2014-02-20 Effects of EGR, Compression Ratio and Boost Pressure on Cyclic Variation of PFI Gasoline Engine at WOT Operation

Seediscussions,stats,andauthorprofilesforthispublicationat:http://www.researchgate.net/publication/274035012

EffectsofEGR,compressionratioandboostpressureoncyclicvariationofPFIgasolineengineatWOToperation

ARTICLEinAPPLIEDTHERMALENGINEERING·MARCH2014

ImpactFactor:2.62·DOI:10.1016/j.applthermaleng.2013.11.013

CITATIONS

2

6AUTHORS,INCLUDING:

HaiqiaoWei

TianjinUniversity

44PUBLICATIONS184CITATIONS

SEEPROFILE

Availablefrom:HaiqiaoWei

Retrievedon:26August2015

lable at ScienceDirect

Applied Thermal Engineering 64 (2014) 491e498

Contents lists avai

Applied Thermal Engineering

journal homepage: www.elsevier .com/locate/apthermeng

Effects of EGR, compression ratio and boost pressure on cyclicvariation of PFI gasoline engine at WOT operation

Mingzhang Pan, Gequn Shu, Haiqiao Wei*, Tianyu Zhu, Youcai Liang, Changwen LiuState Key Laboratory of Engines, Tianjin University, 92 Weijin Road, Tianjin 300072, China

h i g h l i g h t s

� The effects of EGR, compression ratio and boost pressure on cyclic variation was investigated by means of both experiment and simulation.� Cyclic variation will increase with increasing EGR ratio.� It is feasible to increase compression ratio and intake air pressure mitigate the increasing cyclic variation caused by a higher EGR ratio.

a r t i c l e i n f o

Article history:Received 8 June 2013Accepted 6 November 2013Available online 27 November 2013

Keywords:Cyclic variationEGRCompression ratioBoost pressureComputational fluid dynamics (CFD)

* Corresponding author. Tel./fax: þ86 022 2789128E-mail address: [email protected] (H. Wei).

1359-4311/$ e see front matter � 2013 Published byhttp://dx.doi.org/10.1016/j.applthermaleng.2013.11.013

a b s t r a c t

To reduce the cyclic variation is an effective way to improve the fuel economy and combustion of sparkignition engines. In this paper, the effect of exhaust gas recirculation (EGR), compression ratio and boostpressure on cyclic variation was investigated by means of both experiments and simulation for a port fuelinjection (PFI) engine. The mechanisms of these effects were analyzed by computational fluid dynamics(CFD) tools. A Ricardo E6 PFI engine was used as the experimental prototype. The experiments wereperformed with six EGR ratios, three compression ratios and two boost pressures for a given equivalenceratios. The experiments were operated with the same parameters as that of the simulation. Resultsshowed that the simulation work agree well with the experiment. Both simulation and experiment re-sults showed that an increase of EGR ratio will increase cyclic variation for its lower laminar flame speed.However, the combination of EGR with either an appropriate compression ratio or boost pressure canachieve a relatively low cyclic variation.

� 2013 Published by Elsevier Ltd.

1. Introduction

Today, lots of technologies and strategies are being proposedto improve engine performance and to reduce emissions. Amongthem, only those measures which can meet stringent emissionregulations with comparable engine power and economy, can beadopted. EGR is emerging as a promising technology to addressthe increasing demand for pollutant standards withoutcompromising performance in spark ignition engines [1e3].Actually, EGR improves fuel economy due to decreasing pumpinglosses at part load and preventing “engine knocking” at full loadin gasoline engines [4e6]. It was found that with a stoichiometriccharge diluted with cooled EGR, knocking was shifted to a higherindicated mean effective pressure (IMEP). This was mainlyattributed to a lower in-cylinder temperature and pressure whichwas beneficial to reduce flame speed with EGR. However,

5.

Elsevier Ltd.

excessive EGR in an internal combustion engine will cause un-stable combustion and even make fire missing, resulting in sig-nificant cyclic variation, which may be a big problem to cardrivability [7,8].

With EGR, the engine can use a higher boost pressure andcompression ratio to achieve a better performance and higherthermal efficiency for a stoichiometric airefuel mixture. Nowadays,intake boost pressure is a widely used method to improve enginepower output in high-performance internal combustion (IC) engineapplications [9e11]. Meanwhile, variable compression ratio (VCR)is also a promising technique to improve thermal efficiency and fueleconomy [12e14]. Effects of EGR, compression ratio, and boostpressure on gasoline engine performance have been widelyinvestigated. However, few literature report their effects on cyclicvariation.

Cyclic variation has been known as a limit to the operatingconditions of spark ignition engines, in particular, to highly dilutedEGR conditions [15e20]. Previous investigations have shown thatabout 10% increase of the output power will be obtained with thesame fuel consumption [21e23].

Nomenclature

ATDC after top dead centerAFR airefuel ratioBDC bottom dead centerBP boost pressureBTDC before top dead center�CA degree of crank angleCFD computational fluid dynamicsCFM Coherent flamelet combustion modelCOV coefficient of variationCR compression ratio

ECFM extended coherent flame modelEGR exhaust gas recirculationEVC exhaust valve closingEVO exhaust valve openingIC internal combustionIMEP indicated mean effective pressureIVC intake valve closureIVO intake valve openingNOx oxides of nitrogenPFI port fuel injectionSI spark ignitionTDC top dead center

Fig. 1. (a) The experimental single cylinder engine test-bed. (b) Schematic diagram ofthe experimental setup.

M. Pan et al. / Applied Thermal Engineering 64 (2014) 491e498492

Several investigations have been carried out to study cyclic vari-ation in spark-ignition (SI) engines by means of experiment orsimulation. Galloni [7] studied parameters that affect cyclic variationby means of the CFD simulation method and the simulation resultswere validated by the experiment. This paper proposed that meanlaminar flame speed, turbulence intensity, and velocity are the threemain factors that affect cyclic variation. The degree of the influencewas also demonstrated by an empirical equation. M. AkifCeviz et al.[24] investigated the effect of air-fuel ratio (AFR) on cyclic variationin a FORD MVH-418 lean-burn SI engine. Results revealed that cyclicvariation would increase with the increase of AFR. NirOzdor et al.[21] reviewed about cyclic variation in a SI engine and concludedthat flame propagation depends greatly on early flame development,mixture condition, and airflow in the spark region at spark time,which greatly influence the occurrence of cyclic variation. JianjunZheng et al. [25] investigated the effect of compression ratio on cyclicvariation in a direct injection natural gas engine. They found that in-cylinder combustion would be faster with the increase of compres-sion ratio, which decreasing cyclic variation. But the flow field in-formation cannot be given by their experiment.

Simulation of engine working process can help gain insight intothe complex process that occurs in an engine. Parameters that aredifficult to measure experimentally can be estimated by usingcomputational models. Simulation results can provide helpful di-rection for experimental work, as well as optimal control. Thoughthese numerical simulations cannot provide information aboutcyclic variation, CFD analysis has been employed to calculate themean value of laminar flame speed, turbulence intensity, and flowvelocity in the ignition zone at ignition time. The influence mech-anism of EGR, compression ratio and boost pressure on cyclicvariation can be found by analyzing these key factors that affect theearly stages of flame development.

Therefore, this paper takes EGR ratio, compression ratio, andboost pressure together into consideration to study their influenceson cyclic variation through experiments and simulation methods.The coefficient of variation (COV) of IMEP is measured by experi-ment. Reasons are analyzed by simulation method with AVL-FIRECFD code.

2. Engine and experiment setup

The PFI engine used in this study is modified from a Ricardo E6single cylinder engine. It is a variable compression ratio engine,which changes compression ratio by raising and lowering the cyl-inder head, further affecting the volume of the combustion cham-ber. Fuel injection quantity and injection timing are controlled by aMOTEC ECU. The engine was coupled to a DC dynamometer tomaintain a constant speed. Cylinder pressurewas measured using aKistler 6041A water-cooled pressure transducer which was flush-

mounted at the side wall of cylinder head. Then pressure signalswere passed to a Kistler 5018 charge amplifier and finally a NI6123data acquisition. The schematic diagram of the experimental setupis shown in Fig.1(a), and the engine test-bench is shown in Fig.1(b).The specifications of the engine are listed in Table 1.

The engine is equipped with an external boost device to controlintake boost pressure and temperature. The device includes piston

Table 2Engine operating condition.

Engine speed 1000 rpmFuel injection pressure 3.5 barInjection timing 4�BTDCIgnition timing 22�BTDCBoost pressure 1 bar, 1.3 bar, 1.5 barCompression ratio 8:1, 9:1, 10:1EGR ratio 0%, 5%, 10%, 15%, 20%, 25%

M. Pan et al. / Applied Thermal Engineering 64 (2014) 491e498 493

compressor, surge tank, and pressure relief valve. The engine boostpressure is measured at the intake surge tank.

A self-designed cooled EGR control system is used in this study.The system includes EGR valve, EGR cooler, and EGR pipe. Theexhaust gas flows through the EGR cooler and EGR valve then en-ters into the intake pipe. A constant amount of coolant passesthrough the EGR cooler. The mass flow rate of the EGR coolantremains constant in all the tests. The EGR ratio is manuallycontrolled by adjusting the EGR valve.

3. Experiment details and calculation methods

All the tests presented in this paper were carried out at a con-stant engine speed of 1000 rpm and wide-open throttle. It must bementioned that the engine was operated at a steady conditionduring the entire test. The temperature in the conditioning systemwas controlled, and all datawere collected during the process whenthe water and oil temperatures were 70 �C. The injection timingwas fixed and the duration of injection was chosen to obtain astoichiometric equivalence ratio. The experiments were performedat three different intake boost pressures: 1 bar (naturally aspired),1.3 bar, and 1.5 bar. The engine was operated at the same AFR andintake air temperature. More details about the engine operatingconditions are reported in Table 2.

EGR ratio was measured with an UEGO based on oxygen sensorsin both exhaust and intake pipes after mixing. The concentration ofoxygen was measured with an ECM 5220 analyzer whosemeasuring accuracy is 0.5%. The EGR ratio is defined by theformula [26]:

EGR% ¼xO2;amb � x�O2;man

xO2;amb � x�O2;exh(1)

where xO2 ;amb represents the concentration of oxygen in the envi-ronment, x�O2 ;man represents concentration of oxygen after themixing of exhaust gas and intake fresh air, x�O2;exh

represents con-centration of oxygen in the exhaust pipe.

The in-cylinder pressure was measured by a Kistler 6041Awater-cooled cylinder pressure sensor with a sensitivity of 40 pc/bar. The crank angle signal was obtained by a crank shaft encoderthat was rigidly mounted on the front of the engine and connectedto the crank shaft with a flexible coupler. The pressure and crankangle information were recorded by a NI6123 data acquisitionsystem in every 0.5�CA. The cylinder pressure is the mean value of100 consecutive engine cycles. The IMEP values were calculatedfrom the collected cylinder pressure data.

The COVimep is a measure of the cyclic variability in the indicatedmean effective pressure. It is noted that car drivability problemswill arise when the COVimep exceeds 10% [27]. Under normal cir-cumstances, COVimep is expressed as a percentage and is defined as[22,28,29]:

Table 1Specifications of the single-cylinder engine and other controllable parameters.

Engine type Ricardo E6

Number of cylinder 1Bore � stroke 80 mm � 100 mmCompressible ratio 4.5e20 variableDisplacement 0.5 LConnecting rod length 185 mmPiston shape FlatIgnition source Spark plugBoost pressure 1e5 bar VariableEGR ratio 0e90% VariableInlet air temperature 25e400 �C Variable

COVimep ¼ sIMEP

IMEP� 100% (2)

where

IMEP ¼ 1N

XNi¼1

IMEP; simep ¼

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiPNi¼1

�IMEP� IMEP

�2N

s(3)

sIMEP is the standard deviation of IMEP for continuous engine cyclesof N. IMEP represents indicated mean effective pressure for eachcombustion cycle.

4. Numerical procedure and modeling methodology

4.1. 1D simulation and validation

Boundary and Initial conditions for the 3D engine simulationwere evaluated by means of one-dimension (1D) Boost code vari-able computations conducted separately from the 3-D code. Thewhole engine model is shown in Fig. 2. The air firstly flows into theinlet SB1, and then through the CC1 and TH1. The air will mix withthe gasoline fuel that is injected at I1 and into the cylinder of theengine. The mixture of air and gasoline will react and transfer intopower. The combusted gas will flow into CAT1 and PL and theexhaust into the atmosphere through the exhaust pipe.

Fig. 2. The whole engine model in AVL-BOOST. SB e system boundary; CL e aircleaner; R e restriction; TH e throttle; I e injector C e cylinder; CAT e catalyst; PL e

plenum; 1e13 e pipes; MP e measuring point.

0 100 200 300 400 500 600 700

-0.02

0.00

0.02

0.04

0.06

0.08

0.10

Mas

s fl

ow (

kg/s

)

Crank angle ( CA)

CR=10: 1, naturally aspirated CR=8:1, BP=1.3bar CR=8:1, BP=1.5bar CR=8:1, naturally aspirated

Fig. 4. Inlet mass flow of the calculation cases.

M. Pan et al. / Applied Thermal Engineering 64 (2014) 491e498494

In order to validate the reliability and accuracy of the 1D boostmodel, an experiment was conducted under full load to test thespeed characteristic of the engine. Fig. 3 shows comparisons be-tween calculated and measured results of torque and power. Theresults of the one-dimensional simulation are similar to that of theexperiment. The maximum deviation between the measurementand the simulation for torque is about 5%. The existed difference isdue to the reason that the calculation is operated at an ideal con-dition, in which the leakages of air and heat loss are ignored.Therefore, the model is able to give adequate initial and boundaryconditions.

Base on the accuracy and reliability of the 1D model result, itwas adopted to calculate the initial and boundary conditions of the3D moving mesh. Fig. 4 shows the inlet mass flow rate of thecalculation cases. The intake top-dead-center is 360 degree ofCrank Angle (�CA). Transient boundary condition is used toconsider the dynamic effects of airflow in the intake and exhaustports. It can be seen that the intake mass flow rate remain com-parable level with increasing compression ratio, but increases withthe increase of intake boost pressure.

4.2. CFD code and calculating meshes

Intake port, combustion chamber and exhaust port weremeshed with a Fame Engine Plus Tool using AVL FIRE. The FIREprocessor allowed solving the ensemble-averaged governingequations of the flow and the heat transfer within the

600 800 1000 1200 1400 1600 1800 2000 2200 2400 260032

33

34

35

36

Tor

que

(N*m

)

Engine speed (m/s)

Measured Calculated

600 800 1000 1200 1400 1600 1800 2000 2200 2400 26002000

3000

4000

5000

6000

7000

8000

9000

Pow

er (

W)

Engine speed (m/s)

Measured Calculated

Fig. 3. Engine torque and power at different engine speeds.

computational domain. The partial transport equations were iso-lated on the basis of a finite volume method. Intake and exhaustports were included in the 3D computational mesh in order toconsider the influence of intake and exhaust dynamic processes onsubsequent in-cylinder flow and combustion. Calculations began inintake valve opening (IVO) and ended in exhaust valve closing(EVC). The mass flow rate and temperature of the inlet boundarywere set based on the one-dimensional gas-exchange calculations.The fixed wall boundary, including intake ports, exhaust ports,valves, cylinder head, cylinder wall and the piston crown, wereconsidered as adiabatic. The exact numbers of cell in the movingmesh were 359,988 and 349,502 at intake top dead center (TDC)and exhaustbottom dead center, respectively, which was consid-ered to give adequate simulation accuracy. Fig. 5 shows a schematicof the computational mesh with a compression ratio of 8:1 and10:1. Compression ratio varied by raising or lowering the cylinderhead, which changed the volume of the main combustion chamber.

4.3. The spray model

The main reason for applying the spray model was to considerthe non-uniformity of the intake mixture concentration of port fuelinjection engines. The standard Tab model described in Ref. [30]was used for the simulation of primary and secondary atomiza-tion of the droplets. The Dukowicz model was employed to calcu-late the droplets’ heating and evaporation process based on the

Fig. 5. 3D surface mesh of the engine with compression ratio of 8:1 and 10:1.

-400 -300 -200 -100 0 100 200 300 400

0

1

2

3

4

Cyl

inde

r pr

essu

re [

MP

a]

Crank angle [ CA]

CFD simulationexperiment

Fig. 6. Comparison of calculated and measured in-cylinder pressure (n ¼ 1000 rpm,compression ratio ¼ 8:1, naturally aspirated).

Fig. 7. The calculated temperature distribution in the spark-plug region at the sparktime.

10

12

14

BP=1 bar11.37

11.8113.12

8.789.07

9.42

CR=8:1CR=9:1CR=10:1

M. Pan et al. / Applied Thermal Engineering 64 (2014) 491e498 495

assumption that the droplet temperature was consistent. A spray-wall impingement model was used to analyze the interaction be-tween spray andwall based on the assumption that the droplet wallcollision is affected by reflection, and based onWeber number [31].Also, the keε model was used to take the effect of turbulence intoconsideration while the complex gasoline fuel oxidation processwas summarized by a single step irreversible reaction.

4.4. The combustion model

In the present work the well-known coherent flame modelingapproach available in the CFD code AVL FIRE was suitably extendedto accurately predict premixed flame propagation characteristics ingasoline engine configurations. The current CFM combustionmodel was designed for premixed flames where the fuel andoxidizer are perfectly mixed before entering the computationaldomain. Therefore, it was suitable for the simulation of a port in-jection engine. The ECFM (E stands for extended) is a combinedmodel employing a model for the progress of the global reaction,and a model for the composition of the reacted and non-reactedfractions of the fluid. The model solved a transport equation forthe flame surface area density, in addition to the reaction progressvariable [32].

4.5. Model validation

The model validation process was conducted in a Ricardo E6engine, the engine specifications are in given in Table 1, at

Table 3Computational conditions for studied cases.

Engine speed 1000 rpmLoad WOTFuel supply PFIFuel used GasolineIVO 10 �CA BTDCIVC 45 �CA ATDCEVO 51 �CA BTDCEVC 11 �CA ATDCFuel injection timing 4 �CA BTDCInjection duration 90 �CASpark timing 22 �CA BTDC

compression ratio 8:1, with naturally aspirated .The engine speedwas 1000 rpm, and a stoichiometric AFR was used with wide-openthrottle, which was the same as that of experiment. As is shown inFig. 6, the calculated cylinder pressure agrees well with theexperiment result, inwhich themaximum difference is about 5%. Inaddition, predicted location of the maximum pressure difference is1.5�CA. The result is based on the assumption of uniform walltemperature 410 K for the cylinder wall, and the cylinder head andpiston top were 510 K. Moreover, the model does not take theleakage and heat transfer into consideration. As a result, there existsmall difference between the simulation and the experiment re-sults. However, the 3D moving mesh was accurate enough for theprediction for other simulation.

4.6. Modeling methodology

The cyclic variation is usually attributed to the random fluctu-ations of the equivalence ratio and flow field due to the turbulentnature of the flow in the cylinder. Previous studies of these effectshave been analyzed [7,28]. Cyclic variation is a consequence of theearly flame development conditions, and significantly influencesthe subsequent combustion phase. Flame initiation may be affectedby internal and external engine parameters as well. The flame

0 5 10 15 20 250

2

4

6

8 7.387.45

7.74

5.976.53

6.67

4.694.73

5.1

3.533.59

CO

V [

%]

EGR ratio [%]

3.71

Fig. 8. COV of IMEP at different EGR ratios and compression ratios.

0 5 10 15 20 250

2

4

6

8

10

12

14

CR=8:1

10.6

11.66

13.12

8.589.11

6.24

9.42

7.57.74

5.85.16.24

6.67

4.574.75

5.1

3.473.55C

OV

[%

]

EGR ratio [%]

BP=1 barBP=1.3 barBP=1.5 bar

3.71

Fig. 9. COV of IMEP at different EGR ratios and boost pressures.

M. Pan et al. / Applied Thermal Engineering 64 (2014) 491e498496

kernel formation depends on thermo-fluid-dynamic conditions atthe spark location. It is also affected by the spatial and temporalfluctuations of the turbulent flow field inside the cylinder.

The laminar flame speed, turbulence intensity, and mean ve-locity are important factors that influence cyclic variation. In orderto investigate the parameters of combustion and flow that affectcyclic variation, CFD analysis is carried out. The COV of IMEP wascalculated by experiment data, and CFD simulations of the mainfeatures that affect the early stages of flame development were

Fig. 10. The contours of turbulence velocity and v

used to calculate the mean value in the ignition zone at the ignitiontime. Laminar flame speed, turbulence intensity, and mean velocityat different EGR ratios, compression ratios and boost pressureswere chosen as the objective functions.

The simulation in this paper was used to explain experimentphenomenon. Therefore, only some experiment conditions werechosen in the simulation. The computational conditions are givenin Table 3. The simulations were conducted at two compressionratios of 8:1 and 10:1, with boost pressures of 1.3 bar and 1.5 bar.The engine speedwas 1000 rpm, and a stoichiometric AFRwas usedwith wide-open throttle. The fuel injection quantity and ignitiontiming were fixed in each case in order to analyze the effect of EGRon cyclic variation. The EGR ratio in each case was from 0% to 10%,15%, 20% and 25%.

As shown in Fig. 7, the temperature of the spark plug region ishigher than other regions of the cylinder. All the data obtained andtheir analyses were concentrated in this region at the spark time.Since the main features that affect the early stages of flamedevelopment in the ignition zone at ignition time influence cyclicvariation, that ignition region is chosen.

5. Results and discussion

5.1. Experiment result of COVimep

Cyclic variation in the combustion of a SI engine is a commonproblem that needed to be solved to improve performance and toprevent knocking or misfiring in some circumstances. Asmentioned before, COVimep directly affects engine drivability. IfCOVimep exceeds 10%, drivability problems in automobiles will

elocity fields (CR ¼ 8:1, naturally aspirated).

5 10 15 20 25 300.0

0.1

0.2

0.3

0.4

0.5

lam

inar

fla

me

spee

d (m

/s)

EGR ratio [%]

CR=8:1CR=10:1

(a)

5 10 15 20 25 300

1

2

3

Tur

bule

nce

velo

city

(m

/s)

EGR ratio [%]

CR=8:1CR=10:1

(b)

Fig. 11. Comparisons of the calculated mean laminar flame speed and mean turbulenceintensity at different EGR ratios, naturally aspirated.

5 10 15 20 25 300.0

0.2

0.4

0.6

Lam

inar

fla

me

spee

d (m

/s)

EGR ratio [%]

CR=8:1,naturally aspiratedBP=1.3bar, CR=8:1

BP=1.5bar, CR=8:1

Fig. 12. Comparison of the calculated mean flame speed at the spark time in the sparkregion.

M. Pan et al. / Applied Thermal Engineering 64 (2014) 491e498 497

arise. The effects of EGR, compression ratio, and boost pressure oncyclic variationwere discussed in the following. Fig. 8 illustrates theCOVimep for different compression ratios and EGR ratios. Fig. 9shows the COVimep for different boost pressures and EGR ratios.As shown in Figs. 8 and 9, COVimep will increase with the increase ofEGR ratio for the same compression ratio and boost pressure.Without EGR, COVimep does not exceed 4%, which indicates thatcyclic variation is not severe. However, cyclic variation becomeshigher when EGR ratio gradually increases. Because the mixturebecomes leaner with excess air or more dilute with exhaust gasrecycle, the magnitude of cycle-by-cycle combustion variationsincreases [27]. The COV of IMEP exceeds 10% when EGR ratio rea-ches 25%. This EGR ratio has exceeded the stable combustion limitof the engine. On the other hand, when the EGR ratio is constant,COVimep will decrease with the increase of compression ratio andboost pressure. It appears that combustion becomes more stablewhen the compression ratio and boost pressure increase withoutdetonation happening. For a higher compression ratio and boostpressure, cylinder mixture pressures and temperatures are higherat the time of spark. This will improve the growth and developmentof initial flame kernel and speed up the chemical reactions,resulting in an burning rate and a decrease in COVimep [25], Thus,

we can conclude that either compression ratio or boost pressurecan be used as an optimization parameter for the reduction of cyclicvariation. The reason for the reduction of the cyclic variation withcompression ratio and boost pressure increasing is given in thefollowing by the simulation.

5.2. Modeling result of laminar flame speed, turbulence intensityand mean velocity

Considering that cyclic variation is increased by anything thattends to slow down the early combustion process, early combustionprocess is influenced by the combustion and flow parameters suchas laminar flame speed, turbulence intensity and mean velocity.The laminar burning velocity is an intrinsic property of the mixtureand it is a function of the equivalent ratio, pressure and tempera-ture of the reactants, for a gasolineeair mixture diluted by exhaustgas recirculation, boost pressure and different compress ratio [7].Studies have shown that if laminar flame speed, turbulence in-tensity and mean velocity increase in the spark location at sparktiming, cyclic variation will decrease.

As shown in Figs. 10 and 11, the laminar flame speed at sparktiming decreases with the increase of EGR ratio, while the change ofEGR ratio has little effect on turbulence intensity andmean velocity.It can be concluded that laminar flame speed at spark timing has asignificant effect on flame propagation and the throughout thecombustion chamber, thereby affecting IMEP and the subsequentcycle to cycle variation. A conclusion can be achieved that COV ofIMEP increases with EGR ratios increasing.

On the other hand, COV maintains a low value at low EGR ratios(less than 15%), and it is slightly influenced by compression ratio.For high EGR ratios, COV will decrease significantly with the in-crease of compression ratio. This phenomenon suggests that theincrease of compression ratio can remarkably enhance combustionstability at high EGR ratios. The cylinder stroke will increase due tothe increasing compression ratio. The cylinder air temperature andpressure at the end of the compression stroke will increase as theintensity of compactness increases. Thus, the turbulence intensityand laminar flame speed (as shown in Fig. 11) increase, resulting inthe increase of the tolerated EGR ratio, which is a benefit inlowering the NOx emission of the engine.

Another method to lower cyclic variation is boosting intake airpressure. Fig. 12 shows the comparison of the calculated laminarflame speed with the boosted pressures at spark time in the spark

M. Pan et al. / Applied Thermal Engineering 64 (2014) 491e498498

region.When the boost pressure increases, the laminar flame speedbecomes higher. As the intake air density enhances with the in-crease of boost pressure, laminar flame speed increases with theincrease of boost pressure, and cyclic variation decreases.

In summary, the reason why the cyclic variation increases withthe increasing EGR ratio is that the laminar flame speed decreaseswith the increase of EGR ratio. This illustrates that a stronglyinhibited EGR combustion occurs. In addition, cyclic variation de-creases with the increasing compression ratio due to the increasedturbulence intensity at the end of the compression stroke. Cyclicvariation will decrease slightly to with the increase of boost pres-sure. The reason for this is that the intake air density enhances withthe increase of boost pressure. Therefore, the laminar flame speedincreases at spark timing in the spark region.

The conclusions in this paper provide some guidance for futureengine design to lower cyclic variation. Turbocharging and gasolinedirect injection (GDI) techniques can be used in future high per-formance engines, since the GDI technique allows engines to usehigher compression ratios, which can lead to a lower cyclic varia-tion and improve engine drivability.

6. Conclusion

The effects of EGR, compression ratio and pressure boosting oncyclic variation are analyzed on a PFI spark ignition engine at aspeed of 1000 rpm, by means of both experiment and CFD calcu-lation. Based on the data obtained, the following conclusions can bemade from the present investigation:

� Cyclic variation will increase with the increase of EGR ratio.Slight influence of the EGR ratio on cyclic variation is observedat low EGR ratios, while a significant influence of the EGR ratioon cyclic variation is demonstrated at high EGR ratios. When theEGR ratio is higher than 20%, COV of IMEP exceeds 10%, whichmay become a problem for car drivability.� Cyclic variation decreases with the increasing compressionratio for a given EGR ratio. The reason can be explained by thefact that the laminar flame speed and turbulent intensity willincrease due to a higher compression ratio. Meanwhile, a highercompression ratio can support a higher EGR ratio.� By increasing the intake air pressure, the cyclic variation willdecrease. This is due to the reason that laminar flame speed willenhance with the increase of intake air density.� The analysis described in this paper appears to provideguidelines for reducing cyclic variation by combining EGRwith ahigh compression ratio or a high intake air boost pressure. Thus,we can conclude that either compression ratio or the intake airboost pressure can be used as an optimization parameter forreducing cyclic variation and improving performance.

Acknowledgements

The authors wish to express their sincere appreciation for thefinancial support of the NNSFC (No. 51176138 and 12TJZDTJ2800).

References

[1] Khalid Al-Qurashi, Angela D. Lueking, André L. Boehman, The deconvolution ofthe thermal, dilution, and chemical effects of exhaust gas recirculation (EGR) onthe reactivity of engine and flame soot, Combust. Flame 158 (2011) 1696e1704.

[2] M. Ghazikhani, M.E. Feyz, A. Joharchi, Experimental investigation of theexhaust gas recirculation effects on irreversibility and brake specific fuel

consumption of indirect injection diesel engines, Appl. Therm. Eng. 30 (2010)1711e1718.

[3] Haiqiao Wei, Tianyu Zhu, Gequn Shu, Gasoline engine exhaust gasrecirculation-A review, Appl. Energy 99 (2012) 534e544.

[4] G.H. Abd-Alla, Using exhaust gas recirculation in internal combustion engine-a review, Energy Convers. Manag. 43 (2002) 1027e1042.

[5] Deepak Agarwal, Shrawan Kumar Singh, Avinash Kumar Agarwal, Effect ofExhaust Gas Recirculation (EGR) on performance, emission, deposits anddurability of a constant speed compression ignition engine, Appl. Energy 88(2011) 2900e2907.

[6] G. Fontana, E. Galloni, Experimental analysis of a spark-ignition engine usingexhaust gas recycle at WOT operation, Appl. Energy 87 (2010) 2187e2193.

[7] Enzo Galloni, Analyses about parameters that affect cyclic variation in a sparkignition engine, Appl. Therm. Eng. 29 (2009) 1131e1137.

[8] Yun-long Bai, Zhi Wang, Jian-xin Wang, Part-load characteristics of directinjection spark ignition engine using exhaust gas trap, Appl. Energy 87 (2010)2640e2646.

[9] Mustafa Canakci, An experimental study for the effects of boost pressure onthe performance and exhaust emissions of a DI-HCCI gasoline engine, Fuel 87(2008) 1503e1514.

[10] B. Jayashankara, V. Ganesan, Effect of fuel injection timing and intake pressureon the performance of a DI diesel engine e a parametric study using CFD,Energy Convers. Manag. 51 (2010) 1835e1848.

[11] Mustafa Canakci, Combustion characteristics of a DI-HCCI gasoline enginerunning at different boost pressures, Fuel 96 (2012) 546e555.

[12] K. Muralidharan, D. Vasudevan, Performance, emission and combustioncharacteristics of a variable compression ratio engine using methyl esters ofwaste cooking oil and diesel blends, Appl. Energy 88 (2011) 3959e3968.

[13] K. Muralidharan, D. Vasudevan, K.N. Sheeba, Performance, emission andcombustion characteristics of biodiesel fuelled variable compression ratioengine, Appl. Energy 36 (2011) 5385e5393.

[14] Martyn Robert, Benefits and challenges of variable compression ratio (VCR),SAE Pap. (2003), 2003-01-0398.

[15] T. Mantel, Three dimensional study of flame kernel formation around a sparkplug, SAE Pap. (1992). No. 920587.

[16] J.C. Keck, J.B. Heywood, G. Noske, Early flame development and burning rate inspark ignition engines and their cyclic variability, SAE Pap. (1987), 870164.

[17] K. Ishii, T. Sasaki, Y. Urata, K. Yoshida, T. Ohno, Investigation of cyclic variationof IMEP under lean burn operation in spark ignition engines, SAE Pap. (1997),972830.

[18] G.T. Kalghatgl, Spark ignition, early flame development, and cyclic variation inIC engines, SAE Pap. (1987), 870163.

[19] B. Johansson, Cycle to cycle variations in SI engines-the effects of fluid flowand gas composition in the vicinity of the spark plug on early combustion, SAEPap. (1996), 962084.

[20] Syed Ameer Basha, K. Raja Gopal, In-cylinder fluid flow, turbulence and spraymodels-a review, Renew. Sustain. Energy Rev. 13 (2009) 1620e1627.

[21] N. Ozdor, M. Dulger, E. Sher, Cyclic variability in spark ignition engines: aliterature survey, SAE Pap. (1994), 940987.

[22] P.L. Curto-Risso, A. Medina, A. Calvo Hernandez, et al., On cycle-to-cycle heatrelease variations in a simulated spark ignition heat engine, Appl. Energy 88(2011) 1557e1567.

[23] Morteza Fathi, R. Khoshbakhti Saray, M. David Checkel, The influence ofExhaust Gas Recirculation (EGR) on combustion and emissions of n-heptane/natural gas fueled Homogeneous Charge Compression Ignition (HCCI) engines,Appl. Energy 88 (2011) 4719e4724.

[24] M. AkifCeviz a, Asok K. Sen, Alp K. Küleri, et al., Engine performance, exhaustemissions, and cyclic variations in a lean-burn SI engine fueled by gasoline-hydrogen blends, Appl. Therm. Eng. 36 (2012) 314e324.

[25] Jianjun Zheng, Zuohua Huang, Jinhua, et al., Effect of compression ratio oncycle-by-cycle variations in a natural gas direct injection engine, Energy Fuels23 (2009) 5357e5366.

[26] Orian Welling, Nick Collings, UEGO based measurement of EGR rate and re-sidual gas fraction, SAE Pap. (2011), 2011-01-1289.

[27] J.B. Heywood, Internal Combustion Engine Fundamentals, McGraw-Hill, NewYork, 1988.

[28] E. Zervas, Correlations between cycle-to-cycle variations and combustionparameters of a spark ignition engine, Appl. Therm. Eng. 24 (2004) 2073e2081.

[29] Bin Huang, Erjiang Hu, Zuohua Huang, Jianjun Zheng, Bing Liu, Deming Jiang,Cycle-by-cycle variations in a spark ignition engine fueled with natural gasehydrogen blends combined with EGR, Int. J. Hydrogen Energy 34 (2009)8405e8414.

[30] J.K. Dukowicz, Quasi-steady Droplet Change in the Presence of Convection,1979. Informal Report Los Alamos Scientific Laboratory, LA7997-MS.

[31] J.D. Naber, R.D. Reitz, Modeling engine spray/wall impingement, SAE Paper,880107.

[32] ICE Physics & Chemistry, in: AVL FIRE User Manual v.2009.1, 2009.