直喷汽油发动机中NO激光诱导荧光成像检测方案(流量计)

智能文字提取功能测试中

Available online at www.sciencedirect.comProceedings of the Combustion Institute 30 (2005) 2667-2674 2668W. G. Bessler et al. / Proceedings of the Combustion Institute 30 (2005)2667-2674 Proceedingsof theCombustionInstitute www.elsevier.com/locate/proci Quantitative in-cylinder NO-LIF imaging in a realisticgasoline engine with spray-guided direct injection Wolfgang G. Bessler， Max Hofmann， Frank Zimmermann.Gerrit Suck ， Jan Jakobs ， Sascha Nicklitzsch， Tonghun Lee°，Jirgen Wolfrum， Christof Schulz* a Physikalisch-Chemisches Institut (PCI)， Universitat Heidelberg， 69120 Heidelberg， Germany °Volkswagen AG， 38436 Wolfsburg， Germany IAV GmbH， 09120 Chemnitz， Germany Department of Mechanical Engineering， Stanford University， Stanford， CA 94305， USA Institut fur Verbrennung und Gasdynamik (IVG)， Universitat Duisburg-Essen， 47048 Duisburg， Germany Abstract Limiting the in-cylinder nitric oxide (NO) formation is a crucial task in the development of engines withgasoline direct injection. Exhaust gas aftertreatment requires storage catalysts that tolerate a maximumNO flux only， and the frequency of energy consuming catalyst regeneration cycles is directly correlatedwith engine-out NO. We present quantitative in-cylinder imaging measurements of NO mole fractionsin a gasoline engine with spray-guided direct injection using laser-induced fluorescence (LIF). The opticalengine design was kept close to that of a serial four-cylinder engine. Optical access was achieved via sap-phire windows， requiring only minor modifications to the engine block. The engine was operated with com-mercial gasoline and fired continuously. The data interpretation applies the spectral simulation toolLIFSim to calculate pressure， temperature， and gas-composition dependencies of the LIF signal. Tempera-ture-dependent CO2 absorption cross-sections are used to correct for laser and signal attenuation. A sen-sitivity analysis of the quantitative NO concentrations on the different parameters entering the evaluation ispresented. The LIF measurements are compared to results from in-cylinder fast gas sampling through amodified spark plug. The two techniques show good quantitative agreement. The LIF measurements arealso compared to charge-averaged working-cycle-resolved NO chemiluminescence measurements in theexhaust port. NO-LIF imaging results are presented for stratified engine operation with different levelsof exhaust gas recirculation (EGR)， showing the large impact of EGR on in-cylinder NO formation.@ 2004 The Combustion Institute. Published by Elsevier Inc. All rights reserved. Keywords： Nitric oxide； Laser-induced fluorescence； Gasoline direct injection； Combustion diagnostics 1. Introduction ( C orresponding author. Fax： +49 2033793087. ) ( E-mail addresses： christof.schulz@pci.uni-heidelberg. de， christof.schulz@uni-duisburg.de (C. S chulz). ) Gasoline engines with direct injection (DI) pro-vide significantly increased energy efficiency com-pared to spark-ignition engines with multi-pointinjection in the intake port and are considered ( 1540-7489/$ - see f ront m atter @ 2 004 The Comb u stion Institute. Published by Elsevier Inc. All ri g hts reserved.doi：10.1016/j.proci.2004.08.123 ) the most favorable concept for spark-ignition en-gines in the near future. A problem of gasoline di-rect-injection engines is that conventional three-way catalysts do not work with excess oxygen thatis present in the exhaust gases in stratified andhomogeneous lean operation mode. Storage cata-lysts have been developed for exhaust gas after-treatment of DI engines； they rely on successivestorage and reductive operation modes. In thestorage mode， nitric oxide (NO) is oxidized toNO2 and chemically bound upon contact withthe catalyst storage materials which must thenbe regenerated on a regular base by operatingthe engine under rich (reductive) conditions for afew working cycles. Because of restrictions in stor-age kinetics and capacity， the storage activity islimited in both maximum NO flux and accumu-lated amount of NO. Furthermore， the regenera-tion cyclesare fuel consuming，andtheirfrequency should be kept as low as possible. It istherefore crucial in engine development to mini-mize engine-out NO. The requirement for the NO diagnostic ap-proach presented here was to limit modificationsin the engine and its operating conditions to theminimum possible. Sapphire was used as windowmaterial for optical access. Its stability allowed toreduce window thickness such that neighboringcylinders and water-cooling systems required nosubstantial modification. Commercial gasolinewas used as fuel in the continuously running en-gine. This concept， however， also influences thediagnostic approach； e.g.， sapphire windows pro-hibit the use of high laser energy densities at shortwavelengths， and commercial fuel limits the UVtransmission through unburned gases. 2. Background Quantitative imaging of NO concentrationswith laser-induced fluorescence (LIF) has at-tracted significant interest in recent years[1]. Mea-surementss were performedin stablelaminarflames [2] for developing and validating combus-tion models [3]， and in gasoline and Diesel enginesfor in-cylinder diagnostics. The first applicationswere demonstrated in a spark-ignited square-pis-ton engine fueled with propane [4]. More realisticconfigurations included engine operation with iso-octane with multi-point [5，6] and direct injection[7-9]. One-dimensional measurements were per-formed in an engine fueled with regular gasoline[5]. Quantitative NO-LIF imaging was also per-formed in Diesel engines fueled with low-sooting[10] and commercial Diesel fuel [11]. Quantitativeinterpretation of NO-LIF signals in realistic opti-cal engines requires careful assessment of theinfluence of signal interference， pressure andtemperature fluctuations， laser and signal attenua-tion， and calibration. These issues will be briefly discussed in this section. Additional details havebeen recently published [12-15]. 2.1. Selective NO-LIFexcitation and detection Interference is an important issue in in-cylinderNO-LIF. O2-LIF is the main contributor to inter-ference in lean flames. CO2-LIF was recently iden-tified as a broad (200-450 nm) continuum in leanand rich flames [16]. Both the O2-LIF/NO-LIFand CO2-LIF/NO-LIF ratios increase with pres-sure because O2 and CO2 have a different pres-sure-dependence of fluorescence quantum yieldand line broadening compared to NO [13]. In richand non-premixed flames，additional broad-bandfluorescence appears that is usually attributed topolycyclic aromatic hydrocarbons (PAH) [17].The aromatic components of unburned commer-cial fuels strongly fluoresce， and interference withNO-LIF might be present especially in non-pre-mixed engine operation. Excitation wavelength and detection bandpassschemes that minimize interference for NO-LIFhave been investigated in high-pressure flames[13-15]. The optimized strategy that is used hererelies on A-X(0，2) excitation (O12 bandhead at247.94 nm) withblue-shifted detection withinthe (0，0) and (0，1) band at 220-240 nm [8]. Thisscheme provides low O2-LIF interference whileminimizing the detection of PAH-LIF， CO2-LIF， and fuel-LIF， all of which mainly fluorescered-shifted compared to the excitationwavelength. 2.2. Dependence of NO-LIF signal on temperature，pressure， and gas composition The dependence of NO-LIF on temperatureT， pressure p， and species concentration x； hasbeen extensively studied. The main T-dependencearises from the population of the laser-coupledground-state levels. T， p，and x； also affect thefrequency and cross-section of collisions thatNO molecules encounter， influencing the excita-tion efficiency via broadening and shifting ofthe absorption lines [18，19]， and the fluorescencequantum yield via collisional quenching [20，21].These effects are calculated using the LIFSimsoftware package [22]. Temporally resolved temperatures in internalcombustion engines are usually not available.Quantitative NO measurements therefore requireexcitation of transitions that minimize the totalT-sensitivity of the LIF signal. Figure 1 showssimulated NO-LIF signals versus T and p with247.94 nm excitation and fixed NO mole fractions.The strong dependence of the signal on bothparameters is evident； however， the calculationsshow that at high temperatures ≥2000 K the T-dependence is small [9，14]. Note this plot looksdifferent for fixed NO number densities. Fig.1. Simulated NO-LIF signal strength versus tem-perature and pressure with excitation at the A-X(0，2)O12 bandhead. aat247.94 nm， normalizedto)1bar，2000 K. The concentration of colliding species is evenless available than T. The introduced error is.however， small for premixed combustion sinceNO is present only in the post-flame-front gaseswhere the majority species concentrations are rel-atively precisely known. In diffusion flames， therequired corrections can be as large as 25% [3].The influence of local variations in air/fuel ratiosobserved in DI engines has been previously dis-cussed [8]. 2.3. Signal and laser attenuation Electronic excitation of NO requires short UVwavelengths <250 nm [14]. Recent experimentsshow that in high-p， high-T combustion environ-ments， UV light is attenuated by CO2 [23]. Para-metrized data are takenffrom [241. Thecalculation of effective absorption cross-sectionsin inhomogeneous combustion situations is diffi-cult since neither CO2 nor T-distributions areknown. Therefore， an accurate correction is oftenimpossible in practical high-pressure flames， andattenuation remains a dominant source of uncer-tainty. We correct for attenuation of laser andLIF signal based on calculated CO2 absorptivity(for details cf. Section 4.2). 2.4. Calibration With the corrections considered above， NO-LIF is proportional to NO concentration， andabsoluteconcentration requires calibration. Whilein situ calibration has been performed in enginesunder homogeneous lean conditions by addingNO to the feedstock gases [4，8，25] where ~90%of the NO survives the flame chemistry [26]， tech-nical restrictions due to the realistic engine setupprohibited this method in our experiments. In-stead， we use a miniature calibration burner(20 cm long， 8-mm-diameter metal tube) thatwas inserted into the cylinder through the sparkplug hole. A flat premixed methane/air flame was stabilized on the 7-mm-diameter matrix. Thecalibration information is obtained from LIFmeasurements with different amounts of NO upto2000 ppm addedto the premixedgases[9，12，25]. The temperature of this flame was cha-racterized via multi-line NO-LIF thermometry[27]. 3. Experimental The experiments were carried out in a Volks-wagen mass production four-cylinder FSI-enginethat was originally designed for wall-guided di-rect-injection combustion(four valves， bore：76.5 mm， stroke： 75.6 mm， engine displacement：1390 cm’). It was modified for spray-guided oper-ation with a research high-pressure swirl-typeinjection system. The compressionn ratio wasraised to 12.8：1， and the piston shape was modi-fied. A 6-mm-high， 4-mm-thick sapphire ring tha1extends at one side into the side field of the pent-roof section provides optical access to the com-bustion chamber from three directions (Fig. 2).The total window area is comparatively small sothat the wall-heat transfer is similar to the originalengine. The beam from a tunable KrF* excimer laser(Lambda Physik， Av=0.6cm=， 247.94 nm) wasformed into a vertical light sheet of 4×3 mm'with ~30 mJ prior to entering the engine. Foreach laser pulse， the energy was detected in frontof and behind the engine by photodiodes (Fig.3). LIF signals were imaged (Halle， f=100 mmUV achromat) onto an ICCD camera (LaVisionFlamestarIIF). NO-LIF was separated by reflec-tionnbandpass fflters(eight dielectric220-245 nm， 45°mirrors) and two 245 nm short-passfilters (Laseroptik).This filter combination yieldsa detection bandpass at 232 ±8 nm (width at halfmaximum) with a transmission of ~35%at225 nm (NO A-X(0，0) emission) and ~65% at237 nm (A-X(0，1) emission) while suppressingelastically scattered light by six ordersofmagnitude. Fig. 2. Optically accessible four-cylinder engine withrealistic piston and combustion chamber geometry. Fig. 3. Optical setup. 4. Results 4.1. NO-LIF selectivity Excitation scans (247.85-248.00 nm) were mea-sured in the fired engine in homogeneous opera-tion at 20° and 60°crank angle after top deadcenter (CA ATDC) (Fig.4). Calculated NO-LIFspectra can be fitted to the experimental data withLIFSim[22]. Free fitparametersare signalstrength， baseline， and temperature.Backgroundcontributions (fitted baseline over simulated signalmaximum) of 11% and 3% can be determined for20 and 60°CA， respectively， showing the goodinterference suppression achieved with the presentsetup. In stratified operation mode， observed inter-ference levels were slightly higher(~30% at 20°CAand <5% after 40°CA). Negligible signal (<1%)was measured when running the engine withoutignition； thus， LIF from fuel components is almostcompletely suppressed with our detection strategy.The measurements shown in Fig. 4 were taken witha single laser pulse at each excitation wavelengthposition. The scatter in signal intensities therefore Fig. 4. LIF excitation scans in fired engine (homoge-neous charge) and fits of simulated NO-LIF spectra tothe experimental data. reflects the cycle-to-cycle fluctuations in NO-LIFsignal. Nevertheless， fits to the spectra yieldrealistic temperatures (2400-2800 K) via multi-lineNO-LIF thermometry [27]. 4.2. NO-LIF quantification An important parameter entering signal quan-tification is temperature. For the correction of theT-dependence of the NO-LIF signal and laser-and signal-light attenuation due to hot CO2， weassume adiabatic flame temperatures (stoichiome-tric octane/air flame， T~2600 K) for crank an-gles up to the 75% burned mass fraction (MFB，at ~15°CAATDC with1stratified operation).After 100°CA， mixing is assumed to provide ahomogeneous T-distribution， and volume-aver-aged Tcalculated from pressure traces is used.Temperatures are linearly interpolated at interme-diate CA. Accordingly， equilibrium CO2 concen-trations(forr (=1.0) are assumed for theburned gas zone until 75% MFB and interpolateddown to measured exhaust CO2 concentration(~5%) at 100℃A. The absorption path lengthhas further influence on the total absorptivity.We assume a quasi-homogeneous， symmetric， lin-early growing flame kernel， starting at 10% MFB(around -10°CA) that reaches a distance of10 mm to the cylinder walls at 90% MFB (MFBpoints are evaluated individually for each opera-ting condition from the in-cylinder pressurecurves). Using these parameters， we perform cor-rections for laser- and signal-light attenuationbased on T-dependent CO2 absorption cross-sec-tions [24]. The LIF images were corrected for laserand signal attenuation on a pixel-by-pixel basis(i.e.， calculating correction factors for each pointbased on the respective laser and signal pathlengths， and the respective wavelengths). Figure 5 shows the relative variation of theNO-LIF signal of a given NO concentration as afunction of crank angle for a spatial positionin the center of the imaged volume. Dividingmeasured signal intensities by this factor yields LIF ——Transmission -- Total Fig. 5. Signal variation of a fixed NO concentration(stratified engine operation) versus crankshaft positionand its contributions from LIF signal yield (normalizedto 40°CA) and attenuation correction. The calibration(1900K， 1 bar) condition corresponds to a LIF factor of2.1 on this scale. (non-calibrated) NO concentrations. Contribu-tions from the LIF signal yield (Fig. 1) and atten-uation are separated. From this figure， we candistinguish two regimes： Before 40CA ATDC，the total correction is dominated by the laserand signal attenuation， while the LIF signal yieldhas a weak crank-angle dependence. After 60°CA，attenuation plays a minor role， but the LIF yielddepends strongly on T. The application of commercial fuel addition-ally limits the UV transmission through unburnedgases. A correction is not attempted here. Themeasurements presented are therefore restrictedto cases where the major part of the fuel is burned(later than ~20°CA ATDC)， and light attenuationdue to unburned fuel is neglected. Earlier experi-ments showed that signal attenuation due to NO(fluorescence trapping) is negligible [23]. 4.3..Sensitivity and error analysis of NO-LIFquantification Figure 6 shows a typical T-trace used in the dataevaluation at stratified engine operation. A devia-tion from these values induces a variation in correc-tion factors. For each crankshaft position， theT-range has been assessed that leads to a <10%and <25% variation of the correction factors. Anextended T-range means low T-sensitivity. TheLIF signal (Fig. 6A) is relatively T-insensitive forT≥ 2000 K. Knowledge of absolute temperaturesas well as their local distribution in the burned gas istherefore not required at early combustion stagesfor correcting the LIF signal yield. However， laserand signal attenuation due to hot CO2 is very T-sen-sitive， especially at high p and T (around 0CA)(Fig. 6B). The total T-sensitivity is shown in Fig.6C. We find that the effects of Fig. 6A and Fig. 6Bpartially compensate at 40-50°CA ATDC. We be-lieve that real in-cylinder burned-gas temperaturesare within the 25% NO variation limit (Fig. 6C)compared to the assumed temperatures. Similar sensitivity analyses were performed forthe assumed CO2 concentration and the absorp-tion path length. While the NO quantification is relatively insensitive to CO2 concentration withinlimits that are reasonable for post-flame gases，the path length has a major influence on the atten-uation correction at high p (0-30°CA ATDC). Thelack of detailed information on flame growth inour simple approach makes this parameter a majorsource of uncertainty(≥25%) before 30°CA. The overall uncertainty in absolute NO molefractions was estimated assuming the followingparameter uncertainties： T： ±200 K， LIF yield：±10%(mainly due to the uncertainty in gas com-position affecting the quenching correction)， xco，：±1%， flame radius： ±1 cm (both used in attenua-tion correction)， and calibration： ±15%. Due tothe non-linear coupling of the different parametersand their dependence on p， uncertainty variesstrongly with crank angle (Fig.7). At 40-70CAATDC， the uncertainties are around1±20%.Resulting error bars are shown in the figures. 4.4.Comparison of NO-LIF to in-cylinder gassampling A UV-analyzer [28]was used for NO concentra-tion measurements in gas samples extracted fromthe combustion chamber at the spark plug locationvia a fast gas-sampling valve (GSV). GSV andNO-LIF measurements were applied under identi-cal operation conditions with homogeneous charge(=1.0) used to reduce spatial and cycle-to-cyclefluctuations in NO concentrations. The valve hasa sampling time of approx. 1 ms. Sampling isperformed every sixth cycle， and the extractedgases mix before entering the detector. The mea-sured data therefore represent a phase-averagedconcentration over several working cycles.S. Figure 8 shows a good agreement of in-cylinderNO concentrations obtained by LIF (phase-aver-aged over 60 instantaneous NO concentrationfields) and the GSV technique. The remaining dis-crepancy before 30°CA ATDC may be explainedwith additional laser and signal attenuation dueto unburned fuel， which is not corrected for， andby the difference in sample volumes (LIF： 18x4×3=216mm’ dlighti sheetlength xheight× ——AssumedT 10%[NO] change 二25%[NO] change Fig. 6. Temperature sensitivity of NO-LIF quantification (stratified engine operation). Shown are the temperaturevalues used in the data evaluation versus crank angle. The gray areas correspond to the temperature range that wouldresult in a ≤±10(25)% change in evaluated NO mole fraction. (A) LIF signal yield， (B) laser and signal attenuation， and(C) total T-sensitivity. Fig. 7. Error analysis of NO-LIF quantification forstratified and homogeneous engine operation. Fig. 8. Comparison of in-cylinder NO concentrationsobtained with LIF and in-cylinder gas sampling. Homo-geneous (d=1.0) engine operation at 0.3 MPa indicatedmean effective pressure (IMEP)， 2000 rpm， 0%EGR. width， GSV： ca. 500 mm’， ca. 10 mm distancebetween the centers of both sample volumes). 4.5. Influence of exhaust gas recirculation on NOformation NO-LIF imaging was used to investigate theinfluence of exhaust gas recirculation (EGR) onEGR is an important technique to reduce en-gine-out NO. Its effect can be explained with anincreased absolute heat capacity because of mix-ture dilution， lowering the in-cylinder burned-gastemperatures and thereby slowing the stronglyT-dependent NO formation reactions [9]. Mea-surements with 0%， 20%， and 30% EGR (ratioof recirculated exhaust gas mass over total fedcharge mass) at stratified engine operation (over-all o：0.3-0.5 with increasing EGR) are shownin Figs.9-11. Instantaneous NO concentration fields for se-lected crankshaft positionsat20%EGRareshown in Fig. 9.The strong cycle-to-cycle fluctua-tion is evident. Surprisingly， fluctuations increaseat later crankshaft positions. This is qualitativelyvisible in Fig. 9 and is confirmed when calculatingthe standard deviation of the single images， whichincreases from ~30% to ~80% between 20 and80°CA ATDC for all studied EGR conditions.Given the relatively small LIF sampling volumecompared to the cylinder dimensions， we interpretthis observation as an effect of spatially inhomo-geneous NO distributions (fluctuations in flame Fig. 9. Instantaneous NO concentration fields. Stratified(d=0.4) EGR， 2000rpm，0.3 MPa IMEP. The images show an area of 14×4 mm. Fig. 10. Average NO concentration fields for differentEGR ratio. Stratified (=0.3-0.5) engine operation，2000 rpm， 0.3 MPa IMEP. The images show an area of14×4 mm . At 30% EGR the look-up table is enhancedby a factor of 10. Fig. 11. Average NO concentrations from Fig. 10 fordifferent EGR ratio. position and flow field) rather than cycle-to-cyclefluctuations in overall NO production. Phase-averaged NO concentration fields (20images per crankshaft position) are shown inFig. 10. The strong reduction of NO concentrationwith increasing EGR ratio is evident. NO is asymmetrically distributed even in these averageimages (note that the spark plug position is rightbehind the center of the imaged area， cf. Fig. 2).Thelaser travels from left (exhaust side) to right (in-take side)； nevertheless， the observed asymmetry isnot an attenuation effect (which was corrected for inthe data evaluation). Instead，it is more likely tofind areas with high NO concentrations on the leftside of the imaged area (location of exhaust valves).Spatially averaged NO concentrations from Fig. 10are plotted versus crank angle in Fig. 11. MaximumNO concentrations in the imaged area occuraround 30-40°CA ATDC. Note that concentra-tions are probably under-predicted at <20°CAdue to laser attenuation by fuel vapor (cf. Section4.2). At late crankshaft positions， the values are inqualitative agreement with the global NO emissionsof 1200， 780， and 140 ppm NO for 0%， 20%， and30% EGR， respectively， measured with the fastCLD analyzer. Additional operating conditionshave been investigated， results are reported in [29，30]. 4.6. Comparison of NO-LIF to exhaust port fastCLDmeasurements To investigate further the cycle-to-cycle fluctu-ations in NO distribution， measurements werecarried out in the exhaust port with a fast chemi-luminescence detector (CLD) [31]. This allows acomparisonbetweeninstantaneousNO-LIFconcentrations obtained at arbitrary crankshaftpositions and the charge-mass-averaged NO con-centration [32] of the corresponding cycle. Investigations were performed for stratified en-gine operation with different EGR ratio (Fig. 12).The correlation between the two techniques is evi-dent when NO-LIF is measured at 20°CA ATDC.However， it is almost lost with NO-LIF at 80°CA.The observed scatter reflects the strong cycle-to-cycle fluctuations of NO concentration withinthe observed area at late crankshaft positions(cf. Fig.9). Note that the uncertainty in LIF mea-surements is similar at these two crankshaft posi-tions (Fig. 7). LIF probes only a small volumewhile the CLD results represent an average overthe complete cylinder charge. A quantitative com-parison of NO concentrations is therefore notpossible. Instead， the LIF/CLD comparison con-firms the image of strongly inhomogeneous NOdistributions at late crank angles under stratifiedconditions. At the same time， the results showthe importance of temporally resolved NO mea-surements at earlier detection times((around20°CA ATDC)， where observed fluctuations aredirectly correlated to engine out NO. 5. Conclusions Laser-induced fluorescence： of nitric oxideproved a viable technique for quantitative in-cylin- Fig. 12. Correlation of instantaneous NO-LIF concen-trations (averaged over the imaged area) and exhaustport fast CLD measurements obtained from the corre-sponding working cycle. Note the double-log scale. der measurements of NO concentrations in aspray-guided DI engine with minimized opticalaccesses and operation with commercial gasoline.The major source of uncertainty arises from theunknown temperature and flame diameter thatdetermine the correction for laser and signal atten-uation. An overall accuracy of ±20-60% is esti-mated depending on crankshaft position. ThecomparisonVwithin-cylinder sampling showedgood agreement， and the comparison with exhaustport charge-averaged measurements confirms theimage of a spatially very inhomogeneous in-cylin-der NO distribution. The additional spatially andtemporally resolved information of the NO ima-ging measurement is of high value for the under-standing of the popolllluuttaanntt fformation in DIgasoline engines. In this paper， we investigatedthe influence of exhaust-gas recirculation on NOformation under stratified engine operation. Thepeak NO concentrations decrease by a factor of11 when increasing the EGR ratio from 0% to 30%. Acknowledgments The technical support by U. Branczyk (PCI，Heidelberg)， and T. Schindler and D. Lehmann(Volkswagen AG) is gratefully acknowledged.The Division of International Programs at theUS NSF supports the traveling of T.L. References ( [1] J. Wolfrum， Proc. Combust. Inst. 27 (1998) 1 -42. ) ( [2] J.R . Reisel， N.M. Laurendeau， Energy F uels 8 (1994) 1115-1122. ) ( [3] J.B. Bell ， M.S. Day， J.F. Grcar， W.G. B essler， C.Schulz， P. Glarborg， A.D. Jensen， Proc. Combust. Inst. 29 (2002) 2195-2202. ) ( [4] C. Schulz， V. Sick， J. Wolfrum， V. Drewes， M. Z ahn， R. M aly， Proc. C ombust. Inst. 26 (1996) 2597-2604. ) ( [5] M. Knapp， A. Luczak， V . B eushausen， W . H e nt- schel， P . Manz， P. Andresen， S AE Technical PaperSeries 970824 ( 1997). ) ( [6] F. Hildenbrand， C. Schulz， V. Sick， G. Josefsson， I. Magnusson， O. A ndersson， M . Alden， S AE P a per 980148 (1998). ) ( [7] P. Jamette ， P. Desgroux， V. Ricordeau， B. Des- champs， SAE Technical Paper Series 2001-01-1926(2001). ) ( [8] F. Hildenbrand， C. Schulz， M. H artmann， F. Puch-ner，G. Wawrschin， SAE Paper 19 9 9-01-3545 (1999). ) [9] W.G. Bessler， C. Schulz， M. Hartmann， M. Schenk，SAE Technical Paper Series 2001-01-1978(2001). ( [10] J.E. Dec，R.E. Canaan，SAE Technical Paper Series 980147 (1998). ) ( [11] F. Hildenbrand， C . Schulz， J. Wolfrum， F. Keller， E.Wagner， Proc. Combust. Inst. 28 (2000)1137-1144. ) ( [12] W.G. B essler， C. S chulz， T . Lee， D . I. S h in， M . Hofmann， J.B. Jeffries， J. Wolfrum， R.K. Hanson，Appl. Phys. B 75 (2002)97-1 0 2. ) ( [13] W.G. Bessler， C . S chulz， T . Lee， J.B. Jeffries， R.K. Hanson， Appl. Opt. 42 (2003) 2 031-2042. ) ( [14] W.G. Bessler， C. S c hulz， T. Lee， J. B . Je f fries， R . K. Hanson， A p pl. Opt. 42 (2003) 4922-4936. ) ( [15] W.G. Bessler， C . S chulz， T. Lee， J.B. Jeffries， R.K.Hanson， Appl. Opt. 41 (2002)3547-3557. ) ( [16]W.G. Bessler， C . S c hulz，T. L ee， J.B. Jeffries， R.K. Hanson， Chem. Phys. Lett. 375 (2003) 344-349. ) ( [17] A. Cialolo， R. B arbella， A . Tregrossi， L. Bonfanti，Proc. Combust. Inst. 2 7 (1998) 1481-1487. ) ( [18] A.Y. Chang，M.D. D i Rosa， R.K. Hanson， J. Quant. Spectrosc. Radiat. Transfer 47 ( 1992) 375-390. ) Comments Volker， Sick， University of Michigan， USA. Are cyclicvariations in NO formation a significant source for over-all NO levels emitted by the engine? Reply. Figure 12 shows cycle-resolved measurementsof the charge-mass-averaged NO concentration mea-sured in the exhaust port of the engine by a fast chemi-luminescence detector. From the upper diagram wederive the following relative standard deviations of 21，24， and 33% at 0， 20， and 30% EGR rate， respectively.There is indeed a strong cyclic variation of the overallNO emission which increases with increasing EGR-rate. Michael Drake， General Motors， USA. You are mea-suring NO LIF in a narrow region (4×14 mm) in thishighly stratified engine， but you still get good correla-tions with in-cylinder sampling from a different regionin the cylinder and from exhaust gas measurements.How widely applicable do you think this is? Reply. The comparative measurements between NO-LIF and gas-sampling-valve (GSV) technique (Fig. 8)were carried out under homogeneous operating condi- ( [19]M.D. DiRosa， R.K. H a nson， J. Q u ant. Spectrosc. R adiat. T ransfer 52 (1994) 515-529. ) [20] P.H. Paul， J.A.Gray， J.L. Durant Jr.，J.W. ThomanJr.， Appl. Phys. B 57 (1993) 249-259. [21] P.H. Paul，J.A. Gray，J.L. Durant Jr.， J.W. ThomanJr.， AIAA J. 32 (1994) 1670-1675. ( [22] W.G. Bessler， C . S chulz， V . Sick， J . W. Daily，in： Proceedings of the 3rd Joint Meeting of the US Sections ofthe Combustion In s titute， Chicago， 2003， PI1-6.Available from www.lifsim.com. ) ( [23] F. Hildenbrand， C . S chulz， Appl. Phys. B 73 (2001) 1 65-172. ) ( [24] C. Schulz， J.B. Jeffries ， D.F. Davidson， J.D. K och， J. Wolfrum， R .K. Hanson， Proc. C o mbust. Inst. 29 (2002)2725-2742. ) ( [25] C. Schulz， V. Sick， U.E. Meier， J . H einze， W . S tricker ， Appl. Opt. 38 (1999) 434- 1 443. ) ( [26] P.A. B erg， G .P. S m ith， J . B. Jeffries， D .R. C r osley，Proc. C ombust. Inst. 2 7 ( 1998) 1 377-1384. ) ( [27] W.G. Bessler， C. Schulz， Appl. Phys. B 7 8 ( 2004) 519-533. ) ( [28] B. Heller， G. Lach， J.D. Baronick， W. Fabinski， M. Moede， SAE Technical Paper Series 2004-01-1830 (2004). ) ( [29] G. Suck， J . J akobs， S. N icklitzsch， T . L ee， W.G. Bessler， M. Hofmann， F . Zimmermann， C . S c hulz， SAE Paper 2004-01-1918 (2004). ) ( [30] G. Suck， J. Jakobs， S. Nicklitzsch， W.G. Bessler，Max Hofmann， C. Schulz， in： International Sympo-sium o n I nternal Combustion Diagnostics， Baden- Baden， 2004. ) ( [31 ] R. Ford，N. Collings，SAE Paper 1999-01-0208(1999). ) ( [32] J.K. Ball ， C.R. Stone， N. Collings， P roc. Inst.Mech. E ng. 2 13 ( D) ( 1 999) 175-189. ) tions with early fuel injection in order to minimize localvariations in mixture properties and combustion pro-gress. The GSV system can neither perform instanta-neous measurements (valve opening time ca. 1 ms) norcycle-resolved measurements (about 100 cycles are sam-pled). Therefore， the NO-LIF concentration results weresimilarly averaged (data averaged over one millisecondand the data of 20 engine-cycles) in order to compareboth techniques. The NO-LIF technique， however，addi-tionally provides spatially and temporally resolved mea-surements.TThelocal variationsunderrstratifiedconditions are shown in Fig. 9. Under these conditionsa sampling measurement would not yield useful datafor a comparison. The correlation between in-cylinder NO-LIF and ex-haust gas measurement is relatively poor on a cycle re-solved basis. At 20°CA ATDC (Fig. 12) we observe agood correlation on a statistical basis. This is plausiblebecause with stratified-charge operation and the com-bustion (and hence， NO formation) mainly takes placein the vicinity of the spark plug where the NO-LIF sam-pling volume was located. Under these conditions， thevolume observed by NO-LIF represents the total loadrelatively well. Limiting the in-cylinder nitric oxide (NO) formation is a crucial task in the development of engines withgasoline direct injection. Exhaust gas aftertreatment requires storage catalysts that tolerate a maximumNO flux only, and the frequency of energy consuming catalyst regeneration cycles is directly correlatedwith engine-out NO. We present quantitative in-cylinder imaging measurements of NO mole fractionsin a gasoline engine with spray-guided direct injection using laser-induced fluorescence (LIF). The opticalengine design was kept close to that of a serial four-cylinder engine. Optical access was achieved via sapphirewindows, requiring only minor modifications to the engine block. The engine was operated with commercialgasoline and fired continuously. The data interpretation applies the spectral simulation toolLIFSim to calculate pressure, temperature, and gas-composition dependencies of the LIF signal. Temperature-dependent CO2 absorption cross-sections are used to correct for laser and signal attenuation. A sensitivityanalysis of the quantitative NO concentrations on the different parameters entering the evaluation ispresented. The LIF measurements are compared to results from in-cylinder fast gas sampling through amodified spark plug. The two techniques show good quantitative agreement. The LIF measurements arealso compared to charge-averaged working-cycle-resolved NO chemiluminescence measurements in theexhaust port. NO-LIF imaging results are presented for stratified engine operation with different levelsof exhaust gas recirculation (EGR), showing the large impact of EGR on in-cylinder NO formation.
2004 The Combustion Institute. Published by Elsevier Inc. All rights reserved.