sae-Paper-2013-01-2791 Investigation of Flowfield Pattern in a GDI Engine at different speeds Using Numerical Techniques

ABSTRACTNumerical simulations were performed in a four stroke pent-roof SI engine under cold flow conditions at three differentspeeds (1000 rpm, 2000 rpm and 3000 rpm) to investigate theflow characteristics like swirl, tumble, turbulence, massinducted and the formation of liquid film during cold startcondition. The suction and compression strokes weresimulated at the three speeds mentioned. Results show thatswirl and tumble are increased from 1000 rpm to 2000 rpmwhereas the increase is very low between the speeds 2000rpm and 3000 rpm. The computational model was validatedagainst the experimental data available in the literature. Thereis a slight increase in the mass of air inducted as the speedgoes up. Further investigation with liquid film modellingreveal, presence of liquid film mass on the cylinder andintake region. Around 34% of fuel quantity converted to filmmass in the port wall region of intake while using port fuelinjection (PFI) at the entry of Siamese port. The piston regionhas around 8% of fuel as liquid film at compression TDC.Later this may lead to soot and Hydrocarbon emission. Fromthese results CFD code can be confidently used to furtherinvestigation like port orientation, manifold, spray, film andmixture formation and combustion.

KEYWORDSSwirl, Tumble, GDI, Turbulence, Mass Inducted, Liquid FilmINTRODUCTIONAs the new Euro 6 regulations are to be introduced in themarket, the pressure is ever more increasing on the enginemanufacturers to produce cleaner engines with enhanced

performance. As the modern PFI engines are already refineda further fuel consumption reduction is possible by newengine design like combination of good features of Dieselengine with SI engine resulting in GDI engine. The conceptof GDI though very old has not been popular due to thedifficulty in controlling the injection of fuel for varying loads.Due to the recent advances in electronics and control systemthe attention has been focused again on GDI engines. Theintake generated air motion is important in the SI engine. It isespecially of paramount importance in the case of GDIengines, where different flow field generation mechanismsare utilized in the engine[1,2]. The cyclic variation of swirlmotion is one of the factors affecting the combustion qualityin GDI engines[2]. This cyclic variation persists even thoughthe velocity fluctuations become homogeneous during thelate stages of combustion[3]. The tumble motion especiallyplays a greater role in GDI engines. They help in mixturepreparation and transportation. However too much tumble atthe time of combustion can destroy mixture distribution andthe burning rate will be decreased[9]. The cone angle of thefuel injector also affects the spray penetration inside theengine. If the spray angle is reduced the rate of evaporation isreduced[5]. It is due to the denser travel of fuel particles andlesser chance of fuel coming in contact with air. In order toimprove upon this phenomenon as the cone angle of anyinjector is more or less fixed, multiple injection strategieswere tried. The in-cylinder gas density also affects the spraypenetration, spray structure and hence evaporation.[4,6]. It isalso reported in literatures that the timing of fuel injectiontiming and spark timing strongly influences the particulateemission from GDI engines[7, 8]. There are different mixture

Investigation of Flow Field Pattern in a GDIEngine at Different Speeds using NumericalTechniques

2013-01-2791Published

11/27/2013

Ramesh P and James Gunasekaran EAnnamalai University

Copyright 2013 SAE Internationaldoi:10.4271/2013-01-2791

Downloaded from SAE International by Ramesh P, Wednesday, November 27, 2013 10:49:08 AM

preparation strategies for the direct injection engines, likespray guided, air guided and spray guided[9, 10, 11, 12, 13].High pressure multiple injections were also investigated forthe GDI engines and they help in smoother combustion andlesser particulate emission compared to single pulseinjection[14, 15]. The formation mechanism of liquid filmwere studied by various authors and they reported that thefilm formation on the wall is due to stick and spread[16, 17].An attempt has been made in this study to investigate theflow field pattern of a commercially available SI engine forits potential for GDI engine.

GEOMETRICAL DETAILS ANDDESIGNFor the computational analysis and physical validation thedimensions of a commercial SI engine whose geometricaldetails are available in the literature [Kim et al, 16] is takenand a model is created in a commercial geometrical modelerCATIA. The created model is exported to STARCD for thecreation of mesh and application of physical parameters andboundary conditions. Table 1 shows the geometrical detailsof the engine and details of fuel injection.

Table 1. Engine Details and fuel injection

The details and features of the modeled engine are shown inFig.1. Fig. 2 shows the discretized domain of the engine. Inorder to understand the flow characteristics inside the engineat different speeds simulations were conducted 1000 rpm,2000 rpm and 3000 rpm. The Siamese intake port is at the leftand the exhaust port is at the right. The flow towards theexhaust port is assumed to be positive and towards the intakeport is assumed to be negative.

Fig.1. Draft of the Engine Model

(a). Isometric View of the Engine depicting the intakeand exhaust Valves

(b). View through the sectional plane passing throughintake and exhaust Valves

Fig. 2. (a) and (b) - Discretized domain of the Engine

Downloaded from SAE International by Ramesh P, Wednesday, November 27, 2013 10:49:08 AM

VALIDATIONThe model is validated by comparing the calculated CFDresults with the experimental results of Kim et.al. Themeasurement points as done in the experiments (Kim et. al.)are shown in Figure.1. Measurements were done using asingle component LDV along the axial direction (V- line) andon a line which is parallel to both the intake ports and passingthrough a point which is 13 mm below TDC which is denotedas H-line. The spacing between the measurement points is5mm. These locations are clearly illustrated in Fig-1.

Mean Velocity Along the V-LineAs the intake ports are at the left of the cylinder axis the flowof air towards the exhaust side is assigned a positive value (u)and the one which is flowing towards the intake port isassigned a negative (u) value. The x axis is parallel to theports and measurements were done at 60, 120, 180, 240,300, and 360 crank angle (CA) aTDC. Fig. 3 shows thecomparison of CFD results with that of the experimental(Kim et. al.) results. The comparison pertains to themeasurement along the cylinder axis as shown in Fig.1. TheValidation is between the experimental and CFDmeasurement at 1000 rpm. The chart also shows the velocitymeasurement along the same measurement points at higherspeeds 2000 rpm and 3000 rpm. The CFD results arereasonably in good agreement with the experimental data. At60 CA the tumble centre is approximately at 5mm belowTDC. This crank angle the rotating vortices have notdeveloped vigorously yet. This can be understood by noticingthe low values of the velocity (towards the intake) near thepiston crown. At higher speeds at the same crank angle thetumbling vortices have strongly evolved at the same crankangle with the negative velocity reaching a value as high as18 m/s near the piston crown. At 120 CA the predictedvalues along the axis closely matches the experimental value.At 240 and 300 CA also the predicted value closely followsthe trend measured in experiment.

(a). 60 degree CAFig.3.

(b). 120 Degree CA

(c). 180 Degree CA

(d). 240 Degree CAFig.3. (cont.) (a) to (f) - Comparison of Experimental

(Kim et al) and CFD Results at a speed of 1000 rpm andthe Predicted Values at higher Speeds 2000 rpm and

3000 rpm along the V - Line

Downloaded from SAE International by Ramesh P, Wednesday, November 27, 2013 10:49:08 AM

(e). 300 Degree CA

(f). 360 Degree CAFig.3. (cont.) (a) to (f) - Comparison of Experimental

(Kim et al) and CFD Results at a speed of 1000 rpm andthe Predicted Values at higher Speeds 2000 rpm and

3000 rpm along the V - Line

Mean Velocity Along the H-LineFig.4.shows the comparison of mean velocity along the H-line and experimental value. Here also the flow away fromthe intake port and towards exhaust port is treated as positiveand the one flowing from the exhaust port towards the intakeis treated as negative. Except for 60CA the other timingshows a reasonable good agreement with the experimentalvalue. This may be due to insufficient data available for thevalve lift profile and port orientation. At 60 CA along the H-line the velocity shows a negative direction at a distance of 5mm in both directions. At this crank angle the piston is nearthe H-line and descending downwards. This is in clearconformity with the earlier result along the vertical line (V-line). This point to the fact that the incoming air is organizedalong a clockwise rotating tumbling vortex. At 180CA it canbe observed that the vortex has almost stabilized and the CFDresults follow the experimental trend.

(a). 60 Degree CA

(b). 120 Degree CA

(c). 180 Degree CA

(d). 240 Degree CAFig.4. (a) to (d) - Comparison of Experimental (Kim et

al) and CFD Results at a speed of 1000 rpm and thePredicted Values at higher Speeds 2000 rpm and 3000

rpm along the H-Line

Downloaded from SAE International by Ramesh P, Wednesday, November 27, 2013 10:49:08 AM

Fluctuating Velocity Along the V-LineFig. 4 shows the comparison between the experimentalfluctuating velocity and the simulated result along the V- line.The simulated results are in good agreement with theexperimental data at 1000 rpm. At higher speeds thefluctuating velocity is showing a higher value. Thefluctuation is maximum during the valve opening period andcontinuously decreases well in to the middle of thecompression stroke. After this the fluctuating velocity slightlyincreases indicating that the destruction of tumbling vortices.The destruction of tumbling vortices increases turbulenceresulting in a higher velocity as evidenced by Fig. 5. Hencethe higher tumble motion helps in the generation ofturbulence.

Fig. 5. Comparison of Temporal Variation ofFluctuating Velocity along the V- Line with

Experimental Data (Kim etal)

RESULTS AND DISCUSSIONThe main advantage of present port inducted engine is theprecise control of fuel quantity on cycle basis. But they havecold start emission startup. The formations of the film are themain reason for this and are influenced by flow dynamics.The following section discusses the flow characteristics andfilm formation.

Effect of Speed on Swirl and TumbleThe GDI engines are normally designed to be tumbleoriented. The pent roof combustion chamber and theorientation of the intake port geometry helps in the generationof tumbling motion inside the engine cylinder. Fig. 6 showsthe temporal variation of swirl and tumble for the threespeeds mentioned earlier.

The rotation of vortices about the axis Y (Try) is called asnormal tumble about y coordinate and the tumbling vorticesabout the axis X is denoted as cross tumble (Trx) and therotating motion about the cylinder axis Z is called as swirl.Tumble motion inside the engine cylinder can be divided intothree phases as generation, stabilization and destruction. The

generation phase occurs usually during the intake whichproceeds up to 200 CA. The stabilization and spin up phaseoccur due to the upward moving piston. Because of themoving piston the spin up phase enhances the tumble motionagain up to 300 degree. The tumble destruction phase resultsin increased turbulence. This can be confirmed with referenceto Fig 6 where there is a slight increase in the turbulence levelstarting from 250CA and lasts up to 310CA. The Crosstumble denoted by Trx is almost near zero level in thisparticular geometry considered. The Swirl about the cylinderaxis denoted by Srz increases up to 30CA and begins to fadeout and eventually becomes zero at a crank angle of 70CA.From this crank angle it begins to change direction androtates in opposite direction and reaches a maximum value of0.93 around 120CA aTDC. Regarding the effect of speed thetumble is increased as high as two folds between the speeds1000 and 2000 rpm, whereas the increment in tumblebetween 2000 and 3000 rpm is only marginal. This may beattributed to the poor strength of the intake generatedtumbling and swirling vortices which are not sustained duringthe stabilization and spin up phase.

Fig. 6. Temporal variation of Swirl and Tumble insidethe Engine Cylinder for Three Speeds

Effect of Speed on the Quantity of AirInductedThe speed of any GDI engine is controlled by quality ratherthan by quantity, that means the throttle controlled speedadjustment can be altogether eliminated with qualitygoverning. Fig. 7 compares the effect of speed on the quantityof air inducted. It is evident that the accumulated masssteadily increases until 180CA and the mass decreasesslightly until the valve closure period around 45CA afterBDC. The accumulated mass is more at higher speed. This isdue to the inducted air mass comes at higher velocity in to thecylinder and this inertia of the incoming air is more than theinertia of the air mass moving upwards near the piston crown.

Downloaded from SAE International by Ramesh P, Wednesday, November 27, 2013 10:49:08 AM

Fig. 7. Effect of Speed on the Temporal Variation of AirInducted into the Engine Cylinder

In Fig 8(a) at 60CA smaller clockwise rotating vorticesbegin to form. At this stage the incoming air through theintake valves on the exhaust side and near the cylinder linerside interact. By the time of 120CA more air is flowingalong the cylinder liner towards the piston. At 180CA thevortices are well organized and there is a central region wherea low velocity region is prevailing and so is the case at240CA where the tumble motion can be clearly seen. At300CA the tumble motion can still be observed but at areduced strength. From this point the vortices are beingdestroyed and by the time the piston reaches TDC it iscompletely destroyed. The effect of this is increasedturbulence. This is important in SI engines as this turbulenceenhances flame propagation. It is also important where thisturbulence is prevailing inside the combustion chamber.

(a). 60 Degree CAFig.9.

(b). 120 Degree CA

(c). 180 Degree CA

(d). 240 Degree CAFig.9. (cont.) (a) to (f) - Comparison of Flow Field alonga Sectional Plane Passing through Intake and Exhaust

Valves at Different Crank Angles

Downloaded from SAE International by Ramesh P, Wednesday, November 27, 2013 10:49:08 AM

(e). 300 Degree CA

(f). 360 Degree CAFig.9. (cont.) (a) to (f) - Comparison of Flow Field alonga Sectional Plane Passing through Intake and Exhaust

Valves at Different Crank Angles

Liquid Film MassAlso in this investigation, the accumulation of liquid filmmass to the different regions of cylinder (dome, liner andpiston) mand intake system (valve face, valve stem and portwall) with PFI as shown in fig. 1 for motorized cold startcondition at three speeds. Fig. 10 shows the characteristics ofliquid mass formed at different parts of the engine.

(a). DomeFig. 10.

(b). Piston

(c). Cylinder liner

(d). Intake valve faceFig. 10. (cont.) (a) to (f) - Characteristics of Liquid Film

Mass over the region of cylinder and intake system.

Downloaded from SAE International by Ramesh P, Wednesday, November 27, 2013 10:49:08 AM

(e). Intake port wall

(f). Intake valve stemFig. 10. (cont.) (a) to (f) - Characteristics of Liquid Film

Mass over the region of cylinder and intake system.

Fig. 11. 3D view of Liquid film formation over the regionof port wall

The rate of accumulation of liquid mass is more at lowerspeeds compared to that of at higher speeds. The maximumquantity is accumulated before TDC during compressionstroke. The 35 percentage of fuel mass quantity is collected inthe region of port wall of intake system as liquid filmwhereas at other regions the collected mass is less than 8percentage of fuel mass quantity. Fig. 11 shows the 3Drepresentation of liquid film formation over the engine portregion. This is due to the cold spray directly impinging on the

intake port walls and the interaction lead to the formation ofliquid film in this region.

CONCLUSIONComputational Fluid dynamic analysis on a SI engine wasconducted at motoring conditions at different speeds. It isvalidated against the experimental data. Swirl and normaltumble plays are dominant role whereas the cross tumble isvery low. These organized vortices enhance the turbulenceinside the combustion chamber during the late stages ofcompression stroke. The turbulence is more at higher speeds.The port injection at cold start condition results inaccumulation of nearly 35 percentage of injected fuel mass(45.873 mg) in the intake port region itself. This fuel filmpersists until the compression TDC.

Due to the accumulation of film mass in the port region, theengine is starved of fuel vapour in the first cycle leading tomisfire. In the subsequent cycles the film mass acts as asource point for further accumulation of the injected fuel.Due to this cold start emissions will be more during the firstfew cycles. Hence suitable measures like heated port duringinitial stages may reduce the possibility of emission for theport injection.

REFERENCES1. Harada, J., Tomita, T., Mizuno, H., Mashiki, Z. et al.,Development of Direct Injection Gasoline Engine, SAETechnical Paper 970540, 1997, doi:10.4271/970540.2. Li, Y., Zhao, H., Peng, Z., and Ladommatos, N., Analysisof Tumble and Swirl Motions in a Four-Valve SI Engine,SAE Technical Paper 2001-01-3555, 2001, doi:10.4271/2001-01-3555.3. Bergin, M. and Reitz, R., Effect of Flowfield Non-Uniformities on Emissions Predictions in HSDI Engines,SAE Technical Paper 2011-01-0821, 2011, doi:10.4271/2011-01-0821.4. Hutchison, A. and Wicker, R., DISI Fuel Spray EffectsDuring Simulated Injection Timing, SAE Technical Paper2001-01-2026, 2001, doi:10.4271/2001-01-2026.5. Nishidaa K, Jiangping Tinab, Sumotoa Y, Longb W, SatocK, Yamakawa M An experimental and numerical study onsprays injected from two-hole nozzles for DISI enginesProceedings of the combustion Institute Volume 88, Issue 9,September 2009, Pages 1634 to 16426. Karaiskos, E., Arcoumanis, C., and Hale, T., Flow andSpray Investigation in Direct Injection Gasoline Engines,SAE Technical Paper 2002-01-0832, 2002, doi:10.4271/2002-01-0832.7. Maricq, M., Podsiadlik, D., Brehob, D., and Haghgooie,M., Particulate Emissions from a Direct-Injection Spark-Ignition (DISI) Engine, SAE Technical Paper 1999-01-1530,1999, doi:10.4271/1999-01-1530.8. Davy, M., Williams, P., and Anderson, R., Effects ofFuel Injection Pressure in an Optically-Accessed DISI Engine

Downloaded from SAE International by Ramesh P, Wednesday, November 27, 2013 10:49:08 AM

with Side-Mounted Fuel Injector, SAE Technical Paper2001-01-1975, 2001, doi:10.4271/2001-01-1975.9. Kim, S., Kim, Y., and Lee, J., Analysis of the In-CylinderFlow, Mixture Formation and Combustion Processes in aSpray-Guided GDI Engine, SAE Technical Paper2008-01-0142, 2008, doi:10.4271/2008-01-0142.10. Piock, W., Weyand, P., Wolf, E., and Heise, V., IgnitionSystems for Spray-Guided Stratified Combustion, SAE Int.J. Engines 3(1):389-401, 2010, doi:10.4271/2010-01-0598.11. Yamashita, H., Seto, M., Ota, N., Murakami, Y. et al.,Spray Guided DISI Using Side Mounted Multi-HoleInjector, SAE Technical Paper 2007-01-1413, 2007, doi:10.4271/2007-01-1413.12. Fan, Q., Hu, Z., Deng, J., Li, L. et al., Stratified MixtureFormation and Combustion Process for Wall-guidedStratified-charge DISI Engines with Different Piston Bowlsby Simulation, SAE Technical Paper 2010-01-0595, 2010,doi:10.4271/2010-01-0595.13. Suh, E. and Rutland, C., Numerical Study of Fuel/AirMixture Preparation in a GDI Engine, SAE Technical Paper1999-01-3657, 1999, doi:10.4271/1999-01-3657.14. Serras-Pereira, J., Aleiferis, P., Richardson, D., andWallace, S., Mixture Preparation and CombustionVariability in a Spray-Guided DISI Engine, SAE TechnicalPaper 2007-01-4033, 2007, doi:10.4271/2007-01-4033.15. Katashiba, H., Honda, T., Kawamoto, M., Sumida, M. etal., Improvement of Center Injection Spray Guided DISIPerformance, SAE Technical Paper 2006-01-1001, 2006,doi:10.4271/2006-01-1001.16. Reitz, R. and Diwakar, R., Effect of Drop Breakup onFuel Sprays, SAE Technical Paper 860469, 1986, doi:10.4271/860469.17. Reitz, R. and Diwakar, R., Structure of High-PressureFuel Sprays, SAE Technical Paper 870598, 1987, doi:10.4271/870598.

The Engineering Meetings Board has approved this paper for publication. It hassuccessfully completed SAE's peer review process under the supervision of the sessionorganizer. This process requires a minimum of three (3) reviews by industry experts.All rights reserved. No part of this publication may be reproduced, stored in aretrieval system, or transmitted, in any form or by any means, electronic, mechanical,photocopying, recording, or otherwise, without the prior written permission of SAE.ISSN 0148-7191

Positions and opinions advanced in this paper are those of the author(s) and notnecessarily those of SAE. The author is solely responsible for the content of the paper.SAE Customer Service:Tel: 877-606-7323 (inside USA and Canada)Tel: 724-776-4970 (outside USA)Fax: 724-776-0790Email: [email protected] Web Address: http://www.sae.orgPrinted in USA

Downloaded from SAE International by Ramesh P, Wednesday, November 27, 2013 10:49:08 AM