Development of high speed spectroscopic imaging techniques for the time resolved study of spark ignition phenomena

Transcription

1 Development of high spee spetrosopi imging tehniques for the time resolve stuy of sprk ignition phenomen C. F. Kminski, J. Hult, M. Rihter, J. Nygren, A. Frnke, M. Alén Lun Institute of Tehnology, Sween S. Linenmier, A. Dreizler, U. Ms University of Stuttgrt, Germny R.B. Willims University of Southmpton, Engln Copyright fl 2000 Soiety of Automotive Engineers, In. ABSTRACT This pper reports on the evelopment of novel time resolve spetrosopi imging tehniques for the stuy of sprk ignition phenomen in omustion ells n n SIengine. The tehniques re se on plnr lser inue fluoresene imging (PLIF) of OH rils, on fuel trer PLIF, n on hemiluminesene. The tehniques oul e hieve t repetition rtes rehing severl hunres of kilo-hz n were yle resolve. These tehniques offer new pth long whih engine relte ignostis n e unertken, proviing welth of informtion on turulent sprk ignition. INTRODUCTION The present pper reports on the pplition of high spee time resolve spetrosopi imging tehniques using novel multiple pulse lser fility. The fility llows the pplition of ll stnr lser spetrosopi mesurement tehniques, ut t repetition rtes rehing severl hunres of khz. With this equipment it is possile to trk ynmi turulent omustion events in rel time with ovious implitions for yle resolve SIomustion reserh. The system onsists of 4 oule pulse N:YAG lsers whih n e use either iretly or s pumping soure for onventionl ye lser systems. A speilly esigne frming mer is ouple to the lser, ple of tking up to 1 million pitures per seon. In the pper we present vriety of mesurements, onute in purpose uilt turulent ignition ells n n SI engine. For the first time, 2D lser spetrosopi imging tehniques oul e pplie in suh systems in time n yle resolve fshion. The smple results presente here re prt of n ongoing effort to provie tse for etile moels of turulent sprk ignition phenomen. The pper is orgnise s follows. In the first prt n overview is given on the experimentl systems n the spetrosopi imging tehniques evelope. Three ojets of stuy re fouse on: A onstnt volume sprk ignition ell with stnr fuel injetor; turulent ignition ell equippe with high spee fns to imprt homogeneous isotropi turulene on the fuel mixture; n finlly 4-stroke lortory SI-engine. In the seon prt we present results otine from these systems. The tehniques re ple of overing nerly the entire phse of omustion in n SI yle: Here we show results from time resolve mesurements of rekown n susequent r phses uring the sprk event. The ensuing omustion phse oul then e exmine using either of fuel trer PLIF, or hemiluminesene spetrosopy. We furthermore present novel tehnique se on the simultneous pplition of OH-PLIF n emission spetrosopy, offering itionl topologil t tht neither tehnique provies on its own. The tehniques presente llow the stuy of omplex reting flows n turulene-hemistry intertion in iret wy. Phenomen suh s flme kernel evelopment in turulent flow n flme extintion n e visulize n trke in time. We emonstrte the potentil of the tehniques for engine reserh, where they my e use to perform yle resolve mesurements of flme front evelopment n ir fuel rtio. 1

2 CCD 1 Bem splitter Optionl imge intensifier MCP 1 Mirror Bem splitter optis Mss storge Frme store CCD 8 CCD 2-6 Lens mount MCP 8 Mirror Iris Figure 1: The multiple YAG lser luster. Roun elements signify ihroi emsplitters whih re trnsmissive for 1064 nm n refletive for 532 nm. Lrge ues: Frequeny ouling rystls. Smll ues: em umps. The single output em oul optionlly e oule to 266 nm y use of n itionl, single ouling rystl. EXPERIMENTAL LASER / DETECTOR SYSTEM The high spee spetrosopi imging system onsists of luster of 4 N:YAG lsers (BMI) whih n e fire sequentilly t ritrry repetition rtes (with time intervls rnging from ns to out 100 ms etween pulses from ifferent vities) [1]. The output of the lsers is omine using ptente em omintion sheme (BMI) to provie single output em. The omintion sheme employs series of ihroi mirrors n suessive frequeny ouling of the YAG funmentl (t 1064 nm) to hieve em omintion with miniml energy losses (see figure 1). Eh osilltor is equippe with oule pulse option (DPO moe) llowing the Pokels ell of eh vity to e swithe twie uner the urtion of the flshlmp pulses. Thus totl of 8 lser pulses n e extrte. Time seprtions etween oule pulses from eh unit n e hosen to vry etween 25 n 145 μs, etermine y the length of the flshlmp pulses n the gin uil-up times respetively. Pulse energies rehing 600 mj n e extrte from eh vity in single pulse opertion, roun 270 mj re ville in DPO moe t the seon hrmoni (532 nm). As n option single KDP rystl n e inserte into the output empth proviing rition t 266 nm, ielly suitle for etone exittion (see forthoming setions). The multiple YAG lser luster n either e use iretly (etone imging), or, to pump n optionl yelser (OH fluoresene exittion). Couple n synhronise to the lser system is ustom esigne version of ommerilly ville frming mer (Imon 468, Hln Photonis, UK). The system fetures 8 inepenent intensifie CCD etetors (ICCD) with 384 times 576 pixel rrys n 8 it ynmi resolution. Iniviul events were imge onto the iniviul rrys y use of Cssegrnin prism em-splitter uilt into the mer system (see fig. 2). To inrese the overll gin of the system n to mke it UV sensitive n itionl three stge intensifier moule ws tthe to the optil input of the mer prior to the prism em- Figure 2: Operting priniple of the fst frming mer. MCP: Multi hnnel plte imge intensifier. Light enters from the right, psses internl optis n n iris efore impinging on prism imge-splitter. The ltter irets iniviul imges onto the MCP s whih re sequentilly gte. splitter. This intensifier oul e operte t repetition rtes of up to 1 MHz. The eletronis of the mer were moifie to llow flexile triggering of the intensifiers to ritrry externl events n to synhronise it to the lser system. EMISSION STUDIES Diret emission ws reore iretly y the mer, ppropritely filtere for the require mesurement tsk. The mer ws triggere iretly y the ignition eletronis for the two ells use n from rnkshft enoer (CSE) in se of the engine mesurements (see oming setions for more etils). Delys etween susequent emission reorings were then triggere internlly y the mer hrwre. All imges reore here were performe using the optionl imge intensifier t the mer input (see fig. 2) owing to the wekness of emission signls. Exposure times were mthe to the iniviul situtions ut typilly in the rnge of 30 to 100μs. PLANAR LASER INDUCED FLUORESCENCE STUDIES (PLIF) For OH we exite the temperture insensitive Q 1 (8) trnsition of OH in the A 2 ± + (v 0 = 1) ψ X 2 Π(v 00 = 0) eletroni n whih ppers in the 282 nm wvelength region. The exiting lser pulses were otine using the frequeny oule output from ye lser (Continuum) operting ner 564 nm on Rhomine 590 ye pumpe y the seon hrmoni (532 nm) of the multiple YAG lser. The lser light ws forme into sheet pproximtely 50 mm in height n 160 ± 30 μm in with y ylinril telesope. A shemti representtion of the set-up use for PLIF experiments in the omustion ells is shown in figure 3 n in fig. 6. Susequent fluoresene in the v 0 = 0! v 00 = 0 n ner 309 nm ws iretly ollete y the ppropritely filtere frming mer. Two ifferent moes of opertion were employe. In the first moe, 8 exittion pulses were use, n PLIF of OH ws reore y ll hnnels of the mer. In 2

3 Multiple YAG Dye lser 2w.l..l. PD qurtz plte frming mer ye ell.l..l. filter qurtz plte UG11 eletroes Figure 3: Shemti set-up for the PLIF experiments. 2! : frequeny ouling unit, PD: Photoioe,.l.: ylinril lens, UG11: filter to rejet lser stry light. the seon moe, we use only four lser pulses n mer hnnels to reor OH-PLIF. In this se, the remining 4 mer hnnels were free to reor line of sight OH hemiluminesene (iret emission of OH). We gte the mer in suh wy tht 2 s fter eh PLIF imge orresponing OH-hemiluminesene imge oul e reore. As will e seen, this pproh offers itionl struturl informtion on the flme kernel (see isussion further elow). The 2 s time ely is neessry to voi rosstlk etween the hnnels reoring PLIF n hemiluminesene n n e onsiere instntneous on the turulene n overll hemistry time sles previling in the stuie systems. Exposure times were 200 ns for PLIF n 30 s for the muh weker emission imges. Figure 4: Photogrph of the onstnt volume fuel injetion ell use in the present experiments. The sprk plug is seen to the left, n ion sensor is seen on the top of the ell. Heting wire ws use to prehet the ell. 340 nm. Susequent fluoresene ours etween 330 n 660 nm with ro pek roun 420 nm [2]. The signls from etone n OH oul thus esily e seprte. 3-pentnone PLIF The PLIF fuel visuliztion experiments in the engine were performe with isootne s one-omponent fuel. Sine pure isootne oes not show ny fluoresene fluoresent trer must e e. When hoosing trer it is of mjor importne tht the trer follows the fuel uring the ompression stroke. Hene, the trer must hve physil properties similr to the fuel. For tring isootne, 3-pentnone hs eome wiely epte opnt. In the presente mesurements 5% (v/v) of 3-pentnone hs een use s trer. The low trer ontent is not expete to ffet the physil properties of the isootne. For 3-pentnone exittion, 266nm light ws use, orresponing to the fourth hrmoni of the YAG-lsers. This wvelength is lose to the top of the sorption urve n, therefore, the temperture sensitivity is minimize. The spetrosopi properties of the fluoresene from 3-pentnone re lmost ientil to tht from etone (see elow), i.e. ron emission rnging from nm with pek t 420 nm. COMBUSTION SYSTEMS STUDIED 3 ifferent systems were stuie s prt of the present work: A onstnt volume omustion om employing stnr port injetor n sprk ignition system, fn stirre omustion hmer for premixe homogeneous hrges, n 4-stroke sie vlve engine. Fuel injetion om. For the turulent etone PLIF series onstnt-volume omustion hmer ws use whih ws esigne to ensure tht s mny prmeters s possile were ontrollle. These inlue, in ition to stnr engine prmeters, the mixture homogeneity n the gs flow insie the ell. The hmer h pproximtely ui imensions n volume of 167 m3. It ws forme y three orthogonlly interseting ylinril pssges in ui lok of stinless steel (see fig. 4). Full optil ess to the omustion hmer ws provie y 3 fuse sili winows. A moifie sprk plug with elongte eletroes ws use in the experiments to ensure the lser oul ome s lose s possile to the point of ignition. The ell feture piezo-eletri pressure trnsuer n n ion urrent sensor, whih oul e use to monitor the omustion yle [3]. Aetone PLIF For etone exittion in the fuel injetion ell 266 nm light ws use, orresponing to the fourth hrmoni of the multiple YAG lser. For the omine OH PLIF n etone PLIF stuies the sme frequeny oule ye lser operting ner 282 nm oul e use, sine the OH sorption overlps with etone sorption ns. Both exittion wvelengths fll within the S1 ; S0 ; n n whih extens from out 220 to A set-up igrm of the system is shown in fig. 5. Diret-ting mgneti vlves ontrolle inlet n outlet of gses. The vlves, the ignition system n the PLIF system, s well s the logging of pressure, ion urrent 3

4 LOCAL φ AIR PRE HEATER FUEL SEEDER Cmer PRESSURE CURRENT GLOBAL φ SPARK PLUG EXPANSION VESSEL Lser Sheet Figure 5: Shemti igrm for the port injetion omustion om. For etils refer to the min text. n glol equivlene rtio, were ontrolle y PC y mens of 12-it A/D r. Air n fuel were injete seprtely into the hmer. The ir ws prehete in n itionl vessel to pproximtely C prior to injetion. The fuel (methne or propne in the present se) psse n etone seeer prior to injetion. For etils refer to fig. 5. A yle is efine y injetion of prehete ir, followe y the injetion of the gseous fuel few milliseons lter. After user efine ely the mixture is ignite. The length of this ely governs the egree of mixing s well s the mount of turulene in the ell, whih is use y the injetion proesses n eys in time. A few seons fter omustion, the outlet vlve ws opene, n the urne gses llowe to expn into n itionl vessel, where n oxygen sensor ws use to etermine the overll equivlene rtio. The yle ene with the evution of oth vessels to pressure of 20 kp to ensure very low EGR. The overll urtion of suh yle mounte to 90 s. Pressure tres were reore y flush-mounte, wteroole, piezoeletri trnsuer onnete to hrge mplifier. Fn stirre om. For the stuy of turulent flme growth using OH PLIF fn stirre omustion om ws use. Fig.6 shows shemti rwing of the fn stirre turulent ignition ell, whose volume mesure 55 l. Full etils of the system re foun in [4]. The following offers only rief esription of the system. Eletroes entere in the ell were me of shrpene tungsten tips, seprte y istne of 1 mm. The ell ws fille with homogeneous mixtures of methne n ir prior to sprk ignition. The mixture oul e sujete to ontrolle egrees of turulene y justing the rotor spee etween 0 n 5000 rpm (rottions per minute) effeting Figure 6: Fn stirre omustion om for turulent ignition stuies. turulene intensities in the rnge of 0 to 1.9 ms 1 whih were mesure y LDA (Lser Doppler Anemometry). Sprk urtion n energy ontent were ontrolle y rpi ptive regultion of the voltge etween the eletroes oring to the plsm resistne. SI engine. The test ojet ws single yliner, 5 hp, sie vlve engine (Briggs n Strtton) moifie to provie optil ess. Sine it ws of sie-vlve type, reltively few moifitions were require to this en. The originl rnkse n piston remine unltere while the yliner he ws equippe with three qurtz winows. For PLIF stuies the exittion light psse horizontlly through the smller winows on eh sie of the omustion hmer whilst signl etetion ws performe from ove (see fig. 7). A photogrph of the engine is shown in fig. 8. A rnk ngle enoer ws use to provie trigger signls for the lser n mer systems. In this wy the strt of the lser pulse trin oul esily e synhronize to ny esire rnk ngle. Isootne ws use s fuel for the present experiments. Two ifferent types of visuliztion tehniques were performe: ) iret emission reorings of flme kernel hemiluminesene, n ) fuel trer PLIF. For the ltter 3-pentnone ws hosen s trer for isootne. IMAGE PROCESSING To enhne signl to noise rtios in the PLIF imges n to simplify imge segmenttion for susequent nlysis steps we employe n imge post-proessing sheme se on nisotropi non-liner iffusion filtering [5]. In the present se, we filter the imge rw t using the t u = iv(g(jr(g ff Λ u)j)ru) (1) where u represents the intensity of the imge uner onsiertion, n g(jr(g ff Λ u)j) represents lolly ptive iffusive strength. The ltter is me proportionl to the grient ru in the imge itself (fter smoothing with Gussin kernel G ff of with ff, whih is one for stility resons [6]). The priniple of the metho is to smoothen out noise lolly y iffusive flow whilst pre- 4

5 Figure 7: Shemti set-up for engine PLIF experiments. The engine ws 4 stroke engine of sie-vlve type. The yliner top ws reple y qurtz isk through whih fluoresene n emission signls were reore. Optil ess for the lser ws provie y two smll siewinows situte ner the yliner top. Figure 9: Chrteristis of the sprk ignition system use in the fn stirre om. Tres shown here re three overli grphs orresponing to three seprte ignition events. Fst high voltge regultion ws use to keep the ishrge urrent onstnt uring the ishrge phse (lower tre). The upper tre shows the eletroe potentil uring the ishrge. The finl voltge pek isplye on this tre is irrelevnt for sprk evelopment. The exellent reprouiility of the ishrge is evient from the three urrent tres shown. RESULTS venting flow ross physilly importnt ounries. By proper hoie of the iffusion kernel, ojet ounries my e enhne (ut nee not e, if unesirle) n physil grients shrpene. Thus etermintion of ojet ounries n imge segmenttion is gretly simplifie. Figure 8: Photogrph of the moifie 4-stroke engine use for optil ignosti experiments. SPARK INITIATION AND BREAKDOWN The omustion vessels esrie here re sujet of moeling tivities [4, 7] n the results from mesurements s presente here will e use s tse ginst whih moels n e ompre n vlite. For this purpose the ignition systems were esigne to e s preisely efine n reprouile s possile. Of mjor importne ws to keep the energy eposite y the sprk ignition system s onstnt s possile from yle to yle. In the fn stirre om this ws hieve y ptively regulting the voltge rop ross the sprk gp. In fig. 9 this is shown for three typil ignition events. The top tre shows the voltge rop ross the eletroe gp s funtion of time. Two istint phses n e iserne: The lrge spike t the eginning effets the rekown (whih t roughly 150 ns is fr too short to e seen y the eletroe sensor), n the susequently lower voltge, previling for some 200 or so μs, sustins the ishrge hnnel (r phse). The lower tre isplys the plsm urrent for this ishrge whih, fter the initil rekown spike, remins onstnt over the entire event. By justing the urtion of the ishrge, preisely efine energies oul e elivere to the omustion system. Nerly the entire energy eposition into the sprk ourre uring the onstnt r phse whose exellent reprouiility is evient from fig. 9. To provie more etile informtion on the ishrge proess high spee iret emission pitures were reore. Figure 10 shows onseutive sequenes of the rekown n ishrge phses in the fn stirre re- 5

6 Figure 10: Sequentil imging of sprk ignition event. Numers in the imges orrespon to μs fter rekown. Series - orrespon to 4 ishrge events to illustrte the entire sequene of the proess. tor. Four sequenes of 8 imges eh re shown, lele - on the figure (series fetures only two onseutive imges). Eh sequene orrespons to single ignition event, overing the time intervl inite y the figures in the orners (given in μs reltive to the rekown). The length of the sprk hnnel seen orrespons to 1 mm (istne of the eletroes). The 4 sequenes shown re mthe to over n entire sprk ishrge event. Severl fetures re pprent. During the first 8 imges one n follow the uilup of the ishrge hnnel. During rekown (0μs) the luminous intensity is mximum, immeitely ropping to very low vlue fter rekown. One n lerly follow the evelopment n growth of the ishrge hnnel in time. Series ) emonstrtes the exquisite sensitivity n time resolution tht is ffore y the mer system. Exposure times were only 10 ns for eh event overe. Sine the signl rightness is severl orers of mgnitue lrger uring rekown the mer gin h to e lowere oringly on hnnel 2 to mke reoring of sequene ) in the figure possile. Sine the hnnel gins n e ritrrily juste on the mer, ynmi rnge of severl orers of mgnitue n e overe within single time series, something tht woul e very iffiult to hieve with lterntive reoring evies. Between out 20 n 170 μs the fully estlishe ishrge hnnel remins lmost onstnt in size. Note how the sprk hnnel is moving in time: In series ) the sprk is moving ehin the thoe (t 120 μs) leving thoe show in the imge. Suh movement of the thoe spot n only e resolve using the fst sequentil imging tehniques esrie here. Note lso the evelopment of noe n thoe spots in the series, owing to eletri fiel foussing effets t the shrpene Tungsten tips. Series ) lso revels tht the rekown phse is on the orer of ns, muh shorter thn wht the voltge rop ross the eletroes my le to onlue (see fig. 9). The onstnt overll intensity uring the ishrge provies further eviene tht the energy elivery is onstnt uring this phse. TURBULENT FLAME GROWTH Following the sprk n (initilly lminr) flme kernel is initite in homogeneous flmmle mixture. By sujeting homogeneous methne/ir mixtures in the ell to vrying egrees of turulene sprk ignition n e stuie in systemti mnner over wie rnge of prmeters suh s stoihiometry, sprk energy eposition, et. To provie goo sis for moel vlition it ws hosen to monitor ) flme front topologies n ) onentrtion fiels of omustion generte rils. Suh t is well suite for omprisons with numeril moels [7, 8]. Some exmple results of suh runs re shown in figure 11. The imge re overs pproximtely 5 m in height y 6 m in with. Stoihiometries (ffi) n rotor spees (! in rpm) orresponing to the ifferent series re inite in the first piture of eh series. Time instnes re shown elow eh imge. In ) the evolution of the OH onentrtion fiel is shown for stoihiometri lminr flme (fns turne off). In [4] etile hemil kinetis lultions re presente ompring suh flmes to lminr flme lultions. Series ) shows the sme sitution ut with wek turulene levels present (! = 1000 rpm). One n lerly follow the onset of flme wrinkling. Note lso the inrese of the flme spee evient from the reue time ifferene etween onseutive imges, sine the flme surfe is inrese y the wrinkling. In ) the sme sitution is shown s in ) exept tht the mixture is len (ffi = 0:65). In this se the flme spee is muh lower, n the flme is muh more strongly wrinkle thn the stoihiometri flme in ). Series ) is flme sujete to high egrees of turulene, similr to onitions previling in typil SI engines. Here ffi = 1 n! = 3000 rpm. The first two imges pper lk sine the flme kernel is initilly moving wy from the lser sheet. Only in the thir piture oes the flme reh the lser sheet. The flme is very hevily isture y the previling turulene levels. In e) simultneous emission imges re presente orresponing to the series shown in ). The series ws otine in the fshion esrie in the experimentl setion. The enefit of these simultneous line-of-sight reorings is pprent: espite the non-existene of PLIF t in imges 1 n 2 of series ) the flme kernel hs nevertheless egun to propgte s is lerly evient from the orresponing imges in e). The emission signls stem from the entire light emitting flme volume. Comining this t with 2D-PLIF t 6

7 =1 = ms 3.35 ms 5.35 ms 7.35 ms 3.2 ms 4.9 ms 6.6 ms 4.5 ms 7.5 ms 10.5 ms =1 = ms =0.65 = ms =1 =3000 e 1.5 ms 1.5 ms 2.6 ms 2.6 ms 3.7 ms 3.7 ms Figure 12: PLIF imging of etone () n OH () s flmefront mrkers. The lowest row () orrespons to oth imge series superimpose. Note tht the regions mrke y the two tehniques re signifintly ifferent. 4.8 ms A question of interest in the present se ws how well the etone ontours follow the flmefront s mrke y the PLIF-OH imges (see fig. 11). For this purpose we reore OH n etone imge series t simultneous time positions in the onstnt volume ell. Fig. 12 shows suh sequenes for lminr flme. Series ) orrespons to the etone signl n ) to OH PLIF. In ) oth signls re superimpose. The pitures orrespon to unproesse rw t, the stripiness in the fuel istriution is use y em profile vritions in the frequeny oule ye lser whih were not normlise out (the lser em propgte t slight ngle with respet to the imge plne). The stripes re not present on the muh stronger OH signls owing to prtil sturtion of the signls. Interferene from lser stter ws stronger for the etone imges ue to lower signl levels, n this is why the eletroes re prtilly visile on the etone imges. Note the mrke ifferene etween the flmefront rii mrke y the two tehniques evient from series ). The isrepny is ue to the ifferent tempertures n pressures t whih etone is pyrolyse n OH rils re proue. Furthermore, the iffusion oeffiient for etone is lower ompre to methne [10, 11] lthough this is to e onsiere of minor importne in the present sitution. In quntittive t extrtion for numeril moels neglet of these effets oul le to signifint prolems. 4.8 ms Figure 11: Sequentil time resolve imging of turulent ignition events y PLIF (series -) n hemiluminesene (series e) of OH. Series n e where tken qusi-simultneously showing the very ifferent informtion ville from 2D-lser imging n line of sight integrting tehniques. Detils re provie in the min text. oth 2 n 3 imensionl spets of the event re pture. One n think of the PLIF imges s ut through the volume imge y the line of sight emission tehnique. A sttistil nlysis of suh topologil t is urrently eing unertken proviing informtion tht neither tehnique oul ffor, if pplie on its own. FLAME FRONT TRACKING BY PLIF OF OH AND ACETONE Of mjor interest to engine reserh is the pility to provie instntneous informtion on the lol ir/fuel equivlene rtio uring omustion, sine this hs mjor implitions on the effiieny of the proess n yle to yle vritions [9]. Severl tehniques hve een evelope for this purpose [10]. A ommon tehnique is to tg the fuel with fluoresing speies n to monitor susequent fluoresene. In the present se we use etone PLIF to stuy ignition phenomen in the two ell systems esrie in the experimentl setions. For this purpose the (non-fluoresing) fuels were seee with etone whih ws exite using the sme frequeny oule ye lser ner 282 nm (see experimentl setion ove). The vntge of the present tehnique is tht the otine PLIF signls not only re useful to stuy the lol ut lso for simultneously monitoring the flmefront propgtion: In regions were tempertures re suffiient to pyrolyse etone the PLIF signl isppers. Applie to the port injetion system (refer to figures 4 n 5 the etone PLIF tehnique n e pplie to simultneously provie informtion on mixture frtion n flme propgtion. The imge sequene shown in fig. 13 orrespons to stoihiometri mixture of propne seee with out 5% of etone [3]. These pitures were tken with the 4th hrmoni of the multiple YAG lser, proviing muh more homogeneous em profile thn the frequeny oule ye lser use to otin the sequenes in fig. 12. The fuel resiene time (e7

8 e g e f g h f h Figure 13: Time resolve PLIF imges from the omustion om using etone s fuel trer. The etone signl isppers in urnt regions. Figure 14: Diret emission imges of OH for stoihiometries ner the extintion limit. In oth sequenes shown the time step etween suessive imges is 800 μs. The rrow mrks the enter etween the two eletroes, the imge region orrespons to 50 y 75 mm. Here ffi = 0:65,! =2000. ly etween fuel injetion n ignition) in the ell ws 50 ms prior to ignition. The results presente here re first results of n ongoing stuy with the purpose to sttistilly nlyse n orrelte flme spees n flme topology to lol stoihiometries n resiene times. For resiene times > 200 ms nerly lminr flme evelopments were oserve, orresponing to ompletely homogeneous hrges in the omustion hmer. In the presente run one n see tht the flme moves towrs the fuel riher sie t erly stges of flme evelopment (seen on imges - of the sequene where the flme is initilly eveloping into the right, i.e. fuel rih, regions). Numers inite on the imges orrespon to ms ely fter ignition. The imge region orrespons to out 12 y 18 mm in size. FLAME EXTINCTION One of the mjor vntges of the high spee imging system is its ility to highlight the fetures of ynmi phenomen in rel time n to provie physil insight into their origins. In turulent iffusion flmes the phenomenon of flme extintion oul e visulise in rel time n in systemti mnner using the present system [12]. In engine reserh n sprk ignition the extintion of turulent premixe flmes is of funmentl interest [8]. The series in fig. 14 pture suh n event. They orrespon to iret emission reore with the unfiltere mer for len (ffi = 0:65) homogeneous methne/irmixtures in the fn stirre om. Turulene levels were moerte (0.75 ms 1 t! =2000rpm) in oth ses shown. The rrows mrk the position of the sprk gp. Exposure times were 256 μs for eh event, owing to the extreme wekness of the signl levels. The Figure 15: Diret emission (hemiluminesene) imges orresponing from the sie-vlve engine seen in fig.8. In the first imge the rekown of the sprk is seen. Exposure times were 30μs, the ely etween onseutive imges is 250 μs. time ifferene etween suessive imges mounts to 800 μs. Clerly one is le to follow the initition, growth n movement of smll flme kernel in turulent fiel, eoming righter uring the first instnes, ut then filing to grow n eventully fing wy. In series ) lol extintion event n e witnesse (imges 6-8). It is use y turulent eies. The turulene les to onvetive trnsport of retnts towrs the flme, uiling up grients n thus iffusive loss. As onsequene the loss terms strt to ominte over the proution terms (ril speies, het genertion) n the retions ie out. Fig 14 provies iret eviene of these phenomen. CYCLE RESOLVED MEASUREMENTS IN PRACTICAL SI ENGINES Finlly, we prove tht the tehniques evelope n pplie in the lortory systems esrie ove re iretly usle in engine reserh. The following setion esries rnk ngle resolve mesurements of fuel istriution n hemiluminesene. The test ojet ws the single yliner sie vlve engine esrie in the experimentl setion, operting on isootne. Fig. 15 shows yle resolve hemiluminesene from the engine. The mer ws use without filter. In piture ) the rekown event n e seen n the susequent flme evelopment n e followe. The ility to monitor mixture formtion n omustion uring single engine yle n ring mjor ontriutions towrs the unerstning of yle-to-yle vritions. We emonstrte the pility to mke fuel trer LIF mesurements in fig. 16, whih shows the fuel istriution uring single omustion yle. The iniviul imges shown re spe t 50μs intervls n where otine using 3-pentnone s the fuel mrker (see experimentl setion for etils). Imging of this type oul e omine with imging of ril speies proue uring omustion, for exmple NO onentrtions [13] or with OH ril formtion s presente in erlier setions of the present pper. Sequentil, time resolve imging, provies unique informtion on in-yliner flow-hemistry intertions n hs goo ignosti potentil for GDI or HCCI pplitions for exmple. 8

9 Figure 16: Fuel trer PLIF from the sie-vlve engine using 3-pentnone s the fuel mrker. Exposure times were 50 ns, the ely etween onseutive imges is 50 μs. White regions orrespon to regions with high fuel onentrtion, lk regions orrespon to regions where the mixture is urnt. CONCLUSION The present pper reports on novel wy to onut lser se ignosti work in sprk ignition reserh in time resolve fshion. The tehniques esrie offer welth of informtion on the physis n hemistry tking ple uring turulent sprk ignition n re suite to provie input t for numeril simultions. We hve shown results from plnr imging of OH onentrtions, n fuel trer PLIF se on etone n 3-pentnone. For the first time 2D lser mesurements of the flme struture in rel SI engine oul e me using these two tehniques, time resolve uring single omustion yle. The ignosti potentil is enormous, with the potentil to provie insight into the origins of yle to yle vritions, yieling more etile t on sprk ignition physis, s well s funmentl t on turulent flme growth. REFERENCES e f [1] C.F. Kminski, J. Hult, n M. Alén. High repetition rte plnr lser inue fluoresene of OH in turulent non-premixe flme. Appl. Phys. B, 68: , [2] M.J.G. Borge, J.M. Gigurer, n J.Luque. Stuy of the emission of the exite etone vpour t intermeite pressures. Spetrohim. At, 46A: , g h [5] J. Weikert, B.M. ter Hr Romeny, n M.A. Viergever. Effiient n relile shemes for nonliner iffusion filtering. IEEE Trns. on Imge Pro., [6] F. Ctté, P.-L. Lions, J.-M. Morel, n T. Coll. Imge seletive smoothing n ege etetion y nonliner iffusion. SIAM J. Numer. Anl., 29: , [7] C.F. Kminski, J. Hult, M. Alén, S. Linenmier, A. Dreizler, U. Ms,, n M. Bum. Sprk ignition of turulent methne/ir mixtures revele y time resolve plnr lser inue fluoresene n iret numeril simultions. Pro. Com. Inst., 28, in press. [8] J. Wrntz, U. Ms, n R.W. Dile. Comustion. Springer, Berlin, [9] B. Johnsson, H. Neij, M. Alén, n G. Juhlin. Investigtions of the influene of mixture preprtion on yli vritions in n si-engine using lser inue fluoresene. SAE , pges 85 99, [10] H. Neij. Development of Lser-Inue Fluoresene for Preomustion Dignostis in Sprk- Ignition Engines. Ph.D. Thesis, Lun Institute of Tehnology, [11] B. Yip, F.A. Miller, A. Lozno, n R. K. Hnson. A omine OH/etone plnr lser inue fluoresene imging tehnique for visulizing omusting flows. Exp. Fluis, 17: , [12] T. Ding, Th.H. vn er Meer, M.Versluis, M. Golomok, J. Hult, M. Alén, n C.F. Kminski. Timeresolve PLIF mesurements in turulent iffusion flmes, volume 3 of Turulene, Het n Mss Trnsfer. Aihi Shuppn, Jpn, Y. Ngno, K. Hnjlić n T. Tsui (Es.). [13] F. Hilenrn, C. Shulz, V. Sik, G. Josefsson, I. Mgnusson, O. Anersson, n M. Alén. Lser spetrosopi investigtion of flow fiels n no formtion in relisti SI engine. SAE , Journl of Engines, 107: , [3] H. Neij, A. Sitzkoff, R. Reinmnn, A. Frnke, n M. Alén. Applition of two-imensionl lserinue fuel trer fluoresene for ion urrent evlution. Com. Siene. Teh., 140(1 6): , [4] A. Dreizler, S. Linenmeier, U. Ms, J. Hult, M. Alén, n C.F. Kminski. Chrteristion of sprk ignition system y plnr lser inue fluoresene of OH t high repetition rtes n omprison with hemil kineti lultions. Appl. Phys. B, 70: ,