Deschamps v52n6

8/21/2019 Deschamps v52n6

1/17

VISUALISATION DE LA RƒPARTITION DU CARBURANTET DES GAZ BRóLƒS DANS UN MOTEUR Ë4 SOUPAPES Ë ALLUMAGE COMMANDƒ ;EFFET DE LA STRATIFICATION SUR LA COMBUSTIONET LES POLLUANTS.

Une mŽthode indirecte pour cartographier les gaz bržlŽs dans un

moteur ˆ allumage commandŽ a ŽtŽ dŽveloppŽe. Elle est fondŽesur une visualisation ˆ partir de la fluorescence induite par laser(LIF) du mŽlange air-carburant non bržlŽ et ensemencŽ avec dubiacŽtyl. Les gaz bržlŽs provenant ˆ la fois des recirculationsinternes et externes sont observŽs. Ce type de diagnostic estcomplŽmentaire des techniques de LIF utilisŽes pour observer ladistribution du carburant. Ces mesures de concentration sontrŽalisŽes dans un moteur ˆ 4 soupapes avec acc•s optiques, pourune gamme Žtendue de conditions opŽratoires. Celles-ci compren-nent des variations des modes d'injection du carburant et desmodes de recirculation des gaz bržlŽs, provoquant ainsi diffŽrentstypes de stratifications qui correspondent ˆ des dŽgagementsd'Žnergie et des Žmissions de polluants tr•s distincts. Le niveau de"tumble" et l'emplacement de l'Žtincelle sont aussi modifiŽs.L'observation de la stratification telle qu'elle est rŽellement dans lemoteur constitue un moyen pertinent pour expliquer ses perfor-mances. Les param•tres permettant d'optimiser les niveaux deNOx et d'HC peuvent en tre dŽduits, de mme que l'efficacitŽ desstratŽgies de recirculation et d'injection du carburant. Les rŽsultatsdes visualisations ont ŽtŽ confirmŽs par des mesures obtenuesdans un moteur monocylindre muni d'une culasse conventionnelleavec la mme gŽomŽtrie de chambre de combustion.

VISUALISATION OF GASOLINE AND EXHAUST GASESDISTRIBUTION IN A 4-VALVE SI ENGINE; EFFECTS OFSTRATIFICATION ON COMBUSTION AND POLLUTANTS.

An indirect method to map the burned gases in SI engine has beendeveloped. It is based on visualisation by Laser Induced

REVU E D E L’IN STITUT FRAN Ç A IS D U PÉTRO LE

V O L. 5 2 , N ° 6 , N O V EM B RE -D É C E M B RE 1 9 9 7

651

VISU A LISATIO N O F G A SO LIN EA N D EXH A U ST G A SES

D ISTRIBU TIO NIN A 4-VA LVE SI EN G IN E;EFFEC TS O F STRATIFIC ATIO NO N C O M BU STIO N

A N D PO LLU TA N TS*

B. DESCHA M PS a nd T. BARITAU D

Institut français du pétrole1

(1) 1 et 4, avenue de Bois-PrŽau,92852 Rueil Malmaison Cedex - France

* This article was presented at the International Fall Fuels and Lubricants Meetingand Exposition, San Antonio, USA, October 14-17, 1997

TEC H N IC A L N O TEcopyright C 1997, Institut Français du Pétrole


2/17

Fluorescence of the unburned mixture seeded with biacetyl. Bothinternally and externally recirculated burned gases are monitored.

This diagnostic is complementary to the LIF technique applied tomeasure the gasoline distribution. These LIF gasoline and burnedgases measurements are applied in a 4-valve optical access SIengine for a large range of operating conditions. These includevariations of both fuel injection and burned gas recirculation modescausing different types of stratification leading to very distinct heatrelease and exhaust emissions characteristics. Tumble level andspark location are also modified. The observation of the actualstratification in the engine forms a sound basis explanation of theengine performance. Parameters allowing an optimisation of NOxand HC levels can be inferred, and in particular the effectiveness ofrecirculation and fuel injection strategies. The conclusions areconfirmed by measurements in a single engine cylinderconventional head with the same geometry.

VISUALIZACIîN DE LA DISTRIBUCIîNDEL CARBURANTE Y DELOS GASES CONSUMIDOSPOR UN MOTOR DE 4 VçLVULAS DE ENCENDIDOCONTROLADO ; EFECTOS DE LA ESTRATIFICACIîNSOBRE LA COMBUSTIîN Y LAS EMISIONESCONTAMINANTES

Se ha desarrollado un mŽtodo indirecto para cartografiar los gasesconsumidos por un motor de encendido controlado. Este mŽtodose funda en la visualizaci—n a partir de la fluorescencia inducidapor l‡ser (LIF) de la mezcla aire-carburante no consumida ysembrada con biacetil. Se han observado los gases consumidos,

procedentes tanto de las recirculaciones internas y externas. Estetipo de diagn—stico es complementario de las tŽcnicas de LIF quese emplean para observar la distribuci—n del carburante. Estasmediciones de concentraci—n se aplican en un motor de cuatrov‡lvulas con accesos —pticos, y ello para una amplia gama decondiciones operatorias. Estas œltimas incluyen las variaciones delos modos de inyecci—n del carburante y de los modos derecirculaci—n de los gases consumidos, provocando de este mododiversos tipos de estratificaciones que corresponden adesprendimientos de energ’a y emisiones de contaminantes demuy diversa ’ndole. TambiŽn se han modificado el nivel de"tumble" y el emplazamiento de la chispa. La observaci—n de laestratificaci—n tal como se produce realmente en el motor

constituye un modo pertinente para explicar sus resultadospr‡cticos. Los par‡metros permiten optimizar los niveles de NOx yde HC pueden as’ ser debidamente deducidos, del mismo modoque la eficacia de las estrategias de recirculaci—n y de inyecci—nde carburante. Los resultados de las visualizaciones se hanconfirmado mediante mediciones obtenidas con un motormonocil’ndrico provisto de una culata convencional, con la mismageometr’a de la c‡mara de combusti—n.

INTRODUCTION

Diluting the gasoline-air mixture with burned gas isan efficient way to reduce NOx emissions level inspark-ignition [1] as well as in diesel engines. It is wellknown that the formation of nitric oxides is highlydependent on the flame temperature. When recirculat-ing exhaust gases, the unburned mixture is diluted withburned gases, reducing the peak burned gas temperatureand as a result the NOx formation rates. Anothermethod is to run the engine under lean conditions.However both methods affect combustion stability,entailing the problem of irregular ignition and slowerpropagation which results in additional unburned

hydrocarbon emissions. For both methods it isnecessary to favour ignition based upon distributions of fuel and burned gases.

It has already been shown that in a 4-valve enginewith a pentroof chamber, fuel stratification in thedirection parallel to the top of the roof can be achievedby injecting into only one of the two intake ports [2, 3,4]. Biacetyl-LIF combined with LDV and combustionanalyses [4] showed that by coupling the effect of intense turbulence, moderate velocity at the spark andignition on the rich side, the performance of the 4-

valves engine was greatly enhanced. A strongcorrelation level between the local mixture compositionat the spark and the duration of the early phase of combustion was found for two turbulence intensities.Also, the early heat release was favoured by enhancingturbulence. Hence, the lean operating limit of theengine could be extended. This study demonstrated thepotential offered by stratifying the charge for leancombustion. Nevertheless this earlier work needed to beextended by the studying the effect of stratified gasolineon NOx and HC emissions which could be increased bythe propagation of the flame toward the very lean

region. The effect of residual gas distribution might beimportant as well [5]. Also the other known possibilityfor reduction of NOx emissions without altering HCneeds to be investigated: Exhaust Gas Recirculation(EGR). In a lean operating engine, there is a limit toEGR beyond which the combustion process is alteredand additional HC emissions and engine instability areexpected. A possibility for improving tolerance torecirculation might be to stratify the engine charge inburned gases while stratifying the gasoline in theopposite way [6] and [7]. One can imagine thatstratification in the burned gases might be also

accomplished by recirculating a part of the exhaust gasinto one of the two ports for stratifying gasoline.

VISU ALISATIO N O F G ASO LIN E AN D EXH AU ST G ASES DISTRIBU TIO NIN A 4 -VALVE SI EN G IN E; EFFEC TS O F STRATIFIC ATIO N O N C O M BUSTIO N AN D PO LLUTAN TS



652


3/17

This study has been conducted to associate control of gasoline distribution and burned gases distribution in

order to optimise the engine performance includingemissions and tolerance to recirculation. In-cylinderburned gas measurements employing laser sheetexcitation require the presence of markers in the burnedgases. Water and CO2 are naturally present incombustion products. The concentration of burned gashas been measured locally using CARS applied toCO2 [8]. Recently water has been used to visualise theresidual gas by PLIF with an eximer laser as a lightsource [9]. We decided to visualise the unburnedmixture fraction from which we deduced burned gas

fraction: the concept was to visualise fuel on one sideand air on the other side. The complementary imagereveals the burned gas distribution. The aim of thepresent work is to develop an optical diagnostic able tocharacterise the burned gas distribution in order to testpossible solutions of exhaust gas recirculation forreducing emissions without altering combustion.

1 RUNNING ENGINE CONDITIONS

The test engine is a 4-valve pentroof chamber single

cylinder described in earlier work1 [10] and [4]. Itsgeometrical characteristics are similar to the productionengine, despite the presence of optical ports (Table 1).The operating speed was 1200 RPM and the volumetricefficiency in unburned mixture was regulated to 0.6.The ignition was activated 25 CAD before the top deadcentre (TDC). With the convention of TDC at360 CAD, the intake valves open at 681 CAD and closeat 245 CAD. The fuel injectors were mounted in theintake ports, so they pointed directly at the back,upstream side of the valves. The timing of fuel injection

and duration, ignition, and data acquisition arecontrolled by a computer. The synchronisation to theengine is achieved with the use of a crankshaft angleencoder. The engine is connected to an electric motorfor keep the engine speed constant.

TABLE 1

Geometrical characteristics of the engine(in bold the compression ratio for the transparent engine)

Displacement 441 cm3

Bore 82 mm

Stroke 83.5 mm

Compression ratio 9.5/8.4

Rod/Half stroke ratio 3.45

The geometry of the studied engine allows differentpossibilities for varying the flow, injection and

recirculation configurations.

1.1 Flow configurations

In the standard configuration, the 2 intake valvestogether with the pentroof combustion chamber lead toa flow pattern with a major rotating component aroundan axis parallel to the top of the roof (tumble X) withsignificant flow motions remaining in other directions.A description of the flow field has been given by LeCoz et al. [10] using three-dimensional computations

and measurements. Earlier work showed that thetumble component is accentuated by welding 180¡shrouds on the back of the intake valves. With thisconfiguration, the flow pattern is composed of astronger tumble X that increases while the piston isrising until destruction into small turbulent eddies andresulting in higher turbulence level at TDC.

1.2 Intake configurations

The 2 intake valves allow either homogeneous or

stratified gasoline operations, depending on theinjection mode (injecting into both intake ports or asingle port respectively). The injection is initiated at250 CAD and lasts 38 CAD (for dual ports injection) or60 CAD (for single port injection) to obtain astoichiometric mixture. A l-probe in the exhaustallowed verification and adjustment of the equivalenceratio by regulation of the fuel injection duration.

1.3 EGR System and emission control

A return line was welded between the tail pipe andthe intake duct. For a better control of the EGR, hotexhaust gases were water-cooled before beingintroduced in the intake system. Thus, most of the waterin the recycled exhaust gases was condensed and didnot enter the combustion chamber. A thermocouplegave the temperature after the heat exchanger whichwas regulated to the temperature of the admitted flow(between 24¡ and 26¡C). By cooling the exhaust gas weensured the intake temperature was approximatelyconstant over the set of experiments. The cooledexhaust gases were introduced either at a point

upstream of the Y separation of the intake duct, or intoonly one intake duct. To control the EGR rate, two CO2




653


4/17

concentration analysers were installed in the system.The first one measured the molecular CO2 fraction in

the exhaust, whereas the second one gave the CO2fraction in the intake duct. NO2 and HC concentrationswere also measured. Figure 1 illustrates the experi-mental setup for the exhaust gas recirculation andcontrol methods.

Figure 1

Sketch of the exhaust gas recirculation system with themeasuring devices.

The EGR rate is defined as the volumetric fraction of recirculated exhaust to the total intake flow. Hence weare able to determine the EGR rate as:

(1)

where (eCO2)ADM and (eCO2)ECH are the volumetricfraction of CO2 in the intake and the exhaust ductsrespectively. When using 2 intake ducts for recirculat-ing, the equi-distribution of (eCO2)ADM in two ducts wasverified. The difference was less than 0.5%. The outputsignals of the analysers are transmitted to the testcomputer, which allows monitoring and regulating the

EGR rate, a valve system being mounted downstreamof the heat exchanger.

It is important to point out that while recirculatingthe exhaust gases, the mass flow rate of admitted air

was kept constant since the volumetric flow of fresh airinto the cylinder was regulated by sonic orifices. Thevolumetric efficiency defined as the ratio of the mass of admitted gases to the mass of gases that corresponds toan entire fill-up of the cylinder depends on hEGR. Forexperiments with EGR the modified volumetricefficiency lEGR can be easily related to the volumetricefficiency relative to admitted unburned mixture la andhEGR as defined previously:

(2)

This relation was verified experimentally for hEGRvarying from 0 to 25%.

Note that when recirculating exhaust gases, it is thetotal burned gas fraction in the cylinder that acts as adilutent. These burned gases consist in residual gasesfrom the previous cycle and exhaust gas recirculated tothe intake. Residual gases are not taken into account byCO2 analysers.

1.4 Spark plug configurations

To take advantage of the in-cylinder stratification, wealso had to optimise the spark plug location, based on atrade off between a good initial combustion phase andthe wall vicinity. When stratifying the gasoline, thespark plug and the pressure transducer locations(originally 35 mm away from the centre in a directionparallel to the roof ridge) were permuted in order tofavour the ignition process. Deschamps et al. [4] haveshown that tumble motion in the offset spark plug caseis favourable for flame initiation since the low velocity

(1 Ð 2 m/s) at the ignition angle (335 CAD) does notproject the flame kernel on to the wall and since thehigher but moderate turbulence levels for the strongtumble configuration also favours ignition.

2 NEGATIVE PLIF METHOD FOR BURNED

GAS VISUALISATION

The visualisation technique in the present study isderived from the fuel concentration method presentedin previous work and is based on planar laser induced

fluorescence of biacetyl [11]. A frequency tripled YAGlaser sheet (355 nm) 25 mm wide is introduced in the

l l

h

EGR a

EGR

=

-

* 1

1

he

e

e

eEGR

CO ADM

CO ECH

CO ADM

CO ECH

» 1

2

2

2

2

2

( )

( )

( )

( )or

thermocouple

cylinderhead

intakeexhaust

heatexchanger regulation

system for

EGR rate

exhaust gas return line

concentration analysers

CO2

CO2, NOx, HC




654


5/17

combustion chamber in the direction of the pentroof ridge 4 mm below the spark plug. The air is seeded

with biacetyl which fluoresces when excited with theYAG laser. The biacetyl burns so that no fluorescencecan be observed in the burned gas. During the lowpressure part of the cycle, a fraction of the burned gasesremains in the combustion chamber. After the intakephase the mixture in the engine is composed of unburned gases seeded with biacetyl and residualburned gases without biacetyl. The locations where theintensity is lower reveal the presence of residual gas. Acamera collects the fluorescence image through aquartz piston via a 45¡ inclined mirror placed

underneath the combustion chamber (Fig. 2).The image is a 21 x 53.5 mm rectangle digitised to112 x 240 pixels. Image calibration is achieved usingthe fluorescence image of a homogeneous mixtureseeded at the same rate during a cycle with no residualgas (reference image). Using reference images,unburned mixture images and background images, theaveraged unburned mixture image is divided by theaverage reference image after correction from thebackground, gain of the camera and laser intensity. Thedivision by the reference also corrects for spatialimperfection of the laser profile and camera distortions.The burned gas image is finally obtained by subtractionfrom unity (justifying the "negative PLIF" designation).

The accuracy of this method strongly depends on agood choice for the reference as well as the quality of the seeding system. To guarantee of a homogeneousdistribution of the biacetyl in the intake air, a part of theair was diverted through a biacetyl seeding system. Theseeded air (approximately 1.5% biacetyl) wasintroduced far upstream of the intake valves insuring ahomogeneous distribution of the biacetyl in the intakeair. We verified the perfect spatial homogeneity. Sincethe seeding system worked on a carburetor basis, thequantity of biacetyl in the air depended on the airpressure in the seeding vessel and thereby on thepressure in the intake system. The intake pressure wasmonitored so that the pressure in the biacetyl vessel wasthe same for the reference images and the imagescontaining residual gas.

The negative PLIF method was applied, with normaloperation or with skip firing, with propane or iso-octaneas a fuel, and with or without EGR.

When running the engine normally the reference was

acquired before and after cycles with residuals. Anexample with propane as a fuel is given on Figure 3.

Figure 2

Engine description.

Figure 3

Mean intensity in the image versus cycles corresponding to a

percentage for the calculated residual fraction (110 mbar

pressure blockage, normal firing).

For this condition an exhaust blockage was applied toenhance residual gas concentration. 90 reference

images (high signal) were acquired before and after180 images with residuals (lower signal). The images

0

BackgroundBackground

cycles with combustion

calculated residuals fraction 26% M e a n i n t e n s i t y i n t h e i m a g e

temperaturevariation

seeding rate variation

cycles without combustion

Cycle number

0

40

80

120

160

200 300 400100

Y

x

fused silicaside window

mirror

fused silicapiston head

intakeexhaust

laser sheettripled YAG

IntensifiedCCD

iso-octane or airseeded with biacetyl




655


6/17

are acquired 2 CAD prior to the spark activation angle.During the acquisition of images, the engine is not fired

and temperature decreases. This results in an increasingfluorescence signal as shown in Figure 3 (asfluorescence intensity increases when temperaturedecreases). The fluorescence intensity for imagesacquired with residuals (cycle with combustion)decreases slightly due to biacetyl level variations in theseeder. As can be seen, the seeding rate variationbetween the two sets of reference images follows thesame slope as for images acquired with residuals. As theseeding rate decreases constantly over time (less than10 units of intensity for 1000 cycles) it is easily

corrected. Only the first few images of each referenceacquisition set are taken into account for the averagedreference (to avoid the temperature problem). Theresiduals fraction is calculated using the averagedreference image and reported constant on the figure(26%).

Under skip firing operating conditions (1/5), theimages containing reference information and residualsare successively at the same crank angle. Then 3 cyclesfollowed with no combustion. The Figure 4 showsalternatively the mean intensity in the reference image

(grey squares) and in the image with residuals (blackpoints) versus the cycle number. The individualreference image preceding the image with residuals isused to calculate the absolute residual gases rate (lowerblack crosses). Only a small variation of 0.5% isnoticed, which corresponds to a relative error of 3%.

An exhaust blockage pressure was applied to test thesensitivity of the method in both operating conditions(for propane as fuel). In Figure 5, one can see that theresidual gas fraction increases with exhaust pressure,which is quite consistent, and more residual gas aredetected with normal firing. Under skip firing conditionsat 80 mbar two acquisition sequences lead to residualgas rate of 19.6% and 19.4% respectively. At 110 mbar21.0% and 21.2% values were measured. This shows agood relative repeatability of the measurements.

When iso-octane is used as fuel, measurement of themean burned gas distribution also requires theknowledge of the mean fuel concentration in addition tothe mean air concentration. The acquisition consists of recording a block of images for fuel distributionmeasurements (seeding the fuel with biacetyl, not theair) and a block of images for air distribution (seeding

the air only with the same amount of biacetyl as for thereference). In each record 60 images of background

Figure 4

Mean intensity in the image versus cycles corresponding to apercentage for the calculated residual fraction (with noexhaust blockage (skip firing 1/5).

Figure 5

Exhaust blockage pressure effect on residual gas fraction( skip firing, normal firing, enhanced tumble,volumetric efficiency 0.3, f = 1).

(no biacetyl) and 60 reference images (no combustion,air seeded with biacetyl) are acquired. For each record,corrections for laser intensity and gain of the cameraare applied. Then the average of background images,reference images and iso-octane or air images are

obtained. Subtracting the background images anddividing by the reference image lead to the final

Exhaust bloca e ressure (mm of water)

14

18

22

26

30

0

normal firing

skip firing

R e s i d u a l s g a s f r a c t i o n ( %

)

40 80 120

Cycle number

0

40

80

120

160

cycles without residuals

cycles after combustion

16.6% residuals M e a n i n t e n s i t y i n t h e i m a g e

residuals

0 100 200 300 400

Background Background




656


7/17

concentration image of fuel or air. The obtained fuelconcentration is normalised by setting the mean

intensity equal to the concentration given by the lambdaprobe.

At the equivalence ratio f, vaporised iso-octanerepresents 0.0172 f liters. When a volume of air isadmitted, 0.0172 f volume of iso-octane is injected. Sothe concentration of burned gas is given by thefollowing equation:

[BG]xy = 1 Ð Cair [air]xynorm Ð 0.0172 f Cair [fuel]xy

norm (3)

Cair represents the mean air fraction in the image

[air]xynorm the normalised distribution of air

[fuel]xynorm the normalised distribution of iso-octane

f is the equivalence ratio measured with thel Ð probe.

Under EGR conditions (example on Figure 6), somecares must be taken regarding the reference imagequality (related to seeding rate).

In the case of visualisation of residual gas withoutexternal recirculation, it is assumed that the burned gasin the real charge images is replaced by biacetyl seededair in the reference images, so that the division of the

two mean images gives the mean air fractions. This isnot true when the exhaust gases are recirculated sincethe mass of gas in the combustion chamber is increased.

Figure 6

Mean burned gas fraction versus cycles with 15% EGR (onblack corrected images, on grey reference images intensityafter correction of decreasing signal). Mean intensity

corresponds to a percentage for the calculated residualfraction.

Also, as the recirculated exhaust gases increase thepressure in the intake system, the seeding rate for the

EGR images is reduced compared to the correctedreference images which are acquired for a lower la = 0.6.For this reason, the admitted air flow rate for thereference images is adjusted, so that the volumetricefficiency la corresponds to lEGR according to Equation(2). It is verified that by adjusting la to the appropriatevalue, the admitted pressure increased to the pressuremeasured with EGR so the seeding rate is alsoappropriate.

As long as the slope of the straight line, whichconnects the first set of reference images with the

second one, corresponds to the decrease in intensitysignal for the burned gas images, the mean image of thetwo reference image sets will give a mean referenceimage for the total of the burned gas images forquantification purposes. When running the engine withEGR, between 2 and 6 minutes passes between theacquisition of the reference images and the burned gasimages to regulate the EGR rate. During this time thelevel in the biacetyl tank decreases reducing the seedingrate. To account for that, we have measured the timebetween the acquisition of the different image sets, i.e.

first set of reference images Ñburned gas imagesÑsecond set of reference images. Subsequently thedecrease in intensity signal with time was calculated inthe image treatment from the burned gas images, forwhich we have also measured the acquisition time. Thedecrease in signal intensity between the acquisition of the different sets could then be calculated andcorrespondingly be added or subtracted from thereference image sets. This is shown in Figure 6, wherea representative experiment with EGR is illustrated. Ascan be seen, this correction method improved the

reliability of the measuring method.Thus we have an optical diagnostic available for

burned gas as well as iso-octane mean distributionstudies. However, it is not possible to dissociateresidual gas from recirculated exhaust gas or toevaluate a single cycle burned gas distribution. Also itis quite difficult to quantify the accuracy of thepresented technique. Nevertheless, the measurementsfor the distribution of residual gases were repeatable.For experiments with EGR it turned out that thedependence of the seeding system on the operating

conditions presented a problem which was rectified byacquisition procedures.

Cycle number

0

40

80

120

160

M e a n i n t e n s i t y i n t h e i m a g e

cycles with combustion

cycles with no combustion

deriv corrected in gray

calculated burned gas fraction

0 100 200 300 400

Background Background




657


8/17

3 RESULTS

3.1 Residual gas

As measured and described by Galliot et al. [5], theamount of residual gas in the cylinder depends on theexhaust and intake processes. In naturally aspiratedengines, when the intake valve opens, there is a blow-down of the burned gas into the intake port, which goesback into the cylinder later on during the intake pro-cess. During the valve overlap period, there is also areverse flow of the burned gas from the exhaust portsinto the cylinder. The amount of residual gas trapped inthe cylinder therefore decreases with increasing intakepressure. With the presented technique a series of engineoperating points with propane or iso-octane injection inthe two intake ducts were performed. The imageacquisition angle was constantly 333¡CA. Differentparameters which influence residual gas concentrationwere investigated: fuel, injection configuration, flow,volumetric efficiency, and spark plug location.

3.2 Influence of fuel:

propane/iso-octane

Residual gas distributions obtained by running theengine with propane or iso-octane for the reference witha central spark plug and a volumetric efficiency la of 0.6are given in Figure 7. The average residual gas fractionfor propane (12.3%) is lower than for iso-octane (14.3%)octane (14.3%). This is explained by the fact that the

combustion process of propane and iso-octane aredifferent. This influences the pressure in the intake

system and thus the mean residual gas fraction. In bothcases the residual gases are concentrated in the middleand, particularly for the iso-octane case, on the intakeside. This distribution can be explained by the largetumbling motion on the exhaust side of the combustionchamber, being responsible for a good mixing processbetween the residual gases and the intake air, and thesmall vortex on the top of the intake side, which on thecontrary makes the mixing of the residual gases withfresh air more difficult. In previous work [4] it was alsoremarked that the mixing of fuel with air was more

difficult in the middle of the chamber. This can explainthat the distribution of residual gases for an engineoperating with propane is flatter than for iso-octanesince propane/air mixing is perfectly homogeneous.

To summarise flow and fuel distributions appear tobe important parameters for residual gas distribution.

3.3 Effect of iso-octane distribution on

residual gas

Figure 8 confirms the effect of fuel distribution on

residual gas. The normalised fuel distribution isreported on the top of the figure for injecting through 1or 2 intake ducts and the corresponding residual gasprofiles plotted on the bottom show opposite trends. Astratification in fuel of 20% produces a reversestratification of residual gas of 10%.




658

Figure 7

Residual gas distribution for: propane (left) and iso-octane(right) for a central position of the spark plug and a

volumetric efficiency la = 0.6. Intake is on the right on theimage

Figure 8

Residual gas distribution (injection for 1 or 2 intake ducts).

2 pipesinjection

-10 20

1 pipe injection

residuals

iso-octane

N o r m a l i s e

d

i s o - o c t a n e

c o n c

e n t r a t i o n

Roof ridge position (mm)

0.8

1.0

1.2

0.10

0.15

R e s i d u a l s

c o n c e n t r a t i o n

-20 0 200.10

0.15

0.20

iso-octane

propane


M e a n r e s i d u a l s c o n c e n t r a t i o n


9/17

3.4 Effect of enhanced tumble

on residual gas

The residual gas distribution results from its motionfrom intake to compression. It is interesting to visualisetheir organisation during these phases in standard andenhanced tumble configuration. This experiment wasmade with propane to avoid fuel distribution effects.Figure 9 shows burned gas distribution during intakestroke (standard on top/enhanced tumble on bottom).Images are normalised by their mean intensity forgradients to be enhanced. One can see the unburnedmixture (blue) entering the cylinder between the 2

intake ducts located on the right and mixing withresidual gas from the beginning of the intake process tothe BDC. The highest velocity in the tumbling directionis between the 2 intake valves. One can notice that themixing between unburned mixture and residual gas ismore difficult in the roof ridge direction for theenhanced tumble case. This is due to the large velocityin the perpendicular direction for this condition.However, during the compression stroke the increasedturbulence induced by the destruction of the tumbleleads to increased mixing of residual gas in the roof ridge direction.

Figure 10 depicts the mean, minimum and maximumburned gas fraction in the image during the intake andcompression stroke for both configuration. This figureshows the difference in mixing of residual gas with air

Figure 9

Burned gas distribution during intake stroke (standard ontop/enhanced tumble on bottom).

Figure 10

Burned gas fraction during intake and compression stroke(standard/enhanced tumble).

between the two flow configurations. For the referencecase, the minimum, maximum and mean burned gasfraction strongly decrease as soon as the intake valvesopen. At 190 CAD unburned gases have filled thecylinder, burned gas remains around intake valves

because of the tumble dragging the residual gas whichwere near the bottom of the intake part of thecombustion chamber. The velocity level is lowerbecause of the reverse motion of the piston. No moreunburned gas will be admitted, the piston rises and onecan see the burned gas fraction level fluctuating as it iscarried along by the flow. In the enhanced tumble case,the maximum remains at 100% during 30 CAD while ithas already decreased by 71% in the standard case.Also the minimum and mean burned gas fraction (in theimage) is higher than for the standard case. This isexplained by a more directed flow field in the tumblecase. When the piston rises, the burned gas globallyincreases showing that during the intake process,burned gas had accumulated in the bottom of thecylinder. An oscillating phenomena appears with aperiod of 70 CAD. A simple calculation based on thetime it take to a fluid particle to run over a diameterabout half of the stroke determined a mean tumblevelocity of about 13 m/s (also measured by LDV duringthe compression stroke).

The experiments was performed with iso-octane (at333 CAD) and residual profiles in the roof direction

were compared for standard and enhanced tumbleconfigurations on Figure 11. One can see that for the

100burnt gasfraction

intake valve lift

mean

enhancedtumble

CADCAD

20

40

60

80

0-90

standard

minimum

maximum

M e a n b u r n t g a s f r a c t i o n ( % )

0 90 180 270 360 0 90 180 270 360

15 25 35 188 208




659


10/17

enhanced tumble case, the residual gases are morehomogeneous because the turbulence is higher. The

residual gases are also less concentrated due to theperiodic phenomena of tumble or to the better combus-tion which leads to more pressure in the intake ducts.

3.5 Effect of spark plug location

When igniting the mixture at a position near thechamber wall, the residual gas distribution does notchange significantly. As can be seen in Figure 12 wherethe residual gas concentration images for a central andan offset position of the spark plug with a volumetric

efficiency of 0.60 are compared. The main fraction of the residual gas is again located in the centre and on theintake side.

Figure 11

Residual gas distribution with central spark plug. Standard orhigh tumble conditions.

Figure 12

Distribution of residual gas for a volumetric efficiencyla = 0.60; offset spark plug on the left, centred on the right.

However, regarding the profiles, we see that for anoffset position of the spark plug the distribution of

residual gases is more even, and the average fraction islower: 12.5% with regard to 14.3% for a centralposition. A partial explanation to this might be that theengine pressure during the outlet process, whichproportionally influences the scavenging of thecylinder, was slightly higher for the offset position, ascould be verified by the acquired pressure data.

3.6 Effect of volumetric efficiency on

residual gas

By varying the volumetric efficiency, the residual gasdistribution field for iso-octane changes significantly, ascan be seen in Figure 13 for the standard configuration.For higher volumetric efficiency the main fraction of residual gases is still concentrated in a corridor betweenthe exhaust and the intake side but clearly more towardsthe upper part of the image. This higher agglomerationof residual gases in one part of the imaged region mightbe explained by the increasing flow velocity, and gasdensity with increasing volumetric efficiency, whichenhance a possible unsymmetrical in the velocity field.

Le Coz et al. [10] and Deschamps et al. [4] have shownthat the velocity field in the combustion chamber, dueto marginal differences in the two intake ports, is farfrom being perfectly symmetrical. Table 2 shows thatthe average residual gas concentration decreases withincreasing volumetric efficiency. This is due to theincrease in pressure in the intake ducts for largercharges. Also one can see the combined influence of the volumetric efficiency and spark plug location whichincreases either intake or exhaust pressure.

Figure 13

Residual gas distribution for different volumetric efficienciesla and a central position of the spark plug.

max.fraction

min.fraction

l a = 0.6 0.66 0.71 0.75

allumage

centrŽ

14.4%12.4%

-20 0 20

0.05

0.00

0.10

0.15

M e a n r e s i d u a l s c o n c e n t r a t

i o n


centralsparkplug

offsetspark

plug

max.

min.

standard

14.4% 10.5%

M e a n r e s i d u a l s c o n c e n t r a t i o n


enhanced tumble

0.05

0.10

0.15

-20 0 20




660


11/17

For la = 0.6 less residual gases are present due to alarger exhaust pressure as already suggested. At higher

volumetric efficiency, the offset spark plug leads tohigher residual gas concentrations than in the centralspark plug case.

TABLE 2

Variation of residual gas concentration with volumetric efficiency.

Spark plug location effect

la Offset spark plug (%) Centred spark plug (%)

0.30 26.9

0.50 13.4

0.60 12.4 14.4

0.67 11.7 7.6

0.71 9.8 5.4

0.75 9.9 3.9

The investigation of the residual gas visualisation hasdemonstrated that fuel distribution, flow, spark pluglocation and volumetric efficiency have importantinfluences on their distribution. Residual gasconcentration is a result of a complex combination of intake and exhaust pressure, flow motion andcombustion processes. We verified that the amount of residual gases in the combustion chamber decreases forconfigurations which increase intake pressure ordecrease exhaust pressure: volumetric efficiencyinfluences intake pressure. The spark plug locationinfluences the combustion process then the exhaustpressure, especially at high charge. Also it was foundthat the flow motion is responsible for the residualconcentration at spark timing, especially when thetumble is enhanced as residual gas is trapped in thetumbling movement. Finally, residual gas distributionseems difficult to control because of the complexinteraction between parameters. A solution would be tominimise residual gases and replace them byrecirculated exhaust.

4 EXTERNAL EXHAUST GAS

RECIRCULATION

The burned gas distribution is now composed of

residual gases and recirculated exhaust gases, admittedwith the fresh air.

4.1 Injection by two ducts

In a first approach fuel was injected by two ducts tostudy possibilities of stratifying burned gases bychanging EGR configuration. EGR was introducedthrough two ducts with rates of 7.5 and 15%, or oneduct with a rate of 15%. Burned gas concentrationprofiles in the roof ridge direction for the different EGRrates and configurations have been compared withresidual gas (reference conditions) on Figure 14. WhenEGR is introduced through two ducts the distribution of burned gas is relatively equivalent to residual gasdistribution (0% EGR) but the mean concentration in

the image increases when the EGR rises and is alwaysgreater than the EGR rate measured by CO2 analysers.This difference is due to residual gas presence. One cansee that when recirculating exhaust gas through oneduct, burned gases are stratified with a variation of 21%in the roof ridge direction. The steepest slope is encoun-tered in the middle of the imaged plane. This corres-ponds to observations made by Deschamps et al. [4] fora horizontal fuel stratification with the same engine andflow configuration and in the same image plane.

We now know that it is possible to obtain homoge-neous or stratified burned gas distribution whether theexhaust gases are recycled through two or only oneduct respectively. We would like to study now thepossibility of enhancing the stratification factor.

Figure 15 describes the burned gas distribution fordifferent EGR rates (10.15 and 20%) introducedthrough both intake ducts. Figure 16 shows the effect of EGR rate on burned gas concentration profiles for both

Figure 14

Burned gas concentration profiles for different EGR rates and

configurations to be compared with residuals (standardconditions).

0.10

0.15

0.20

0.25

M e a n b u r n t g a s c o n c e n t r a t i o n

-20 0 20

15% EGR 1 pipe

15% EGR 2 pipes

7.5% EGR 2 pipes

0% EGR





661


12/17

Figure 15

Burned gas concentration distribution for different EGR ratesand configurations to be compared with residuals (standardconditions).

EGR configurations (reference conditions) and theeffect of spark plug location on burned gasstratification. Figure 16a and 16a' show evidence thatthe burned gas distribution becomes more homo-geneous when recirculating exhaust gas by both intakeconducts. Note that the larger amount of burned gasesis concentrated towards the side for which we haveverified a higher agglomeration of residual gases.

When recirculating EGR through one duct, for acentral position of the spark plug (Fig. 16b and 16b'),each of the three EGR rates leads to a clearstratification. The highest concentrations for all threeEGR rates are not encountered at the upper end of theimaged region, but rather at a location, whichcorresponds to +15 mm on our scale. This is morepronounced for the lowest EGR rate where residual gasconcentrations are at their highest percentage at thislocation. For 15% and 20% EGR rates the stratificationdecreases because the contribution of residual gasdecreases considerably (see Table 2 for correspondinglEGR (0.60; 0.667; 0.706; 0.750) to hEGR (0, 10, 15 and

20%). The stratification, which was about 27% forEGR rate 10% leads to 20% for 15% and 20% EGR.

For an offset spark plug residual gases were morehomogeneous and less decreased at high volumetricefficiencies (corresponding to EGR rate, on Table 2).Burned gas stratification seems to increase withincreasing EGR rates. For 10%, 15% and 20% EGRrates the stratification factor is 12.5%, 15% and 22%.The greatest slopes are again encountered in themiddle of the imaged planes, but they are less steepcompared to Figure 16b. The highest concentrations are

encountered on the side of recirculating. The differencein the profiles for a central and an offset spark plug

Figure 16

Burned gas concentration profiles for different EGR rates andconfigurations to be compared with residuals (standardconditions).

position are mainly due to differences in residual gasdistribution. This difference vanishes when the EGRrate increases.

In fact while recirculating more exhaust gases, the

contribution from residual gases decreases. This isconsistent with getting a more homogeneousdistribution of burned gas when recirculating throughboth intake ducts. Also one can compare the meanburned gas fraction in Figures 16 a, b and c andconclude that when recirculating, regardless of theconfiguration, as the EGR rate is increased the lessvariation there is in the mean burned gas fraction.

The residual gas contribution for an offset spark plugand different EGR rates through one duct are ispresented in Figure 17. This figure compares the mean

burned gas concentration measured in the image (1) tothe concentration which should be measured if residual

0.05

0.10

0.15

0.20

0.25

0.60

0.80

1.00

1.20

0.05

0.10

0.15

0.20

0.25

-20 0 20 -20 0 200.60

0.80

1.00

1.20

0.05

0.10

0.15

0.20

0.25

0.30

EGR rate

0%

10%15%

20%


0.60

0.80

1.00

1.20

1.40

a

b

c

a'

b'

c'

max.fraction

min.fraction

hEGR = 0 0.10 0.15 0.20




662


13/17

Figure 17

Comparison of measured burned gas concentration at EGRrate hEGR (--), and measured residuals (--) at equivalentvolumetric efficiency lEGR. lEGR = f (hEGR) and burned gasfraction = hEGR also reported with the calculated residualcontribution (- -) following Equation (2). Standardconditions, EGR by one intake ports, offset spark plug.

gases were not present (2) (hEGR = burned gasconcentration). The gap between (1) and (2) should bedue to presence of residual gas. The molar fraction of residual gas is easily deduced from measured burnedgas fraction xbg and EGR rate following the Equation (4)which is drawn in Figure 17.

(4)

The measured residual gas for volumetric

efficiencies corresponding to the EGR rates are alsoplotted on Figure 17. One can see that the calculatedresidual gas contribution is smaller than measuredresidual gas. One can explain this by the fact thatresidual gases are measured in a plane while thecalculation includes the entire chamber volume.Another explanation would be that EGR modifiescombustion process and can lead to less residual gases.For the experiments without EGR the intake pressurecan be estimated to be higher since the heat release andthe engine pressure, which affect the intake conditions

were higher. Therefore the mean residual gas fractionshould be lower in the EGR conditions than the one

measured at corresponding volumetric efficiencies. Therecirculation of exhaust gas scavenges the engine in the

sense that the amount of residual gases is considerablyreduced. Now we can say that it is possible to producehomogeneous burned gas distribution whenrecirculating EGR through both ducts and to stratifythem when recirculating through one duct. As the rateof recirculation is increased, the more controlled theburned gas distribution will be since the residual gasdistribution vanishes.

By enhancing the tumble a stratification of 23% inburned gas is amplified to 33% as shown on Figure 18for 15% EGR.

Figure 18

Effect of enhancing tumble on burned gas stratification.

4.2 Injection by one duct

By injecting fuel into only one of the two intakeducts when stratifying the engine charge with burnedgas, a complementary fuel stratification can beachieved. Injecting fuel in the duct opposite to therecirculation one amplifies the stratification (Fig. 19).Moreover a homogeneous distribution of burned gascan be achieved when recirculating EGR through theinjecting duct. Figure 20 shows that 3 levels of stratification are obtained 0, 7.5% and 17% respectivelywhen recirculating in the injecting duct, in both or theopposite duct.

Burned gas stratification control is achieved by acombination of favourable parameters: maximumrecirculation in one duct, injection in the opposite pipe,

and enhanced tumble. Homogeneous distribution of burned gas is obtained when recirculating exhaust gases

-20.00 0.00 20.00

0.80

1.00

1.20

1.40

standard

enhanced tumble

2 pipes

1 pipe15 % EGR


N o r m a l i s e d b u r n t g a s c o n c e n t r a t i o n

xx

res

bg EGR

EGR

=-

-

h

h1

0 10 20 30%0

10

20

30

0.40

0.50

0.60

0.70

0.80

hEGR = [bg]

calculated residualscontribution

m e a

s u r e

d b g

c o n c

e n t r a

t i o n

lEGR

l E GR

M e a n b u r n t g a s c o n c e n t r a t i o n ( % )

hEGR

measured residuals

concentration athEGR




663


14/17

Figure 19

Effect of fuel stratification on burned gas distribution.

Figure 20

Effect of EGR configuration on burned gases distribution andfraction ([bg]). Fuel is injected by one duct.

and injecting fuel in both intake pipes or whenrecirculating exhaust gases and injecting fuel in thesame intake pipe. The higher the EGR rate, the betterthe distribution of burned gases can be controlled.

4.3 Tolerance to EGR

Tolerance has been determined with a stabilitycriterion of MEP of 0.07 for all possible combinationsof EGR, injection, spark plug location and flowconfigurations at optimum spark advance. Three keyparameters emerge from tolerance results reported onFigure 21 along with NOx emissions: flow field,recirculation configuration and fuel concentration atignition location. Enhanced tumble, stratified fueldistribution (with favourable spark location) andrecirculating exhaust gas in two ducts favour

recirculating capabilities. The configuration whichmeets the 3 favourable conditions referred as B is the

Figure 21

Effect of exhaust gas and flow configurations on toleranceand NOx emissions.

most tolerant configuration (35% for stoichiometry and31.5% at an equivalence ratio of 0.8). The sameconfiguration except central spark plug (referred as A)

was also very favourable (29%). In standard flow field,the B' configuration analogous to B takes up to 24.5%and 25.7% at respective equivalence ratio 0.8 and 0.9.Configuration C with recirculation and injection inopposite intake pipes tolerates 23% while configurationD combining recirculation and injection in the sameintake pipe does not take more than 17% EGR. To ourknowledge, configuration D is the most homogeneousin term of burned gas distribution and configuration Cthe most stratified for standard flow cases. D is moredifficult to ignite than C because of a lack of oxygen.We did not find any reason for configuration B' beingmore tolerant except that homogeneous burned gasdistribution implies a better cycle to cycle stability foran ignition or flame propagation.

4.4 Emissions

Without recirculation (Fig. 22), enhancing thetumble increases NOx emissions: a high temperature atthe end of the combustion phase adds further to a lowerdilution by residual gas. Igniting combustion near thewall, the flame rapidly hit the wall reducing flame

interface and increasing thermal transfer so NOxemissions are reduced. Also we noticed that igniting in

EGR by1 pipe

standard enhanced tumble

EGR by2 pipes

EBR by1 pipe

EGR by2 pipes

EGR

EGR

EGR

EGR

EGR

10 20 30 40

Tolerance % EGR

0

200

400

600

N O x ( p p m )

A

B

B'

CD

*

*

EGR

*

*

* ***

* *

*

*

**

*

*

*

*

*

*

*

****

* **

*

*

-20 0 200.7

0.9

1.1

1.2

1.3

1.0

20.7%15% 23.5% EGR

injection

[bg]=23.8%[bg]=19%[bg]=18.4 %


EGR by the pipe opposite to injectionEGR by the pipe of injection

EGR

EGREGR

N o r m a l i s e d b u r n e d g a s c o n c e n t r a t i o n

EGR by both pipes

0.8

-20 0 200.7

0.8

0.9

1.1

1.2

1.3

1.0

EBR

EBR

Roof rid e osition mm

injection in both intake pipe

EG R

injection in the pipe opposite EGR

N o r m a l i s e d b u r n t g a s c o n c e

n t r a t i o n




664


15/17

Figure 22

Effect of exhaust gas recirculation on NOx emissions/comparison with no recirculation configurations.

a stratified zone of fuel distribution reduces NOx. Thelowest emissions are obtained when the gasoline is

stratified while igniting in rich zone. Actually in thiscase NOx are created earlier and have time torecombine before exhaust valve opens. NOx emissionsare further reduced by exhaust gas recirculation anddecrease globally with increasing recirculation rate(Fig. 21). Points are however dispersed due toconfiguration differences. Configurations with a highertolerance have not systematically lowered the NOxemissions level. For example configuration D whichonly tolerate 17% EGR has the lower emissions levels.However a low tolerance may lead to high HCemissions levels. Note that enhanced tumble decreasesNOx emissions when recirculating while it increasesthem without: this is an important effect of dilution.The better configuration in term of NOx/tolerancetrade-off is the configuration B.

To summarise, tolerance is maximised by highturbulence level, recirculating by two ducts and with astratification of gasoline. An important reduction of NOx emissions without deterioration of HC emission isexpected. Some configurations where visualisationshowed high additional residual concentrations exhibita good efficiency for NOx reduction. However the

tolerance to recirculating may lead to high HCemissions level.

Pollutants emitted by the optical cylinder are ques-tionable. The mean temperature is lower, dead volumes

are more important and absorption by oil does not exist.Therefore, we decided to test the same good configur-ations on a conventional single cylinder head with thesame geometry. Tests were made at recirculating rateaccepted by all configurations: 15% at optimum sparktiming. Configurations can be split into 2 groups.

The first one with NOx + HC emissions lower than5000 ppm is the configuration with recirculatingthrough both intake pipe. In this case the emissions arefurther reduced to less than 3000 ppm by enhancing thetumble. Configuration A and B selected in the optical

engine study are included in this group.The second group is for the recirculation through

only one duct for which stratification of gasolinereduces emissions to less than 6000 ppm. It was seenthat stratifying gasoline leads to a natural stratificationof burned gases. A high local equivalence ratio leads toa low temperature in burned gases at the maximumcrank angle pressure. Stratifying burned gases slowsdown flame propagation at the end of combustionwhich is unfavourable for HC when the fuel is homo-geneous because some fuel cannot burn at the end.

The diagram NOx/HC is drawn on Figure 23. Itconfirms that the reference configuration with no

Figure 23

NOx/HC diagram for conventional single cylinder. Effect of introducing 15% EGR on NOx and HC emissions.

HC (ppm)

1 5 % E G R b y o n e p i p e

enhanced

tumble

0 2000 4000 6000 8000

250

750

1250

1750

2250

0

500

1000

1500

2000

2500

N O x ( p p m

)

GBR

GBR

f = 1

f = 0.8

f = 1

stratified fuel ignitionon rich side

enhancedtumble

1 5 % E G R b y 2

p i p e s

basis configuration

GBR

GBR

GBR

A

Bcentralsparkplug

10 20 30 400

500

1000

1500

2000

2500

standard

enhanced tumble

offset spark plug

ignition in a stratified zone

stratified fuelignition in

a rich zone

No EGR

N O x ( p p m )

Tolerance % (EGR max)

with EGR




665


16/17

The diagram NOx/HC is drawn on Figure 23. Itconfirms that the reference configuration with no

recirculation is the worst for NOx. As found in theoptical engine, one verifies that NOx reduction isachieved by stratifying the fuel with the spark on therich side. Adding 15% exhaust gas further reducesNOx. Nevertheless this NOx reduction is alwaysaccompanied by a deterioration of HC which isamplified when recirculating into only one duct.Enhancing turbulence increases air/fuel/burned gasesmixing stability and accelerates the flame front. Thisleads to reduced HC emissions. In the case of recirculating exhaust gases through both ducts, the

enhance tumble notably compensates the HCdeterioration.

CONCLUSIONS

A method for the visualisation of burned gases inengines, on the basis of Planar Laser InducedFluorescence using biacetyl as a tracer is developed.The admitted fresh air is seeded with biacetyl andimages of the engine charge are taken, containing theburned gases for which the biacetyl burned in the pre-

ceding cycle. Hence the intensity image of thefluorescence shows a lower signal at locations wherethe residual gas are present in higher concentrations.For quantification, a reference image is used,corresponding to a homogeneous charge withoutburned gases. The technique, which allows a normaloperation of the engine (i.e., no skip-firing), gives goodqualitative results for the evaluation of burned gasfractions.

The investigation of the residual gas visualisation hasdemonstrated that fuel, fuel distribution, flow, spark

plug location and volumetric efficiencies have animportant influence on their distribution. Residual gasconcentration is a result of a complex combination of intake and exhaust pressure, flow motions andcombustion process. We verified that the amount of residual gases by the combustion chamber decreases forconfigurations increasing intake pressure or decreasingexhaust pressure: volumetric efficiency influencesintake pressure. The spark plug location influences thecombustion process, and then the exhaust pressureespecially at high charge. It was also found that theflow motion is responsible for residual concentration at

spark timing, especially when the tumble is enhancedas residual gases trapped in the tumbling movement.

Finally residual gas distribution seems difficult tocontrol because of the complexity of interaction

between parameters. A solution would be to reduceresidual gas and replace them by recirculated exhaustgases.

Burned gas stratification control is achieved by acombination of favourable parameter: maximumrecirculation in one duct, injection of fuel in theopposite duct, and enhanced tumble. Homogeneousdistribution of burned gas is obtained whenrecirculating exhaust gas and injecting fuel in bothintake ducts or when recirculating exhaust gases andinjecting fuel in the same intake duct. The higher the

EGR rate, the better the distribution of burned gasescan be controlled.

Tolerance to recirculation is maximised through highturbulence level, recirculating through two intake portsand stratification of gasoline. An important reduction of NOx emissions without deterioration of HC emission isexpected. Some configurations where visualisationshowed high additional residual concentrations exhibita good efficiency for NOx reduction. However,tolerance to recirculation may lead to high HCemissions level.

Our conclusions have been confirmed bymeasurements in a conventional single cylinder headwith the same geometry.

ACKNOWLEDGEMENTS

This work was sponsored by the Groupement Scientifique Moteur (Peugeot SA, Renault SA and IFP)and the Ministre fran•ais de la Recherche et del'Industrie Ñ ƒtude VPE E5 1-GSM n¡ 4. The authorswish to acknowledge V. Ricordeau for his technical

support, B. Cousyn and A. Floch for fruitfuldiscussions.

REFERENCES

1 Toda T., H. Nohira and K. Kobashi (1976), Evaluation of Burned Gas Ratio (BGR) as a predominant factor to NOx.SAE Paper 760765.

2 Kijota Y., T. Akishino and H. Ando (1992), Concept of leancombustion by Barrel-stratification. SAE Paper 920678.

3 Kuwahara K., T. Watanabe, J. Takemura, S. Omori, T. Kumeand H. Ando (1994), Optimization of in-cylinder flow and

mixing for a center-spark four valve leanburn engineemploying the concept of barrel-stratifcation. SAE Paper940986.




666


17/17

4 Deschamps B., R. Snyder and T. Baritaud (1994), Effect of flow and gasoline stratification on combustion in a 4-ValveSI engine. SAE Paper 941993.

5 Galliot F., W.K. Cheng, C.O. Cheng, M. Sztenderowicz, J.B.Heywood and N. Collins (1990), In-Cylinder measurementsof residual gas concentration in a spark ignition engine. SAE Paper 900485.

6 J. Stokes, T.H. Lake, M.J. Christie and I. Denbratt (1994),Improving the NOx/fuel economy trade-off for gasolineengines with the CCVS combustion system. SAE Paper940482.

7 Tabata M., T. Yamamoto and T. Fukube (1995), Improvingthe NOx/fuel economy for mixture injected SI engineassociated with EGR. SAE Paper 950684,

8 Lebel M. and. J.M. Cottereau (1992), Study of the effect of residual gas fraction on combustion in a SI engine usingsimultaneous CARS measurements of temperature and CO

2concentration. SAE Paper 922388.9 Johansson B., H. Neij, G. Juhlin and M. AldŽn (1995),

Residual gas visualization with laser induced fluorescence.SAE Paper 952463.

10 Le Coz J.F., S. Henriot and P. Pinchon (1990), Anexperimental and computational analysis of the flow field ina four-valve spark ignition engine - Focus on cycle-resolvedturbulence. SAE Paper 900056.

11 Baritaud T. and T. Heinze (1992), Gasoline distributionmeasurements with PLIF in a SI Engine. SAE Paper 922355.

Final manuscript received in October 1997



V O L 5 2 N ° 6 N O V EM B RE D É C E M B RE 1 9 9 7

Documents

Deschamps v52n6