i
__ bokmi~dow~ iito~__puatinalobuiztmLpioVi eisfar more detaifllthan either of than for
-~~K
-= ____
______ --
In the present aiproch the cosbustor jlowfield, represenfed- s-chematically in Figure
~L1, -is broken-'down -into-thre najor com~ponetsi a directed flow, -which is treated as para-
onO asume tobe eprseted by well wstirred--reactr()
bolc, reirclaton and a
turbuln sha ae ln he dividing streamline whiich-Spts t tertwregions.
heserlaye sevs steopin ein -btwen the otirtwo-mod~1 components;
fluxesa of seisndnegacostishalaefomheoudar onditions -on the
-two coptatoa rein. ntert msr based -on- the -quasiglobal moel (51 can
be inclded in theffomlaitions for -b.th-thii:directed flow-and well-stirred reactor re-I
gions, although- for the miodular moadel cdalculiations described -in this report, the recircu-
lation -region welsirdreacto.r for'nMulation has been restricted to a global finite-rate
chemistry model; The 'directed flow is assue to be fully turbulent, with the turbulent
viscosity defined by a twoequation-turbulence-model (6J. A key difference between this
modulr model -and themodlsof Roet, ta 31_ and- Switbenbank, et al. 141 is the
provsionfor-he shear l--ayer coupling region i h urn oe.'hog h so
this;- elemenft of-the model, thedison fthmsslubewn the directed flow afid
the
ecicutio- region Is comiputed iteratively -rather -than sipecified empirically
Furthroe the dire fd-liQ-reg 'in detail as a two-dimensional- parabolic
-s------ted
flowfield1 iather .thiz through a -one~dimthisional brpoximaton4 allowing tht use-of -de-
tad-dc ttoh fb miin and chemcal reactions in-this -region of the -oxnustor;
liON-RECIRULATING,
- VI"',OUS MAIN FLOW
LIUDFUEL 7 TYPi1CAL-
VELOCITY PROFILE
DIVIDING STREAMLINE
FLAMEHOLDER INJECTION
2. M(O5LAR-HML FORHULATION
Thebasceeme t of themodular model-formulation are a parabolic finite-difference
~.o. uatinatehniue(71# used f]or the directed-flow portion of the anfalysis, ad a
stirred renotor -computationh which-reipresenits 'larqe-acale recirculiation regions. -These
alement -~r mdules, are coupled together through a 'simplified representation of th
turbulenit sheiz -layr-which exists between thedA-1redted flow and the recirculation region.
mocu
Uin
th a0approachr the s--hear layer- repieSentation is used to' dfine the ~gradiant
spcia, ad ethapy- bewen hetwo regions of, tire -flow,; thuse providing
in eloity
boththei -boundary co :nditiofis on the directed'flow and the stirred reactor feedrae
-~- ~- __
5,R = . .=-'
.--..--..
S_4
N-3
rates for species and enthalpy. in the modular model, the net inflow of specieu and en-
dividing streamline, so that for the modular model the energy and species conservation
equations for the stirred reactor can be written
Energy
00
0
2:lRCW V CT) I ds %(4)
0
where RCx) :.epresents the radial location of the dividing streamline, and
Species
The second major element of the modular model for a sudden-expansion combustor is the
formulation for the directed flow portion of the combustor flowfie'-. It is assumed that
the-boundary layer approximations apply to this part of the flowfiald, so that the de-
scribing- 6quations are parabolic. For a -steadk,-axisymmetri6 flow these equations may he
wrltten:-
Global Continuity
arou arov-o(5
3x r
A-
4A
4A
4 ~ ~ ~ ~ - t____ ---
- Z~L4 pecies Dfson for the ith -pecie
$ ~ aj 4-i- (7)
Momentum Equation
Pu au v au I o u (8) '
and the
Energy Equation_
Pv pvT aH(Er 1) hi+ 1
W ar r r 7r- r (s a1
r Z
in which Pr and Sc represent the Prandtl and Schmidt numbers, respectively. These equa-
tions, along with expressions for the enthalpy
p - IOPT (a wi (11)
i 1 w)
can be solved, given an expression for the turbulent eddy viscos:'y UT =PVT. In the
present formulation, this is defined by the two-equation -turbulent kinetic energy model
[61, in-whih transport equations are written foir the turbulent kinetic energy, k, and its
dissipation-rate, =e. in boundary layer form, these equations can--be written:
Turbulent Kinetic Energy
-2
Pa + 0 a I a 't PC-2
IneroTh th
Pe Te rtr nde deedifloprtions thcae odlae
ofte mo volurbt-
ric chemical kinetic production rate terms appear in the species transport equations.
These terms can be evaluated usin4-a full- hydrocarbon chemical kinetics scheme based on
the quasiglobal kinetics model [5], whichi has as a key element a subglobal oxidation step
-- atCfCnj 11
A T"' POJ
=n~ C 121 exp [-/R.
with the constants, A, b. and Ei/R defined in Table 1, where P must be given in atmospheres,-
Trin degrees Kelvin and 1] denotes molar concentration; coupled to this subglobal step are
the intermediate reversible reactions also given in Table-I.
t_ ~- ~
29-5
REACTION A E/R
- FORWARD
key feature of the modular model that allows the inclusion of a fuel injection module.
This is particularly true in the case of liquid fuel injection, for at the fuel/air ratios
appropriate for ramjet combustor operation, the liquid fuel streams initially occu'py a
T, very small portion of the overall combustor cross-sectional area. The liquid fuel injec-
t-on model-makes use of a combination of empirical information and turbulent mixing calcu-
lations. For example, the fuel jet penetration from the wall is- computed through the use
of an empirical penetration correlation 181, using a breakup time correlation [9) to esti-
alate-the downstream-distance at which penetration is to be computed. That is, it is
assumed- that the fuel jet has turned parallel to the airflow at the axial position at
which-the- initial fuel jet has broken up into droplets, as given by the braakup time cor-
relation and the local airflow velocity.
Since the basic modular model formulation- invol -ves an axissmetric flowfield, indi-
vidual fuel jets cannot be-resolved, and it is assumed that the liquid fuel spray forms an 1
annulus whose cross-sectional area may either -be-specified -or computeid based on an assuimed
fuel spray-bulk velocity. A bulk spray evaporati -oncoirrelation is-then, usle to compute
the fuel. vaporizatibn. rate;. this correlation 110]1 is, a fuficti6n- of the initial v..locity.a
and- tempekature differeacebetweei the fuiel spray and the surrounding- air stream. Spread-
"ainof -thd-.fuel jet is computed through-use of a turbulent mixing hypothesis as for-the
mixang-process in the remainder of the parabolic flow. Figure 2 sh ows the-results of a
computation of_ the fuel injeto~rcs fothreefl inetrs, located in the com- 3
bustdi Inlt -wall,-along, the-centerline, and in a-midstream position. Shown'.are the -com-
puted c6nitours of the fuel mass fraction, or, with the vapo r-phase--fuel shown as the solid
li ne and'the liquid-phase fuel as the dotted line, as a function of both axial and-radiil
- -positon ±n~the combustor inlet. For these calculations the air inlet velocity was
apprxmtl 70 tsca at eaue of1600K; the overall fueljAir equivalence
ratio -was 40.6. The results shown in Figure 2 -proiide a good example of the detail of-the
ful njctonprcss available through use of this aspect of modularmodeling.
_P:
____ -aN
_ f
II
-I. --
0.6.
OV0
7 r 0
0 07
a0 a .
.7 ... .
0
S 0 0.5
0-
00.10.
0 0
0 0! 0.2
AXIAL POSITION. FT
An initial state for the stirred reactor must be specified in order to begin the
iteration procedure. This state is reasonably arbitrary, e.'cept that a reacted tempera-
ture level-must be specified. The shape of the dividing streamline separating the direc-
ted flow and the recirculation zone must also be specified, along with the shear stress
level, expressed as a "skin friction" coefficient. The initial stirred reactor state, in
conjunction with a model for the shear layer between the recirculation zone and the direc-
ted flow then defines the boundary conditions for the parabolic directed flow calculation.
Dire~cted flow computations are carried out to the axial station at which the end of the
recirculation zone has been defined; as part of these computations the diffusive flux of
species and energy across the dividing streamline is computed. These fluxes then define
a new set of stirred reactor "feed rates", i.e., species and energy fluxes into the recir-
culation zone region. Convergence of the procedure is defined by the change in the
stirred reactor feed rates from iteration to iteration; each species and the energy flux
muit change less then 0.11 before the convergence criteria are satisfied. If they are
not, the species and energy fluxes, and the overall diffusive mass flux and physical re-
circulation zone volume are used to compute a new stirred reacte- state. The pressure
required for this computation is taken to be the arithmetic averagezof the pressures com-
pu ted as part ofthe directed flowi solution at the- beginning and-end of the recirculation
i| region. The new stirred reactor state is then used to define new directed flow boundary
conditions, and the computation is restarted -from the dump station. When convergence is
achieved, typi6ally in 3-5 iterations, the combustor calcul-tion is continued t6&the
specified-combustor exit station.
1:In -
-. _--
describing the coupling between the stirred reactor and directed flow elements of
the model, 'it is useful, for clarity, to consider a simplified formulation for the stirred
- - reacior, -using a global finite-rate kinetics model. In this model, the fuel consumption
rate is given-b-the eXpiessin -
12
de 'v. 0 A exp (-DI/RT) (16) _
-M7 M -
-2-D
'Ie (),
0ec~~ (17i
31172
°dt VS(T) 12)
*~~N d2 =0 %n - 0
+ m/41
d1I
*19
- 20+
lioj%5Iv(T
(0
M(21
state and Eq. (3) are solved to obtain the new steady-state values of the active species,
i.e., the values obtained when dfli/dt - 0.
WaR NON1-RCUL I
No STIRIRED REACTOR
-- _ICALU LATI#ON M. Vk
C=O4
C /2 (24)
where and U are evaluated as the average of their values along the flowfield centerline
and along the streamline 0 = #p. As noted above, temperature and species mass fractions
are assumed to vary linearly across the shear layer. If #v is defined as the streamline
immediately "outside" the recirculation region (i.e., within the directed flow) and Vr is
a (an imaginary) streamline imediately "inside" the recirculation zone, then, for species
i
air(25)
#%i~ aM
where air and Tr are the values of the species mass fractions and temperature obtained
from the stirred reactor solution.
AL
= d f(x)da d (x)dx (29) _
Be= or
_ 132)
72077-
-a .
typical ramjet ct~stion _chather. and the lack of an inlet- sectIon, the -ode "= assump-
Soultie t necssay
math te moila one~prdi inswith-these data arenm ncesrl
a~opratefora
vLS ranjet c~or . EMave. ctatzrnedi's data, uhich inelmde raan
radial profiles of mean velocity, axial and radial turtole-t in-tensity cam. nts, and
totael pressure, are detailed emush to provide- a critical- tevt of a-deflaw tnabaility.
The parameters, in addition _to wail bmudary cwtditions and inlet redtions. -that
=as-- be specified -for a- noiular dr-utatoa Aabids: recirculatim zone sire and shsp=e,
sheaJrr 4i~ thickness And gruxth rate, dividing z b-
sear Stenstoef-
c~a'~
ecruato zn t1fl - kinetic - sgy--- level.,n reczxcalazne turtlest
is....t~m
ecthscle ~ia 'rc1 bad establishbed4 -for typical sme-xp=~-
andir imstask -that di~e reml~n sbape was wel represeated -y a parabolic
a.... arrange-d such that atz the steUp. d.r/dr = , sad %&ichlntercepts the srall at a dirao
9 2 step heights 1--m the 1 e- t#ls recz-rculseion zone length. correlatioan %ax based o
axnrvys results -U131. This Wor also Showed that- the dirm r-anlin shear stress
coef.±e" --
a a strong, effect on the 'p..cd sai pesr dinstrbui, iie th
±uxbolent kinetic ener-gy dispipation rate length s=l ~ fr the rucltinn
had a stronag effect on the rate ot tanspor- int and o=_t of te rear-'atio rmsisn.
-be zhar layer in-iia' t ~'ctiess and rate of nowtb Lad ittss oaveal1 ef fect. -te
wa~ considered in the earlier wock.
Althuag Chatrrredi *s date shwa reatta-t point at 9.2 step heights fe the
'-e and radicate an escsimtially parabolic rc-namne - ape. coparisn of the
tca these data cold nort he reproduced vith the parabolic zrimrculation z- IvdAryv
thr- h.ad bee =M&d Lanstea, an atandefd rectrenzatioo Zo e h ad to_ he asopted.
bolc _- zdfie z=--~
ot shapes adopted far the -ibiptati.mae flatV'
theextnsthapes thate e effem-ti'y deennrcirculatrZon Zenefor pu~sf , heomr-
patios the stir-red reactor Voume is kept amthn - .61 ft distme inaicated- T !!e d-Ata:-
effect othe serlayer hondizp- the divti screamlize relay to that of a wal
bdcaduy layer. A atailed descri4ptiont Of. thOeffaie thhe aiin Ma- Parwm-
tera for this floefiel-d iS; giroelseahere 1141- It is ipOtaxt to note tha the S2djest-
t fthe xodel prse.mters carrind aut for this case %-4meaith -rfernme to the will
stattc oressure end cafterline Velocity MeaSurves=: ?he deald r~ ffl a ver =o
Use is the moelf iteration.
0:04-
q 0*
.. T7:1.
lm-==-
model predicts a greater wiring rate and shorter recirculation zone length than tie data
indictate:- however, no effort-w-as sae to optzaize the unified model, calculais for these
data (tvuh.brewladiiai to- the wall boundary conditions-use- i-n the co- -i
puctationJ - it is -intexuating to A thatfor X/D - 2 and VID -4, the thified model
velocity profile shapseprudition axe siviler to- those the aofular model prodwes. up to
tie dividing streasline- _;modler-odel of course, provida no detail within1 the r--c
Eiveato rein As fin sf Abesteepening Of th'. velocity profil in the
neWir il region between ttb - *and =I ! 8 that Is predictd -ybt9oel.Fgr
This prvde uthreidecethat a eshear -layer eaata-rinhythizdto
erirt im the modular model fommiatln= of the flruftld is In fact a feature of. this flow.
raial'tt balet KID - £ are stcn iz Yiqure a. Although the
-ieieegytuinfor
ellintic ,wdel~ produces a slghtly- lowr-level of turtalent kinetic energy than does the
rdular appoach. both calculations Are An reasohatly god aresaet with the-data. These
oaparisons indicate that or this cofgrtothe ZodUlar Zapprouatf is Capable of Pro-
I iding predictions of the details-of =-he flaufield that cre in agremet ith t ,-' the
eernennal data and with predictioax-oetaned f--= unified techniques. wt~*
they vrvide evidence thatt fl- lyer anvw
d recir-culation . models incldd! the
siodular approach are adequma regresentations nf bohof these regons
r
"Co-a i-WMZX ---- N
1- 0 ~M=46ra~
WIN
*L
AKT
I IN
FMa ~
Z
Fi~ odlaran ~ u----e--Wth-at-o------ei
3.2 RIACTING FLOW APPLICATZOWN
in the premixed cosbustor investigation, the reaction rate for the JP-;4 fuel was modeled
us-Ing a global-expression derived for propane. However, for these calculations the thdra-
modynamic coefficients and mblecular weight of the fuel were those appropriate for JP-4.
2 PRP11OICTIONS
* krO.1O0
~0.6 4
0 TIAYA. REF. 15
0.5 -
0.4-
DA/RE
0.3-L
0.2
PT2 0.0
flo I u.=- -1 -v
so LEGEND;
6 MEASURED (qh) a I
0 MEASURED V'IC) 1TT
80- 110 'R
I~l
9C. % M2 -=0.49
40 JP-4 FUEL
WALL INJ9CTION/S ORIFICES
A3lA2 0
20
END OF COMBUSTOR
0I I -
0 1 2 a 4
XID
3 THEORETWrAL
- ~ ... MEASURED
THEORETICAL =1.8611
MEASURED = 1.70 (*R)
FIGURE 10. Comparison of Modular model Prediction With Measured Dump Combustor
Performance, Liquid Fuel Injection. Data From Ref. 18.
W_ 4-
-72 - - _
loih ~q-YO
FUEZL/8 ORIFICOR
3.011)
0~~ 012-1350TAI~t
0
0. 1.08111 1 50
OVEAL ATION
COMPUT.ENE
Daa.1o 0e.18.00
4.8-
3.0-
0 peet.t1 0.I
PENETRATION HIEIOITINLET-RADIUS.11' g-o
Thera, i~s -no currently :available:ifinite-rate chemical kinetics model which cax orovie
adeqateresuts or omplex'hydracarbon.-xuels 'under fuel-rich- cond-tions -at temperatures__
at which the effects of dissobiatiori maiiifest themselves. The modular-model- foxmulatidi-
ued-for the computations described ih-thisl-ipet includes as options citiftar Cn~-ep
~ __2. ~ -
finite-rate model or the quasiglobal kinetics model. In the former case, the products of
-combustion are the fully-reacted species H2 0 and C02 , so that the effects of dissociation
i"1-#are-ignored, and the overall reaction rate-is underpredicted. The quasiglobal model used
in the modular formulation is valid for complex hydrocarbon fuels and includes the effects
cZ dissociation, but is not appropriate for equivalence ratios much above stoichiometric.
Other models have been proposed for fuel-rich conditions, iuch as the four-step model of
Dryer and Glassmin [20-, but this-model ignores dissociation effects. Thus it is diffi-
cult to obtain reliable estimates for the chemical kinetic rates that pertain in combus-
tores in which significant regions of the flowfield-are highly fuel-rich. This difficulty
is reflected in the perforvane results shown in Figure 13. While the trend of combustion
efficiency as a function of feel penetration distance obtained through the use of the one-
step global kinetics model is as expected, the magnitude of the overall change is rela-
tively small and probably incorrect. The reason that the magnitude of these results can
be expected to be incorrect is that the global finite-rate model can be expected to sig-
nificantly overpredict the recirculation zone temperature for fuel-rich conditions, since
the only effect of fuel-rich operation that this model recognizes is a dilution of the
products of combustion with excess fuel for 0 >> 1. With this model, the reaction rate
is linear with respect to fuel concentration, whereas evidence exists that for fuel-rich
conditions reaction rates do not continue to increase with fuel concentration, and, as
noted earlier, dissociation effscts are also ignored.
g1.0i-
: i :i ONE-STEP GLOBtALFMIITE-RATF
O ~KINTllICS . OVERALL
lMO'VEL
it0.4- RATIO,
EOUIVttLrNGII 1G:
.0
O -
< -: ---
In
response to th ed ust outline# adetailed moe-of tesae-_qaso
liquid-fueled ramjet-ctombustor has bendeveloped. This radel makes use of the nodular
=: -s ---
- iconcept, in which the overall combustor -flowfiield is broke down into a number rfrgon
_ _: > _ - }eaich of which is analyzed in detail using a technique suitable for the specific part of
_-; .b -- +ithel flowfield. These ana y-ticl -subelfe~nts--, -r modulas-, are then coupled together itera-
.... tively. through their Sbou:ndar-cofnditicfos;
4yield EoEN The Rthexamples
per given data
ntal i this paper
for a haveofshown
variety con-
--:=-: 7,-figurations .anid-thai:--th4- model :prvides a useful1 t~ol for the analysis and -interpretation
-- A
REFERENCES
1- Hutchinson, P., Khalil, E. E., and Whitelaw, 3. H., "Measurement and Calculation of
Furnace-Flow Properties," journal of Energy, Vol 1, No. 4, July-August 1977, pp.
212-219.
2. Abou Ellail, H. M. M., Gosman, A. D., Lockwood, P. C., and Megahead, I. E. A.,
"Description and Validatica of a Three-Dimensional Procedure for Comustin:n chamber
Flows," Report FS/77/27, imperial Colleae, October 1977.
3. Roberts, .R. A., Aceto, L- D., Kollrack, R., Teixeira, 0. P., and Donnell, J. M., *An
Analytical Model for Nitric Oxide Formation in a Gas Turbine Combustor," AIRA Journal,
Vol. 10, No. 6, June 1972, pp. 820-826.
4. Swithenbank, J., Poll, I., Vincent, M. W., and Wright, D. D., "Combustion Design
Funametal,"Fouteeth vmom International) on Combustion, The Combustion
Institute, Pittsburgh, 1973,pp 627 636.
5. Edelman, R. B. and Fortune, 0., "A Quasi-Global Chemical Kinetic Model for the Finite
Pate Combustion of Hydrocarbon Fuels," AIAA Paper 69-86, 1969.
6. Launder, B. E., Morse. A., Rodi, W.. and Spalding, D. B., "Prediction of Free Shear
Flows -A Comparison of the Performance of six Turbulence Models," Free Turbulent
Shear Flows, Vol 1, Conference proceedings, NASA Report No. SP-321, 1973, pp. 463-519.
I 7. Boccio, J.1 L., Weilerstein, G., and Edelman, R. B., "A Mathematical Model for Jet
Engine Combustor Pollutant Emissions,'- NASA CR-M2208, GASL TR-781, General Applied
Science Laboratories, 1973.
8. Catton, I., Hill, D. B., and McRae, R. D., "Study of Liquid Jet Penetration in a
Hypersonic stream," AIM- Journal, Vol. 6, No. 11, November 1977, pp. 2084-2089.
9- Clark, B., 'Break-tip of a Liquid Jet in a Transverse Flow of Gas," NASA TN-D-2424,
August 1964.
10. Ingebo, R. and Foster, H., 'Drop Size Distribution for Cross-Current Break-up of
Liquid Jets in Air Stream," NASA TN-4087, COxtober 1957.
11. Chaturvedi, fl.C., "Flow characteristics of Axisymmetric Expansions," Journal of the
Hydaules
ivsio, Pocedigs of the American Society of Civil Engineer, Vol. 89,
liarsha, P. T. and Edelman, R. B., ' Applicatioin of Modular Modeling to Ramjet Perfor-
mance Prediction," AIAA Paper 78-944, Fourteenth Joint Propulsion Conference, Las -
Veglas, Nevada, July 25-27, .1978.
3. Drewry, James E., "Characterization of Sudden-Expansion Dump Combustor Flowfields,'
AVAPL -,R-76-1,2, Air Force Aero Propulsion Laboratory, July 1976.
14. Marsha, P. T. and Edelman, R. B., "Mixing ena Comburstion in High Speed Air Flows,"
AFOSER 8G0-1005, SAX-80-020-CP, Science Applications, Inc., Canoga Park, CA, July
15. Craig. R. R., Drewry, J. E., and Stull F. , "Coaxial Dump Combustor Investigatior,"
AIMA Paper 78-1107, AlMASAP. 14th Joint Propulsion Conference, July 25-27, 19718.
16. Airbreathing Combustor Developent Cocmdittee, "ecommended Ramb;Irner Teat-Reporting
Standards," CPIA Publicatich 276, chemical Propulsio. information Agency, March 1976.
- -- 17. Edelman, R. B. and Mrsha, PT. *L Lmnar axnd Tu.rbulent Gas Dynamics in Combustore-
Current Status,* Progress in Energy and Combu~stion Science, Vol. 4, No. 1, 1976, pp.I
1-62.
/T
29-18
ACKNOWLEDGEMENT
Research-partially supported by Air-Force Office Of
tract F49620-80-C-0082 under the direction Scientific Research under Con-
of Dr. B. T. Wolfson and by Air Force Aeropro-
pulsion Laboratory under Contract P3 3615-77-C-2062,
John Smith, technical monitor.
DISCUSSION
LGlasspoole
How long do your computer calculations take?
Author's Keply
The computation time depends primarily on the complexity of the
chemistry model used in the computation.
When a one-step global finite-rate model is used, a complete combustor
CD)C 7600 computer; this involves typically ive iterations of the oeal computation
reue.Wnth
requires 4-5 minutes on a
uailbal ealed
kinetics model is used, typical computation times are of the order
of 15 minutes on a CDC 7600.
ZZV
-35
-- - - ~ -"- I-----t-- --- = - - -- - . ...- IF
- -I
THE MATHEMATICAL MODELLING OF AIR-AUGMENTED ROCKETCOMBUSTORS: A PROGRESSREPORT- -
J.C.P. Bjrclley
L.T. Glasspoole
A.S. Wilson
Procurement Executive, Ministry of Defence
P-opellants, Explosives and Rocket Motor Establishment
Westcott, Aylesbury. Duckinghasshire HPIB 0HZ
The theoretical bases for the calculation of axisysivistricand three-dimensional ducted flows are set
out and suitable solution procedures are outlined. Progress is illustrated by presentation of results
from examples of axisyrtric ducted -flow calculations and directions for future work are indicated.
Nomenclature
E. Activation energy for forward rate coefficient of j'th reaction - equation 17.
H Total enthalpy.
2 2 2 ,
k Turbulent kinetic energy j(u' v' + w- ) where u' v', wi are the
fluctuating components of velocity.
kf1 , kbj Forward and backward reaction rate coefficients for j'th reaction.
i Chemical species i
020
Mi Conversion factor for chemical reaction rate (= 6.023 x 1 ).
Nmer of reabtions.
p Pressure.
__ r Radial co~inae
I
-- - -- - -- - --
1=z
IME
a-
T Temperatur.
x Axial coordinate.
o Transverse coordinate.
O Mass density.
aI k' cm0,cc. 0 Turbulent exchange coefficients for enthalpy, turbulent kinetic energy,
species, turbulent energy dissipation rate, and 4 . respectively.
INTRODUCTION
The successful design of an air-augmented rocket vehicle encompasses a number of complex problems.
The one which is addressed here is that of assessing the perforiance of the secondary combustion chamber.
The design and realistic testing of such a unit may be expected to be a difficult, time-consuming and
expensive exercise, involving as It does the provision of flight test or flight simulation facilities.
Such a progr of work was considerably beyond the limit of current funding. Instead a odest theo-
retic l Investigation was initiated, aimed at gaining insight into the problems of cobustor ignition ard
permitting trends in combustor behaviour to be established. It was decided that the investigation would
proceed on two fronts. Firstly, use was to be We of existing expertise and techniques to develop a
computer code to calculate axisyrmetric, non-recirculatory, ducted flo-. This comuter code was to be a
extension of the capabilities of existing free-flow computation techniques, the hope being hat significant
progress could be attained for a ooparatively small expenditure of effort. Secondly. and .ith a longer
time-scale, %orkwas to be initiated on a computer code designed to calculate three-dimensional ducted
flow, with due allowance for flow asymetry and reeirculation. The latter emphasis is necessary to enable
the effects of, e.g. flight angle-of-attack and inlet geomtry to be accounted for. This paper reports
progress on these topics.
NUMERICAL MODELLlNG OF FLI-S
in carrying out a mainly theoretical study of flows with combustion the overall flow behaviour cwy be
ascertained by examining the behaviour of such features as the values of velocity, pressure, temperature
and chemical composition at points throughout the flow. In addition the way in which the (steady state)
flow develops as it passes through the flow region is influenced by te processes of mixing, so that the
turbulent characteristics of the flow mt also be considered. ea baviour of p.prties such as these
may be d-scribed by a sit of simultaneous partial differential equations, together with additional rela-
tions for the state variables. Further expressions for the rates of production of various quantities are
required, for example the production of cherical species my be determinei from the law of mass action
applied to a suitably chosen chemical scheme, utilizing appropriate thermodynamic and reaction rate data.
Formulation of the flow behaviour in this way permaiithe ioble to be cast into a tractable ma n-i_tieal
form. Transformation of the differential equations into firite differene for makes it possible to ii
approximate the relationships between theContinIosly varying flow properties by discrete values defined
at a finite nu er of points within the flow region. l -enepressedin this way the problem is ideally
suited to efficient solution by digit;al cosuteri and details of how this approach has been folloed for
-_ -R
-~ ~- 4
jilC
axisyntric and three-dimensional flows are given below. rnly after completion of this process may
practically Interesting quantities (e.g. thrust) be derived.
AXISYMKETRIC FLOW
m
A simle geometry was assumed for the axisynsetric flo study (Fig. I). The air strea was assumed
to be parallel and uniform at the duct inlet, while the primary nozzle wall was assumd to taper to
infinitesimal thickness at its lip, thus avoidin 9 probles of both flow asy metry a-nd recirculation. Wall
boundary layers upstream of the duct inlet plane were neglected. The flow was nermitzed to be s-usonic/
subsonic touter/inner streams, respectively), subsonic/supersonic or supersonic/s perscnic, but ti.e
problimi oVfalculatihg choked subsonic flow were essentially deferred by limiting consideration to a
parallel-kidied-duct for this initial stage of the investigation.
Governing equations
The equations governing the flow include those of conservation of axial menzx-, energy and chemical
species, written in familiar boundary layer form,
l
iere x , r are the axial and radial co-ordinates, respectively. u and v are axial and radia coo-
nents, respectively, of the velocity, . is the density, p the pressure. H the total nthap- and
r , 4i are,respectively, the mass fraction and rate of formatior of the ith chemical specie-. In
addition cll is the turbulent exchanoe coefficient ratio for enthalpy, a the tuu~lant er&.an-e
coefficient ratio of mass for species ; Wile V. is the effective eddy v;isisity, k s e
2
kinetic energy of turbulence j(u + vA) wh-re U, , v' are turbulent fluctuations of the W-ocity
components) and c1l the turbulent exchange coefficient ratio for k . The- lost t -- e terms cm. he
right of equation -2) represent differen tial turbuilent diffusion with respect to toteS e.thlr7o c-emici
m
species, mean flow kinetic energy and turbulent kinetic energy, respectivel.- In this work it was assu ed
that the turbulent exchange coefficient for mass 'turbulent Sd=-idt nter) had the sa vale for e c
chemical species and that this was the sa as tne value of the turbulent exchange coefficen-t. enhalp
tr
(turbiulent Prsn-tl number), i-e. that thte turbulent Leis numer was unity-
Padial pressure variations In the flow field ware accmdated by way of an for-
-roximate the
radiali mentim; equation.
am considerationa of local coninuity, a: outlined in ef. 1, together with a global ceo-tinuity statemint.
h.I, ).
Theloa gas density Wup-oenbe recovered f rcw - ldea gas estion of state __
p T * 18
IR
ideeN sthe molecular wight of the V th chemical species.
Turbulent flow modelling
Closure of she flow equations (0-3) is achieveAd W writing the turbulent eddy viscosity coefficient,
7i , as a fwictjon of the turbulent kinetic energy. kc and the eddy eersy dissipation rare. e . in1
dAform
- K
Ahere CD0 is an emirical constant- The descriptior of the turbulent b&?=evi.oer of the flow is ccqiVletsd
by way of tio additiIonal tran-sport equations for k a-'a c
;c 'I
ru+- Tat - V (12)
In equation 12 a-is the turbulent exchtange coeficien ratio for wile C, and C., a qii
Cal constants. Equatiofl5 10-12 comrise one of stersi we-mlestabisd utc"truec t odels
(see. e-. Rea- 2. this version heiwncos7 si-ce~L f enat =y i~ i ~ ~ c r
tOe cantitua thnCal species an-d evaluating cen flow prcp'erties from a tnaledge of the- local c-hemical
comosition. For each chencal species -- hefrzoornaic iaflties n - entropy L e
Gibbs free eergy G; ame given bi
To
To
w; hii - T S 5
adiere the specific heats arc caiculatmd from the nthoroer polyriwiael
--ue Gibbs freteftrg chafge &~Per -nixt of tm in ecac.ticn is vowted fron eqaio i
yields the ewal-liie C=tans- -pt via the ,partlim VzA-.iCM
tfj
... Cut-= 1
rV 5
m Uk Wyp
-g 4zL
__ ___16
AL;
The net reaction rate for the j'th reaction is given by the law of naas action as
it. e f t jI tTf-i
Fe .I -IT[' r fj (5
Ma
In eqisation (15) the reacticon rate At. is given in kmle3 ;3s if tte rate coefficients k., k*=
are in the al-alecule-secnd units. Aenity ias in%kg zi and X 6-223 xp~I*- e forward Me
coefficient Mly be written%in Arrheaitus for-m
Bonary conitions
The flaw bowndary conditionts ame readily speifiad at the dre-t axis u~cm. becale a, tuasae
symetry of tle Flow. it is sufficient tol tn-e ftlt
ninoes of the turbalen e~t vi#s-osity coeflicLent ina the- near-wall reoie aacre I..tm Prome
curs of tfs 3 a A, 61 te for tevntccte-- the laws eqiwip ft zf
Wptin Of CaLculatilt ether adiabatic Or Contnt heat flu codiirsi MCThs M-1ainse
can,
Uwall 1his also enlsidered to he 0=-p anO! alyWacie
-~-In met s- %:zy sMz finite Caffcwzne u-v-~ --F. tee Csed o wf- tar-
x . reacnauil
rmaauiog fias, e.q. t. ;.33 f mysralsob tr t± th i n e f n
m-of wt'- screws ofba. etra
Coe=~ t ia fa in a or-ary Cume- w-- a =W-cbZr nw fty
ct%!W =41a i.- --- W by ms= inlesulerc
t vbn. ,-Ow =.=
:s rsose
so~scs a pnCS S~TM-y. In Via. f m a am 1 ~ m
9 mbr v=Nti~s a- ts *ezr-aIRso it. ith b- dcci*d,4 r - maae -t &O- at bent so iri
ascension t e seirodery aabr. Quwcn atz all upcfe
to pi and ast~
jg or: ice, and the of zero SzeCPS4n Cdioni ias 120E at ' swr- of me f ao
_ ~tzJ
I7
ft -- -- ___X
_ _ 77~~Ci
Scyliafticai -;ar waorname (xr~e) the uramson of rj- scalar Viun:zy 4can be mj-rnswc
-as fi-s
3!x 3 2&)
Io i the dnlt. I (w.w~ arm Zbe aecar -r . i the *wr-*_ vscsfe and
aisthe eang onfficicit enuf EPrat'rri _e. __'for the i v Mesw ter- -S a
fluo of I'-e field wejlatles s evaicte a=- 'T., -l Taze, -!he i't Su=c
n Wr
etuauto (fl) are t- Velocity *teM anoer s= r.ck etrr. * the
=& itn
awerv dissiWaflc ratef C * te oca nthsay ad the spci ass '-4r-os F. is! -- 16l Mee
tw~b'an visweLzy is g1;.r by eqctava IB0. lifel state w - Wes of preswwr * c-sw- a-d
rin T an fttezf~e f'au g,- Qs:~ MM =
0*±
h astwi 43t
5l a t thio=
s Pe Wemare
~
-=~ ~ 5 u~~ _ -_00
IMe-
3 ~iiris~sr ~MR.
th 'I. it~ cii~a*e ls~feE ~M~ s
Ailc.42% zbe
afs ai~lSecic Iouu AsaPlusa..
itel %; eMN
en -
ws 0- a
Solution Procedare
S s
Equtio -ese toellminate a timequation
usd~stitcte trim equaion 33. Is linearisfld 25. If xb rcsultlng equation,
with respct w A a--- itis 'rith
Setti. possible, by
in (321
t i cse! a-me Oensure Sfl 1 and nilocitv flela a-&wazr
3) Equation 21 Is sole as in ,ao for k aff ic fo taw eddyv Vi;city
laced using tion. 8 p; Is!tir
~) The Pskulry M wst- -acalaed. from iAica the trat,
-s a -aced,_ by meu cfeuai
5) The- for-=
Whteo s ftactions F_ are solves- fo*am
treflcm. rhe r_ are-- Sesi". a also for elm-t cove-
acijae toczl With eleflea- comru-ancm wi th the CZrans
The second examle is aaln a hiolly supersonic flow, but of a more representative scale and with a
realistic primary flow composition. The outer flow is assumed to be an undiffused Mach 4 airstrea.
Details of the input Conditions are given in Table 4: turbulence model constants are again as listed in
Table 3. Two calculations were made with this flow gemetry, with and without chemical reaction, to
illustrate the effect of Combustion of the flow. The reaction mechanism used for the reacting case is
shown in Table 5, values of the rate coefficients being taken from Ref. 7 (thermodynamic dat in all cases
Considered here were derived from-the JAIAF Thermochemical Tles, Ref. 8.). Trx results of these caicu-
latias-are shom-at isometric diagrams of pressure (Figs 5 and 8 for the reacting and inert flows
respectively)-and temperature (Figs 6 and 9). The shock pattern apparent in the pressure diagrams is
illustrated schematliatli In plan form in Fig. 7. Comparison of the teqerature fields. with a l0 wit hou
reactioni shows that ilition of the flow in this case occurs downstream of the Intersection of the outer
shock after reflection from the duct wall.
CONCLUDlNG
MARKS
Considerable progress has ben made towards the capability to calculate ducted rocket flows, as the
illustrative exles testify. lowever, this success is as yet limited to flows with axial syntry,
which do --t undergo choking, and., moreover, has not been achieved without difficulty. Early stages in
the investigation were marked by problems of stability of the calculation procedure, associated particu-
larly with the prediction of pressure. These difficulties have not been entirely eradicated and further
investigatiorn will be necessary. Flows which undergo choking, either physically or because of thermal
effects, present a special problem for the axisy=metric flow program. The marching in-grat;on tecknique
is ntrinsically unsuited to the analysis of such flows, in i.9ch conditions at a downstream station
control the uistrea flow, unless so fore of iterative adjustmi of, e.g., the initial mass flow can be
carried out (as is done for example in nozzle flow calculations. se Ref. 5). This aspect of the problem
will require further work.
As far as the three-dimensional analysis is concerned, a major factor likeiy to inflixece the effect-
I iveness of such a treatment applied to the secondary coustion chas-er flow is the coutational effort
required. Since the treatment is rot only over a three-dimensional field but also iterative by nature (to
account for flaw recirculation for examle) not only will ti required computer stor ge be large, the many
grid points involved also ioply a larger numer of individual co putations per iteration. resulting in a
provides physically realistic answers- future Work will need Concentrate on Couting efficiency while
2 Latider.S-E., a pldng. D.8., "Mathematical models of turbulence", Academic Press, London and
4 Gibson. M._ iad Spalding, L3, "Application of the k- model of turbulence to -indary layers
wihstrenhmise pressure gradients, large temperature variations and trantspiration cooling",
Im-perlialC.ollege. London , A-c-h==ic.al
Eng'lring Deatm-ent Rept~ EHT/rtN/A/32 (1971)
5 ;if A-S.. "A user's guide to cmuter programs for the calculation of conditions in flames aid
* noz-ies". t.mvibllsh--d #1M() Report (172)
6 arkv. A-P., "Nixing of supersonic jets Including the effects of transverse pressure gradient using
difference"mthods", NAS Terlcai Note ill -602 (1971)
7 ,.e
Flom...A. 1-34a
O.iE.. Jons.td
(t19a4 G.., "Reaction rate coefficients ifor flinm calculations", Comb-ustion and i
ii-
+++-+ I
+- Ib
sw+ -wo_
r-U#
30-9
TABLE I
u~ 4+ (2 LuU v I
rx 3x r T7 r ax a
+22I2
v~~ r L
2v v
~ ~~~~~~
'r.i , aa r
alua +~)x ,a _a LPa
2i W_
r a x 'r rFr
2
410- T r _r e r2 r r
k 4p
c (c1 -~ c elk
SIaf 2a
ox
~ir'~
+ r
a Orj
+
r2aB
a- fe a-
Be j ~ L)
i .1V, + w2
UI2
j
+
o
1
o
Ki
i
where P, p - t ( 2 ±.12-(r,)
P+ +2.a)
ax r. r rI a J ~ re a
2 (2 )2 + 2uva
awa2
and p 2waw ) -vI + wrBe r
rx r r Be
~~r3
_A-_
04MIN
__
30-10
TABLE 2
Ineut conditions for example 1
N20.0 0.7856
H2 1.0 0.0
C2 0.0 3.27 x 10-
H20 0.0 6.95 x 10-3
02 .00.2071
TABLE 3
Turbulencemodelconstants
C0 0.09 1.0
C1 1.62 am 1.0
C2 1.92 ak 1.0
cc 1.3
TABLE 4
Inputconditionsfor example2
Primry
zzleAir
n stream
Radius. m 0.01 0.0376
Velocity, ms1I 2500 1400
Pressure. Nm 2 5.066 x io5 1.013 x 105
Temperature, K 1500 300
Chemical species concentrations (mole fraction)
20.1083 0.7856
"2 0.1904 0.0
CO -44385 0.0
CO2 0.0893 3.27 x 10-4
H20 0.1734 6.95 x i-
02 1.322 x 10-9 . 0.2071
H1.013 x 10 0.
ON 1.014 x 10~ 0.0
064710 x i1 0.0
HO2 1.000 x 10 ~ 0.
-- 7
______________4!
1ABLES5
ChemiuaLreaction che=
2 3 x1071 exp(900ir)-
H + H+ M OR+ 3x101 T--
OH
+ MH0H2 2
1~i 7~
CO0+C0 1 1 7 -10~ exp(-2200/T)
O+ H2- H0+ H 1.9 x1-5T.
OH 11 i~o10-1 T
13
exp(-182%/T)
~ 3~16o1 4 Tep(-4480/T)
2- OH + 0 2.4 x10-10 exp(-8250/T)
CO+ OR - CO2 11H 2.8 'c jOeV 3
1. ep30
OH 4t OR 11H20
+0 1 i01 aXp(-550/-)
CO +O0 - OD 4 122
2O 2 + . x1 p(-24000/T)
HM-
+ 0F0+ + M2 x10
32
eX(50/T)
H+H2 -OH + OH 4 x ICO~exp(-950/T)
H1+1102 - H2+0 4 x101 exp(-35o/T,
H2+H2 11 02O
H 10-12
112+HO~
110+ O 1 o12exp(-94OO/-)
Jh
C*Pyr:9tVotrler HMSO.London,
1981IZ-.
3042
c
0
-a-.
ci) U)
cv~ I 4-.
~I)
- I E
-
0 0(~) I
0
0)
I 0)
- I,
(~) 'I 0
I % I
V I ' '4-
I ~ I
I (~)
I I
t I I cv
A XII E
'I
E
U'
II II U)
II II x
0
'-0)
04-IL~
II
U)
___
__________
c%J x
0i
IN'
S0
UEL
C 0
OrO
30-14
--- - - - - - -- -- - - - - - - -- - -- -----
- ---
'Cc
- -2
Cc Z
Ir I
N L)
- -- - - - -- - - - - - - - - - -- - - - - -
- - - - - - - - - --- - - - ---- - - -
41o
- - .. . .- |
30.15
0 M
dcE
SC.oinet oaia jt
00i
-HM
M 0-17
~ t~f
*=4K
Fig. 6 diagramso
Isomeric example Atmstr il
k - - c
--
LI
~~IR
Ic
I0
f L4-
nmn
ON i
fr
Fi. I~eJcdnwus
B fpesuefil -arQecs
WI~~~hO~~t chmclrecin
mit View a
Rl x D
_--- MU 4~-
"S'
HH11rr-a
~~r
Il
gy._P.
A.F. Scblader and J. -Users
Departmcet of Mechanica- Engineering
Universite Lava!
1uebec, CANADA, CII 7M
nmcr
Further tests tere then made to verify the model, sone of which were wish sta InJe-
fion. The coda was fond to be satisfactory for sit condtions examined.
XOIwECLAMCU
1. Mr- A0ID'S
The use of a baffled recirculation zone and the absence of both film coolies and dilu-
tion air resenbles, In general., the characte; of a ram-Jet. On the other bZud, the system
might be sald ro be representative of at Isolated pri--ry zone of a gas-turblne combusotr.
Because of its pr*-msxed nature It may not be unreasonable to presume that the system
"a reaction rate controlled - a fact to a large degree substantiated b; the uniform Ity of
gas analysis noted in the subsequent experlets. Both temperature a=d species measuremns
Mere derived from gas analyses carried out at a series of planes downstrean from the combus-
tor. These values were tb--used to derive the siple combustion model which was later
substantiated by a idt-tioal t-ests. both with and without. atea= injection.
f
Ipitrea
-propatne
is given in Fig. I.
baffle to the co-buator
Separately metered air and fuel are premised upstrean of the Inlet
The !uel injector was a simple cylindrical distributor. 0.75 a
of the conbustor baffle. fitted with injection
(32 holes of 1.3 am diameter). At all
holes normal to the air stream--
conditions great care was taken to ensure that the
was InJected in the gaseous phase thereby obviating the need for any evaporation
time.
To ascertain the degree of pre-wix, carbon dioxide (selected because of its denasity
similarit7 to propane) has Injected into the air stream and the mixture strength was deter-
gined by sam4Int at nmerous poits (30) at the combustor tnlet - the plane of the pri-t
ry bffle - Teje esfa Indicated that no single value differed from more thn-2.71fo
the=an value. This was considered Satisfactory.
a ubr Iz
in are listed In the table of Refereces. -
..... - -i [=:r
--- = =: =- ==++-"
- -
-7
i -----. -- = z - -
+ -
TEST RIG
fue ,
.30 .8.-.70 a
=:+.:
"+ .- ength =(el
section "A-A*" FG
The combustor Itself was al 90* sector of an annular chamber (hydraulic dlm - 0,127 a)
being totally lined with ceramic and having a recirculation zone sahtlied with both a pri-
mary and secondary baffle. Downstream of the recirculation zone was another refractory-
lined sector contalining the sampling ports which were situated at distances of 129 am,
163 m, 214 me, 262 == and 379 mm from the primary baffle. Air velocities were select d so
as to give residence times commensurate with those which obtain within ram-Jet and gas-
turbine combustors (i.e. 2-8 ms).
The cold residence times were calculated at Inlet conditions based upon the diaeter
and e n velocity within the recirculation zone. Not residence times were based upon the
wean velocities obtained at the Inlet and any of the various planes of =easurement.
Cas analyses were withdrawn from one or o:her of the five planes mentioned above.
Three samplers were used in any one plane and each sampler was fitted with 3 sampling po-rts.
All the samples vwre combined prior to analysis to give a representative mean analysis for
any Riven plane. The occasional independent check substantiated the validity of this tech-
nique. -he samplers were water cooled, (the outlet te perature from the sampler was still
sufficiently high to prevent water condensation, Toutlet = 100*C) and the lne from the sam-
plers was steam-traced up to the entry to the gas analysis apprastus. This latter was a
double instrument, prepared for rapid gas chromatography. Deterlna-ions were made of
C3BH(FID), C02 CO- H 2 , 02 & U 2 . As a check on the accuracy of the sample, r- psrlsou was
made of the derived value of equivalence ratio and tLat based on the air and f.el meter i -
The mai-m difference between the two values !or any test was 32 which Is welithIn the
norma!ly accepted value of a5,OZ.
Heat losses from the system were assessed by measurement of the temperature of the
outer metal liner supporting the ceramic. In no case did the heat loss exceed 0.6222. This
was considered to be sufficixntly low so as not to merlt carrection.
SN0, =easurements are not Included. The f y nessurmuents which were made showed them to
Jim substantial agreement ith the empirical predictions of Lipferr 13) and retsch er et
- JA]. E x was deliberately nt considered for the model since (a) the Intenrion was to
keep the 'combustion' as simple as possible and (b) prndictions using the empirical tech-
niques of 131 ant ill appear to be quite satisfactory for NOx determination at the exhaust.
• .+- M
313
first stage, a multi-step reaction is assumed for the combustion of the carbon monoxide and,
if necessary, hydrogen. Equations are generally included to enable the formation of NOx to
be calculated. Typical-examples Are given in [5-to 12]. A-poesible exception to the
could be treated as a
3j-
assumption of the hydrocarbon bieakdown mechanism is methanewhich
multi-reaction scheme -nvolving the initial breakdown. An alternative to the above complex
modelling ii the saple, -onveaient, one-step reaction mechanism (so-called global reaction)
as portrayed by several workers [13 to 16]. One justifiable objection to this simple ap-
proach is that it'does not enable the-pollutants to be separately assessed (i.e. CO and
VBMO).
The pre .eit model represents a compromise between these two types. The reaction is
represented as 11 three step mechanism using propane as fuel -
This mechanism includes a limited oxidation rate of the fuel, since it is active
throughout the flame, and is represented by the following three rate equations:
-dfC 3 38/dt - A 1 (P/RT)nl-l. fC 3 RSnil.l f0 2 al e-El/RT (4)
0 n 2- 2+ a O C "
dfCO 2 /dr - A 2 (P/RT) 2 1 fCO 0( t) = fO2 2 .fH2 2 e E2/RT (5)
- u
dfH2O/dT - AI../RT)n fO 2 0 -6 3 /RT (6)
The second equation was proposed by Hottel [17; to describe his results for a stirred
* reactor.
The results from the present seriea of experiments were employed to determine values of
ths various constanti for Eqns. (4) to (6). If the three equations are valid, then at any
plane within the combustur, it should be possible to specify the amounts of C308, C0 2 , CO,
H2, 02, H20, and N2.
In order to derive the values of the various paramaters, test results were selected
that had a common equivalence iatio (0) within a narrow margin of fluctuation. Four groups
of these results have been formed on the lean side and three on the rich, with the :ollowing
mean 0 va. uea (Table I). The results ir these groups are mean vlues for a range of tests in
which 0 varied very slightly (t0,01).
from the equation representing T vs. i. This gives then the following rates:
dfCO,/dT - dfCO2 /dT dT/dr (7)
The derivative dT/dT is common to all three of ove equations. Taus the rates,
molar fractions, temperature and pressure in equati ), (5) and (6) were known fiom the
experimental results. The values of the parameters .ese equations were then obtained
by Isolating one parameter on the left side of one o! these equations, and taking for the
remaining parameters on the other side, values given in the -litetature for the first trial
[17 to 20). Theoe are given in TUble 1I. The, the iszleted paraseter with its calculated
value ,'ssshifted to the right aide of the equation and another parameter was transferred to
the left. This cyclic procedure was repeated for all parameters until the rates expressed
by the equations (4), (5) and (6) matched the values derived from the experimental results.
Since the variables (rate, molar fractions, pressure and temperature in equations (4),
(5) and (6) deoeed a wide range of fiae temperarule and 0, it was possible to demonstrate
by the above calculations that the parameters A ard E, are functions of T only whereas the
parameters n,a and e depend upon T and to a somewhat lesser degree upon 0. The hypothesis
of variable orders of reactions and variable activation energy R was put forward and discuss-
• d by Levy and Weinberg [21). This, however, is nor considered in the global reaction
mechanism where a slow rate-determining step is assumed to Jomihate the mechaiism [22]. The
results obta.ined
lesser degree of here.
0 havernfirmed this
been need to hypothesis, and
represent the polynomial(A,functions
parameters n, a, c. of T and
R)i4, to a
equationa
(4), (5) and (6). For the parameters the following expressions were obtained.
N-
* - ~ -~---- 'AA
= ~~S-'~~
IM
Table II
Data _taken from Literaturc and Used to Asses the Values of
the Parameters Used In the-Solution of Euns. (4). (5) & (6)_
Ifunction of Rih 2j of 0
4
A3 a exp[-14,7894 + 6,11699 x 10-2 T - 5,20148 x 10-STZ + 2,40499 x 10-8T3 -4,259o3 x 101 T~ 1
E3 -1.62708 x 104 - 21,8015 T + 1,34966 x 10-2 T2 + 2,22496 x 10-7 T3
n3--0,120686 + 0,0413256 0 + 1,79418 X 1 0 -3 T - 4,00207 - 10-7 -2
a3 -1,24458 - 0,0475454 0 - 1,0372 a 10-3 T + 3,67014 . 10-7 T2
3
A3 ap (37,9777 - 0,)494814 T + 3,47914 x 10-5 T2-
e 7,04779 x 10-9 T 1
E3 5,05266 x 10'1- 91,892 T + 0,071018 72 - 1,48654 x 10-5 T3
(P,435549 + 0,405397 0 + 4,12306 x 10-4 T + 1,96293 X 10-4"
3,74 - 7,74854 ft- 3,69989 X 10-3 T + 5,2833 x 10-3 0 T + 1,85779 02
5,5344 x i0 7 - 8,77687 X 10-022
-By analysing the foregoing expressions obtained for the parameters, one sees -that two
of equations are required to cover the range of reaction conditions investigated
-~-era
(Table 1), -one set for the-lean aide, i.e. 0 i51-and one set for the rich side, i. e. 4
0' . This has also been reported by-other workers [20].
-zzI
The parameters (A, n, a, el E) have the following numerical values for tne given
values of 0 and T. 3
Table III
Variation of the parameters with k and T
_d .__ 1.0 60 00
5 10 t 8 1
Aj 4,45406 43,1x0 6 2,05x10 13.xlO 43,17x10' ,05x1 14,62,106 l7,96xIO12.13xl0O
E2 929 16850 28430 929 16850 2840 ± 5 241 19
n2 1,49 1,85 1,96 1,46 1,61 1,87 1.31 1,57 1,83
a12 0,49 0,49 0,60 0,41 0,48 0.59 0,58 0,84 0,87
I phenomenon can be given at this time. In fact, all the values of the parameters were de-
I rived from best-fit equations, and must be Considered more as working tools rather than
beinE necessarily representative of the rcel phenomnna. However, the strong influence of
temperature In
these parameters has certainly a fundamental explanation which may be close
to that given by Levy and Weinberg 1211. They postulated that global meehanias are vali
only at temperatures close to the maximum heat release ra:.s and above. At lover tempera
tures it i3 almost ce:-tain that updicals assume an impor~a~t ro'~a. The values noted for
the experimentally det~rmined parameters probably reflect this latter Influence.
To determine the composition and temperasure profiles with the proposed model, a numer-
ical (simultaneous) Integration of Eqns. (4, 5 & 6) is required. ThIs integration was
rather complex and the convergence problems were dealt with by the law of propagation of
errors 1231. For most experinents, the observed temperatures were over 1000 K and for this
reason no attempt was made to develop the model bilow this temperature. Therefore, it was
assumed that between the inlet and the position marking 1000 K (as predicted by the model)
the temperature rise would be linear. The final step of the integration procedure was dnter-
mined by the position at which the C02 concentration corresponded to the theoretical.
adiabatic, dissociated flame temperature. This latter value was computed using the data
given in (24).
Altogether 10 steps of integration were used to cover the raage of C02 obtained fora
particular condition. The program for the integration of the rate equations consisted of
calculating the residence ti.0, the distance, the flame temperature by entpalpy balance, and
the composition (CO2. 01 2 02, N', H2 0, Ar, C3 HB) for a given in.~rease of C02 UP to its
msx~mum value. The rat equationis gave the values of 002, H2 0 and C3 Hs_; the re:.aining Com-
pone beIng defn by mass balance.
4. EXPERIMENTAL POIBTS AND DERIVED PARAMETERS
The data reported here are representative and Illustratr- typical results obtained from
a such larger amount of work [2]. A single expariment consisted of operating the rig at a
given condition long enough fox gas samples to be obtained from each of the five planes
downstream of Lhc recirculation zone. For each such experiment,, fuel and air-maa flows
were determined, as also inlet temperatures and pressures, humidity, external wall tempera-
tures, static and total pressures.
To establish the correlaticn parameters, 17 experiments were nade. These are tabulated
below (Table IV). In addition-ta these teats, a further 18 experisents wera performe' at _
various cdtsn. A major objective of the thesis 121, from ivhich this wori% is c ---
ad,
was to Investigate the affects of steam Injection upon premixed combustion. Many, but not
all, of these later [181 experiments concerned steam addition, and the model' was used to
make predictions, at_
Pigure 2 gVes-an example of avweak mixture test (expt. No. I)which was used to estab-
Al lish the' pasadteis- of-Table :111. the scatter of the model prediction lines is duec to the
IL
!' Z;WN- A-
averaging of values taken from all the axoerittenta Ron. 1-17. ?Igurc 3111 unZraes the
34raUlli ror a rich mixture (expt. No. 15). The remaining data exhibited similar scatter.
Table IV
3
6a a t 3 T .m :M3
Exp. No. 0 kg/s kg/s k~a K IExp. 110. 0 kg/s . kg/a kPa K
1. 0.697 0,218 9,73x10'3 103 312 1r 10 0.967 0,2741 1,76xl10' 1F03 298
2 0,705 0,284 1.28z10-2 103 31 11 0,979 0,269 1 67X10O-2 102 311
2
3 0,839 0,293 1.71~i3-
313 12 j1.016 0,180 1 14xl0- 103 309
2
4 0,852 0,177 9,68xl0-3 103 311 13 1.102 0.238 1,63xl10 I102 306
5 0,852 0,229 1,24xl102 103 305 14 1,077 0,175 1.23:10-2 103 310
6 0,877 0,271 1,51xl102 103 293 15 1,205 0,.181 1 ,39 10-'2 I103 309
7 0,890 0,237 1,36x10.2 102 306 16 1,2161 0:2161 1,70X10-2 103 311
8
9
0,919
0,934
0,292
0,178
1,71110-2
1,G6x1102
1-3
102
305
305
17
1
1370178
0
1 1____ 1,49x10-2 103 310
00,01c
V 0z
S.,.r, .o ,7 It
4.3
behaiou
to Fis 2
At thispoint t waso onsdrdta l aaeeso h oe a esetbi
Prior
to th stazt~r
in 0.7gton xe et erSm-aadeuigoar
Utobf ~ tesLt mte icc
Fro
reictio. A tapna se of prlbesecnsanorthmear conustitentst Ced in0 P0gs 5f
th 6 n generathe agreement ood ta
istee 2nd tlaremarimal asoin in Pars. god
Additionally,
he finalpointoftheintegraionofthemodl_(foreachofthemajorcons
4
2400
100
N I 1.0 C
~200Oi ;
16000r it 0 40.70- [
800 _
12007
_ _ _ _- d__ ?200
SS S 0 o*. 02 03 0 1 J..
02 0Xm6cc 0,6 0,8 ao, C
0.1.1
2200
I I
0. 030.3 0 0.2!
*. copu I-a. f S d.
t~- CV
-- of f- --- vtzi=MS -
A.'
It will be-seen :trhat (Fig. 7) the component predictions are acceptably good both for
the propane/air and propane/air/steam. The differences between the two tests are also
314 ,predictedquite quantitatively. Because the unburned components are quite small, the
effects are not so obviousifn terms of temperature (Fig. 8). In this latter case the
measured temperatures, although in reasonable agreement with the model predictions, do not
illustrate the differences suggested by the predictions.
These results (plus those of the 15 remaining tests) confirm the validity of the model
within this geometry. The effects of steam on combustion, both from the viewpoint of com-
bustion analysis and model prediction/anaisis will be presented elsewhere. Suffice to say
that :he prediction scatter is very similar to that illustrated here.
The present model represents a compromise between the simple one-step 'global' reaction
model and the complex multi-equation kinetic schemes. It predicts all the stable substances
found In hydrocarbon/air mixtures with the exception of NO X . This latter may be estimated
independently by one or more of the empirical methods suggested earlier [3 or 4J or by other
published techniques. The accuracy of prediction within the limits of the experimental
conditions investigated is very satisfactory, both in terms of components and t'mperature.
Because of its relative simplicity, the computing time is small (in comparison with
many kinetic schemes) and it offers potential for combining with a suitable mixing nodal.
Like most other correlation formulaet the validity is always In some doubt due to the
limited experimental data. In particular, it would be nice to test the data at pressures and
inlet temperatures other than atmospheric. Unfortunately this is not possible with the
existing test facilities.
An attractive possibility would be to attempt to apply the model to predict the perform-
ante of a pre-mixed, pre-vaporised gas turbine combustor. Unfortunately, to date, the
published information has been insufficient to enable predictions to be made.
Ram-set combustors also offer attractive possibilities for this type of model since
there is neither fila-cooling or dilution problems. The aemi-prevaporisation/premix of a
ram-jet may make the model directly applicable. In the present work the results have been
presented in terms of a distance function, X. but it may well be more satisfactory to
express the results in terms of residence times.
7. CONCLUSIONS
1. A three-step model has been proposed to describe the combustion with a fully premixed
propane/air system. Some 510 different experimental res6lts were used to define the
model.
2. After the above derivation, the model was used to predict a further 540 experimental
points comprising propane burning in air or air + steam. The predictions are in sensi-
ble ajreement vith' the experimental values even though those conditions having steam
addition are outside the frame of the model calibration.
3. The model is capable of predicting CO, H2O. CO, H 2 , C 3 H8 , 02, N2 and Ar.
4. There is potential application for the model both in the field of ram-jets and, the gas
turbine.
8. ACKNOWLEDGEMENTS
The financial assistance of the Canadian Natural Sciences and Engineering Research
Council is gratefully icknowledged.
REFERENCES
1. Roy. P., Sehlader, A.?., & Odgers. J-. "Premixed Combustion in a Baffle Combutfl and
the Effect of Steam Injection", J. of Engineering for Power, pp. 387-393 Oct. 1976.
2. Roy, P., 7'odelage et Etude de la Combustion pr m-langie avec et sans Injection de Va-
peur:, Thise prfsenthe I l'Ecole des graduis de l'uiversit! Laval pour l'ohtention
do grade 'pbiloaophise-doctor', 1980.
3. Lipfert. F.M., "Correlation of Gas Turbine Emission Data", ASHE Paper No. 72-ST-E0.
_ 4. Kretschmer D., i Udgers, J., "Modelling of Combostors: The Effects of Ambient Condition:
3- upon Performance", ASME Paper No. 73-IAGT.-6.
5. Edelman, R.B. & Fortune, U., "A Quasi-Global Cheical:*inetie Model for the Finite Rate t.
f-Combustion o Hydrocarbon Fuels', AZA, Paper 69-86, i969
/7
6. Mellor. A.M., Blger, R.W.. Shialer, R.A. & Tuttle, 3.N., "Enissioni fro= Aircraft Fuel
Nozzle Flames", Report No. PURDU-CL-75-C4, July 1975.
8. Pratt, D.T., Wormeck, 3.3., "CRECK - A Computer Program for Calculation of Combustion
Reaction Equilibrium and Kinetics in Laminar or Turbulent Flows", Washington State
University Report WSU-ME--IEL-76-1, 1976.
9. Spalding, D.B., & Serag-Eldin, M.A., "Prediction of the Flow and Combustion Processes
in a Three-Dimensional Combustion Chamber", ISABE, Munich. March 1976, 489.
10. Edelman, R., & Economos, C., "A Mathematical Model for Jet Engine Combustor Pollutant
Emissions", AIAA/SAE, 7th Propulsion Joint Speciailitt Conference, Paper No. 71-714,
June 1971.
11. D'Soura, N.V., & Karim, G.A.. "An Analytical Study of Methane Oxidation in a Steady -
Flow Reactor", Comb. Science & Tech., 3, 1971, p. 83.
12. Kollrack, R. & Aceto, L.D.. "Recirculation Effects in Gas Turbine Conbustors", 95th
Winter Annual Meeting of the ASME, Pape'r No. 74-WA/GT-3. Nov. 18, 1974.
13. Longwell, J.F., & Weiss, M.A., "High Temperature Reaction Rates in Hydrocarbon Combus-
tion". I. & E.C., 47, No. 8, p. 1634.
14. Avery, W.H., & Hart, R... " Conbustor Performance with Instantaneous Mixing', 1. & E.C.,
A5, No. 8, 1953, p. 1934.
15. Essenhigh, R., "An introduction to Stirred Reactor Theory Applied to Design of Co=bus-
tors, Comlustion Technology", Pennsylvania, State University, Some Modern Developments.
1973.
16. Kretschser, D., & Odgers, 3., "Modelling of Gas Turbine Combuastors - A Convenient Reac-
tion Rate Equation". ASE, J. Eng. for Power, July 1972, p. 173.
17. Hottel, H.C. et al, Tenth Symposium (International) on Combustion, p. 111, The Co=bus-
tion Institute, 1965.
18. Knzlov, G.I., Seventh Symposium (international) on Combustion, p. 142, The Combustion
institute, 1958.
19. Avery, W.N., and Hart, R.1;., Ind. Eng. Chem., Vol. 45, p. 1634, 1953.
20. Kreaschner, D., and Odgers. 2., journal of EngineerIng for Power, ASME, p. 173. July
1972.
21. Levy, A., and Weinberg, F.3., Combustion and Flame, Vol. 3, p. 229, 1959.
2 Fristrom, R.M., and Westenberg. A.A.. Flame-Structure p. 326, McGraw-Hill, 1965.
26. JARAF Thermochemical Tables. 2nd Edition, N.B.S., U.S. Dept. of Commerce, June 1971.
I
I Nif,
I m t -~~ im mm•mm ld
31RIO
DISCUSSION
R.Monti
I wonder if the ve-ry larpe ariation of the "constants" A,.E. with temperatures up to 3-4 order of magnitude for
A,. does not make your model.questionable.
Asn'xors Reply
I do not think the model is questionable within its constraints. Variations of E. A and n have been noted by various
workecrs other than ourselves. An example is cited in the ter,(Ref.12). I believe that it niejit stemn from
representing arnultitue. f reactionsb yasingle "globar" reaction. The one-equation model (Ref 16) also has a
variable E and a. but this equation has bad a very widespread application to a large number of experimental condi-
tions and types of comnbustion equipment. Also. I believe that Essenhgh accepts this concept. For this reason. Ida
not think that the model is questionable from this vriewpoint. It is certainly an enmpirieal model with pseudo-
costlants, but that does not necesarily make it unreliable as a tool.
.J.Winter
Whfat do you onticipate will be the effect of inlet temperature and pressure on the ~cntnsdrved for the
model?
I I
Author-s Reply
It is difficult to answer this question without the experimental data. However. if I had to make a prognostieetion of
the effects of inlet temperature and pressure. I would utilize the c-orrections sikrgecsted for our ealier one-step
equation. These may be found in References 16 and 20. The- basis for this thlazglst would be the similarities of the
4 form of the present equations Ic the older one. I would again stress that it is only a suggestion and experimental
* confirnation would be requited to give a meas; re of confidence.
tet
IA
ALs
BELGILM ITALY
Ctmdnraten AGARD Va1 Arna~lutk Miore-
LW~bzpedo IhViv AaxmW U1ffsio dd Ddlcpto NaziuicaWMAAKD
Qualtier Mine U.ntbcth 3. Pwr* Adcnnr
am 41-90c. 1HOBierRomJW
CANADA WIIDR
Ddceface In(.nouW= Scstitr See sc%=M
Dtpfmenr of %ational We=n NETHERIJWDS
Ottawa. Osarlo KIAOMehrad cktw oAA
DENMARK Naicaui Arnnspnc La9=tuy.lOS
atasDeftmec Raert Board PDO. Pon 1:6
OfnvpdeKwree 260 A.C. DdP
0 iNORWAY
FRK'CE Nonqian Deft-= Reseat Estbmeza
OSN-EL jnecia Main Lhw
V9 Aenu delIsniDW= toet P.O. Box 25
92320 Chbli sot a gmt M?00KIM~e
GEXMANY -PORTUGAL
If coias --f the orjginal ubliafftn -r not avail",le v- tbine eMks.te fonga"Mg1 t Vreardf
ISBN 92-835-0312-X
donRNVIetoAR
tdxii±~
of
~or~~ duo=d
~ Ncks the
:Scnfaouinc hztDoC~a -
Ccwes. Fmffci,- cnpn re
______P LUXMBOj
NOWAY
&rr-Rrt:h esttdxe*
M~rb
C -L N-2aD Ke.
I Ann:C AARDMAn4N~
Rmrz *f a-ndNwotnARGIDI
=1d l
If m
UNMEWSTATFS
A=-: RnrWD=-:&=a&~
nmri
-r~flIISE
Z.-
MA __
IS 2S5-32-