1-s2.0-S1540748910003202-main

8/11/2019 1-s2.0-S1540748910003202-main

1/7

CFD analysis of the HyShot II scramjet combustor

C. Fureby a , , M. Chapuis a , E. Fedina a , S. Karl b

a Defense Security Systems Technology, The Swedish Defense Research Agency FOI,SE 147 25 Tumba, Stockholm, Sweden

b Institute of Aerodynamics and Flow Technology, German Aerospace Center, DLR, 37073 Goettingen, Germany

Available online 21 September 2010

Abstract

The development of novel air-breathing engines such as supersonic combustion ramjets (scramjets)depends on the understanding of supersonic mixing, self-ignition and combustion. These aerothermochem-ical processes occur together in a scramjet engine and are notoriously difficult to understand. In the presentstudy, we aim at analyzing the HyShot II scramjet combustor mounted in the High Enthalpy Shock TunnelGo ttingen (HEG) by using Reynolds Averaged Navier Stokes (RANS) and Large Eddy Simulation (LES)models with detailed and reduced chemistry. To account for the complicated ow in the HEG facility a zonalapproach is adopted in which RANS is used to simulate the ow in the HEG nozzle and test-section, provid-ing the necessary inow boundary conditions for more detailed RANS and LES of the reacting ow in theHyShot combustor. Comparison of predicted wall pressures and heat uxes with experimental data showgood agreement, and in particular does the LES agree well with the experimental data. The LES resultsare used to elucidate the ow, mixing, self-ignition and subsequent combustion processes in the combustor.The combustor ow can be separated into the mixing zone, in which turbulent mixing from the jet-in-crossow injectors dominates, the self-ignition zone, in which self-ignition rapidly takes place, and the turbulentcombustion zone, located towards the end of the combustor, in which most of the heat release and volumetricexpansion takes place. Self-ignition occurs at some distance downstream of the injectors, resulting in a dis-tinct pressure rise further downstream due to the volumetric expansion as observed in the experiments.The jet penetration is about 30% of the combustor height and the combustion efficiency is found to be around83%.

2010 The Combustion Institute. Published by Elsevier Inc. All rights reserved.

Keywords: Scramjet; HyShot II wind-tunnel experiments; Large Eddy Simulation; Supersonic mixing; Self-ignition

1. Introduction

The development of reliable hypersonic ightvehicles requires the solution of many technicalchallenges associated with the comparatively smallnet thrust at supersonic or hypersonic ight speeds.One of the more essential issues is the design of an

air-breathing propulsion system capable of operat-ing over the wide range of Mach (Ma) numbersdesired to facilitate the advancement of high-speedight and space access vehicles. For ights in thesupersonic (3 < Ma < 5) regime a ramjet, in whichthe ow is decelerated to subsonic levels before itenters the combustor, is favored whereas for ightsin the hypersonic (5 < Ma < 15) regime a super-

sonic combustion ramjet (scramjet), in which owthrough the engine remains supersonic, is pre-ferred, [1]. Below Ma 3 a turbojet engine can be

1540-7489/$ - see front matter 2010 The Combustion Institute. Published by Elsevier Inc. All rights reserved.doi:10.1016/j.proci.2010.07.055

Corresponding author. Fax: +46 8 5550 4144.E-mail address: [email protected] (C. Fureby).

Available online at www.sciencedirect.com

Proceedings of the Combustion Institute 33 (2011) 23992405

www.elsevier.com/locate/proci

Proceedingsof the

CombustionInstitute
http://dx.doi.org/10.1016/j.proci.2010.07.055mailto:[email protected]://dx.doi.org/10.1016/j.proci.2010.07.055http://dx.doi.org/10.1016/j.proci.2010.07.055mailto:[email protected]://dx.doi.org/10.1016/j.proci.2010.07.055

8/11/2019 1-s2.0-S1540748910003202-main

2/7

used to accelerate the vehicle to ramjet speed,whereas a rocket motor is required to acceleratethe vehicle to scramjet speed. To facilitate the con-tinuous operation from subsonic to hypersonicspeeds dual-mode ramjet/scramjet engines havebeen proposed [2]. Although the ramjet technology

is mature enough for working engines, the scramjettechnology, and in particular the scramjet combus-tor technology [3], is critically relaying on improvedunderstanding of the aerothermochemistry in thescramjet engine.

At scramjet conditions fuel is supplied to thesupersonic air-stream through the combustorwhereafter mixing, self-ignition and combustionare supposed to complete the process of super-sonic combustion. Most experimental researchon supersonic combustion has been conducted inground based research facilities [46], but recentlya few scramjet powered vehicles have own for afew seconds [78], generating pioneering data. Inthis study, we use numerical simulations to repro-duce and analyze, in detail, the processes takingplace in the HyShot II combustor under ight-likeconditions. Flight test data [9], and experimentaldata from the High Enthalpy Shock Tunnel Go t-tingen (HEG) of the German Aerospace Centre(DLR) [10], in particular, will aid in validatingthe computational models and aid in the analysis.Reynolds Averaged Navier Stokes (RANS) mod-els, using presumed-shape Probability DensityFunction (PDF) combustion models [11], andLarge Eddy Simulation (LES) models, using thePartially Stirred Reactor (PaSR) turbulence chem-istry interaction model [12], will complement eachother in simulating the ow in the HEG nozzleand test-section, and in the HyShot II scramjetcombustor. Accurate predictions of mixing, self-ignition and subsequent combustion in theHyShot II scramjet combustor requires that theintake ow is accurately predicted, and thus thatboth the nozzle and test sections are included inthe numerical model. A zonal approach isadopted in which RANS is used to predict theow in the nozzle and test sections thus providingaccurate inow boundary conditions to the moredetailed combustor RANS and LES.

2. HyShot HEG experimental studies and ighttests

In the HyShot II ight experiment [8], a two-stage Terrier-Orion Mk70 sounding rocket carriedthe payload to an apogee of 315 km after which the

vehicle was turned and gravitationally acceleratedduring the descent phase. Hydrogen (H 2) super-sonic combustion data was taken at ight Ma num-bers between M = 7.6 and 7.8 in an altitude rangefrom 35 km down to 23 km. Post-ight analysiswas performed in the free piston shock tunnel T4at the University of Queensland [9], and in theHEG [10], and here we consider HEG test condi-tion XII, representing the HyShot II freestreamconditions at an altitude of 32.5 km. The owpathgeometry of the wind-tunnel model is identical tothe ight conguration and consists of an intakeand two combustion chambers as shown in Fig. 1.The intake ramp is a wedge of 18 half angle andis 100 mm wide whereas the cross-sectional dimen-sions of the combustion chambers are9.8 75 mm2 . Gaseous hydrogen is injected per-pendicular to the combustor air-stream at four fuelinjectors 58 mm downstream of the combustionchamber leading edge. Each combustion chamberis 232 mm long, and attached to a single-sidedexpansion forming the thrust surface. Betweenthe intake ramp and the combustion chamber, sideand oor bleeds are used to spill the boundary lay-ers from the oor and the side-walls of the intake,allowing the leading edge cowl shocks to pass thecombustion chamber. Compared to the ight teststhe HEG experiments offers the possibility of con-trolled conditions and more detailed measure-ments, which is the main reason for selecting theHyShot experiments as the target for this study.In the HEG combustor, 14 pressure transducersand 10 thermocouples (cf. Fig. 1) are used to mea-sure the pressure and wall heat-ux.

3. Mathematical and numerical modeling

Due to the complexity of simulating the react-ing ow in the HyShot scramjet in the HEG facil-

Fig. 1. The HyShot II wind-tunnel model with fuel injectors and data acquisition system.

2400 C. Fureby et al. / Proceedings of the Combustion Institute 33 (2011) 23992405

8/11/2019 1-s2.0-S1540748910003202-main

3/7

ity under ight-like conditions, a zonal approachhas been adopted [13], in which different simula-tions have been performed for the different partsof the HEG facility as shown in Fig. 2a. The rstsimulation aims at evaluating the HEG nozzleow conditions (velocity, pressure, temperature

and composition) discharging into the HEG test-section, where the HyShot model is mounted. Thissimulation was performed in an axisymmetricdomain (with 20,000 cells) using the non-equilib-rium ve-species, ve-step reaction mechanismof Gupta et al. [14], and a steady-state compress-ible RANS model implemented in the TAU code[15]. The results agree well with experimental dataand provide the boundary conditions for the sec-ond simulation of the HEG test-section contain-ing the HyShot wind-tunnel model [13], whichwas performed in the planar domain of Fig. 2a,with 115,000 cells, adapted to the resulting shocksystem and boundary layers. The results from therst (nozzle ow) simulation are used as fareldboundary conditions for the second (test-section)simulation in which the 18 inclined intake rampand combustion chamber are modeled as no-slipand isothermal walls. Wave-transmissive outowboundary conditions [16], are used at the oorbleed outlet and at the HyShot scramjet combus-tor outlet. To emulate ight conditions the entireconguration was mounted at 3.8 angle-of-attack. The same computational model as usedfor the rst (nozzle ow) simulation is used. Thevelocity, temperature, pressure and compositionproles at the cut-plane x = 0.360 m, shown inFig. 2b, will act as the inow conditions to theHyShot combustor, the analysis of which is theprimary objective of this study. For the HyShotcombustor simulations both RANS and LESmodels have been used to capture different aspectsof the physics and to evaluate these models.

The RANS and LES models are based on thesame reacting ow model in which the mixture isassumed to be a linear viscous uid with Fourierheat conduction and Fickian species diffusion. Inthe RANS Blottner curve-ts with the Wilke mix-

ture rule and a modied Eucken correction withthe HerningZipperer mixture rule were appliedfor thecalculation of viscosityand thermal conduc-tivity [13]. In the LES, the viscosity is computedaccording to Sutherlands law and the thermal con-ductivity and species diffusivities are computed

using theviscosityandconstant Prandtl andspeciesSchmidt numbers, respectively [16]. The mixturethermal and caloric equations of state are obtainedunder the assumption that each species is a ther-mally perfect gas, with tabulated formation enthal-pies and specic heats, respectively. The reactionrates are computed from GuldbergWaages lawof mass action by summation over all participatingreactions, with rate constants obtained from amodied Arrhenius law. For RANS, the nine-spe-cies, 19-step Jachimowski mechanism [17], is usedwhereas for the LES the seven-species, eight-stepmechanism of Davidenko et al. mechanism [18], isused, resulting in the same ignition delay timeprediction.

The RANS model used [11], is based on theensemble averaged version of the governing equa-tions with the Reynolds stress tensor and ux vec-tors modeled by the two-equation Wilcox k xand the one-equation SpalartAllmaras models[19]. Here, only the SpalartAllmaras results arediscussed. The ensemble averaged reaction rates,including the turbulence chemistry interactionterms, are modeled by a presumed-shape PDFmodel based on a Gaussian distribution for thetemperature and a multivariate b-PDF for the spe-cies mass fractions [11]. In addition to the ensembleaveraged mass, momentum, energy, species andturbulence equations this requires the solution of two additional equations for the variances of thetemperature and the sum of the species mass frac-tions [20]. The equations are solved using an expli-cit multi-stage Runge-Kutta scheme using a point-implicit treatmentof thechemical source terms andlocal time stepping for convergence acceleration.The AUSMDV ux vector-splitting scheme andthe MUSCL gradient reconstruction algorithmsare employed.

Fig. 2. Computational models. (a) Schematic of the HEG and zonal partitioning of the computational model and (b) theresulting combustor inow velocity, pressure and temperature proles. The proles are normalized by v0 = 1800 m/s,T 0 = 1459 K and p = 53.0 kPa.

C. Fureby et al. / Proceedings of the Combustion Institute 33 (2011) 23992405 2401

8/11/2019 1-s2.0-S1540748910003202-main

4/7

The LES model used [12], is based on the low-pass ltered version of the governing equationswith the subgrid stress tensor and ux vectorsmodeled by the mixed model [21]. The lteredreaction rates, including the subgrid turbulencechemistry interaction terms, are here modeled by

the PaSR model [22], in which the ltered reactionrates are evaluated as the product of the reactionrates of the ne structures and their volume frac-tion. This model has also been carefully validated,cf. [12] and references therein, with very goodresults for a wide range of applications. The equa-tions are solved using a fully explicit nite volumediscretization, based on the C++ library open-Foam [23], utilizing a monotonicity preservingux reconstruction algorithm [24], and two-stageRunge-Kutta time integration.

4. The HyShot II supersonic combustor model

Thecomputational model of the HyShot scramjet combustor used is presented in Fig. 3. At theinow (x = 0.360 m) Dirichlet conditions are usedfor all variables with proles ( Fig. 2a) resultingfromthe test-section simulation. Isothermalno-slipwall boundary conditions are used to represent thetop and bottom walls of the combustor whereassymmetry conditions are used in the spanwisedirection, thus neglecting the inuence of the side-walls. The wall temperature is xed to 300 K toaccount for the short test time in the HEG facility.The exhaust nozzle was only partially included upto a plane 0.110 m downstream of the end of thecombustion chamber (at x = 0.760 m) at which allvariables areextrapolated. Fuel injection was mod-eled by partially includingtheinjectors using a totalhydrogen pressure of 2.97 bar and a total tempera-ture of 300 K at the injector inow. For the HEGexperiment simulations the global equivalenceratio was / 0:43 but for the ight tests it wasslightly lower. The RANS computations makeuse of all symmetries of the combustor and henceonly one quarter of the computational domainshown in Fig. 3 was modeled, whereas in the LESthe domain shown in Fig. 3 is modeled. For RANS,unstructured grids, adapted to the density, temper-ature and pressure gradients in the ow, are usedwith about 0.8 million points, whereas for LES a

coarse and a ne grid with 12.5 and 26.2 millionpoints, respectively, are used. For both RANSand LES performed grid renement studies indi-cate that the grids used are sufficiently ne to cap-ture the key ow features.

5. Results from combustor simulations

The composite Fig. 4 presents selected aspectsof the reacting ow in the HyShot II combustorfrom the LES computations. The wall pressure inFig. 4a reveals that initially the pressure increasesslowly with increasing distance from the transversefuel jets to increase more rapidly between 30 and60 D downstream of the transverse fuel jets to peakat about 100 D. Higher wall pressures are alsoobserved beneath the bow-shock, forming a hoodover the transverse fuel jet and beneath the fuel jets.The more rapid pressure increase further down-stream is caused by volumetric expansion due toexothermicity and chemical reactions. The largediffusivity of H2 allows it to rapidly mix with thefreestream air, forming a reactive mixture aroundthe high momentum H 2 lled jet core seen inFig. 4c.

The velocity in Fig. 4b shows the high-speed(ight Ma 7.2) ow entering the combustor,the transverse H 2 jet, the bow-shock and the com-plex ow structures developing downstream of theinjection points. The jet-to-freestream momentumux ratio, J qH2 v2H2 =q air v2air , is 1.1 and themean H 2 jet penetration prole is close to that dis-cussed in [25], and the H 2 jets are observed to pen-etrate to 30% of the combustor height beforeself-ignition (Fig. 5). The axial velocity in thecombustor is somewhat reduced compared withthat at the inow due to the combined blockingand redirection effects caused by the transverse jets, helping to stabilize the ame.

The time-averaged transverse H 2 jets typicallyconsist of a counter-rotating vortex pair and ahorseshoe-vortex, whereas instantaneously theyconsist of smaller and topologically more complexvortexstructures as seen in thesecond invariant, k2 ,of the velocity gradient in Fig. 4c. These typicallyconsist of small bent S-shaped vortices (side arms)with their lower parts aligned with the ow andtheir upper parts curling over the jet forming the

Fig. 3. Computational model of the HyShot II combustor used in the RANS and LES analysis.


8/11/2019 1-s2.0-S1540748910003202-main

5/7

neck (circumferential rollers) of the counter-rotat-ingvortexpair. This feature is observed experimen-tally by Ben-Yakaret al. [25], and theypropose thatthe side arms are stretched by increased shear stres-ses in the regions of steep velocity gradient. Thesevortical structures seem to arise from Kelvin Helmholz (KH) instabilities in the jet shear layers just beneath the bow-shock. Since the transverse jets contain all the H 2 , mixing dominates duringthe rst 20 to 40 D, whereby air is entrained intothe vortex structures and H 2 diffuses into the air,resulting in a combustible mixture around the jets.Further downstream, between 30 and 60 D the H 2and air are sufficiently mixed to burn if the temper-ature is sufficiently high. In this region, self-ignitionoccurs intermittently in lean low vorticity regionswith temperatures close to the laminar self-ignitiontemperature but also due to shockshock interac-tions occurring at lower temperatures. The volu-metric expansion causes the S-shaped side armsandspanwise rollers to combine into X-shaped vor-tices, dominating the self-ignition region. Due to

volumetric expansion, vortex stretching, baroclinictorque and self-diffusion, the vortex structureseventually develop into longitudinal vortices, dom-inating the downstream part of the combustor.These vortices grow in size with increasing distancefrom the injection point due to the volumetricexpansion, and when theyreach the end of the com-bustor, the gradual expansion increases the veloc-ity, presented in Fig. 4b, causing a forwarddirected thrust on the thrust surface.

In Fig. 4d, the iso-surfaces of the H 2 mass frac-tion (gray) and the heat release, dened as thesource term in the transport equation for the sen-sible enthalpy [26], conditioned on k2 supports theprevious description of the self-ignition process.Heat release occurs locally beneath the bow-shockas H 2 and air are mixed when the transverse H 2 jets impinge on the airow through the combus-tor. However, the heat release ceases furtherdownstream, between 10 and 25 D, due to insuffi-cient mixing. Further downstream, where H 2 andair are well mixed, self-ignition, aided by recircu-

Fig. 4. Composite gure of the reacting ow in the HyShot II combustor: (a) wall pressure and an iso-surface of the H 2mass fraction, (b) axial velocity cut through a fuel injector, (c) iso-surface of the second invariant of the velocity gradient,k2, , colored by the temperature and (d) iso-surfaces of the H 2 mass fraction (gray) and the heat release conditioned on k2.

Fig. 5. Comparison of LES and RANS predictions in terms of contours of the instantaneous and mean axial velocity, ~vxand hvxi , and temperature, ~T and hT i , and numerical schlieren images superimposed on semi-transparent H 2 massfraction distributions. Shown is also a frequency spectrum of the near-wall pressure at three locations (P1: x = 0.45 m,P2: x = 0.55 m and P3: x = 0.65 m).


8/11/2019 1-s2.0-S1540748910003202-main

6/7

lated combustion products, due mainly to inter-acting shocks, occurs, causing the S-shaped vorti-ces and the spanwise rollers to rapidly develop X-shaped vortices. The chain-branching stepH + O 2 M OH + O is particularly important forself-ignition as its products react with H 2 to pro-

duce H, that continues to react, producing moreradicals, until the resulting pool of radicalsreaches a critical level, whereby a rapid exother-mic reaction occurs.

In Fig. 5, results from LES and RANS predic-tions are compared. The perspectives show con-tours of the instantaneous velocity, ~vx, (rear) andtemperature, ~T , (front) from LES (top) and themean axial velocity, hvxi , (rear) and temperature,hT i , (front) from RANS (bottom), at 11 cross-sec-tions. The time-averaged LES result in similar con-tours as obtained by RANS and is therefore notpresented. From the LES results the unsteady nat-ure of the ow is evident, with the jets resulting in acomplex wake that interacts with the surroundingow. On average,each jetresults in a velocity decitextending throughout the combustor in RANSwhereas in LES it dissolves at about three quartersof the combustor. The cold H 2 jets extend furtheraft in the RANS results compared to the LESresults in which they break-up and self-ignite ear-lier. In both cases a thin boundary layer can be seenat both the top and bottom walls. The side viewsshow numerical schlieren images, obtained fromthe vertical gradient of the refraction index com-puted from the LorenzLorenz equation [27],superimposed with semi-transparent contours of the H 2 mass fraction. From the schlieren imagethe bow-shock is clearly visible as is the Ma discbarrel shock and separated boundary layer just infront of the transverse H 2 jet. The boundary layersshow signs of intermittent separation, and the jetpenetration is larger in the LES predictions com-pared to the RANS predictions. The transverseH2 jet seems more diluted from the LES predictionsthan from the RANS predictions but is alsounsteady and strongly three-dimensional. The fre-

quency spectra of the near-wall pressure at P1(x = 0.45 m), P2 (x = 0.55 m) and P3(x = 0.65 m) along the jet centerline are also shownin Fig. 5 . Close to the injectors, at P1, the spectrumconsists of several peaks (including the Helmholzfrequency of 12 kHz) whereas further downstream,

at P2, frequencies at 7 kHz (corresponding the tocombustor eigenfrequency as obtained from astand-alone combustor eigenmodel calculation)dominate. Close to the combustor exit (atx = 0.650 m) the dominating frequency decreasesto about 125 Hz.

In Fig. 6, predicted and measured axial prolesof (a) the time-averaged wall pressure, h p i , and (b)wallheat-ux, hhi , arecompared.Thewall pressureand heat-ux are compared at lines on the bottomwall of the combustor between injectors and6.0 mm off centerline between injectors, respec-tively. For the time-averaged wall pressure theexperimental data show a sudden increase betweenx = 0.50 m and 0.53 m followed by a slowerincrease up to the end of the combustor, atx = 0.65 m, after which the pressure drops rapidlyto the exhaust pressure. The RANS predictionsshow an almost linear increase from the combustorinlet to the combustor exit, missing the sharp pres-sure increase indicating combustion, and under- oroverpredict the wall pressure by up to 25%. TheLES predictions are in better qualitative agreementwith the experimental data, in particular showing apressure rise although in two phases and startingsomewhat too early. For the time-averaged wallheat-ux a sudden rise is observed at aboutx = 0.52 m, corresponding well to the location atwhich the pressure rises due to combustion. Theaverage level in the rst part of the combustoragrees well with the laminar heat-ux predictedusing a Blasius prole and a wall temperature of 300 K, but the transverse H 2 jet introduces somepeculiarities after x = 0.42 m. The RANS predic-tions typically overpredict the heat-ux whereasthe LES predictions typically underpredict theheat-ux, with between 15% and 10%, respectively.

Fig. 6. Comparison of predicted and measured (a) wall pressure between jet injectors and (b) heat-ux along a line6.0 mm off centerline between jet injectors.


8/11/2019 1-s2.0-S1540748910003202-main

7/7

Both the pressure and heat-ux predictions aregenerally within the experimental uncertainty. Inaddition, the combustion efficiency, gc 1 R Aq j~vj~Y H2 dA=q j~vj~Y H2 Ainj, at the outlet is found tobe about 83%.

6. Summary and concluding remarks

Here we use RANS and LES to analyze theHyShot II scramjet combustor at freestream con-ditions representative of the wind-tunnel experi-ments in the HEG shock tunnel correspondingto ight conditions at 32.5 km altitude. To pro-vide accurate inow boundary conditions to thereacting RANS and LES combustor simulationsseparate RANS computations of the HEG nozzleand HEG test sections were carried out insequence before the reacting RANS and LEScomputations were performed. The LES resultsreveal a very complicated ow pattern dominatedby the transverse jet-in-cross ow and the associ-ated fuelair mixing, the self-ignition processand further aft in the combustor a region withfully developed turbulent combustion. Between30 and 60 jet diameters downstream of the injec-tors self-ignition occurs due to hot spots causedby colliding shocks supported by recirculatedhot combustion products. This intermittent hotspot formation leads to a longitudinally oscillat-ing self-ignition zone that participates in creatingthe unsteady features observed in the LES results.The qualitative average ow features of LES werefound to be similar to the results of the RANS.Both the qualitative and quantitative agreementof the surface pressure distribution and surfaceheat-ux to the experimental data was improvedby the use of the LES model compared to RANS.The results show that RANS analysis is sufficientto capture the main ow features at limited com-putational cost whereas the application of theLES model can lead to signicant improvementof the prediction of detailed ow features.

Acknowledgement

The HEG tests were carried out by A. Gardnerand additional funding was provided by the Euro-pean Space Agency Grant TRP 17001/02/NL/MV, and the LES computations were supportedby the Swedish Defense Material Agency.

References

[1] E.T. Curran, S.N.B. Murthy (Eds.), Scramjet Pro- pulsion, Progress in Astronautics and Aeronautics ,vol. 189, AIAA, Washington, DC, USA, 2000.

[2] D. Andreadis, The Industrial Physicist 10 (2007) 26 29.

[3] F. Ladiende (Ed.), AIAA J . 43 (3) (2010), in press(special Issue on Scramjet Combustion Technology).

[4] W. Waidmann, F. Alff, U. Brummund, M. Bo hm,W. Clauss, M. Oschwald, Space Technol. 15 (1995)421429.

[5] J.P. Drummond, G.S. Diskin, A.D. Cutler, Fuel Air Mixing and Combustion in Scramjets, Technol-ogies For Propelled Hypersonic Flight , WorkingGroup AVT 10, Final Report, NATO Research andTechnology Organization, 2001.

[6] T. Sunami, P. Magre, A. Bresson, F. Grisch, M.Orain, M. Kodera, AIAA 2005-3304, 2005.

[7] Available at: .

[8] A. Paull, H. Alesi, S. Anderson, The HyShot ightprogram and how it was developed, in: AIAA AAAF 11th Int. Space Planes and HypersonicSystems and Technologies Conference , Orleans,France, 2002.

[9] M.K. Smart, N.E. Hass, A. Paull, AIAA J. 44(2006) 23662375.

[10] J.M. Schramm, S. Karl, K. Hannemann, J. Stree-lant, AIAA 2008-2547, 2008.

[11] P. Gerlinger, H. Mobus, D. Bruggemann, J. Com- put. Phys. 167 (2001) 247276.

[12] C. Fureby, Philos. Trans. R. Soc. A 367 (2009)29572969.

[13] S. Karl, K. Hannemann, J. Streelant, A. Mack,AIAA 2006-8041, 2006.

[14] R.N. Gupta, J.M. Yos, R.A. Thompson, K.P. Lee,A Review of Reaction Rates and Thermodynamic and Transport Properties for an 11-Species Air Model for

Chemical and Thermal Non-equilibrium Calculationsto 30000 K , NASA RP 1232, 1990.[15] T. Gerhold, O. Friedrich, J. Evans, M. Galle, AIAA

1997-0167, 1997.[16] T.J. Poinsot, S.K. Lele, J. Comput. Phys. 101 (1992)

104129.[17] C.J. Jachimowski, An Analytical Study of the

HydrogenAir Reaction Mechanism with Applicationto Scramjet Combustion , NASA TP 2791, 1988.

[18] D.M. Davidenko, I. Gokalp, E. Dufour, P. Magre,AIAA 2006-7915, 2006.

[19] D.C. Wilcox, Turbulence Modeling for CFD , LaCanada, California, USA, 1998.

[20] S.S. Girimaji, Combust. Sci. Technol. 78 (1991) 177 196.

[21] R. Bensow, C. Fureby, J. Turb. 8 (54) (2007) 117.[22] M. Berglund, E. Fedina, C. Fureby, V. Sabelnikov,

J. Tegne r, Finite rate chemistry les of self ignition ina supersonic combustion ramjet, AIAA.J ., in press,doi:10.2514/1.43746.

[23] H.G. Weller, G. Tabor, H. Jasak, C. Fureby,Comput. Phys. 12 (1998) 620631.

[24] D. Drikakis, C. Fureby, F.F. Grinstein, M. Liefen-dahl, in: F.F. Grinstein, L. Margolin, B. Rider(Eds.), Implicit Large Eddy Simulation: Computing Turbulent Fluid Dynamics , Cambridge UniversityPress, Cambridge, UK, 2007, p. 94.

[25] A. Ben-Yakar, M.G. Mungal, R.K.R.K. Hansen,Phys. Fluids 18 (2006) 026101.[26] T. Poinsot, D. Veynante , Theoretical and Numerical

Combustion , Edwards, 2001.[27] L.A. Vasilev, Schlieren Methods , Keter Inc., New

York, 1971.

http://www.nasa.gov/missions/research/x43-main.htmlhttp://www.nasa.gov/missions/research/x43-main.htmlhttp://dx.doi.org/10.2514/1.43746http://dx.doi.org/10.2514/1.43746http://www.nasa.gov/missions/research/x43-main.htmlhttp://www.nasa.gov/missions/research/x43-main.html

Documents

1-s2.0-S1540748910003202-main