AFRL-PR-WP-TP-2006-248 FUEL-AIR INJECTION EFFECTS ON COMBUSTION IN CAVITY-BASED FLAMEHOLDERS IN A SUPERSONIC FLOW (POSTPRINT) Captain William Allen, Dr. Paul I. King, Dr. Mark R. Gruber, Dr. Campbell D. Carter, and Dr. Kuang–Yu Hsu JULY 2005 Approved for public release; distribution is unlimited. STINFO COPY The U.S. Government is joint author of the work and has the right to use, modify, reproduce, release, perform, display, or disclose the work. PROPULSION DIRECTORATE AIR FORCE MATERIEL COMMAND AIR FORCE RESEARCH LABORATORY WRIGHT-PATTERSON AIR FORCE BASE, OH 45433-7251
Transcript
AFRL-PR-WP-TP-2006-248
FUEL-AIR INJECTION EFFECTS ON COMBUSTION IN CAVITY-BASED FLAMEHOLDERS IN A SUPERSONIC FLOW (POSTPRINT) Captain William Allen, Dr. Paul I. King, Dr. Mark R. Gruber, Dr. Campbell D. Carter, and Dr. Kuang–Yu Hsu JULY 2005
Approved for public release; distribution is unlimited.
STINFO COPY
The U.S. Government is joint author of the work and has the right to use, modify, reproduce, release, perform, display, or disclose the work. PROPULSION DIRECTORATE AIR FORCE MATERIEL COMMAND AIR FORCE RESEARCH LABORATORY WRIGHT-PATTERSON AIR FORCE BASE, OH 45433-7251
i
REPORT DOCUMENTATION PAGE Form Approved OMB No. 0704-0188
The public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, searching existing data sources, gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information, including suggestions for reducing this burden, to Department of Defense, Washington Headquarters Services, Directorate for Information Operations and Reports (0704-0188), 1215 Jefferson Davis Highway, Suite 1204, Arlington, VA 22202-4302. Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to any penalty for failing to comply with a collection of information if it does not display a currently valid OMB control number. PLEASE DO NOT RETURN YOUR FORM TO THE ABOVE ADDRESS.
1. REPORT DATE (DD-MM-YY) 2. REPORT TYPE 3. DATES COVERED (From - To)
July 2005 Conference Paper Postprint 09/01/2004 – 07/31/2005 5a. CONTRACT NUMBER
In-house 5b. GRANT NUMBER
4. TITLE AND SUBTITLE
FUEL-AIR INJECTION EFFECTS ON COMBUSTION IN CAVITY-BASED FLAMEHOLDERS IN A SUPERSONIC FLOW (POSTPRINT)
5c. PROGRAM ELEMENT NUMBER 62203F
5d. PROJECT NUMBER
3012 5e. TASK NUMBER
AI
6. AUTHOR(S)
Captain William Allen and Dr. Paul I. King (Air Force Institute of Technology) Dr. Mark R. Gruber and Dr. Campbell D. Carter (AFRL/PRAS) Kuang–Yu Hsu (Innovative Scientific Solutions, Inc.)
5f. WORK UNIT NUMBER
00 7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) 8. PERFORMING ORGANIZATION
REPORT NUMBER Air Force Institute of Technology WPAFB, OH 45433
Propulsion Sciences Branch (AFRL/PRAS) Aerospace Propulsion Division Propulsion Directorate Air Force Research Laboratory Air Force Materiel Command Wright-Patterson AFB, OH 45433-7251
Innovative Scientific Solutions, Inc. Dayton, OH 45440
AFRL-PR-WP-TP-2006-248
9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10. SPONSORING/MONITORING AGENCY ACRONYM(S)
AFRL-PR-WP Propulsion Directorate Air Force Research Laboratory Air Force Materiel Command Wright-Patterson AFB, OH 45433-7251
12. DISTRIBUTION/AVAILABILITY STATEMENT Approved for public release; distribution is unlimited.
13. SUPPLEMENTARY NOTES Conference paper published in the Proceedings of the 41st AIAA/ASME/SAE/ASEE Joint Propulsion Conference and Exhibit, published by AIAA. The U.S. Government is joint author of the work and has the right to use, modify, reproduce, release, perform, display, or disclose the work. PAO case number: AFIT/PA 060861; Date cleared: 06 September 2006. Paper contains color.
14. ABSTRACT The effect of direct fuel and air injection was experimentally studied in a cavity-based flameholder in a supersonic flow. Cavity- based fuel injection and flameholding offer an obstruction-free flow path in hydrocarbon-fueled supersonic combustion ramjet (scram jet) engines. Additionally, this study included characterization of the operational limits (i.e., sustained combustion limits) over a variety of fuel and air flow rates. The cavity rearward ramp includes 10 spanwise injection ports at each of 3 axial stations configured to inject air, fuel, and air, respectively. Planar laser-induced fluorescence (PLIF) techniques were utilized to collect planar distributions of the OH radical at various axial locations within the cavity under different flow conditions. A high-speed emissions camera was used to evaluate the combustion across the cavity. Direct injection of both fuel and air provided additional capability to tune the cavity such that a more stable decentralized flame results. The addition of air injection provided the most improvement over the baseline case (fuel only) near the upstream portion of the cavity close to the cavity step.
The effect of fuel and air injection was experimentally studied in a cavity basedflameholder in a supersonic flow. Cavity based fuel injection and flameholding offer anobstruction-free flow path in hydrocarbon fueled supersonic combustion ramjet (scramjet)engines. The characterization of cavity-based fueling systems is still largely unavailable.Therefore, the subject of this investigation was to expand the cavity based fueling systemsuch that both fuel and air are directly injected. Additionally, this study includedcharacterization of the operational limits (i.e., sustained combustion limits) over a variety offuel and air flow rates. The cavity is recessed below the surface with a 90-degree rearward-facing step and a trailing ramp with a 22.5 degree ramp angle. The cavity rearward rampincludes ten span-wise injection ports at each of three axial stations configured to inject air,fuel and air respectively. Planar Laser-Induced Fluorescence (PLIF) techniques wereutilized to collect planar distributions of the OH radical at various axial locations within thecavity under different flow conditions. Furthermore a high speed emissions camera wasused to evaluate the combustion across the cavity. Direct injection of both fuel and airprovided additional capability to tune the cavity such that a more stable decentralized flameresults. The addition of air injection provided the most improvement over the baseline case(fuel only) near the upstream portion of the cavity close to the cavity step.
This material is declared a work of the U.S. Government and is not subject to copyright protection in the United States.
AmericanInstitute of AeronauticsandAstronautics2
I. Introductionypersonic fl ight hasoffered and wil l continue to offer significant payoffs for both the military and civilianpopulace. Flight at higher Mach numbersis conducive to business in the global market placeas both high
priority packagesandpeoplecanbetransportedacrossgreat distancesin shorttime. Military leaderscanutilizethistechnologyin thewar on terror which fundamentally requiresa quick responseto neutralizesinglesignificant threatswithin narrow time windowsin order to capitalize on intelligence. Higher fl ight velocitiesenablegreater distancesto be coveredwithin acceptable responsetimes. Given that today’s military continuesto strugglewith downsizingandbaseclosureand realignment, hypersonicvehiclesoffer thepotential to rampdownoverseasoperationswithouta detrimental effecton responsetime to overseas targets. Furthermore, thehigh kinetic energy could also beappliedto weaponsystemswhere targetsare neutralized using the kinetic energy of the warheadrather than chemical ornuclearenergy. Theapplication of SCRAMJET technologycould alsoreducethecostof space accessby providingspace vehicles with a fraction of the required escape velocity. For these reasons, SCRAMJET technology is thesubjectof researcharoundtheglobe.1,2
A. BackgroundFlight in thehypersonicregimehasbecomemore common sincea GermanV2 rocketexceededMach5 in 1949.
Several countriesnow have rocket programs that provide accessto spaceand thesevehicles encounterhypersonicenvironments. Traditionally, however, rocket propulsion has been applied to realize hypersonic fli ght. Suchsystems,from a historical standpoint, require morefuel and oxidizer to satisfy a desire to fly fartheror faster. Oneof the primarydisadvantagesof rocket propulsion at least for atmospheric hypersonicfli ght is that they must carryall of their oxidizer on board. This in addition to the increased number of components required to store andtransport the oxidizer to the combustionchambercontributessignificantly to the overall weight andcomplexityofthe vehicle. The increased weight translates simply into larger vehicles or decreasedpayloads. SupersoniccombustionRAMJET (SCRAMJET) engineswould negate the need to carry oxidizeron boardof theaircraft asallof the oxygen needed for combustion would be garnered from the atmosphere. AnotheradvantageSCRAMJETengineshave overrockets is their ability to bethrottled. Thrust levelsfor solid rocketsarebasedsolely upondesignandcurrentliquid rocketshave limitedthrottleability.
Supersoniccombustion is inherently a diffi cult event. Generally speaking, combustion is an exothermicchemicalprocesswhich requires fuel, oxidizer, initiation energyand time for the chemical reactionto take place.The last key ingredientis not easy to come by given supersonic flow through the SCRAMJET. The simplerelationshipbetweentime distance andvelocity would tend to suggestincreasing the lengthof theengine to allow agreater time for combustion to takeplacegiven the velocity of the flow through the engine. Howeverthis wouldincreasethe weight of the enginethereby decreasingan aircraft’s payload. Furthermore, it hasbeennoted that thethrust to drag ratio of an engineis approximately proportional to the ratio of thecombustor’sdiameter to its length.3
This provides additional incentive to keep the combustor length to a minimum. A significant challenge in thegenerationof SCRAMJET propulsioniscompleting thecombustionprocesswithin theengine. Combustion requiresthat fuel is introduced, mixed with the oxidizer in a sufficient quantity and thenprovided with energyto start thereaction process. As noted previously, this requires a fini te amount of time which, givencorevelocity throughtheburner, can be relatedto distance. Since large distancesare not feasible several techniqueshavebeenemployedboth computationally andexperimentally to assist the combustion process. First, obstructionsand/or fuel injectionschemescanbe introduced into the supersonic flow causing disruptionin the boundarylayer andthe formation ofshockwaves. Previouswork hasshown this creates a regionof high turbulence thatcanbecomparedto a regionofeffective mixing at leaston a qualitative basis. Secondly, a cavity canbeintroducedto theflow creatinga subsonicflow region therebyincreasing the residence time and creating a region of heatedgases to aid in the combustionprocess.
B. Previous ResearchCavity based fuel injection and flameholding offer an obstruction-free flow path in hydrocarbon fueled
SCRAMJET engines. Suchflame holding cavities canprovide the benefitof relatively long residence timesand,coupledwith a direct cavity fuel injection scheme,can provide robustflameholdingwith minimal dragpenalties inthe presenceof significant changes in the freestream flow field. However, detailed information regarding thebehavior of these devicesnamely their optimal shape and fueling strategies, combustion stability and interactionswith disturbancesin the main air flow is largely unavailable in the existing literature.4 Previous studies haveconcludedthatvariations in geometryaffectdifferentaspectsof the flow in andaroundthecavity. Key geometriesthat affectcavity flowf ieldsandthereforeits suitability asa flameholderare asfollows: lengthto depthratio, offset
H
AmericanInstitute of AeronauticsandAstronautics3
ratio andaft rampangle. The lengthto depth ratio categorizes cavitiesaseitheropenor closed. Theshearlayer ofanopencavity spansthe entire cavity length whereasthe shear layer attaches to the bottom wall of a closed cavitydueto thecavity’s increased length. Typically, L/D<10 definesanopen cavity while L/D>10 is considereda closedcavity. Studieshave shownthat open cavities imposea smaller dragpenalty on a supersonicengine.4 Previouslowspeed combustion studiesfoundoptimum flameholdingperformancecoincided with a cavity with its length to depthratio sizedfor the minimum aerodynamic drag. Longer cavities produced vortex sheddingthat resulted in cavityoscillationsand unstable flames and shorter cavities lacked sufficient air entrainmentto sustain combustion.5 Asnoted before, if thecavity lengthincreased suchthat thecavity wasclosed(L/D>10)anevengreaterincrease in dragwould occur. Cold flow calculationsperformed by Baurle and Gruber for variousgeometriesshow that cavitylength determinesmassentrainmentandcavity depth determinesresidence time.6,7 Changesto theoffset ratio alsocausedrastic changesto theflowfield. As offset ratio is increased above unity a strongexpansionfan takestheplaceof a compression waveat the forward cavity wall. Additionally, increasingthe offset ratio seemsto influence thevortexstructurewithin thecavity. During a nonreactive flow studyperformedby Gruberet al., curvedwaveswereshown emanating from the cavity’s forward wall at anoffset ratio of one. Thesewavesmay be the product of theoscillatorynature of the cavity vortex. However, when the offset ratio wasincreasedto two for the same aft rampanglecurvedwaveswere not generated. The aft rampangleis anotherkey parameteraffecting cavity flowfields.Gruber et al. studied the flowfield in and around several different geometric configurations underMach 3 flowconditions. The study was non-reactive andincludedbothschlierenandshadowgraph photography.Furthermore,acomputational fluid dynamics (CFD) routine wasexecuted for variouscavity geometries. Residencetime (τ) wasreduced from CFD data. Starting from a steady state solution the fluid is markedand the simulation is steppedforward in time while the markedfluid is monitoredasit exits the cavity. The dragcoefficient presentedin theirstudyis thedragforce normalized by thefreestreamdynamic pressure andthe cavity fore wall area6. As theaft wallangle (θ) is reducedfrom 90° a more stable, two-dimensional flowfield is formed. The separationwave at theforward cavity step changesfrom compressive to expansive asθ decreasesfrom 90̊ to 30˚ to 16°. Additionally,reductionsin theaft ramp anglefrom 90-30-16° resultedin higher dragcoefficientsandlower residencetimes,bothof which could be considered detrimental to an effective flameholder. However,theresulting stableflowfi eld froma decreasedaft ramp angle could justify a decrease in residence time and an increasein drag coefficient. “Ingeneral, decreasing the aft wall angleshouldpromoteboth a moreacoustically stablecavity flow (and subsequentstableburning) andimproved entrainmentbecausethe shearlayer impingesdeeper into thecavity.”6 This trendhasbeen verified in reactivestudies. After several injection sites were studiedfor a fixed cavity geometry, a widerrangeof sustainedflames wasestablishedusingcavity rampinjection.4
Numerousstudieshavebeenaccomplishedregardingflow overopencavitiesasit is anoften-seenconfiguration. Thereareseveral flow trendsthat should benoted. First, rectangular cavities are usually characterized by a level ofunsteadiness. This unsteadinessis observed asoscillations in pressure,densityand velocity in and around the cavity. Unsteadiness introduces anothercomplicating elementinto the cavity flow dynamics and it has beennoted thatcavity flow canbe very threedimensional, especially off centerline. Secondly,the creationof a lobed recirculationzoneis commonly noted. Figure 1 showsthe pressure contours and stream traces derived from a standard two-dimensional eddy-viscosity-based CFD turbulence model. Notice that twocounter rotating lobes are formed for each of the geometries used in thesimulation. Decreasingthe aft ramp angle appears to decrease the size of thesecondary lobe. However, for both L/D and each aft ramp angle studied aprimary and secondary vortex was generated. It has beennoted in previoussubsoniccombustor simulations that the sizes of the vortices alternate in time.Cavity flow is furthercomplicatedby the three dimensionality of the flow. Thesimulation results shownin Figure 1 are basedon thecavity centerlinewheretheflow tends to be two dimensional in the x-y (streamwise-transverse) plane.However, becauseflow is three dimensionaladditional structures most likelyexist in the x-z (streamwise-spanwise)plane. This aerodynamic featureofcavities presents both challengesand benefits to its use as a cavity basedflameholder. The regionof recirculation will provideadditional residencetimefor combustion to takeplace. However the dual vortex structure may require morecomplicatedfueling schemestoprovide a uniform combustible mixture throughoutthecavity.6
Figure1 – StreamTraces(Ref 5 Mach3)
AmericanInstitute of AeronauticsandAstronautics4
Fuelingstrategies mustbe derived to ensurea robust flameholderfor both the subsonic andsupersonic modes.First, consider the air entrainment rate for boththe subsonic (high backpressure) andsupersonic (low backpressure) cases. Figure 2shows a representative shadowgraphimagesforeach case. Notice the shear layer reattachmentis on theaft ramp face for thepurely supersoniccase(low backpressure) and that it is separatedfrom the cavity in the subsonic/supersonic case(high backpressure). This differencesubstantiallyaltersthefreestreamentrainmentwhich could bea mixtureof fuel and air, and effectively increasesthevolumeof thecavity in thehighbackpressure case. It hasalsobeenshownthat mixing is enhancedwithin thecavityby theshock train developed at a high backpressure. A flameholding cavity wasdesigned,fabricatedandtestedbythe Air Force Research Laboratory at Wright-Patterson Air ForceBasein Ohio throughthe Propulsion SciencesBranch (AFRL/PRAS). Thedesign is shownin figure 3. The cavity hasa length of 2.60 in, a depthof 0.65inchesandanaft rampangleof 22.5˚. It is anopencavity given its L/D of 4.7 andit hasanoffset ratio of unity. Severalfuel injection strategieswere studiedby Gruberet al.4 for boththehigh andlow backpressure cases.Mixing studiesinvolving indirect injection showedhigherjet penetration given the high backpressurecondition. This equatedto areducedentrainment into the cavity. They also showed that entrainment into the cavity relies largely on diffusionthrough the shear layer and the interactionbetweentheshearlayerandtheaft rampface.Direct injection throughF4 andF5 portswerein general better cavity fueling schemes.However, cavity fueling was still dependenton the shear layer interaction with the aftramp. As noted before the flameholder mustbe effective during dual-mode operation.Several cavity combustion tests wereconductedduring the transition from low tohigh backpressure. The only fueling schemethat produced sustained cavity combustion with the presenceof theshocksystemwasF5 (aft rampinjection). The influence of the shock system and shearlayer on cavity fueling isminimized by fuel injection from F5. Despite the increasedrobustnessof the flameholder using aft ramp injectionfueling schemes,Gruber et al. noticed that somefuel injection pressuresresulted in localizedcombustion regions.This suggestedthat the cavity may be too largefor efficient mixing andcombustionfor the conditionstested. Adragpenalty is paid for the inclusionof a cavity basedflameholder. Fromthis standpoint, it is importantto ensurethat thecavity sizeis kept to a minimumandthereforeefficient useof cavity volumeis essential. The combustionstudyaccomplishedby Gruberutilized PlanarPLIF configured to detect the presence of the hydroxyl radical. Forgivenfreestreamconditions,imageintensitywasgreatestfor a singlefuel flow rate. This indicatesthatgivena fuel-only flow thefuel flow ratemustbetunedto obtain maximum utilization of thecavity volumewith minimum flameoscillations. Aft rampfueling strategiesappear to offer the bestfuel/air distributionwithin the cavity as well asawide rageof fuel flow ratesoverwhich combustion may besustainedwhencompared to theotherfueling locationsstudied. The fuel flow rate canbe optimizedanddeviationsfrom this optimalpoint lead to a flame with increasedoscillations and large spatial gradients.4 Fueling the entire cavity from a single streamwise location can becomplicateddueto theaerodynamicsof the cavity vortices. Fuel mustbetransportedfrom theinjectionsiteforwardto the cavity stepby meansof thesestructures.
C. Current StudyThe subjectof this investigationwasto expandthe cavity basedfueling system suchthat both fuel andair are
directly injected. It wasproposed that this method would provide a uniform fuel air distribution within the cavityover a wide rangeof fuel flow rates andfreestreamconditions therebyresultingin anefficient, robustflameholder.Additionally, this studyincluded characterizationof theoperational limits (i.e., sustainedcombustion limits) overavarietyof fuel andair flow rates. Both advancednon-intrusive diagnostics (i.e.PLIF) andtraditional methodswereused to characterizethecombustionandflowfield conditions.
Figure2 – Cavity Flow Conditions(Ref 4)
Figure3 – Cavity Geometry
AmericanInstitute of AeronauticsandAstronautics5
II. Experimental SetupA. Test Facility
The AFRL/PRAS Large-Scale Supersonic Combustion Research Facili ty is an in-house facility capable ofallowing studiesof theenhancement andcontrol of fuel-air mixing in supersoniccombustorswith conventionalandstate-of-the-art non-intrusivediagnostictechniques.Thetunneldesignprovides optical accessfrom up to threesidesof the test section through fusedsilica windows which provide excellent transmissive propertiesin the ultravioletwavelengths. The nozzle sidewalls,as well as the top and bottom walls of the test sectionare equippedwithconventionalstatic pressure andthermocouple taps. Furtherdetails of the testfacility aredescribed elsewhere.10
A two-dimensional converging-diverging Mach2 nozzlesection, configuredwith an asymmetricnozzle,is usedto developthedesired inlet conditions. The facili ty nozzle is configured with nozzleblocks to createa 2-inch highby 6-inch wide exit to createtheMach2 flow through the testsection. The testsectionis equipped with insertstocreate a constant-areaisolator section 7 inchesin length. The constant areaisolatorallows the tunnelto function inramjet, scramjetand dualmodes. In theramjetconfiguration, thebackpressureis raisedto movetheshock structurecompletely into the isolatorsection creating purely subsonic flow in the testsection. Lowering the backpressuremoves theshockstructure into thetestsection. Lowering thebackpressurefurther createspurely supersonic flow inthe testsection. The isolatorsection is followed by an insert creating anexpansion sectiondiverging at 2.5 degrees29.125inches in length.
B. Test ProcedureThecavity, shown in Figure 4, is recessed
from the surface with a 90-degree rearward-facing step, and the trailing edge isconfigured with a 22.5-degree ramp. Thecurrentflameholderconfigurationhasa depthof 0.65 inchesand a length of 2.60 inches.Fuel and air injection is accomplishedthrough three sets of injection sites locatedalong the aft ramp. All injectors are directedparallel to the cavity floor. Each spanwiserow of injectorsis fed from a single manifoldand can be configured to inject either air orfuel. This fueling schemeallows the fueloxidizer to be obtained from main twosources: direct injection and free streamentrainment. Theupper (A2) and lower rows(A1) of injectors were configured to inject air and consist of 11 orificeseach with a diameterof 0.078 in. Themiddle row (F1) wasconfiguredto injectethyleneand consistsof 10 orificeseachwith a diameter of 0.063inches.Injectorcenterlineswere located0.35, 0.55and0.75inchesvertically abovethecavity floor.
The fuel and air injection systemwas automated and interfacedwith a computer basedcontroller and datacollection system. The injection pressurewas regulated with a dome loaderand controlledremotely with an air-actuated isolation valve. A pressuretransducer and thermocouplewere used to measure the pressure andtemperatureof the injectant. Additional pressureand temperature data was gatheredusing a bank of PressureSystems,Inc. strain gagetransducers and Type-K thermocouplesdistributed aboutthe test facilit y. All datawasrecordedin a computer for future analysis. The massflow rate of gaswasmeasuredusing a bankof Tylan massflow controllers. Thesemassflow controllers are manufacturedto output air given their full scale rating which ismeasuredin Standard Liters PerMinute (SLPM). Becauseone of thesecontrollers wasconfiguredto measuretheflow rateof ethyleneas opposed to air, a correction factor of 0.6 wasapplied in accordance with the Tylan massflow controller usersmanual. The ethylene fuel wasintroducedinto the cavity using a 200 SLPM full scale massflow controller andtheair wasmetered by a 500SLPM massflow controller.
As with similar studies performed at this facilit y, the flow through the testsectionwasstabilizedat eithera lowor high backpressure condition. Both conditions wereestablishedby manipulatinga valve downstreamof the testsection. Restricting flow increasedthebackpressuresimulating the ignition transient at low Machnumbers. On theother hand,opening thevalvedecreasedthe backpressureand simulatedhigher fli ght Machnumbersandsupersonicflow through the combustor. Only low backpressure cases were presented here becausethe associated flowdynamicspresentthegreatestchallengesfor mixing and combustion.
Figure4 – CavityHardware
AmericanInstitute of AeronauticsandAstronautics6
C. Non-Intrusive Flow DiagnosticsOH-PLIF is usedto track the presenceof the hydroxyl radical producedduring the combustion eventwithin the
cavity. For laser diagnostics using the OH-PLIF technique,a Lumonics Hyperdye dye laser is pumped with thesecond harmonic of an injection-seededSpectra Physicsneodymium doped yttrium-aluminum-garnet (Nd:YAG)laser (GCR-170). Thedyelaser output is frequency-doubled usinganInradAutotraker III. For hydroxyl excitation,the dye laserwas tunedto 587 nm so that the frequency-doubledradiation matchedthe wavelength for the Q1(8)transitionof theA2Σ+-Χ2Π (1,0)band.
The lasersheetis formedusing a pair of lenses,a plano-concavecylindrical lens(~150 mm focal length)andaplano-convexspherical lens(1000mm focal length). This arrangementresults in a sheetheightof approximately2inches. The transmittingand receiving optical hardwareare positioned on a transversing table allowing remotepositioningof themeasurementvolumeat anydesiredstationin theflow field.
A Princeton Instruments PIMAX Charge-Coupled Device(CCD) digital camerawith a 512 by 512 pixel arraywasusedto detect the fluorescence. For OH LIF detection, fluorescence from the A-Χ(0,0) and (1,1) bandswasisolatedusingUG-11 and WG-295filters. The camerarecordsnon-time correlatedimages during thetestcondition.The camerais programmed to capturean imagewith eachlaserpulse. However, the frequency of the laserpulse is100 Hz, faster than the refresh rate of the camera. The camera collects an image at the next laser pulse afterrefreshing, leading to the non-time correlation of the images. A benefi t of this is the imagesavoid creatingtheimpressionof, or faili ng to detect, harmonic behaviors in theflow.
The profile or cross-flow visualization places the lasersheet on the centerline of the test section. End viewimagesare collectedat stations1, 2 and 3 located at 0.125, 1.5 and 2.5 inches aft of the forward cavity steprespectively. Becauseof limited visualaccessthrough theendof the test section, thecamerais placeat anangle tothe sidewindowof thetestsection. Becausetheimageswerenot corrected,distortionwasevident.
A second optical high speed camera wasplacedperpendicular to the flow. This camerarecordedlight emittedwithin thevisible spectrum at capture rates at approximately 3000framespersecond.The intensity of eachpixel isthe productof line integrationacrossthespanof thecavity. Areasof increasedintensity wereassumedto correlateto areasof increasedcombustion activity .
III. Results and DiscussionA. PLIF Data Analysis
PLIF imaging is accomplishedby excitingatomsandmoleculesusing a two-dimensionalareaof laser light. Thelaser light energyis absorbed by the atoms andmolecules which, in turn, canpotentiall y decayback to the groundstate. This releaseof energy is imagedat a right angleto the path of excitation onto a two-dimensionaldigitalcamera. As expected,the intensity of the image dependsupon the chemical composition and local physicalpropertiesof theflow. Thisstudywill assume that increasedimageintensityis a function of increasedconcentrationof OH. In other words, higher signal implies higher concentration.8 Non-intrusive techniques namely PLIF andhigh speeddigital emissionsvideowasutilized to provideflow characterization data. Theresult of PLIF diagnosticswasa seriesof approximately 100 iagesand theproduct of the emissionsdiagnostics was recordedin *.avi format.Theseimages were reduced primarily through the use of imaging software (Image J version1.23 and PDViewversion5.0) to determine the mean and standarddeviation of a seriesof chronologically capturedimages. Meanimageswere use to characterize the intensity and concentration of hydroxyl radicals given high speedemissionscameraandPLIF diagnostics respectively. Standarddeviationresults were consideredto bea qualitative measureofunsteadinessandtherefore flameinstabili ty.
Thecavity wasconfigured for air injection through the lower injection rows(A1) and for fuel injection throughthe centerrow (F1). Thetest conditionswere nominally at Mach2 with a stagnationpressureand temperatureof 80psiaand 580°F with low backpressure (i.e. purely supersonic flow through the testsection). Fuel was injectedat35%, 50%and75% of full flow of the fuel masscontroller (120SLPM). This resultedin fuel flow ratesof 38.4,60and90 SLPM respectively. Baseline caseswererun for all fuel flow cases(35%,50%and75%)andPLIF imageswere takenat all stations. The mean baselineresults are shown in Figure 5. Stations 1 through3 are labeledrespectively andunlessotherwisenoted,all images arepresented on the sameintensityscale (1800-6000) to allowunbiasedcomparison. Thescalepresentedabovedefinespixelswith a value of 1800to berepresentedby blackandpixels with a valueof 6000to be represented aswhite. Pixels between1800and 6000will be shown in shadesofgrey. The spanwisecenterline of the cavity can be imaginedasa vertical line locatednearthe right-hand side ofeach image. Given that the images abovewere acquired at three different streamwiselocationsthroughoutthecavity, these imagesprovide information as to where combustion was occurringwithin the cavity. An efficientcavity should exhibit evidenceof combustion reactions, hydroxyl radicals (OH) in this case, throughoutits volume.
AmericanInstitute of AeronauticsandAstronautics7
The presenceof OH is indicated by increasedpixel intensity (white regions) within thephotograph. This study will assumethat thepresence of OH is proportional to thecombustion reaction rate that is taking placeat thegiven section. However,it is importantto note that the presence of OH at themeasuredlocation could be the result of theproduction at another location andsubsequent diffusion and/or transport to themeasured location. This is due to therelatively long life of the hydroxyl radical.Notice that the intensity is highestat stations1 and 2 given the 32% fuel flow whencompared to their respective stations athigherfuel flow rates. This is most notablefor station 1 becausebasedon this scalevery little intensityis notedatstation one for both increases in fuel flow above 32%. Furthermore, the overall intensity at stations1 and 2decreaseswith increasesin fuel flow rate. This indicates that as fuel flow increasesabove32%, combustion wasnegatively affectedat streamwisestationsforwardof theaft ramp (stations1 and 2).
The baseline caseexhibits the sametrend observed in previous research. Specifically, a cavity that is directlyfueled is optimally tuned for a single fuel flow rate.Increasesor decreasesfrom this “optimal” level lead tolocalized regionsof combustion which canbe interpretedasinefficient useof the cavity volume. From this standpoint,whenfuel wasinjectedat 38.4 SLPM (32%),thecavity wasoptimally tuned given the fuel only injection schemesstudied and shown in Figure 5 because evidence ofcombustion was noted at all stations. Furthermore, thereadershould note that themostsignificant changein imageintensity as a function of fuel flow was noted at station 1near theforwardcavitystep.
For the next study, air wasdirectly injected through thebottom injection ports (A1) into the cavity to study itseffectson combustion. This wasaccomplishedusinga massflow controller with a full scale capability of 500 SLPM.The same fuel flow rates and injection locations wereutilized for ease of comparison. Figure 6 showsthe effectsof air injection given a constantfuel flow rate of 32% (58.4SLPM). Air is injected at 50% (250 SLPM) in addition to the baseline (fuel only) case. Figure 6 shows animprovementin cavity combustionmostnotably at station1 whereasvery li ttle changeis notedat stations2 and3.This effect demonstrates that the directinjection of air through the bottom row ofinjectors can provide anothermechanism tooptimize the combustion process with thecavity. Howeverasshownabove, at this testpoint, increases in air injection do notnecessarily result in improved combustionthroughtheentire cavity becausecombustionat stations2 and 3 remain largely unchanged.The mostnotable increasein combustion wasat station1 nearthecavity step.
Similarly, fuel wasintroducedat 50% (60SLPM) and air was injected at 50% (250SLPM) and 75% (375 SLPM) in addition tothe baseline (fuel only) case through A1.Figure 7 showsthe effect of increasedair flow
Figure6 – 32% FuelFlow
Figure7 – 50% FuelFlow
Figure 5 – Baseline PLIF images
AmericanInstitute of AeronauticsandAstronautics8
given a constant fuel flow rate of 50% (72 SLPM). The increase in airfl ow from the baseline caseto 50% (250SLPM) air injection flow rate causedan increasein combustion at station 1. However the continuedincreaseinairflow from 50% to 75% (375 SLPM) resulted in a decreaseincombustion at station 1. Image intensity remainedsteady forstation2 and 3 given all air loadings applied at this test point.Although there was insufficient resolution given the data todetermine the airflow ratethat provided the optimumutilizationof cavity volume for this fuel loading, when the air flow wasat50%, station 1 exhibited thehighestconcentration of OH amongconditionstested. In the same way that stations2 and 3 wereminimally affected by the introduction of air at 32% fuelloading,combustion at stations2 and3 at 50%fuel loadingseemto beindependentor weak functionsof introducedair flow.
Fuel wasintroducedat 75% (90 SLPM) andair was injectedat 85% (425 SLPM) through A1 in addition to the baseline (fuelonly) case.Figure8 showstheeffect of increasedair flow givena constantfuel flow rateof 75% (90 SLPM). The combinationof this fuel loading and theintroductionof air demonstratedsimilartrends compared to the 32% and 50% fuel flows. The greatestincreasein intensity was evidenced at station 1 althoughstations 2 and3 incurreda slight intensity increase giventhe increasedair flow.
This fueling scheme,fuel injection at F1 andair injection at A1, producedanincreasein combustionat station1in each of the threefuel flow rates. Figures5through8 showthat given direct air injectioncavity combustion can be optimized forvarious fuel flow rates. However, as notedbefore, combustion is not necessarilyimproved uniformly throughout the cavity.The inconsistency in cavity combustionthroughout the volume is a product of thecomplexities of mixing, variations in localtemperature and pressure and three-dimensionalcavity flowfields among a hostof other parameters. Figures1b and1c showthe streamtracesof cavitieswith comparablegeometry to the experimental hardware.Notice that two counter-rotating lobedstructures are commonly found in such aconfiguration. This structure complicatesthe fuel andair transportmechanism especially near the cavity step. Asnoted before, mass(air, fuel and productsof combustion) is transported at differentrates between the freestream/cavity shearlayer/aft vortex and the forward vortex/aftvortex. Previous aft ramp, direct fuel-onlyinjection studies have concluded that forhigher fuel flow rates a fuel rich region isformednearthe cavity step.4 This region,as implied, is not populated by acombustible mixture and thereforecontributesto the overall inefficiency ofthe cavityvolume.
The addition of air injection throughA1 servedto aid combustionat station 1whencompared to the baseline(fuel only)case. This observation was notedpreviously and evidencewas presented in
Figure9 – RichCavity Combustion
Figure10 – 32%FuelFlow
Figure8 – 75% FuelFlow
AmericanInstitute of AeronauticsandAstronautics9
figures5 through 8. As noted in Figure 9, theregion sampledat 0.125 inchesfromthecavity step(station 1) hasthepotential to become “r ich” in theabsenceof sufficient air injection given thefuel loading. This tendency at station1to becomerich is offset by the direct air injection. Similar to the positive combination of fuel injectantandcavityvortex,air injectednear the bottom cavity floor is complimentary to the local flowfi eld and improvesair transporttoward thecavity step. However, this injection schememerely providesanothermechanismto optimizecombustionover a rangeof operating conditions. Increasing air injection without bounddoesnot alwaysequateto improvedcombustion. Referencethe imagesshownin the first row of Figure 7 corresponding to station1. Increasing the airflow from 50% to 75% at a fuel flow rate of 50% shows a decrease in combustion as inferred from OHconcentration.
Air injection at theA1 sitesignificantly alteredcombustion neartherear-facing step. A follow-on investigationwas initiated to further characterize theeffects of air injection (A1) on combustionnear thecavity step. Thechosenlaserplanewas locatedat 0.25 inches aft of the step,normal to the freestreamdirection and willbe referred to as station 1a. Several fuelflow rateswere studied,however only thelow fuel rate results are presented in thispaper. Optimum combustionat an arbitrarylocation is defined by steady, uniformcombustionthroughout thearea. Therefore,mean images with near constant highintensity and standard deviation imageswith constant low intensity should berepresentative of optimum combustion.Figures 10 and 11 present the mean andstandard deviation respectively of imagestaken at a fuel flow rateof 32% and variousair injection massflow rates. The scale for eachfigureis included and takesthe form of (black-white). Figure 10shows a gradual increase in the combustion present at station 1a given increasesin air flow. Thereis very lit tledifferencebetween the mean imagesacquired at an air flow of 40% through 90%. Figure 11 is the standarddeviation of all imagescollected at this testpoint. The imagesfrom air injection between30% and90% areverysimilar for both themeanand standard deviations. Stable combustion appears to betaking placeat station1afor airinjection above15% as indicatedby the meanand standard deviation images. For 32% fuel flow, 50% air flowseemsto optimally tune the cavity evidencedby a relatively uniform bright meanand a relatively uniform dimstandarddeviation.
B. Luminous Flame EmissionsA high speedcamera was positioned normal to the flow such that the entire cavity profile was visible. This
methodprovidedanoverall view of combustion asevidencedby thepresenceof luminouspartsof the flamewithinthe cavityandwasused to furtherextendthecombustion informationextracted from thePLIF diagnostics. No visible lightemissions are shown in black whileincreasesin flameemissionsarereflectedbyincreasesin intensity (white). Data wastaken at three fuel flows, various air flowsandboth high and low backpressure. A redreference line was added at the samelocation for each image. This line wasintended to define the cavity boundaries,however it must not be taken as an exactrepresentation of the cavity. Specificimageswithin a table will be identified bythe following: (row, column).
Figure10 – 32%FuelFlow
Figure11 – 32%Fuel Flow
Figure 12 – 32%FuelFlow
AmericanInstitute of AeronauticsandAstronautics10
The series of images shownin figure 12 werederived from the mean of all imagesacquired at 32% fuel flow,low backpressure andincreasingair flow throughA1. Increasedair flow decreasesmean combustion throughoutthecavity. Thepresenceof a strongshear layerflamethat extendsalmosttheentire lengthof thecavity is shown at 0%air injection. As air flow increasestheshearlayer flamedraws into theaft rampcombustionregionandis no longerclearly evidentat 70% air flow. Furthermore,asair flow increases the combustion regiondecreasesin streamwiselength toward the aft ramp. The presence of a strongshear layer flame on the fuel only case (image (1,1)) is anindicatorof nearoptimumcavitycombustion.Sincethisoccurswith no air injection,theadditionof moreair shouldtend to leanout the global cavity mixturefurther reducingoverall cavitycombustion.
The figure 13 was derived from the meanof all images acquiredat 50% fuel flow, low back pressure andincreasingair flow through A1. Increasedair flow increasescombustion throughoutthe cavity. Note that at 0% air flowcombustion is localized near the aft ramp,but the existence of a shear layer flame isevident. As air flow is increasedthe shearlayer flame extends farther from the aftramp toward the forward step.Furthermore,at higher air loadings a non-reactive region is formed at the middle(streamwise)of thecavity.
Figure 14 shows the standard deviationimages for 50% fuel flow, various airinjection rates through A1 and lowbackpressure. It is obvious that thereference line does not exactly coincidewith the cavity boundaries. However, itoccupies a fi xed location and served as avalid referenceframe. All images,with the exception of (1,1) and (3,3), display a common attribute. They eachexhibit a very consistentcombustionregionin theshear layer. This region is locatedby its low intensity. A strongshear layer flame is considered to bea good indicator of an effectiveflameholdingmechanism. Such a mechanismserves to sustain combustion within thecavity through the production of hotbyproducts of combustion. These hotproducts are re-circulated by the cavityvortex structure and provide thermalenergy to promote combustion.Additionally, a shear layer flame is wellsuited to transfer energy in the form ofheat to the freestream flow furtheringcombustionreactionsoutsideof thecavity.Air injectionthrough A1 continuesto havea beneficial effect on cavity combustion.Given 50% fuel flow, combustionfill edthe entire cavity volume at nearly everyair flow rate resulting in effective useofcavity geometry.
Figure14 – 50% FuelFlow
Figure 13 – 50%FuelFlow
AmericanInstitute of AeronauticsandAstronautics11
Figure15 wasderivedfrom themeanof 200imagesacquired at 75%fuel flow, low backpressureandincreasingair flow through A1. Combustionat thefuel only case is localized near the aftcavity ramp. The sequential addition ofair produced the following structures:formationof shearlayer flame, extensionof the shearlayer flamefromtheaft rampto the cavity step,and the addition of acombustion zone near the cavity step.The formation of thesestructures basedon controllable parameters (i.e. fuel andair flow rates) allows the cavity to betuned to best serve as a flameholderthroughoutvarious operating conditions.
IV. ConclusionsAir injection from thebottominjection site (A1) servedto tune thecavity for optimumcombustion for eachfuel
flow rate. That is, for a given fuel flow rate,air injection flow ratescan be increasedor decreased to produceastable,uniformcombustionregion throughout thecavity. Cavity aerodynamicshave shown thatmorefreestreamairis entrained by thecavity givenhigh backpressure.Previousfuel only studies havebeen limited to lower fuel flowrates especially at low backpressure, dueto this limited air entrainment. Therefore,this fueling scheme,whereairandfuel aredirectly injectedinto thecavity, signif icantly increasestheoperating limits of thecavity flameholder.
The additionof air injection serves to lean out fuel rich lobesshownto exist near the cavity stepallowing forcombustionthroughoutthecavity thereby increasingits efficiency. Injection at A1 producedthe greatestregionofimpactnearthecavity step. Without air injection, thecavity step region contributes very lit tle to theoverall cavitycombustion. Air injection through the top spanwiserow of injectors (A2) minimally affected global cavitycombustion. However, a localizedregionof influencewasvisually noted.
Efficient combustion canbe characterizedby a strong, steadyshearlayerflameandglobalreaction, Increasesinfuel flow, for theappropriateair flow, producedsignificantheat asevidencedby the increase in temperaturewithinthe cavity. The cavity steptended to retain the heatof combustion more so than the aft ramp, due to the coolingeffectsof the air andfuel flow throughtheramp.
Acknowledgments
The authorswould like to acknowledgethe contributions of W. Terry, D. SchommerandH. Meicenheimerfortheir technical support on this effort. The supportof the Air ForceResearch LaboratoryandtheAir Force Instituteof Technology arealsoappreciated.
References
1 Mercier,R., McClintonC., “HypersonicPropulsion – Transformingthe Futureof Flight,” AIAA paper 2003-2732,AIAA /ICAS International Air andSpace Symposiumand Exposition, Dayton, OH 14-17 July2003
2 Boudreau,A., “Statusof US Air ForceHyTech Program,” AIAA paper2003-6947,12th AIAA InternationalSpace PlanesandHypersonic Systemsand Technologies,Norfolk, VA 15-19 December2003
3 Youngbin,Yoon. “Blowout Stability Limits of a Hydrogen jet FlameIn a Supersonic,Heated,Coflowing AirStream.” CombustionScienceandTecnology Vol 97 pp 137-156.
4 Gruber,M.R., Donbar, J.M., Carter, C.D. and Hsu, K-Y., “Mixing and Combustion StudiesUsing Cavity-Based Flameholdersin SupersonicFlow,” ISABE-2003-1204
5 Mathur,T., Bil lig, F., “SupersonicCombustion Experiments with a Cavity-BasedFuel Injector,” Journal ofPropulsion and Power, Vol. 17,No. 6, 2001,pp.1305-1312.
AmericanInstitute of AeronauticsandAstronautics12
6 Gruber,M., Baurle,R.,Mathur,T., and Hsu, K., “FundamentalStudiesof Cavity-BasedFlameholderConceptsfor Supersonic Combustors” , Journal of Propulsion and Power, Vol. 17,No. 1, 2001,pp.146-153.
7 Ben-Yakar, A., Hanson,R. K., “Cavity Flameholders For Ignition and Flame Stabilization in Scramjets:ReviewandExperimental Study,” AIA A paper98-3122,July 1998.
8 Lee, M.P., McMi lli n, B.K., “Planar FluorescenceImagingof a TransverseJet in a SupersonicCrossflow,”Journal of Propulsion and Power, Vol. 8, No. 4, 1992, pp. 729-735.8 Hsu, K.-Y., Goss, L.P., and Roquemore,W.M., “Study on Trapped-Vortex Combustor-Effect of Injection on Flow Dynamics,” Journal of Propulsion andPower, Vol. 14, No. 1, 1998,pp.57-65.
9 Gruber, M.R., Nejad, A.S., “Development of a Large-Scale Supersonic CombustionResearch Facility,”AIAA paper 94-0544,32nd AerospaceSciencesMeetingandExhibit, Reno, NV, 1994
