You are on page 1of 9


Defense Technical Information Center Compilation Part Notice
TITLE: Application of Astrium' s CryoROC Code to a Single Injector Problem. A Contribution to the RCM-3 Mascotte Test Case [60 bar] DISTRIBUTION: Approved for public release, distribution unlimited

This paper is part of the following report:
TITLE: 2nd International Workshop on Rocket Combustion Modeling: Atomization, Combustion and Heat Transfer held in Lampoldshausen, Germany on 25-27 Mar 2001 To order the complete compilation report, use: ADA402618
The component part is provided here to allow users access to individually authored sections
f proceedings, annals, symposia, etc. However, the component should be considered within

[he context of the overall compilation report and not as a stand-alone technical report. The following component part numbers comprise the compilation report:
ADP012355 thru ADP012373


Application of Astrium's CryoROC Code to a Single Injector Problem A contribution to the RCM-3 Mascotte Test Case (60 bar) Josef Gbrgen. The motivation for this work thus came from two points: on one hand. 2D. It's rather a question of whether the code's predictions are efficient. finite volume SIMPLE algorithm (pressure correction) turbulence controlled (Eddy Dissipation - Concept) kinetically controlled (Arrhenius) multi-step global reaction schemes * hydrogen/oxygen * * Lagrangian particle tracking (Stochastic Separated Flow model) * implicit Stone solver - multi-class. One the other hand allows the workshop a further assessment of the CryoROC computation results by comparison with experimental data and other computations. CryoROC treats the gaseous phase by solving the Favre-averaged Navier-Stokes equations extended by the species continuity and k-s turbulence Capabilities Capabilities (cont. The CryoROC spray combustion CFD-code was developed and intended to simulate the flowfield and the heat exchange within existing and future cryogenic rocket thrust chambers.. Deutschland Introduction. which are anyway included in the spray initialisation process? The numerical code CryoROC Thrust chamber flows of cryogenic hydrogen/oxygen rocket engines are characterised by the coexistence and complex interaction of various physical phases. Astrium's advanced multiphase Navier-Stokes solver. H. species diffusion * compressible . CO 2. 81663 Mtinchen. 02. Motivation and Objectives This paper presents a numerical analysis of the RCM-3 Mascotte test case (60 bar) with CryoROC. Oliver Knab Astrium GmbH. momentum and heat coupling with gas phase supercritical LOX gasification model Grid * * structured non-orthogonal curvilinear e Table 1: Specification of CryoROC (Cryogenic ROcket Combustion) Code .1- .and supersonic o turbulence models - standard k .) o * viscous heating. CO. the computational approach so far excluded too much effort on precise modelling the vicinity of the injector head with its single elements. answering the question whether it is necessary to resolve the injectornearfield flow phenomena en dctail to predict global engine characteristics as the wall heat transfer accurately? Or if that is only of minor importance because there are only negligible differences a short distance further downstream the injector head. For that purpose. N 2. i. together with a dispersed oxygen droplet phase (2nd phase) have to be resolved efficiently. fast and precise enough regarding the whole thrust chamber and its global characteristics. whether the code is capable to resolve the complex combustion phenomena in detail near the injector head. with wall functions 2 layer model * .sub-. standard Jannaf property data base for gaseous combustion species (Gordon & McBride) compressibility effects o multi-gaseous species consideration (H2. . H20.. discrete particle injection and sequential tracing approach mass.e. This procedure is intended to analyse and to evaluate the impact of the injector-nearfield flow evolution on the overall combustion process. IP34. A reactive multi-species gas mixture (1 st phase). The CryoROC code is an important tool used in the thrust chamber layout process at Astrium. it would be an interesting question...) * chemical reaction models - porous walls and crack simulation coupling with Astrium's RCFS (Regen- erative Coolant Flow Simulation) code Numerics . trans. bi-propellant. axisymmetric. OH. Space Infrastructure. when the computational mesh is adequately refined in that area.etc.

H2 H liquid/liquid... because a detailed resolution of the single injector elements is not possible within that context.g. oxygen droplet tracking is performed every 70 to 200 gas phase iterations. The hydrogen. 0%0 SH2 66single "I LOX element 2. CryoROC allows to simulate multiple discrete phases in a Lagrangian frame of reference. of course. [1 2 . For a more detailed description of the models the reader is referred to [1]. Computational Results Since the CryoROC code so far has only been applied to entire thrust chamber configurations.e. [4 4.0 + bH + cOH) basing on a turbulence (EDC) and/or kinetically (Arrhenius) controlled combustion model. see e. H2 single element 3. both central and upwind differencing schemes are applied.[ 3]. primary atomisation processes and 3-dimensional effects have to be excluded to enable an efficient handling. As a thumb rule. oxygen as well as hydrogen. The approach chosen corresponds with concept #1 in Figure 1 and is referred as standard in the following. In addition to solving transport equations for the continuous gas phase.: o - ~ ~ "---- C 00 :O LOX injector row LOX * approach for axsymmoetric modeling 3: 3 liquid/gas. Here the H. boundary fitted grids and solved by an implicit algorithm. as well as droplet-to-gas phase turbulent interaction. some simplistic assumptions are necessary to take. H and OH). Table I gives a survey on the most important modelling features of the CryoROC software. CryoROC computes the trajectories of these discrete phase entities by integrating their force balance. Applied to the "Single Element Problem" The other major assumption made for standard thrust chamber computations lies in the fact that both. Gaseous and dispersed phase calculations are coupled in a loosely manner. O. the spray initialisation has to get adapted. H20. premixed thrust chamber mr O O section of SOo0 . The latter include appropriate modifications accounting for compressibility effects and handle the near wall region optionally by a logarithmic wall function approximation or by a two-layer approach. Hereby. The set of equations is discretized according the finite-volume methodology for non-orthogonal. Up to now. in order to receive a certain degree in reliability and efficiency. CryoROC allows for transient LOX droplet heat-up. co-axial liquid/liquid. a single-step. but above all does that enable CryoROC to simplify the -I- . In particular. 1. are injected in the chamber as dispersed and liquid droplets. as already hinted in Figure 1. In particular for LH2/LOX systems. It might not be quite as appropriate for single element calculations. and LOX are perfectly mixed throughout the injector element diameter so that the global mixture ratio fits exactly. co-axial Figure 1: Spray Initialisation Concepts I. source terms in the respective governing equations are not updated simultaneously. i. Because of the finite number of injector elements inside an injector row. This has several advantages. The reaction mechanism of cryogenic hydrogen/oxygen systems is represented by 5 species (H. global reaction scheme is employed (H2 + x02 z: aH. is evaporated instantly.equations.3. These second phases consist of spherical particles representing propellant droplets of different sizes being dispersed in the continuous gaseous phase. this approach is assumed to be rather appropriate and realistic for axisymmetric thrust chamber computations. Firstly. supercritical LOX gasification. .

have a dramatic effect on numerical stability and convergence behaviour. characterised by gaseous or dispersed hydrogen. This ensures the applicability of the concept #1. which could be too high. injection concept #3 is the standard for spray combustion codes. The Figures 2 and 3 show the computational mesh. With concept #2. As one can recognise from Figure 1. Both measures. have been applied and their results compared. But what is more important here. Moreover. Reducing the incoming turbulence should lead to less diffusion and therefore to a flow field similar to concept #2. pre-mixed or co-axial injection. co-axial). is the fact that changing boundary conditions from wall to incoming mass flux condition.e. but as it is shown. however. it may be suitable to reduce the 02 .2 m downstream. in order to simulate the gaseous H2 inflow. As illustrated by other test case contributions. To take into account for thrust chamber applications' unusual single element configuration. but after about 0.boundary condition at the injector head. Best results. Especially the convergence criterion is only reached about a factor 10 later. This is probably due to the specified inflow turbulence (5%). but less appropriate to simulate axisymmetric spray combustion within a multi-element combustor. To elongate the (unrealistic) combustion zone. This becomes even more clear in Figure 7. the flame front is clearly resolved and the flame angle coincide fairly well with that resulting from Abel-transformed emission imaging (= 40). Propellant inflow is realised solely by spray initialisation. i. the temperature levels became very similar in all three computations. the anchoring of the flame front. The flame could be resolved best by the concept #2 (dispersed H2. different correcting measures are applied. One can recognise different patterns at the start due to the different spray initialisation. to overcome the delayed combustion onset in concept #1. as far as we think to know. Concept #3 at last removes that restriction too. co-axial) showed the worst convergence behaviour (about a factor 10) and is therefore ruled out for complex thrust chamber simulations. the way that now gaseous H2 and dispersed 02 are injected co-axial in accordance with the injector geometry (Figure 1). The important point here to notice is the fact that these discrepancies will vanish after a certain distance. it can be stated that this approach is far too time consuming and therefore ruled out for complex thrust chamber design applications. Concept #2 and #3 are assumed to be more realistic approaches to simulate axisymmetric single injectors. by the way.e. the concept #2 gives room to a more detailed resolution of the injector-nearfield. the mesh resolution still isn't fine enough to resolve the LOX post and the taper geometry and this. Here. so that the wall remains closed and only an adaptation of the conservation equations' source terms is needed. these boundary condition infringements could be avoided. is not captured closer to the wall in neither computation (see Figure 4). it proves that one can satisfy industrial needs without resolving the flow phenomena in the injector near-field. the flame length seems to be realistic and compares well with what is known from experimental observations. The alternative concept #3 looks alike. violate the inflow momentum ratio between fuel and oxidiser when a real gas approach is not taken into account. different methods have been elaborated and applied. which is not so much a surprise. i. three different propellant initialisation concepts. Figure 4 reveals the temperature flowfield for all three initialisation concepts. The OH mass fraction contours (Figure 5) give the same tendency. As expected. Besides. is assumed to be responsible for the fact that the onset of the combustion process.e. i. the far-field of the injector is only slightly affected by the individual -2- . Since we have to keep in mind the accessibility of the code to actual design problems with "overnight" reaction times demanded. delivers us concept #2. H2 and LOX are now initialised co-axial but still as droplet sprays with instantaneously evaporating H2 . where the (cross sectional averaged) axial temperature profiles are shown: the discrepancies exist at the start-up. Apart from the standard concept #1. in order to contribute to the Mascotte test case and to apply the CFD code to a single injector problem. but mixture happens much sooner which leads to a much smaller combustion zone compared to concept #2. The concept #3 (gaseous H 2. concept #1 is not suited to resolve the flowfield phenomena in the vicinity of the injector. therefore. The development of the temperature profiles along the chamber towards the throat is shown in Figure 6. The most popular are to increase the LOX droplet injection velocities (SNECMA) or to split up the oxygen inflow into a liquid and a gaseous portion (CNES). Conclusion The single element Mascotte test case RCM-3 has been calculated with Astrium's multi-phase Navier-Stokes code CryoROC. After all.mass mean diameter and hence to adopt the spray characteristics to the assumed pre-mixed conditions.

Wiedmann D. that since there aren't any reliable experimental results available yet. Knab 0.02 003 X [mI 004 0.40 Figure 2: Computational Mesh 512x 101 I I I 0. AIAA 99-2459. Estublier D. Effective Calculation of Multiphaseflow Fields in Liquid Rocket Thrust Chambers. 1999.. Injector Near-Field and Boundary Layer -. [2 ] Preclik D.spray initialisation concept and this outlines the applicability of concept #1 (dispersed H.00 0. Note.. AIAA-2001-3407.04 0. Design Supportfor Advanced Storable PropellantEngines by ROCFLAMAnalvses. Frihlich A. 3- . 1995 [131 Preclik D.30 0. Cryogenic Rocket CalorimeterChamber Experiments and Heat Transfer Simulations. AIAA 95-2779.06 Figure 3: Zoom of the Computational Mesh.. An Eulerian-Lagrangian Approach to Spray Combustion Modellingfor Liquid Bi-PropellantRocket Motors.01 0. and Kretschmer J. References [1] Knab 0.00 0.05 0.. and Wennerberg D. AIAA 98-3440.20 x [m] 0. a thorough assessment of the CFD results seems to be obsolete at this point. 2001 0.10 0. and Wennerberg D. 1998 [4] Wennerberg D.. pre-mixed) to complex thrust chamber simulations. Oechslein W..

12000 1 1800 1 700 1800 1500 1300 " t' 61100 T [K] 00000 X 8in 200 9 ~1400 "1200 J" ' ..10 0.15 T [K] Figure 4: Temperature Contours for Spray Initialisation Concept 1 (top). "~p.00 0.P. 2 (middle) and 3 (bottom) -4- .' 300 7 * 600 3400 3200 3000 2800 000.15 T [K] 3800 3600 3400 3200 3000 2800 e 0.50.10 0.05 X [i l0.

075 0.070 t Pvl- I0.0001 x Vhr0. 0.055 0.15 0. 0.005 0.035 0. i0.045.OH mf No-0 0.020 0.070 0.065 0.025.01 0 0.050 0.030' 0.055: 0.000 0.035( 0.0 15' 0.030 0.065 0.015 010 U *0.05 Figure ~ ~~ ~ otur ~ o ~ocp ~ tp.060 0.060 0.050 0045.2 ~ 5: O mdl)ad3(otm ~ asFato ~ ~ ~ ~ O -5-8 .005' L ~0.025 0.040! 0. 0.05 x t - 01 OH mf !0.040 0.080.

50 mm cc-axI. liq i q.11q. 1..30 i 30 I / ___ -- 1..... . / -20 -20 : it 1000 2000 Temperature [K] 3000 4000 0 1000 2000 Temperature [K] 3000 4000 "300 30 x= 150 mm 30 x = 200 mm m o d.. 2. .klpmml~d•lS X= 100mm 30 .. Iiq g I 1. L u..don&MBdde I ..-..I. G f.- /* 20 --- a.. . c -. 00 __ ... 1.50mm 2. f0 l g 0 - 10 10 I "... L -200 mm ý15 mm 3.Id.. -2.ikq. -o .d 20 - - - -. dxx=20 .iq. W...2 50 m . liq .mm 2m2 .. 3.gscal al. p. ...Gordom&McBride.x= 50 . . 0q.. 2- m i xed .g.i 20 = l ...a .=30 0mm 20 MM x250 -3.1iq.lq. _. .i i ..._ _mb_•_ iOt ...0 -10 -10 I! .3 3 ..go go I! -10 -10C I -20 -30 1 1 1 1 -20-30• 1000 2000 Temperature [K] 3000 4000 1000 2000 Temperature [K] 3000 4000 x = 250 mm Figure 6: Temperature Profiles at Different Axial Locations -6- x = 300 mm ..1 10 10 l_ .• 0 mm 11•.ic -10 -10 i - 10 10m -20 =1 0 m -20 0 1000 2000 Temperature [K] 3000 4000 0 1000 2000 Temperature [K] 3000 4000 x= 50mm 30 1. fiz. i....1..gas. 20 --.1lkh co Go. liq. . ---3._____ . p r1. ld I. 300 mm 3.. p rnmix -. ____. co .

._ - 900--700 • g -600 Gordon & McBride.1 0.1500 1400 1300 _ o.liq.. liq.._.2 0.. _/ / .4 0. co-axial _ _ _ _ _ _ 800 500 400 300 200 100 00 0....5 -7- .liqprmixed 2..gas. Ik~l. liq..... 1200 1100 1000 . 3..3 x [m] Figure 7: Cross Sectional Averaged Temperature Profiles 0.. 1 d I. co-axial -----.