1

Future Directions of Supersonic Combustion Research:Air Force/NASA Workshop on Supersonic Combustion

Julian M. Tishkoff *Air Force Office of Scientific Research

Bolling Air Force Base

J., Philip Drummond CNASA Langley Research Center

Tim Edwards vAbdollah S. Nejad v

Aero Propulsion and Power DirectorateWright-Patterson Air Force Base

AbstractThe Air Force Office of Scientific Research,the Air Force Wright Laboratory AeroPropulsion and Power Directorate, and theNASA Langley Research Center held a jointsupersonic combustion workshop on 14-16May 1996. The intent of this meeting was to:(1) examine the current state-of-the-art inhydrocarbon and/or hydrogen fueled scramjetresearch; (2) define the future direction andneeds of basic research in support of scramjettechnology; and (3) when appropriate, helptransition basic research findings to solve theneeds of developmental engineering programsin the area of supersonic combustion and fuels.A series of topical sessions were planned.Opening presentations were designed to focusand encourage group discussion and scientificexchange. The last half-day of the workshopwas set aside for group discussion of theissues that were raised during the meeting fordefining future research opportunities and

directions. The following text attempts tosummarize the discussions that took place atthe workshop.

NomenclatureA areaa speed of soundCf skin friction coefficientD1 Damkohler first number, L/utc

D2 Damkohler second number, ηc ∆hc/ Ht

Ea activation energyep flow distortionHt total flow enthalpyht specific enthalpyL combustor lengthM Mach numberMc convective Mach number, (U2-

U1)/(a1+a2)n overall reaction orderP pressureq dynamic pressure, 1/2(ρu2)R0 universal gas constantr velocity ratio, U2/U1

s density ratio, ρ2/ρ1T temperature

* Program Manager, Associate Fellow, AIAAC Senior Research Scientist, Associate Fellow, AIAAv Senior Research Scientist, AIAA Member

https://ntrs.nasa.gov/search.jsp?R=20040105532 2018-07-14T08:05:50+00:00Z

2

tc characteristic combustion timeU, u flow velocityw laminar burning rateZ altitude∆hc heat of combustion

ηc combustion efficiency

ρ density

Subscripts0 free stream condition1, 2 stream 1, stream 24 isolator entrance conditionad adiabatic flame conditionavg average valuemax maximum value

IntroductionThis paper summarizes the discussions held atan Air Force/NASA Workshop on SupersonicCombustion, in Newport News, Virginia onMay 14-16, 1996. The purposes of theworkshop were: (1) to review current design,performance, and testing practices forscramjets -- supersonic combustion ramjetsused in high-speed airbreathing propulsionsystems; and (2) to investigate the applicationof novel analytical methods, includingexperimental, theoretical, and computationalapproaches, to improve scramjet designs.

Recent programs for developing high-speedaerospace vehicles that utilize airbreathingpropulsion provided the motivation for thisworkshop. Many of these programs werediscussed at the recent AIAA 7th InternationalSpace Planes and Hypersonics Systems andTechnologies Conference held in Norfolk,Virginia on November 18-22, 1996. Despite ahigh level of activity and financial investmentin scramjet development for high-speed flight,no operational example of a scramjet currentlyexists. The cancellation of the United StatesNational Aero Space Plane (NASP) program

reflects the difficulties in developing this modeof propulsion successfully.

The intention of the organizers of theworkshop was to provide a unique forum inwhich the developers and testers ofpropulsion technology could interact directlywith members of the research community.The workshop was organized to intersperseformal presentations with open discussion inorder to find common ground between twoprofessional activities that otherwise mightnot have opportunities for such direct contact.To facilitate these interactions anddiscussions, invitations to attend theworkshop were extended to approximatelysixty participants, as summarized in Table 1.These participants were invited because oftheir experience and records ofaccomplishments in areas of research andtechnology relevant to scramjet design andtesting. The organizers recognized that manyother scientists and engineers possessknowledge and capabilities appropriate to theworkshop but believed that an excessivelylarge number of participants would hinder theinteractions. The presence or absence of anyscientist or engineer in Table 1 therefore doesnot represent anyone’s opinion about theprofessional merits of participants versus non-participants.

The workshop was conducted over a 2-1/2day period. The first two days were devotedto presentations and related discussions. Thetopics and presenters are listed in Table 2.The body of this paper will review thesepresentations. This paper also may containsome additional ideas and comments that theauthors have assembled since the workshopwas held, but the primary content reflects thepresentations and related discussions at theworkshop. On the last half day of the

3

workshop an open discussion was conductedin which general suggestions were made for

future activities. A brief summary of thesesuggestions is given in Appendix A.

Table 1. Workshop Invitees

Name AffiliationMr. Griffin Anderson NASA Langley Research CenterDr. Fred Billig Applied Physics Laboratory, Johns Hopkins

UniversityDr. Garry Brown Princeton UniversityDr. Dennis Bushnell NASA Langley Research CenterDr. Harsha Chelliah University of VirginiaDr. S M Correa GE Research CenterDr. E. T. Curran WL/PO, Wright LaboratoryDr. Stephen D'Alessio Applied Physics Laboratory, Johns Hopkins

UniversityDr. Paul Dimotakis California Institute of Technology.Mr. Glenn Diskin NASA Langley Research CenterDr. James F. Driscoll University of MichiganDr. J. Philip Drummond NASA Langley Research CenterDr. Craig Dutton University of IllinoisDr. Raymond Edelman RocketdyneDr. Tim Edwards WL/POSF, Wright LaboratoryDr. Fokion N. Egolfopoulos University of Southern CaliforniaDr. John Erdos GASLDr. G. M. Faeth University of MichiganDr. Alan Garscadden WL/CA, Wright LaboratoryDr. Peyman Givi State Univ. of New YorkMr. Edward S. Gravlin WL/POP(HyTech), Wright LaboratoryMr. Wayne Guy NASA Langley Research CenterDr. R K Hanson Stanford UniversityDr. William Heiser HQ USAF/DFAN Department of AeronauticsDr. Casey Jachimowski NASA Langley Research CenterDr. Ajay Kumar NASA Langley Research CenterDr. C K Law Princeton UniversityDr. Ron Lehrach United Technologies Research CenterDr. Frank Marble California Institute of TechnologyDr. Atul Mathur Rocketdyne Division, Rockwell International.

CorporationMr. Chuck McClinton NASA Langley Research CenterMr. Bob Mercure NASA HeadquartersLt Col Richard Moore WL/POP, Wright LaboratoryDr. Abdollah Nejad WL/POPT, Wright Laboratory

4

Dr. G. B. Northam NASA Langley Research CenterDr. Elaine Oran US Naval Research LaboratoryDr. Gerald Pellett NASA Langley Research CenterDr. S B Pope Cornell UniversityDr. David Pratt University of WashingtonDr. David Riggins University of MissouriMr. Kenneth Rock NASA Langley Research CenterDr. Clay Rogers NASA Langley Research CenterDr. Klaus Schadow Naval Air Warfare CenterDr. Joseph A. Schetz Virginia Polytechnic Inst. and State UniversityDr. Munir Sindir Rocketdyne Division, Rockwell International

CorporationDr. Mike Smith NASA Langley Research CenterDr. Louis Spadaccini United Technologies Research CenterDr. Scott Thomas NASA Lewis Research CenterMr. Michael Thompson Applied Physics Laboratory, Johns Hopkins

UniversityDr. Julian M. Tishkoff AFOSR/NAMr. Carl Trexler NASA Langley Research CenterDr. David Van Wie Applied Physics Laboratory, Johns Hopkins

UniversityMr. Randy Voland NASA Langley Research CenterDr. Robert W. Walters AeroSoft, Inc.Dr. P J Waltrup Applied Physics Laboratory, Johns Hopkins

UniversityDr. James Weber WL/POPDr. Al Wieting NASA Langley Research CenterDr. Michael Winter United Technologies Research Center

5

Table 2. Workshop Agenda

1. Engine Design Issues (May 14, Morning Session)

Speakers:Dr. Fred Billig, Johns Hopkins University, Applied Physics LaboratoryMr. Chuck McClinton, NASA Langley Research CenterLt Col Richard Moore, Wright LaboratoryProfessor David Pratt, University of Washington

2. Ground Based Testing (May 14, Afternoon Session)

Speakers:Mr. Michael Thompson, Johns Hopkins University, Applied Physics LaboratoryMr. Randy Voland, NASA Langley Research Center

3. Fuels and Fuel Systems (May 15, Morning Session)

Speakers:Dr. Tim Edwards, Wright LaboratoryDr. Lou Spadaccini, United Technologies Research CenterMr. Chuck McClinton, NASA Langley Research Center

4. Injection and Mixing (May 15, Morning Session)

Speakers:Dr. Abdi Nejad, Wright LaboratoryProfessor Garry Brown, Princeton UniversityProfessor Paul Dimotakis, California Institute of Technology

5. Combustion Chemistry (May 15, Afternoon Session)

Speakers:Professor Ed Law, Princeton UniversityProfessor Harsha Chelliah, University of Virginia

6. Diagnostics and Simulation of High-Speed Flows (May 15, Afternoon Session)

Speakers:Dr. Michael Winter, United Technologies Research CenterDr. Munir Sinder, Rocketdyne

6

Engine Design IssuesThe workshop began with a review of currentpractices for designing scramjet engines.Practical system issues such as missionrequirements, integration of the inlet/isolator,combustor, nozzle, airframe, fuel systemspecification, and cooling concepts wereaddressed. The objective of this session wasto discuss global design challenges associatedwith both cryogenic and hydrocarbon-fueledscramjets with the intent of identifying basicresearch opportunities to impact scramjettechnology needs. However, at the time of theworkshop, the Air Force had already defined anational program to develop technologiesrequired for the development of a fixedgeometry scramjet engine capable of operationover Mach 4 - 8 flight regime usingconventional JP-based hydrocarbon fuels.Therefore, the majority of the discussioncentered around technical challenges associatedwith the development of tactical missiles usingstorable fuels capable of acceleration fromMach 4 and cruise at Mach 8.

For this discussion, high speed vehicles weredivided into the following two classes: a)aircraft or man-rated; b) expendable. Thechoice of high speed propulsion system(airbreathing, and rocket) hinges on manydesign and mission requirements. Factors suchas size, weight, design complexity,maintainability, longevity, storability,production and life cycle costs, and logisticsupportability were identified to be just asimportant as the performance characteristics(speed, range, and efficiency) of thehypersonic vehicle. Billig [1] listed some ofthe characteristics of hypersonic air-breathingvehicles, see Table 3. It is interesting to notethat the combustor length remains virtuallyconstant at 2-6 ft for the three classes ofhypersonic vehicles, suggesting that

supersonic combustion processes areinherently mixing-limited. The trade-offstrategy to attain high combustion efficiency ismuch more complex in supersonic combustors,where shear losses can drastically reduceengine performance. Simply adding combustorlength for optimization of mixing/combustionefficiency is usually not the prudentengineering solution.

The choice of air-breathing ramjet enginecycles depends on the flight Mach number.For example, at lower flight Mach numbers(M < 5 - 6) the subsonic integral rocket-ramjet is the preferred cycle. At high Machnumbers (M > 6 - 7) the scramjet cycle is thepreferred mode of operation. However, atactical missile -- an expendable, low cost, lowweight, and therefore fixed geometry flow pathdesign capable of operating at high flight Machnumbers M > 6.5 using conventional storableliquid hydrocarbon -- must operate as a ramjetat low flight speeds and as a scramjet athypersonic speeds. Fortunately, if adequatecombustor-inlet isolation is provided, thescramjet will function in a subsoniccombustion mode at low Mach numbers withslightly lower efficiency than that of aconventional ramjet. However, ahydrocarbon-fueled scramjet designed tooperate efficiently at Mach 7 - 8 using a fixedgeometry flow path has not been shown tooperate efficiently at Mach 4 flight conditionswithout resorting to use of massive auxiliarypiloting [2], or without the use of largeamounts of stored reactive oxidizer, e.g.,chlorine trifluoride [3]. An interesting exampleof a massively piloted scramjet concept is theDual Combustor Ramjet (DCR) which wasdesigned and tested at Johns HopkinsUniversity, Applied Physics Laboratory, andis schematically shown in Figure 1. This is anaxisymmetric design in which the forebody

7

serves as the initial compression surface of thesupersonic inlet. In this concept, the incomingflow is divided into eight segments at the cowllip. Four smaller inlets supply air to asubsonic dump combustor. They operatesupercritically (the normal shock isswallowed) to avoid the interaction of thenormal shock with the flow entering the largerinlets that feed the supersonic combustor. Inorder to provide stable combustor operationover a wide range of flight Mach numbers, theflow passages to the subsonic combustor havean increasing cross sectional area in thestreamwise direction. The major portion ofthe air is captured by the four larger inlets andthe external cowl compression surface andturned supersonically inward toward theengine axis. Captured flow is spread radiallyto form an annulus of supersonic flow thatsurrounds the outlet of the dump combustor.The aft sections of these supply ducts haveslightly diverging flow passages in thestreamwise direction, which effectively act asthe combustor-inlet isolator. When thepropulsion system is operating at a highequivalence ratio and/or at low flight Machnumbers, the isolator section can sustain ashock train with a pressure rise equivalent tothat of normal shock. In this mode ofoperation the combustor inlet Mach number isless than one, and the mean Mach number atthe combustor exit is either sonic orsupersonic. At lower engine equivalence ratiosand/or higher flight Mach numbers the isolatorshock train pressure rise is equivalent to thatof an oblique wave structure. With theinlet/isolator operating in the oblique shockmode, the mean flow Mach numberthroughout the scramjet is supersonic. Thisdual-mode engine operation has been discussedfully in the literature [4-6].

The issue of coupling combustor burner

characteristics to vehicle cooling requirementsis very important. The endothermicity ofhydrocarbon fuels requires vehicle structuralcomponents to act as a heat exchanger/thermalcracking reactor. The composition of thecracked products depends on the time-temperature history of the cracking processthroughout the vehicle structure. Changes inchemical composition or the state of the fueldirectly affect burner operationalcharacteristics; the time required for a radicalpool to reach flammable conditions is linearlydependent on concentration, quadraticallydependent on pressure, and exponentiallydependent on the temperature. Therefore,precise control of the thermal cracking processis essential to the production of the desiredfuel conversion (constituents) at the burnerentry throughout the flight trajectory.However, the coupling of the heatexchanger/reactor to the combustor is notwithout its engineering challenges. Many testsof heat exchanger reactors have shown severeacoustic instability, leading to catastrophicfailure. Tests of regeneratively cooledstructures with endothermic fuels feeding acombustor have shown system instabilitiesbetween the two systems. The source ofthese acoustic instabilities may be the fact thathydrocarbon fuel remains near thethermodynamic critical point within the heatexchanger, where thermodynamic propertiessuch as density, viscosity, latent heat, ratio ofspecific heats, and speed of sound show largevariations with respect to small changes intemperature and pressure.

Mixing and heat release are significantengineering challenges in supersonic flows.However, when the engineer considers allaspects of the system design, mixingoptimization, and/or combustion efficiencymay not be the driving factors. Thus,

8

combustor and isolator lengths may not dictatethe internal duct length. Since the internal dragcan reduce the performance of a scramjetengine significantly, combustor designs withlarge surface areas should be avoided.Furthermore, the designers are usually carefulin using intrusive injectors. Aside from thesevere cooling requirements, the base and wavedrag of many hyper-mixers render themineffective in a practical device. Therefore,one must optimize and balance system overallperformance, (i.e., maximizing net positivethrust), at the expense of not achievingcomplete mixing.

To develop a scramjet, designers require adesign strategy. The following process wasproposed: (1) start with a conceptual vehicledesign; (2) optimize the design by sensitivityanalysis; (3) select inlet(s), and conduct inlettests, preferably in conjunction with theisolator, combustor, and injector components;(4) analyze the experimental data to updatethe cycle analysis codes to assess theperformance potential of the scramjet design;(5) optimize the combustor/injector designconcept. In order to implement this designstrategy, accurate models for predicting jetpenetration, mixing, combustion, heat transfer,

and combustor-inlet interaction are required.To develop such models, research efforts mustbe ongoing for better understanding of thephysics of supersonic combustion to evaluateand update the empirical design models usedby the engineers. Free jet, semi-free jet, anddirect connect tests must be conducted insufficient detail to allow meaningfulassessment of the performance and operationalcharacteristics of the design and generation ofbenchmark data to aid with the developmentand validation of the analytical tools.

Ground TestingThe objective of this session was to introduceand discuss the state of testing andmeasurement technology used for assessmentof scramjet performance in ground basedfacilities. The speakers outlined testprocedures, instrumentation and measurementaccuracy requirements, analytical modeling ofthe aerothermochemical processes, and erroranalysis procedures used for performancetesting of the scramjet flow path.

Conventional ramjets and scramjets designedfor Mach 6 - 8 flight push the limit of longduration (~ seconds to minutes) ground testdirect-connect or free-jet test facilities.

Mission FlightMach #

PropulsionSystem

Flow PathGeometry

Fuel FlightDuration

Vehicle length(ft)

TacticalMissile

6 - 8 Dual CombustorRamjet and/or

Rocket

FixedGeometry,passivelycooled

Liquid HC,Slurry,

Solid HC

10 -12Minutes

Overall 5 - 15Combustor 2 - 5

Nozzle 2 - 5

Trans-atmospheric

Missiles

0 - 25 Dual modeRamjet/Scramjet +many low speed

options

VariableGeometry

Liquid H2

Liquid O2

20 - 30Minutes

100 cycle

Overall 100- 200Combustor 2 - 6Nozzle 50 - 80

HypersonicCruise

0 - 80 - 15

Mach 6-8Turboramjet

M 15 scramjet

VariableGeometry,ActivelyCooled

Mach 6-8,HC

Mach 15,Liquid H2

M = 6 - 8, 1 - 3 Hours

M = 151 Hour

Overall 100- 200Combustor 2 - 6Nozzle 50 - 80

Table 3. General Characteristics of Hypersonic Vehicles

9

Higher speed flight conditions (M >10) aresimulated in pulsed facilities that can generateflight enthalpies in excess of M = 15conditions, but only for a few milliseconds. Inthis session most of the discussion centeredaround testing scramjets in direct-connect andfree-jet facilities. Figure 2 is a schematicillustration of a direct connect test facility.These facilities are relatively straightforwardand are composed of the following keyelements: (1) a high pressure air source; (2) anair heater (vitiated/arc-heated/pebble-bed/gasfired heat exchanger) for proper simulation offlight enthalpy; (3) a facility nozzle for propersimulation of combustor/isolator inlet Machnumber in direct-connect tests or flight Machnumber in free-jet tests; (4) a combustorand/or scramjet test article; (5) a loadmeasuring system for thrust measurement; and(6) a steam calorimeter for estimation ofcombustion efficiency. Typical scramjetcombustor entrance properties [7] are depictedin Table 4. In theory, it is desirable toduplicate or match these properties as closelyas possible. However, practical requirements -- such as: power generation; fabrication ofhardware to sustain the pressure; and facilityand model cooling requirements for testing atflight enthalpy, which increase linearly withfacility size (mass flow rate) and quadraticallywith flight Mach number-- may preventduplication of all flight parameters. Andersonet al. [8] defined pressure, temperature,velocity, gas composition, and characteristiclength scale as the primitive variables thatdescribe the scramjet flowfield. Voland andRock [9] have pointed out that, since completeduplication of the flight parameters in groundtest facilities may not be possible, then onemust identify parameters that impact thephysical processes of supersonic combustion.It is generally agreed that these parameters are:flight Mach number, total enthalpy, Reynolds

number, Stanton number, Damkohler first andsecond numbers, and the wall enthalpy ratio.Voland and Rock observed that the process ofmatching flight total enthalpy and Machnumber allows proper simulation of theDamkohler second number D2 -- the kinetic-to-thermal energy ratio. If the flight dynamicpressure is not matched due to powerrequirements or facility constraints and massflow limitations forces, testing a smaller scaleengine becomes necessary. Then theDamkohler first number D1 -- the ratio of flowresidence time to chemical reaction time -- isnot simulated properly. If combustion iskinetically limited, then ignition delaycharacteristics of the fuel and the reactiontimes become a critical issue, and propersimulation of D1 becomes critical. However, ifthe combustion is mixing limited, propersimulation of D1 is not an issue. Dynamicpressure and geometric scaling also affect theratio of the inertial to viscous forces(Reynolds number). Recall that the Reynoldsnumber was identified as an importantparameter to match in ground testing ofengines. When scaling reduces the flow pathsize excessively, then one should question theextrapolation of the results due to mixing,shock-boundary layer interaction, boundarylayer thickness, injector nozzle dischargecoefficient, etc.

With few exceptions, instrumentation in thesefacilities is rather conventional and is limitedto electromechanical devices for measuringpressure, temperature, gas composition,thrust, and combustion efficiency. Complexityand safety requirements compound thedifficulty of incorporating advanced laser-based diagnostic techniques. Most often,steam calorimetry is used in long duration testfacilities to quantify the amount of energyrelease, hence combustion efficiency. Several

10

accurate measurements must be made toaccount for a proper energy balance from theheater to the calorimeter exit plane. Theseinclude: temperatures and flow rates of air,make up oxygen, fuel (heater and combustor),quench water, total temperature at the exitplane of the calorimeter, and heat loss throughthe facility nozzle and combustor walls. Inthis technique, water is injected downstreamof the combustor exit plane to rapidly quenchchemical reactions. The precision of the totaltemperature measurement at the calorimeterexit plane significantly impacts the analysisand the results. Stevens and Thompson [10]schematically illustrate the procedures usedfor analysis of an arc-heated facility, Figure 3.They also point out that various issues, such

as precise determination of the heaterstagnation condition, facility nozzle effectiveflow rate and discharge coefficient, combustorentrance and exit conditions, and calorimeterexit conditions are extremely important forprecise estimation of scramjet combustionefficiency using a steam calorimeter.

In general it is recommended that, in additionto steam calorimetry, other measurementssuch as thrust, combustor pressuredistribution, skin friction, Pitot pressure, andgas sampling should also be attempted. Table5 shows the relative accuracy of the derivedcombustion efficiency (ηc) and the skinfriction coefficient (Cf) as functions ofmeasured parameters.

Free Stream Conditions Isolator entrance Conditions

Mo Zo (Kft)Po

(psia)To(oR)

Uo(ft/sec)

qo(lbf/ft2)

hto(BTU/lbm) M4 Ao/A4 P4/Po

P4

(psia)T4

(oR)U4

(ft/sec)

3 47.95 1.868 390 2904 1694 133.3 1.529 2.86 7.8 14.51 744 20344 57.48 1.183 390 3872 1910 264.3 1.945 4.91 15.7 18.57 930 2885

5 65.72 0.7978 390 4840 2011 432.8 2.363 6.92 24.9 19.86 1102 3799

6 73.30 0.5569 394 5839 2020 646.7 2.767 8.91 35.3 19.65 1279 4770

7 80.07 0.4049 397.7 6844 2000 902.3 3.143 10.85 47.0 19.03 1451 5757

10 95.50 0.1984 406.1 9879 2000 1918.2 4.143 16.49 89.6 17.78 1958 8744

15 114.25 0.0857 424.8 15155 1945 4561.0 5.502 25.23 185.9 15.94 2880 13908

20 137.76 0.0319 460.8 21040 1287 8824.3 6.650 33.11 313.6 10.02 4074 19468

26.9 178.21 0.0067 480.5 28865 425.8 16629.8 7.688 40.12 472.9 3.20 5187 27205

Table 4. Typical ramjet/scramjet freestream and combustor inlet conditions

Measurement ηc Cf

Static Pressure Good Fair

Temperature Good Poor

Water Concentration Good Very Poor

Total Pressure Poor Good

Velocity Very Poor Very Poor

Table 5. Measurement Sensitivity

11

Fuels and Fuel Systems"Fuel is becoming the integrating factor of thecomplete {high-speed vehicle} system” --E. T.Curran in [11]. "The problem is, we don'tknow how to make the scramjet combustorwork efficiently using conventional fuels atlow flight speeds corresponding to end-of-boost" -- F. Billig at this workshop.

There have been many recent workshops [11-13] and books [14, 15] in the supersoniccombustion area that included discussions offuels issues. A general consensus is: storableJP-type hydrocarbon fuels can be used up toMach 6-8, although the upper end of thisrange will be a significant technical challengethat will require chemically reactive"endothermic" fuels. Lou Spadaccini ofUnited Technologies Research Center briefedthe workshop on endothermic fuels [16].Liquid methane could be used to somewhathigher Mach numbers, but speeds in excess ofabout Mach 10 will require liquid hydrogen.

Air Force perspectiveWith the demise of NASP, the Air Force (AF)has focused its high-speed propulsion efforton storable hydrocarbon-fueled vehicles.Storable-fueled hypersonics is viewed as animportant technology for the AF for variousfuture missions [17]. However, hydrocarbonfuels have significant shortcomings insupersonic combustion when compared tohydrogen, notably relatively long ignitiondelays and limited cooling capability [12, 13].One issue that needs to be addressed inpractical engine design is the transition of thefuel injection and combustion processes thatoccur as the fuel temperature rises in thevehicle cooling passages. Early in the flight,cooling requirements are minimal, and the fuelis injected in a liquid phase. As the flightprogresses and the flight speed increases, fuel

may be heated to be well above itsthermodynamic critical point. In bothadvanced gas turbines and scramjet engines,the fuel may be partially reacted (cracked ordehydrogenated) through its use as a coolantbefore reaching the combustor. It is ofsignificant AF interest to determine the effectof this change in fuel character on thecombustion process. It is anticipated that thispartially reacted fuel will burn as well as, say,ethylene, with some claims that thecombustion properties (such as ignition delayor reactivity) may approach or exceed that ofhydrogen. Appropriate questions that need tobe addressed are: (a) will the ignition delay ofa partially cracked or dehydrogenated fuelunder engine conditions approach that of (e.g.)ethylene or even hydrogen; and (b) how willthe combustion efficiency/reactivity of a fuelchange as it is heated and is partially reacted inthe fuel system. The first step in kerosene-range hydrocarbon fuel combustion is oftencracking of the C12-level molecules to C1-C3

species. How will the combustion process beaffected if these cracking reactions occur"upstream" of the combustor?

The use of fuel as a coolant in advancedengines can lead to thermal and catalyticreactions in the fuel, yielding H2, CH4, C2H4,C2H6, etc. [16, 18-21]. As these partiallyreacted, hot (e.g., 1200 0F/650 0C) fuels areinjected into a gas turbine or scramjetcombustor, it is appropriate to consider howthe hot, partially reacted state of the fuelmight affect the combustion process. As theliquid fuel is heated at pressure, it becomes asupercritical fluid with significantly differentphysical properties, such as density andviscosity [29]. This could be expected tosignificantly change injection behavior [30].As the fuel begins to react in the fuel system,chemical changes in the injected fluid could

12

also affect the combustion process. Forexample, ignition is considered to be a "radical-poor" process [22], and ignition delay isaffected by radicals present due to air vitiation[23]. Are there sufficient radicals present inthe "reacted" fuel at ~ 1200 0F to reduceignition delay in a similar manner? In somecases, the reacted fuel can contain large molepercent levels of H2, especially forendothermic fuels such as methylcyclohexanethat are dehydrogenated. Does this H2 contentimprove the ignition delay? Note again thatthe relatively long ignition delay time ofhydrocarbons relative to H2 is a key limitationfor hydrocarbon-fueled scramjets [12, 13].There is evidence from shock tube tests thatthe ignition delay of hydrocarbons is reducedby the presence of hydrogen, but still is ordersof magnitude larger than that for purehydrogen [24]. The kinetics of combustion arealso of interest. Does the reacted fuel burn ina manner similar to its measured stableconstituents, or does the presence of(significant?) amounts of hydrocarbon radicalschange the reactivity? Another factor affectingcombustion is that significant fractions ofhydrogen could be generated in the fuel fed tothe combustor either by fuel dehydrogenationor by "steam reforming" a fraction of the fuel{CxHy + xH2O → xCO + (x+0.5y)H2}. Anissue that may be significant is the effect ofcoke particles or soot precursors in the reactedfuel on combustion. Coking is a significantissue for high temperature fuels [16, 18, 25],and some fuels may form aromatics as part ofthe cracking process. Supercritical fuelincreases the solubility of coke precursors(oligomers) from catalysts [26, 27]. How willthese fuel-borne aromatics, particulates, andoligomers affect soot formation (and thusradiative heat loads in the combustor andemissions)?

The consensus at the workshop appeared tobe that the answers to most of these questionsare not known. To obtain this information, itwas suggested that the effects of changes inthe fuel must be studied in a realisticsimulation of the scramjet combustionprocess, i.e. one that represents the diffusivenature of the combustion. One sub-scalepossibility is co-annular or opposed-jetburners [28] that would burn hot, partiallyreacted fuels. Premixed combustion devicesappear to be inadequate to address theimportant issues.

NASA Perspective on FuelsThe NASA Langley Research Center (LaRC)has been examining both hydrogen andhydrocarbon-fueled hypersonic vehiclesconcepts, including dual-fueled (H2 + HC)vehicles. Dual-fueled systems haveadvantages, as demonstrated by the dual-fueled Apollo missions. Chuck McClintonbriefed the workshop on the status of LaRC’sscramjet work in these areas. NASA studieshave confirmed the Mach 7-8 limit forhydrocarbon-fueled vehicles. NASA work, asdiscussed above, has shown the ignition,combustion, and cooling difficulties ofhydrocarbon fuels. Hydrogen is a much betterscramjet fuel, except in the areas of volumetricfuel energy density and logisticalsupportability. Published NASA vehicledesigns for both hydrocarbon-fueled [31] andhydrogen-fueled [32] vehicles were mentioned.NASA is supporting the Air Force HyTechprogram with analysis and modeling, althoughthe primary focus of NASA/LaRC’s work isflight tests of a H2 dual-mode scramjet system.

Combustion ChemistryThis portion of the workshop addressed theidentification of detailed chemical kineticmechanisms for scramjet combustion and the

13

reduction of those mechanisms to producekinetic models for combustor design codes.Discussions were also directed at approachesto model turbulence-chemistry interactions.

The computational complexity of solvingturbulent fluid transport equations provides astrong incentive for simplifying thedescription of chemical kinetics as much aspossible in combustor design codes. Thedegree of success of such simplificationsdepends on the information that is required foreach calculation. For example, equilibriumchemistry is adequate for calculations of non-optimum, non-critical global performance andhas been used successfully for suchapplications as predicting overall energyrelease in internal combustion engines.However, the accuracy of simplified orreduced chemistry must be scrutinizedcarefully for other calculations.

An example of the limitations of simplifiedchemical kinetic models can be found in thecalculation of laminar flame propagation usingone-step global chemistry [33]. Equation 1provides an Arrhenius expression to representone-step model for the laminar burning rate w:

w ~ Pn/2 exp[-Ea/2R0Tad] (1)

where symbols are defined in theNomenclature.

The simplest form of the one-step expressionwould have n as a constant. However, even ifn were treated as a pressure-dependentvariable, this expression can be shown to bedeficient.

To test the validity of eq. (1) with n as avariable, Egolfopoulos and Law [33] measuredthe laminar flame propagation of methane-

oxygen-nitrogen mixtures in a counterflow,twin-flame configuration. Figures 4-5 showthe behavior of the laminar burning rate w andthe overall reaction order n, respectively.According to eq. (1) w should exhibit amonatonic, exponential dependence onpressure. Figure 4 does not confirm thisdependence. Figure 5 shows that the exponentn is always less than 2, which is incompatiblewith n as a constant. n also has considerablevariation with pressure and even can assumenegative values. Thus, eq. (1) is a poorestimator of laminar flame behavior.

The physicochemical basis for the deficiencyof eq. (1) lies in the inability to account for thecomplexities of the competition between two-body chain branching reactions and multibodytermination reactions in determining flamepropagation. The presence of nonreactivethird bodies to serve as collision sites in thetermination reactions makes these reactionsparticularly sensitive to pressure.

If more complex reduced chemical mechanismsare needed, then how are they to be derived?The essential first step in producing reducedkinetic mechanisms is the identification ofcomplete chemical reaction mechanisms forrepresentative fuel combustion conditions.For hydrogen fuel, this process isstraightforward. For example a completechemical reaction mechanism for H2-O2-CO2

can involve 13 species and 27 reaction steps.However, for hydrocarbon fuels, it is morecomplex and difficult. Even a simple methane-air mechanism can include 16 species with 40reaction steps, while hydrocarbon-aircombustion chemistry can involve 40 specieswith 100 reaction steps for more complexhydrocarbon species. In hypersonicapplications, with fuel needed for coolingpurposes, the identification of specific fuel

14

components represents the initial challenge.For example, recent testing of endothermicfuels suggested that ethylene was a majorproduct of endothermic catalytic reactions[16]. However, more recent results, asdiscussed by Edwards in this workshop,contradict this choice. Processes such a sootformation remain elusive because of theircomplexity.

A second obstacle to the measurement ofcomplete reaction mechanisms is limitations inreactor and diagnostic capabilities. Kineticsmust be measured under thermodynamic andfluid dynamic conditions that simulate highspeed propulsion environments. Note that,measurement capability must be adequate forall critical species.

Two steps are generally used for simplifyingchemical kinetic mechanisms: development of“starting” mechanisms and “reduced”mechanisms. The starting mechanismrepresents a subset of the detailed mechanism,obtained by elimination of elementaryreactions to diminish the number of totalspecies in the system by as much as 90%.Further simplifications of the startingmechanism may be achieved by theintroduction of systematic “reductions” basedon the chemical and flow time scales of theproblem. Since calculation times ~ (numberof species)2, these simplifications can producedramatic savings in computational time.

Two approaches have been identified toproduce starting mechanisms:

1. “Systematic” approaches first use intuitivearguments to eliminate noncritical speciesand then use sensitivity analysis to reducethe number of reaction steps. Peters [34]introduced steady-state or partial

equilibrium approximations to achievesuch simplifications. This approach raisesconcerns that the results may be specificto the type of flame being calculated.

2 Automated procedures. This systematicapproach produces mechanisms that spanthe full range of known experimentalresults and should not be unique to anyindividual experiments. Automatedreaction procedures have been suggestedby Lam [35], Chelliah [36], and Pope [37].This approach has been applied tounsteady zero dimensional (homogeneous)systems but not as yet to combustioninvolving diffusive transport. Figure 6[36] illustrates the application of thisapproach to predict heat release in anonpremixed counterflow methane-airflame. This figure shows a comparisonbetween a 16 species, 40 reaction stepstarting mechanism and two systematicallyreduced mechanisms, developed byintroducing steady state approximations.A 31% representative saving incomputational time may be expected fromsuch reductions. Similar calculations areunderway for oblique detonation wavecombustion.

A strategy was suggested to implementautomated reduction:

1. Obtain a detailed, comprehensive data basefor C7-C12 aliphatic fuels.

2. Select a surrogate fuel.

3. Derive [34-37] an appropriate reducedmechanism for the intended application.

Table lookup procedures were suggested in theworkshop as an alternative to embedded

15

solution of reduced chemical kinetic equationsin combustion calculations. Lookup methodsare linearly proportional to the number ofspecies, as opposed to the quadraticdependence noted above for reducedmechanisms. The utility of lookup proceduresdepends on computational efficiency. Pope[37] recently has suggested novel methodologyto accelerate table lookup.

Turbulence-chemistry interactions represent amajor complication, coupling chemical kineticbehavior to fluid transport. Shear flows insupersonic combustion will produce strainrates, with a correspondingly large variation inscales affecting ignition, flame stability, anddiffusion. Compressibility introducesadditional complications, in which dilatationprovides a wave source that impactscombustion.

DiagnosticsThe focus of this portion of the workshop

was on the application of current opticalmeasurement techniques to scramjet researchand testing. Eckbreth [38] describes thefundamental principles on which thesetechniques are based. Hanson [39] providesan overview of imaging methods in combustionflows.

Table 6 summarizes the techniques that werediscussed. Each of the five techniques hasbeen applied under actual or simulatedpropulsion system testing environments.However, with the exception of fuel plumeimaging, none of them can be considered to bea standard in current testing facilities.

The benefits of the techniques in Table 6 canbe appreciated by comparing them to thecurrent state of capability for high speedpropulsion system testing. Particularly athigh flight Mach number conditions (M > 10),ground-based testing is limited to transientfacilities such as shock tubes. In such facilities

TECHNIQUEPARAMETERS

MEASURED ADVANTAGES DISADVANTAGES COMMENTS

Fuel Plume Imaging(Lorenz-MieScattering)

Plume Geometry,Mixing Efficiency

Strong Signal,Experience With

Application

Need To Introduce µm SeedParticles, Behavior At Flame

Front

Initial Difficulties WithSeeding Overcome

By Silica Dry SeedingTechnique

Rayleigh Scattering Plume Geometry,Temperature, Mixture

Fraction

Simplicity,Multiparameter,Multidimensional

Information

Lower Signal-Noise RatiosThan Lorenz-Mie Scattering,

Background Interference

Iodine Fluorescence Time-Averaged (20 s)Velocity, Pressure,

Temperature, SpeciesConcentration

Multiple Parameter,Multidimensional Data

Errors in Interpreting Time-Averaged Parameters From

Time-Averaged Data,Expense, Alignment

Has Provided ValuableData for Scramjet CodeValidation, But LimitedTo Nonreacting Flow

ConditionsCoherent Anti-StokesRaman Spectroscopy

Multiple Major Species,Temperature

Mature, Quantitative,Multi Parameter.

Instrumentation Can BeRemote FromMeasurement

Single Point, Low SignalStrength

Has Been DemonstratedIn Operational TestingEnvironments, Such AsThe Plume of an F100

Turbofan Engine

Planar Laser-InducedFluorescence

Multiple Minor Species,Pressure Temperature,

Velocity

Time-ResolvedMultidimensional,MultiparameterMeasurements

Expensive, Requires CarefulAlignment, Although LessThan CARS. Difficult to

Quantify

Attempts To Apply toScramjet and Hypersonic Testing Have Met With

Mixed Success

Table 6 Non-intrusive Diagnostics Measurement Techniques

16

time per test can be limited to milliseconds,with only a few tests per day. Thus, dataacquisition becomes a primary factor inestablishing the duration and cost of anytesting program.

Methods currently used in scramjet testinginclude electromechanical devices such asthermocouples and pressure transducers forquantitative information, photographic andvideographic image recording, spontaneousemission spectroscopy, and mechanicalsampling. The electromechanical devices arepoint measurement techniques, so that anextensive array is needed to determine time-resolved spatial variations in temperature andpressure. These measurements are intrusiveinto the flowfield if they are mounted onprobes. Otherwise they are restricted tosurface characterization.

Image recording can be based on emitted lightor on shadowgraph or schlieren approaches,which utilize a light source. Images recordedin this manner provide path-averaged,qualitative interpretations of flowfieldbehavior. Attempts have been made to expandschlieren capability by spectrally-resolvedrecording (color schlieren).

Sampling and spontaneous emissionsspectroscopy have provided data oncombustion chemistry. Sampling is a pointmeasurement that is not temporally resolved.The extraction of the sample also can allowadditional chemical reactions to occur in thesample, representing a source of error in themeasurement. Spontaneous emissionspectroscopy shares the path averaginglimitations of image recording methods andrequires assumptions regarding excited statepopulation fractions.

As indicated in Table 6, recent advances in

laser-based measurement techniques offer thehope of overcoming many of the limitationsnoted above for current measurements.However, the application of these methods topractical testing environments is in its infancy,and lessons learned thus far show that theapplication process generally will not bestraightforward or easy.

In Table 6 the results for fuel plume imagingwere based on seeding the injected fuel withsilica (SiO2) particles as scattering sites forlight. As noted in the table, initial problemswere encountered by approaches to createthese particles through the reaction of silanewith oxygen according to the followingreaction mechanism:

H2 + SiH4 + 2O2 → H2 + SiO2(s) + 2H2O + heat

These approaches produced some problemswith nonuniform particle size and particleagglomeration. Furthermore, water vaporwould then be present in the fuel as aconsequence of this chemical reaction.Subsequently, an alternative approach inwhich uniform size silica particles were seededdirectly into the fuel prior to injectionremoved this difficulty. For both approachesparticle agglomeration and residues proved tobe problems. Particle vaporization also was aconcern, and the technique worked best fornoncombusting tests, in which fuel wasinjected into nitrogen gas.

Experience with the optical methods of Table6 has indicated some common areas ofconcern:

1. Windows. Optical access is a major designconcern for a test apparatus. Properdesign requires windows to be sufficientlydurable and not to alter system behavior.

17

If cleaning or replacement is required, thenease in performing these maintenancefunctions should be incorporated into thedesign of the apparatus.

2. Noise and vibration. Noise should beconsidered in the general electromagneticsense. Electrical noise from flowgenerating equipment and frominstrumentation can be a significantinterference to low amplitude signalgeneration. The Lorenz, Mie, andRayleigh scattering techniques in Table 6are examples of elastic scattering, in whichthe signal radiation is at the samewavelength as the incident radiation.Therefore, spectral filtering methods toreduce background radiation may be moredifficult than with the laser-inducedfluorescence and coherent anti-StokesRaman measurements. Mechanicalvibrations represent a serious challenge tooptical alignment and the durability ofoptical components and lasers.

3. Extreme thermodynamic conditions. Thehigh temperatures and pressures associatedwith combustion testing represent a hazardto both measuring equipment and humanoperators. In some previous testsmeasurement system design required acapability for remote adjustment andalignment while measurements were beingperformed.

4. Environmental and safety regulations. Thesafety requirements for propulsion testfacilities and those for the operation oflaser-based measurement instrumentationare not always directly compatible.Recent experience at one scramjet testingfacility that was located outdoors near anairport necessitated enclosing laser beams

to shield them from nearby aircraft. Theseconsiderations impose additional designrequirements that are not directly relatedto the measurements.

The potential benefits of the methods listed inTable 6, as well as other methods currentlyunder study, must be taken into considerationin future scramjet development programs.Although these methods generally have notbeen utilized sufficiently to make theirapplication easy as yet, in some cases uniqueand extremely valuable data have beenobtained that could not be measured by anyother means. The rapid advancement ofoptical measurement technology, includingsuch developments as fiber optics and diodelasers, should facilitate their adoption in thefuture. Furthermore, routine usage shouldsimplify measurement practices, so theparticipation of Ph.D spectroscopists will notbe required.

SimulationThe role of computational fluid dynamics(CFD) in the design of a hypersonicpropulsion system was described by Sindir inthis session. The application ofcomputational techniques to major scramjetcomponents, including the inlet/isolator,combustor, and nozzle, was first discussed.The relevant flow physics in each componentwas considered, followed by the currentapproaches for analyzing that flowfield.Deficiencies in the current approaches werethen described, and new technology requiredto deal with these deficiencies were discussed.The experiments and data needed to validatethe computational tools applied to eachcomponent were also discussed. Followingthe discussion of the analysis and design ofindividual engine components, modeling of theintegrated flow path was considered.

18

CFD has several roles in the design of ahypersonic propulsion system. It primarilyserves as an engineering tool for detailed designand analysis. In addition, results from CFDanalyses provide input data for cycle decksand performance codes. Finally, CFD hasseveral uses in engine test programs used todevelop an engine concept. CFD is first usedto guide the test setup and to determine theproper location for the placement ofinstrumentation in the engine. It has alsoproven to be an effective tool for determiningthe effects of the facility on testing, forexample, the effects of contaminants in acombustion heated facility on an enginecombustor test. During and following a test,CFD is useful to predict flowfieldmeasurements as a complement to measureddata.

The inlet/isolator of a scramjet engine suppliesthe combustor with a required quantity of airat a specified pressure, velocity, and flowuniformity. The physics of the flow in aninlet are characterized by:

1. Moderate strength shock waves2. Shock-boundary layer interactions3. Flow separation in unfavorable pressure

gradients4. Compressibility effects5. Transition to turbulence6. High leading edge thermal loads7. Possible unstart

Computational analyses of inlets typicallyemploy codes that solve the Euler equations orEuler iterated with the boundary layerequations for viscous effects for initialanalyses. More detailed calculations utilizeeither the parabolized Navier-Stokes equationsor the full Navier-Stokes equations if

significant flow separation must be considered.All of the calculations typically solve thesteady-state equations so that the simulationscan be completed in reasonable times.Turbulence is modeled using either algebraic ortwo-equation turbulence models withempirical compressibility corrections and wallfunctions. Transition models are not currentlybeing employed. Thermodynamic propertiesare generally determined by assuming that theinlet flow behaves as a perfect gas orequilibrium air. Calculations are conducted onfixed grids of 100,000 to 2,500,000 points inmultizone domains. A limited degree ofdynamic grid adaptation is employed whennecessary. Typical run times range from a fewminutes to 50 hours on a Cray C-90 computer.

A typical high-speed inlet calculation bySindir is shown in Figure 7. The inlet shownin the figure utilizes side wall compression toachieve the desired outflow conditions into acombustor. The flow in the inlet is modeledusing a full Navier-Stokes code with analgebraic turbulence model. The calculation isconducted on a grid of 240,000 points.Computed pressure contours aresuperimposed on the picture of the inlet. Theplot in the figure shows a comparison betweenthe computed wall pressures plotted as afunction of downstream distance and measureddata. The agreement between the computationand the measured data is excellent. Data awayfrom the inlet walls is not available forcomparison.

Based on the current state of the art for inletcalculations and the future technology needs,the following advancements are needed. Moreefficient parabolized and full steady-stateNavier-Stokes codes with a factor of fiveincrease in run time efficiency are needed.Significant improvements are also required for

19

temporal Navier-Stokes codes for the analysisof unsteady inlet flowfields, including inletunstart. Improvements should occur withalgorithmic advancements, with one promisingarea being multigrid methods [40]. Continuingadvancements in computer architectures willalso enhance code speed. Improved methodsfor dynamic grid adaptation would alsoenhance the ability of computationalalgorithms to capture flowfield features.There is a serious need for the development ofadvanced transition and turbulence models.This is likely the most limiting area foraccurate modeling of inlet flowfields.Promising work is now underway to developnew algebraic Reynolds stress turbulencemodels with governing equations that can beefficiently solved [41, 42]. For nonequilibriumflows, the differential Reynolds stressequations must be solved, however, andfurther work is necessary for this to be donemore efficiently. Advances in large eddysimulation, with the development of subgridscale models appropriate to high-speedcompressible flow, may also allow thistechnique to be applied to inlet flows in thefuture [43]. Finally, work is needed todevelop improved transition models for inletflows, particularly with flows exhibitingadverse pressure gradients.

Experiments must also be conducted toprovide code validation data for inletflowfields. When these experiments areconducted, more extensive wall pressuremeasurements are required, along with detailedwall heat transfer and skin friction data. Thereshould also be an accurate definition of theshock structure present in the inlet flow.Finally, in addition to the wall pressuremeasurements, in-stream measurements arecritical for code validation. Initially, velocityprofiles would be very useful. Pressure and

temperature profiles are also needed.Measurements of these quantities in high-speed compressible flow are quite difficult,stretching the state-of-the-art in flowdiagnostic techniques. To accurately measurethese quantities in inlet flows, significant workwill also be required to develop nonintrusivediagnostic techniques to collect the requiredvalidation data.

The flowfield in the combustor of a scramjetengine is characterized by much of the flowphysics of the inlet, but it is furthercomplicated by:

1. A wide range of flow velocitiesinhomogeneously distributedthroughout the combustor.

2. Small and large scale vortical flows (formixing).

3. Separated flows (for flameholding)4. Complex mixing phenomena.5. Finite rate chemical reaction (that may

equilibrate).6. High temperatures and heat fluxes7. High degrees of anisotropy and

nonequilibrium transfer of turbulenceenergy.

8. Interactions between turbulence andkinetics that affect chemical reactionsand the turbulence field.

Computations of combustor flowfieldstypically employ codes that solve either theparabolized or full Navier-Stokes equations,depending upon the region of the combustorbeing modeled and the degree of flowseparation and adverse pressure gradient beingencountered. Steady-state methods arenormally used with limited unsteady analysesfor mixing studies or the analysis ofcombustion instabilities. Turbulence is againmodeled using algebraic or two-equation

20

models with empirical compressibilitycorrections and wall functions. There is alimited use of models to account forturbulence-chemistry interactions based onassumed probability density functions.Thermodynamic properties are determinedutilizing perfect gas or, in some cases, real gasmodels. Chemical reaction is modeled withreduced reaction set finite rate models. For thehydrogen-air reactions occurring in a hydrogenfueled scramjet, a typical reaction mechanismincludes nine chemical species and eighteenchemical reactions, although other mechanismsare employed as the case dictates [44].Hydrocarbon-fueled scramjet concepts aremodeled with more complex mechanisms thatmust be further reduced to allow practicalcomputations. Calculations in each case aretypically conducted on fixed structured gridsof 200,000 to 2,500,000 points in multizonedomains. Typical run times on a Cray C-90computer range from 30 to over 300 hours.

The results of a calculation of the near-field ofa transverse fuel injector design utilized in ascramjet combustor is shown in Figure 8 [45].Conventional scramjets utilize streamwise fuelinjection in the lower Mach number regime toproduce the desired heat release schedule inthe combustor. In the higher Mach numberregime, some transverse injection is utilized toincrease mixing in order to achieve the requiredheat release schedule with shorter combustorresidence times. The flow near an aligned pairof transverse fuel injectors downstream of arearward facing step is diagrammed in theFigure 8. In this study, air mixed with a smallamount of iodine injected at Mach 1.35 is usedto simulate the fuel. The iodine allows theinjectant to be measured and tracked as itmixes with upstream air initially introduced atMach 2. A comparison of the measured andcomputed mole fraction of injectant in a

streamwise plane cutting through the center ofthe injectors is also shown in Figure 8. Theagreement between the experimental data andthe computed results is quite good.

Many of the future technology needs forcombustor simulations follow from the needsfor inlets described earlier, but many of theadditional requirements will be more difficultto achieve. For combustor modeling, a factorof ten improvement in the efficiency ofsteady-state and temporal Navier-Stokescodes will be needed to carry out the requiredcalculations with the necessary accuracy anddesign turn-around time. Multigrid methodsagain offer promise for significantly enhancingconvergence rates, but the application ofmultigrid methods to reacting flows alsoresults in additional challenges for successwith the method [40]. Current research toapply multigrid methods to high speedreacting flows has resulted in a significantimprovement in convergence rates over singlegrid methods. Dynamic grid adaptation willbecome even more important for capturing thecomplex flow structure in combustors, inparticular the shock-expansion and vorticalstructure in the flow. Proper resolution ofvortical flow requires very high resolution toconserve angular momentum. Again, there is aserious need for improved turbulence modelingin high speed reacting flows, both to model theturbulence field and to properly couple theeffects of turbulence on chemical reaction andreaction on turbulence. Promising work isagain taking place in this area using severalapproaches. Techniques using velocity-composition probability density functionshave been successfully applied toincompressible reacting flows, and this work isnow being extended [46], to modelcompressible reacting flows. Work is alsounderway [43] to apply large eddy simulation

21

(LES) techniques to compressible reactingflows. Subgrid scale models for the LES ofthese flows are currently being developedusing methods previously applied formodeling the full range of flow scales. Finally,further work is needed to simplify themodeling of chemical reaction in combustorflowfields. Methods for systematicallyreducing the number of reactions in a fullreaction mechanism are required to reduce thecomputational work [47]. A number ofpromising methods are under development.They were discussed in a previous section.

As with the modeling of inlet flowfields,experiments are also required to provide datafor the validation of combustor codes. Inaddition to the data required for validating inletmodeling, combustor code validation willrequire extensive temperature and speciesconcentration measurements, as well as thecorrelations of these quantities with each otherand with velocity for validation of advancedturbulence models. Measurements of all of therequired flow variables are more difficult toobtain in the reacting flow environment of ascramjet combustor. Significant work willagain be required to develop nonintrusivediagnostic techniques suitable for making therequired measurements.

The flowfield in the nozzle of a scramjetengine is characterized by much of the flowphysics of the inlet and combustor, butadditional requirements include the modelingof:

1. Strong aerodynamic and chemical non-

uniformities.2. Very high velocities and high

temperatures.3. Significant divergence and skin friction

losses.

4. Changing thermochemical state.5. Potential relaminarization of the flow.6. Energy-bound chemical radicals that

will not relax in the nozzle.7. Excited vibrational states and their

relaxation.

Computations of nozzle flowfields are usuallyconducted with Euler codes or Euler codesiterated with boundary layer calculations forinitial engineering design studies, and witheither parabolized or full Navier-Stokes codesfor more detailed studies. Steady-statemethods are normally employed. Turbulenceis modeled by algebraic or two-equationmodels with empirical compressibilitycorrections and wall functions. Perfect gas or,when necessary, real gas models are used todetermine thermodynamic properties.Chemical reaction is modeled with reducedkinetics models as utilized in the upstreamcombustor flow. Finite rate analyses are stillrequired in the nozzle to assess the degree ofreaction that continues to take place and todetermine the extent of recombinationreactions that add to the available thrust.Calculations for complete nozzles aretypically carried out on structured grids of100,000 to 500,000 nodes grouped inmultizone domains. Typical run times rangefrom 1 to 40 hours on a Cray C-90 computer.

The results of a simulation by Sindir tooptimize nozzle performance are given inFigure 9. A parametric study is performed ona three-dimensional nozzle using a distributionof inflow profiles that are given in the figure.Profiles are characterized in terms of the flowdistortion, given by ep = Pavg/Pmax. Mass andstream thrust are held constant for all of theprofiles. Simulations using each profile areconducted using a 3D Euler code. The effectsof the various flow profiles are characterized

22

in terms of nozzle efficiency, thrust, andthrust vector angle. Plots of nozzle efficiencyand thrust vector angle vs. the distortionparameter are also given in Figure 9. Clearly,nozzle performance is greatly affected by flownon-uniformity. Efficiency tends to increasewhen the distortion parameter becomes morenegative with increasing pressure toward thecowl side of the engine. Therefore, high inflowdistortion, oriented appropriately, canfavorably affect nozzle performance.

Future technology needs for nozzlesimulations, even though less demanding,follow very similar lines to the requirementsfor combustor simulations. A factor of fiveimprovement in the efficiency of the steady-state Navier-Stokes codes is needed. Dynamicgrid adaptation will also be useful forcapturing shock structure and resolvingpossible wall separation due to shock-boundary layer interactions. There is a needfor improved turbulence models for describingnozzle flows. Algebraic Reynolds stressturbulence models offer significant promise fordescribing these flowfields [41, 42]. Thereduced kinetics models currently beingapplied to nozzle flows appear to bereasonably accurate, although some furtherwork to improve the description ofrecombination may be warranted. Validationrequirements for nozzle codes are similar tothose required for combustor codes.

Injection and MixingThe critical issues of fuel injection and mixingin a scramjet combustor were discussed in thissession by Nejad, Brown, and Dimotakis. Anumber of key issues for efficient fuelinjection, mixing and combustion were firstconsidered. The shear/mixing layer flow wasthen discussed to provide a mechanism for abetter understanding of the fundamental

physics of fuel-air mixing and combustion. Anumber of conventional fuel injectionstrategies were then described followed byseveral new less conventional techniques.Finally, an appraisal of these injectionstrategies were made.

There are several key issues that must beconsidered in the design of an acceptable fuelinjector. Of particular importance are the totalpressure losses created by the injector and theinjection processes, that must be minimizedsince they reduce the thrust of the engine. Theinjector design also must produce rapid mixingand combustion of the fuel and air. Rapidmixing and combustion allow the combustorlength and weight to be minimized, and theyprovide the heat release for conversion tothrust by the engine nozzle. The fuel injectordistribution in the engine also should result inas uniform a combustor profile as possibleentering the nozzle so as to produce anefficient nozzle expansion process. Atmoderate flight Mach numbers, up to Mach10, fuel injection may have a normalcomponent into the flow from the inlet, but athigher Mach numbers, the injection must benearly axial since the fuel momentum providesa significant portion of the engine thrust.Intrusive injection devices can provide goodfuel dispersal into the surrounding air, butthey require active cooling of the injectorstructure. The injector design and the flowdisturbances produced by injection also shouldprovide a region for flameholding, resulting in astable piloting source for downstream ignitionof the fuel. The injector cannot result in toosevere a local flow disturbance, that couldresult in locally high wall static pressures andtemperatures, leading to increased frictionallosses and strict wall cooling requirements.

23

Compressible shear/mixing layers and jetsprovide a good model problem for studyingthe physical processes occurring in high-speedmixing and combustion in a scramjet. Mixinglayers are characterized by large-scale eddiesthat form due to the high shear that is presentbetween the fuel and air streams. These eddiesentrain fuel and air into the mixing region.Stretching occurs in the interfacial regionbetween the fluids leading to increased surfacearea and locally steep concentration gradients.Molecular diffusion then occurs across thestrained interfaces. There has been asignificant amount of experimental andnumerical research to study mixing layer andjet flows [48-56]. For the same velocity anddensity ratios between fuel and air, increasedcompressibility, to the levels present in ascramjet, results in reduced mixing layergrowth rates and reduced mixing. The level ofcompressibility in a mixing layer with fuelstream 1 and air stream 2 can beapproximately characterized by the velocityratio, r = U2/U1, the density ratio, s = ρ2/ρ1,and the convective Mach number, Mc = (U2-U1)/(a1 + a2) where a is the speed of sound.Increased compressibility reorganizes theturbulence field and modifies the developmentof turbulent structures. The resultingsuppressed transverse Reynolds normalstresses seem to result in reduced momentumtransport. In addition, the primary Reynoldsshear stresses responsible for mixing layergrowth rate also are reduced. The primarymixing layer instability becomes three-dimensional with a convective Mach numberabove 0.5, reducing the growth of the largescale eddies. Finally, the turbulent eddiesbecome skewed, flat, and less organized ascompressibility increases. All of these effectscombine to reduce the growth rate of themixing layer and the overall level of mixing thatis achieved.

Several phenomena result in the reduction ofmixing with increasing flow velocity, includingvelocity differential between fuel and air, andcompressibility. Potentially, the existence ofboth high and low growth and mixing rates arepossible, and the engine designer with anunderstanding of the flow physics controllingthese phenomena can advantageously usethese effects. The shock and expansion wavestructure in and about the mixing layer caninteract with the turbulence field to affectmixing layer growth [48]. Shock andexpansion waves interacting with the layerresult from the engine internal structure.Experiments have shown that the shocks thatwould result from wall and strut compressionsappear to enhance the growth of the two-dimensional eddy structure (rollers) of amixing layer. This effect is most pronouncedwhen the duct height in the experiment and theshear layer width become comparable. Wavesmay be produced by the mixing layer itselfunder appropriate conditions. Localizedshocks (often termed shocklets) occur withinthe mixing layer when the accelerating flowover an eddy becomes supersonic even whenthe surrounding flow is subsonic. When theoverall flow is supersonic, the eddy shockletswill extend as shocks into the flow beyond theindividual eddies. These shocklets can retardeddy growth due to increased localizedpressure around the eddy.

The growth of a mixing layer produces adisplacement effect on the surrounding flowfield. This displacement in confined flowproduces pressure gradients that can affect thelater development of the mixing layer,typically retarding growth. When chemicalreaction occurs in a mixing layer, resulting inheat release, the growth of the mixing layer isretarded in both subsonic and supersonic flow[48, 49]. The effect of heat release can also

24

vary spatially as a function of the localstoichiometry and chemical reaction.Dimotakis noted that the retarded growth inboth instances can be reversed, however, byallowing the bounding wall to diverge relativeto the initial wall angles where retarded growthwas noted [50].

Several options are available for injecting fueland enhancing the mixing of the fuel and air inhigh speed flows typical of those found in ascramjet combustor. Nejad discussed the twotraditional approaches for injecting fuel includeinjection from the combustor walls and in-stream injection from struts. The simplestapproach for wall injection involves thetransverse injection of the fuel from wallorifices. Transverse injectors offer relativelyrapid near-field mixing and good fuelpenetration. Penetration of the fuel streaminto the crossflow is governed by the jet-to-freestream momentum flux ratio. The fuel jetinteracts strongly with the crossflow,producing a bow shock and a localized highlythree-dimensional flow field. Resultingupstream and downstream wall flowseparations also provide regions for radicalproduction and flameholding, but they can alsoresult in locally high wall heat transfer.Compressibility effects noted earlier formixing layer flows also are evident in themixing regime downstream of a transverse jet.Compressibility again retards eddy growth andbreakup in the mixing layer and suppressesentrainment of fuel and air, resulting in areduction in mixing and reaction. Noncircularorifice injectors, including elliptical and wedgeshaped [59] cross-sections, produce a weakerbow shock and reduced separations, resultingin lower losses and wall heating problems.The lateral spread of the fuel jet is alsoenhanced, and overall mixing is improved,although there is some reduction in transverse

penetration.

Improved mixing has also been achieved usingalternative wall injector designs. Wall injectionusing geometrical shapes that introduce axialvorticity into the flow field has beensuccessful. Vorticity can be induced into thefuel stream using convoluted surfaces or smalltabs at the exit of the fuel injector.Alternatively, vorticity can be introduced intothe air upstream of the injector using wedgeshaped bodies placed on the combustor walls.When strong pressure gradients are present inthe flowfield, e.g. at a shock, vorticity alignedwith the flow can be induced at a fuel-airinterface, where a strong density gradientexists, by virtue of the baroclinic torque. Fuelinjection ramps have proven to be an effectivemeans for fuel injection in a scramjet engine.Two ramp injector schemes are diagrammed inFigures 10 a & b. Fuel is injected from thebase of the ramp. The unswept rampconfiguration provides nearly streamwiseinjection of fuel to produce a thrustcomponent. Flow separation at the base ofthe ramp provides a region for flame holdingand flame stabilization through the buildup ofa radical pool. The ramp itself producesstreamwise vorticity as the air stream shedsoff of its edges, improving the downstreammixing. The swept ramp design provides all ofthe features of the unswept ramp, but thesweep results in better axial vorticitygeneration and mixing. A novel variation onthe swept wedge injector, termed the aero-ramp injector, is also shown in Figure 10c. Itutilizes three arrays of injector nozzles atvarious inclination and yaw angles toapproximate the physical swept ramp design.The aero-ramp injector has many of thefeatures of the swept ramp design without thelosses associated with an intrusive device. Acomparison of the two injectors is given in

25

Figure 11, where transverse fuel penetration,lateral spread, plume area, and mass fractiondecay are shown. While transversepenetration and plume area are reduced withthe aero-ramp, lateral spread and mass fractiondecay are nearly the same as those for theswept ramp injector.

In-stream injection also has been utilized forfuel injection in a scramjet. Traditionalapproaches involve fuel injection from thesides and the base of an instream strut.Transverse injection results in behavioridentical to transverse fueling from the wall.Injection from the base of the strut results inslower mixing as compared to transverseinjection. A combination of transverse andstreamwise injection, varied over the flightMach number range, often has been utilized tocontrol reaction and heat release in a scramjetcombustor. As noted earlier, however,streamwise injection has the advantage ofadding to the thrust component of the engine.To increase the mixing from streamwiseinjectors, many of the approaches utilized toimprove wall injection, including noncircularorifices, tabs, and ramps, have been utilizedsuccessfully. Several new concepts haveemerged as well. Pulsed injection using eithermechanical devices or fluidic oscillationtechniques have shown promise for improvedmixing. A fluidic approach using a Hartmann-Sprenger tube, shown schematically in Figure12a, offers a possible means of producing arapid pressure oscillation with large amplitudeby means of a geometrically simple device.Fuel injection schemes integrated with cavitiesalso provide the potential for improved mixingand flameholding. One possible design isshown in Figure 12b. This integrated fuelinjection/flameholding device, utilizing fuelinjection into a cavity and from its base,integrates the fuel injection with a cavity that

provides flameholding, flame stabilization, andmixing enhancement if the cavity is properlytuned.

Even with these results regarding the behaviorof mixing layer flows and a number oftechniques for enhancing fuel-air mixing, anumber of issues remain to be studied. Indeed,a controversy still exists that questionswhether fuel-air mixing will even be a problemin a scramjet engine in flight. The issue iswhether or not the turbulence present in theatmosphere and ultimately present in the inletflow will provide sufficient turbulent mixing offuel and air in the combustor. Since all of thework to study high-speed mixing flows hasbeen conducted (or simulated) using a different(earth bound) environment, the need forenhanced mixing still remains unresolved.

Concluding RemarksThe presentations and discussion periods ofthe workshop resulted in a number ofinterchanges between engine developers andmembers of the associated researchcommunity that provided a betterunderstanding of the efforts in each topicalarea, in keeping with the workshop objectives.The status of the overall engineering effort wasdescribed, as were critical needs for successfulextensions.

The status of current research in supersonicmixing and combustion was described to theengineering community. A number of plansfor future research were discussed. In manyinstances, the current work and future researchplans were consistent with the engineeringneeds. In other instances, however, needsbecame apparent that are not being addresseddirectly.

While the procedures for engine design are well

26

established and fruitful, the inability to makethe all of the necessary measurements clearlynecessitates further work to develop additionalmeasurement and diagnostic tools.Measurement sensitivities for several criticalengine parameters are given in Table 5.Weaknesses requiring improved measurementtechniques and devices are also pointed out inthe table. Accurate in-stream measurements ofvelocity, temperature, pressure, and chemicalspecies in engine flow fields using nonintrusivediagnostics are also critical to develop asuccessful engine design. A summary ofdiagnostic capabilities for laser-basedinstrumentation applied to scramjet testing isgiven in Table 6. The problems for eachapproach described in the Table must beresolved, or new methods must be developedwhere necessary. Future approaches must bebased on inexpensive and robust technologies,and the resulting instrumentation must beuseful in hostile testing environments.

Simulation and modeling capabilities must beextended to allow more routine application torealistic engine geometries. An order ofmagnitude increase in computational speedmust be achieved before engine design codescan meet this challenge. Multigrid methodsappear to be one approach for achieving thisgoal, but significant work is needed beforeapplying this method to high-speed reactingflows. Improving computer architectures,particularly parallel processors, also willprovide some of the needed enhancement.Turbulence modeling also requires significantwork. Research is needed not only to improvethe capability for modeling the flowfieldturbulence, but also to describe the interactionof turbulence with chemistry in a compressiblereacting flow. For the analysis of enginecomponent flows, large eddy simulation mayprovide a means for computing (rather than

modeling) a larger proportion of the scale ofthe flow. To model chemical reaction of fueland air in an engine, reduced kinetic modelsmust be developed to reduce computationaltime required for solving the species equations,particularly for hydrocarbon fuels. Tosupport hydrocarbon-based scramjet enginedevelopment, a comprehensive data base forC7 - C12 aliphatic fuel components underscramjet conditions should be developed. Inaddition, surrogate hydrocarbon fuels shouldbe selected based upon available informationabout endothermic behavior and catalysis.And finally, the aliphatic fuels data baseshould be utilized to derive suitable startingand reduced mechanisms for candidate fuels.

Several approaches for fuel injection in thecombustor were discussed. Designs utilizinggeometries or flow alterations that inducestreamwise vorticity to enhance mixing appearto be most promising. Losses induced by theinjection process reduce the efficiency of theinjector in most designs, however. Futurework to optimize the injector design formaximum mixing enhancement with minimumlosses will be needed. Work to relate findingsfrom simulations or ground based testing toactual conditions in flight should be included.

Issues regarding the thermochemical andtransport behavior of the fuels were alsoraised. A better understanding of the state ofhydrocarbon fuels as their temperatureincreases in vehicle cooling passages isimportant for design. Changes in the state ofthe fuel can affect the reactivity of the fuel andthe resulting combustion efficiencysignificantly. There is a lack of understandingof the physical processes that may contributeto these effects. To understand thesephenomena, changes in the fuel state must bestudied in a realistic simulation of the scramjet

27

preheating and combustion processes. A co-annular or opposed-jet burner that would burnhot, partially reacted fuels represents onepossible relevant experiment for such studies.Traditional premixed combustion devicesappear to be inadequate to address theimportant issues.

AcknowledgmentThe authors wish to thank all of theparticipants in the workshop for volunteeringto attend and for their valuable contributionsto the discussions. The authors especiallyexpress their gratitude to the presenters at theworkshop, not only for the presentations, butalso for their comments to improve ourattempts to summarize their thoughts in thispaper.

References1. Billig, F. S. "Propulsion Systems FromTakeoff to High Speed Flight," Reprint fromHigh-speed Flight Propulsion System, Vol.137 of Progress in Astronautics andAeronautics, AIAA.

2. Kay, I. W. et al., "Hydrocarbon FueledScramjet, Vol. VIII, Piloting and FlamePropagation Investigation," AFAPL TR 68-146, May 1971.

3. Heins, A. E., et al., "HydrocarbonScramjet Feasibility Program Part III. Free JetEngine Design and Performance," AFAPLTR-74, Jan 1971.

4. Billig, F. S., "Combustion Processes inSupersonic Flow," Journal of Propulsion andPower, Vol. 4, May-June 1988, pp. 209-216.

5. Burnette, T. D., "Dual Mode Scramjet,"The Marquardt Corporation, AFAPL-TR-67-132, June 1968.

6. Harshman, D. L., "Design and Test of aMach 7 - 8 Supersonic Combustion RamjetEngine," AIAA Propulsion Specialist Meeting,July 1967, Washington, DC.

7. Billig, F. S., “Research in SupersonicCombustion,” AIAA Paper 92-0001, Jan 1992

8. Anderson, G. Kumar, A., and Erdos, J.“Progress in Hypersonic CombustionTechnology with Computastion andExperiment,” AIAA Paper 90-5254, Oct.1990.

9. Voland, R. and Rock, K., “ConceptDemostration Engine and Subscale ParametricEngine Tests,” AIAA Paper 95-6055, April1995.

10. Stevens, R. Thompson, M. W., “SteamCalorimetry Uncertainty Analysis for Direct-Connect Arc Tunnel Combustor tests, GWP #107A,” APL-NASP-94-016, Jan. 1995

11. JANNAF Ramjet SubcommitteeWorkshop, "Fuels for Future High-SpeedFlight Vehicles," CPIA Publication 518 (2vol.), June 1988.

12. Northam, G. B., ed., "WorkshopReport: Combustion in Supersonic Flow,"22nd JANNAF Combustion MeetingProceedings, CPIA Publication 432, Vol. 1,1985.

13. Beach, H. L., "Supersonic CombustionStatus and Issues," in Major Research Topicsin Combustion, Hussaini, Kumar, and Voigt,eds, Springer Verlag, NY, 1992.

14. Heiser, W. H., Pratt, D. T., HypersonicAirbreathing Propulsion, AIAA.

28

15. Murthy, S. N. B., and Curran, E. T.,eds, High-Speed Flight Propulsion Systems,AIAA Progress in Astronautics andAeronautics, Vol. 137, AIAA, 1991.

16. Sobel, D. R., and Spadaccini, L. J.,"Hydrocarbon Fuel Cooling Technologies forAdvanced Propulsion," ASME-95-GT-226,June 1995.

17. New World Vistas, Air and SpacePower for the 21st Century, Aircraft andPropulsion Volume, 1996.

18. Ianovski, L. S., "Endothermic Fuels forHypersonic Aviation," Symposium on Fuelsand Combustion Technology for AdvancedAircraft Engines, AGARD-CP-536, pp. 44-1to 44-8, Sept. 1993.

19. Ianovski, L., Sapgir, G., "Heat andMass Transfer to Hydrocarbon Fuels atThermal Decomposition in Channels ofEngines," AIAA Paper 96-2683, July, 1996.

20. Lander, H., Nixon, A. C., "EndothermicFuels for Hypersonic Vehicles," Journal ofAircraft, Vol. 8, No. 4, pp. 200-207, 1971.

21. Favorskii, O. N., Kurziner, R. I.,"Development of Air-Breathing Engines forHigh Speed Aviation by Combining Advancesin Various areas of Science and Engineering,"High Temperature (translation of TeplofizikaVysokikh Temperature), Vol. 28, No. 4, pp.606-614, 1990.

22. Warnatz, J., "Rate Coefficients in theC/H/O System," in Combustion Chemistry,W. C. Gardiner, ed., Springer Verlag, NewYork, pp. 197-360, 1984.

23. Spadaccini, L. J., Colket, M. B.,"Ignition Delay Characteristics of MethaneFuels," Prog. Energy Combust. Sci., Vol. 20,pp. 431-460, 1994.

24. Faith, L. E., Ackerman, G. H., Heck, C.K., Henderson, H. T., Ritchie, A. W. Ryland,L. B., "Hydrocarbon Fuels for AdvancedSystems," AFAPL-TR-70-71, Dec. 1971.(AIAA Journal, Vol. 4, pp. 513-520, 1966).

25. Edwards, T., "Recent Research Resultsin Advanced Fuels," ACS PetroleumChemistry Division Preprints, Vol. 41, No. 2,pp. 481-487, 1996.

26. Savage, P. E., et al, "Reactions atSupercritical Conditions: Applications andFundamentals," AIChE Journal, Vol. 41(7),pp. 1723-1778, 1995.

27. Dardas, Z., et al, "High-Temperature,High-Pressure in Situ Reaction Monitoring ofHeterogeneous Catalytic Processes underSupercritical Conditions by CIR-FTIR," J.Catalysis, Vol. 159, pp. 204-211, 1996.

28. Pellett, G. L., Northam, G. B., Wilson,L. G., "Counterflow Diffusion Flames ofHydrogen, and Hydrogen Plus Methane,Ethylene, Propane, and Silane, vs Air: StrainRates at Extinction," AIAA Paper 91-0370,Jan. 1991.

29. Edwards, T., "USAF SupercriticalHydrocarbon Fuels Interests", AIAA Paper93-0807, January 1993.

30. Chen, L.-D., Sui, P.-C., "AtomizationDuring the Injection of Supercritical Fluid IntoHigh Pressure Environment," paper presentedat 1994 IUTAM on Droplets and Sprays, 6-10 Dec. 1994, Tainan, Taiwan.

29

31. Petley, D. H., Jones, S. C., “ThermalManagement for a Mach 5 Cruise AircraftUsing Endothermic Fuel,” Journal of Aircraft,Vol. 29(3), pp. 384-389, 1992.

32. Pegg, R. J., Hunt, J. L., Petley, D. H.,“Design of a Hypersonic Waverider-DerivedAirplane,” AIAA Paper 93-0401, 1993.

33. Egolfopoulos, F. N. and Law, C.K.,“Chain mechanisms in the Overall ReactionOrders in Laminar Flame Propagation,”Combustion and Flame, Volume, 80, pp. 7-16,1990.

34. Peters, N., “Numerical and AsymptoticAnalysis of Systematically Reduced ReactionSchemes for Hydrocarbon Flames,” NumericalSimulation of Combustion Phenomena,Lecture Notes in Physics, Vol. 241, pp. 90-109, 1985.

35. Lam, S. H., “Reduced ChemistryModeling and Sensitivity Analysis,” 1994-1995 Lecture Series: Aerothermochemistry forHypersonic Technology, Von KarmanInstitute for Fluid Dynamics, 1995.

36. Chelliah, H. et al., “Reduced KineticMechanisms for Application in CombustionSystems,” Peters and Rogg, editors, SpringerVerlad, 1993.

37. Pope, S. B., “Computationally EfficientImplementation of Combustion ChemistryUsing In-Situ Adaptive Tabulation,” CornellUniversity Report FDA 96-02, 1996.

38. Eckbreth, A.C., Laser Diagnostics forCombustion Temperature and Species, SecondEdition. Combustion science and TechnologyBook Series, Vol. 3, Gordon and Breach

Publishers, SA, 1996.39. Hanson, R. K., “CombustionDiagnostics: Planar Flowfield Imaging,”Twenty First Symposium (International) onCombustion, The Combustion Institute, pp.1677-1691, 1986.

40. Edwards, J. R., “Development of anUpwind Relaxation Multigrid Method forComputing Three-Dimensional ViscousInternal Flows,” AIAA Paper 95-0208, Jan.1995.

41. Adumitroaie, V., Colucci, P. J., Taulbee,D. B., and Givi, P., “LES, DNS and RANS forthe Analysis of High-Speed TurbulentReacting Flows,” Annual Report, NASAGrant NAG 1-1122, 1994.

42. Girimaji, S. S., Galilean, A., “InvariantExplicit Algebraic Reynolds Stress Model forCurved Flows,” NASA CR-198340, pp. 1-28June 1996.

43. Colucci, P. J., Jaberi, F. A., Givi, P.,and Pope, S. B., “Filtered Density Functionfor Large Eddy Simulation of TurbulentReacting Flows,” Submitted for publication,Physics of Fluids.

44. Jachimosksi, C. J., “An AnalyticalStudy of the Hydrogen-Air ReactionMechanism with Application to ScramjetCombustion,” NASA TP-2791, Feb. 1988.

45. Eklund, D. R., Northam, G. B., andHartfield, R. J., “A Detailed Investigation ofStaged Norman Injection into a Mach 2 Flow,”Proceedings of the 27th JANNAFCombustion Meeting, Cheyenne, Wyoming,Nov. 5-9, 1990.

46. Delarue, B. J., and Pope, S. B.,

30

“Application of PDF Methods toCompressible Turbulent Flows,” Submittedfor publication, Physics of Fluids.

47. Mass, U., and Pope, S. B., “SimplifyingChemical Kinetics: Intrinsic Low-DimensionalManifolds in Composition Space,”Combustion and Flame, V. 88, pp. 239-264,1992.

48. Dimotakis, P. E., “Turbulent Free ShearLayer Mixing and Combustion,” High SpeedFlight Propulsion Systems, Chapter 7,Progress in Astronautics and Aeronautics, vol.137, 1991.

49. Drummond, J. P., Carpenter, M. H.,and Riggins, D. W., “Mixing and MixingEnhancement in High Speed Reacting Flows,High Speed Flight Propulsion Systems,”Chapter 7, Progress in Astronautics andAeronautics, vol. 137, 1991.

50. Givi, P., and Riley, J. .J., “SomeCurrent Issues in the Analysis of ReactingShear Layers: Computational Challenges,”Major Research Topics in Combustion,Editors: M. Y. Hussaini, A. Kumar and R. G.Voigt, pp. 588-650, Springer-Verlag, NewYork, NY,1992.

51. Drummond, J. P., and Givi, P.,“Suppression and Enhancement of Mixing inHigh-Speed Reacting Flow Fields,”Combustion in High-Speed Flows, pp. 191-229, Editors: J. Buckmaster, T.L. Jackson andA. Kumar, Kluwer Academic Publishers,Boston, MA, 1994.

52. Goebel, S. G., and Dutton, J. C.,“velocity Measurements of Compressible,

Turbulent Mixing Layers,” AIAA Paper 90-0709, January 1990.53. Smith, S. H., Lozano, A., Mungal, M.G., and Hanson, R. K., “Scalar Mixing in theSubsonic Jet in Crossflow.” AGARDComputational and Experimental Assessmentof Jets in Crossflow, Winchester, pp. 6.1-6.13, 1993.

54. Hall, J. L., Dimotakis, P. E., andRosemann, H. “Experiments in Non-reactingCompressible Shear Layers,” AIAA Paper 91-0629, January 1991.

55. Samimy, M., Zaman, K. B. M .Q., andReeder, M. F., “Effect of tabs on the Flow andNoise Field of an Axisymmetric jet,” AIAAJournal, Vol. 31, No. 4, pp. 609-619, 1993.

56. Cox, S. K., Fuller, R. P., Schetz, J. A.,and Walters, R. W., “Vortical interactionsGenerated by an Injector Array to EnhanceMixing in Supersonic Flow” AIAA Paper 94-0708, January 1994.

Appendix A

Research Problems for Future Work

• Affordability - minimize weight,size,complexity, part count for a givenmission profile

• Methodology for optimization

• Materials and structures

– scaling of leading edges tominimize drag

– “cheating” by injecting liquid,ablating, etc.

• Inlet design

• Fuel characterization

• Unsteadiness

• Fuel injection and mixing

31

• Recovery of kinetic energy to producethrust

GROUND BASED TESTING

• Influence of contaminants on ignition-vitiation; combustion and expansionbesides ignition; effects on radiation

• Turbulence - no current data

• Nonequilibrium

• Boundary layer effects

• Scaling to flight conditions

• Complementary experimental programfor relevant phenomena

• Utilization of pulsed facilities− elimination of vitiation effects

FUELS AND FUEL SYSTEMS

• Experimental program to determineenergy yields of fuels

• Creation of kinetics data base for long-term use

• Low temperature starting and pilotingsystems

– trimethyl Al additives

– GASL micro rocket

– plasma torch

– embedded ramjets

– gelled fuels (GASL)

• Improvement of fuel specific ...

• Nano particle carbon particles

• Micro encapsulated fuels

INJECTION AND MIXING

• Exploitation of longitudinal vorticity formixing enhancement

• Interaction between injectors

• Minimization of losses

• Thermodynamic state of fuel at injection

• Cold flow studies?

• Curvature-induced Rayleighdestabilization; role of pre-existingturbulence

• Systems studies to optimize, but notnecessarily minimize, losses

• German-Russian work on three injectorclasses-micro pylons

• Relationship to flame holding

COMBUSTION CHEMISTRY

• Compile and validate kinetic data base atthree levels

– detailed

– skeletal

– reduced

• Ignition enhancers

• Liquid-phase kinetics; supercriticalkinetics

• Recombination kinetics

– catalytic additives

• Role of soot

• Combustion at high strain rates

• Unsteadiness

• Incorporating kinetics mechanisms indesign codes

• Development of subscale experiments

– Russian results by Baev?

– opposed jet burner?

DIAGNOSTICS

• Skin friction measurements

• Heat flux measurements, includingradiation

• Detailed measurements of boundary andinitial conditions

• Application of non-intrusiveinstrumentation to free jet tests

• Measurement of velocity profiles

• Determination of measurementuncertainties

32

• Turbulence intensity levels(concentration in supersonic flow)

• Instantaneous measurements todetermine turbulence-chemistryinteractions

• Pressure-sensitive paint to measuresurface pressures

• Mapping of total pressure and totaltemperature

• Design of well-posed experiment

• Concentration measurements

– mean and fluctuating

– spectrally resolved

SIMULATION

• Stochastic models

• Sensitivity to unsteadiness

• Algebraic closure models

– stress

– scalar flux

• Solvers for particle methods

– improved efficiency

• Well-posed validation experiment

• Preprocessing

– adaptive gridding

• Solvers

– increased efficiency (factor of 10)

– provisions for real time (dynamic)grid adaptation

– domain and functiondecomposition capability formassively parallel and/ornetworked computers

• Physical models

– turbulence/chemistry interactionmodels

– testing of higher orderphenomenological turbulencemodels

– assessment of LES techniques forrealistic geometries and flowconditions

– testing of fast reduced kineticsmechanisms

BILLIG’S COMMENTS

• Inlets-Isolators

– streamline tracing

– analogy between C-I-I &aerodynamic phenomena

– shear high temperature reduction

– sweep

– starting

• Fuels

– densification

– additives

– storability

– toxicity

– rheology

• Fuel Preparation

– heat pipes (open-closed)

– plasma generators

• Injection - Mixing

– subsonic imbedded zones(cavities, steps, bases)

• Ignition− radical generators

• Combustion-Combustors

– physical vs. thermal throats

– shear, high temperature

– recombination kinetics

– transpiration

• Nozzles

– shear, high temperature

– recombination

– exploitation of non-uniformentrance flow

33

Figure 1 Schematic illustration of Dual Combustor Engine

Figure 2. Schematic Illustration of Direct Connect Combustor Test Facility

34

Figure 3. Schematic Illustration of Steam Calorimetry Data Analysis Procedure

Figure 4. Laminar Burning Rate vs. Pressure

35

Figure 5. Overall Reaction Order vs. Pressure

Figure 6. Mass Fraction and Temperature vs. Mixture Fraction

36

Figure 7. Typical Side Wall Compression Inlet Calculations

37

Figure 8. Calculations of the near Field of a Transverse Fuel InjectorConfiguration

Numerical Simulation

Experimental Results

Schematic of Tandem Fuel Injection Behind a Step in Supersonic Flow

38

Figure 9. Optimization of Nozzle Performance Based on Nozzle Inflow Profiles

39

Figure 10 (a, b, c). Schematic Illustration of Ramp Fuel Injectors for Scramjet Engines a) Unswept, b) Swept, c) Aero-Ramp

Figure 11. Comparison of Ramp and Aero-Ramp fuel Injectors(fuel mass fraction contours)

40

(a)

(b)

Figure 12(a, b). Schematic Illustration of Pulsed and Cavity Injector-FlameholdersConcepts. a) Hartmann-Sprenger Tube, b) Integrated Injector-Flameholder