Energy Conversion and Management 50 (2009) 3129–3139

Contents lists available at ScienceDirect

Energy Conversion and Management

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

A comparison between EGR and lean-burn strategies employed in a natural gas SIengine using a two-zone combustion model

Amr Ibrahim *, Saiful BariSustainable Energy Centre, School of Advanced Manufacturing and Mechanical Engineering, University of South Australia, Mawson Lakes SA 5095, Australia

a r t i c l e i n f o a b s t r a c t

Article history:Received 13 June 2008Received in revised form 21 January 2009Accepted 21 August 2009Available online 18 September 2009

Keywords:EGRSI engineLean burnNatural gasNO

0196-8904/$ - see front matter � 2009 Elsevier Ltd. Adoi:10.1016/j.enconman.2009.08.012

* Corresponding author. Tel.: +61 8 8302 5123; faxE-mail address: amralihi@yahoo.com (A. Ibrahim).

Exhaust gas recirculation (EGR) strategy has been recently employed in natural gas SI engines as an alter-native to lean burn technique in order to satisfy the increasingly stringent emission standards. However,the effect of EGR on some of engine performance parameters compared to lean burn is not yet quite cer-tain. In the current study, the effect of both EGR and lean burn on natural gas SI engine performance wascompared at similar operating conditions. This was achieved numerically by developing a computer sim-ulation of the four-stroke spark-ignition natural gas engine. A two-zone combustion model was devel-oped to simulate the in-cylinder conditions during combustion. A kinetic model based on the extendedZeldovich mechanism was also developed in order to predict NO emission. The combustion model wasvalidated using experimental data and a good agreement between the results was found. It was demon-strated that adding EGR to the stoichiometric inlet charge at constant inlet pressure of 130 kPa decreasedpower more rapidly than excess air; however, the power loss was recovered by increasing the inlet pres-sure from 130 kPa at zero dilution to 150 kPa at 20% EGR dilution. The engine fuel consumption increasedby 10% when 20% EGR dilution was added at inlet pressure of 150 kPa compared to using 20% air dilutionat 130 kPa. However, it was found that EGR dilution strategy is capable of producing extremely lower NOemission than lean burn technique. NO emission was reduced by about 70% when the inlet charge wasdiluted at a rate of 20% using EGR instead of excess air.

� 2009 Elsevier Ltd. All rights reserved.

1. Introduction

Natural gas is one of the cleanest economically available fuelsfor internal combustion engines. Studies around the world haveshown that engines running on natural gas emit significantly loweremissions compared to engines running on conventional fuels. Forinstance, Baldassari and coworkers [1] compared natural gas anddiesel engine emissions, they showed that SI natural gas engineemissions of THC, NOx, and PM were significantly lower than thatof the diesel fueled engine with a reduction of 67%, 98%, and 96%respectively. Compared to gasoline engine emissions, anotherstudy showed that natural gas SI engines have the potential toachieve a reduction in CO, CO2, NOx, and non methane hydrocarbonemissions of 90–97%, 25%, 35–60%, and 50–75% respectively [2].Catania and coworkers [3] showed that natural gas engine emis-sions have less impact on the global warming than gasoline emis-sions, taking the global warming potential of the methane intoaccount, the authors concluded that the natural gas fueled engineshowed a carbon dioxide equivalent reduction of 15–24% with re-spect to gasoline. In addition to its lower pollution impact, natural

ll rights reserved.

: +61 8 8302 3380.

gas is available in many parts of the world that have poor oil re-serves. Using natural gas as an alternative clean fuel will decreasethe dependence on imported oil in these countries. Furthermore,the world reserves of natural gas are larger than the petroleumoil, thus the research in utilizing natural gas in engines representsan investment for the future. Recently, environmental and eco-nomical concerns have motivated many governments to expandin natural gas infrastructure in order to be feasible to passengervehicles as well as stationary engines.

One of the natural gas engine combustion technologies, whichbegun in the early 1980s, is the ‘‘lean burn” combustion technique.This technology became dominant in gas engine industry as it ledto high engine efficiency accompanied with longer durability andlower cost. Today after almost a quarter century of continuous leanburn engine development and investment, most of the conven-tional gas engines operate with lean burn mode. According to theEngine Manufacturers Association, USA 2004, over 80% of all heavyduty stationary natural gas engines sold in the USA employ leanburn combustion technology [4]. Most of the research conductedin the lean-burn strategy basically focused on extending the max-imum burning lean limit in order to reduce NOx emissions to sat-isfy the increasing emission restrictions. That usually wasachieved by designing fast-burning combustion chambers and/or

Nomenclature

B cylinder bore, mk rate constant, m3/kmol sL distance between cylinder head and piston, mrc compression ratioSl laminar flame speed, m/sSp mean piston speed, m/sT temperature, KXb burned gas fractionZ mole fractionDh combustion angle, radDhb rapid burning angle, radDhd flame development angle, radho crank angle at start of combustion, radm kinematic viscosity, m2/s

Subscriptsb burnedu unburned

Abbreviationsbsfc brake specific fuel consumptionEGR exhaust gas recirculationMBT maximum brake torquePM particulate matterSCR selective catalytic reductionSI spark ignitionTHC total hydrocarbonTWC three way catalystWOT wide open throttle

3130 A. Ibrahim, S. Bari / Energy Conversion and Management 50 (2009) 3129–3139

employing the stratified charge concept, usually by using either acombustion pre-chamber or direct fuel injection. Recently, laserignition systems have been developed in order to ignite extremelylean fuel air mixtures, which require high ignition energy.

Currently, increasingly stringent ambient air quality standardsdemand engine emissions to be extremely low; see Table 1 [5].In order for the engine under the lean burn mode to produce lowerNOx emissions, it has to operate with a leaner mixture. In otherwords, the engine has to operate near the misfire limit to producerelatively lower NOx emissions. As the engine operates near themisfire limit, the engine stability deteriorates, the hydrocarbon(HC) and CO emissions increase, and the engine efficiency de-creases. Another way to control NOx emissions is to retard thespark timing, which also leads to a decrease in engine efficiencyand an increase in HC emissions. Therefore, it seems that any ef-forts towards a future decrease in NOx emissions would lead toan increase in HC emission and a decrease in engine thermal effi-ciency. At the end, a compromise must be made between the in-crease in NOx emissions and the decrease in engine efficiency. Ithas become obvious that it would be difficult for the conventionalgas engine operating on lean burn mode to meet the stringent fu-ture emission standards especially for NOx emissions withoutusing exhaust gas after-treatment.

The current emission reduction technologies used for the NOx

emission after-treatment in lean burn engines such as the selectivecatalytic reduction (SCR) devices are expensive and add some com-plexity to the engine use. For example, the SCR technique consistsof ammonia storage, feed, injection system and a catalyst. In thissystem, the ammonia is injected in the exhaust gases upstreamof the catalyst. In order for this system to operate properly, a cer-tain exhaust gas temperature range must be maintained [6]. Inaddition, an oxidation catalyst would also be necessary to reduceboth the HC and CO emissions.

It could be concluded that in order for the engines to meet thefuture emission standards, some alternative techniques must beinvestigated and developed. One of these alternative techniquesis the use of a three way catalyst (TWC) to reduce NOx, HC, andCO emissions. The three way catalyst technology was developed

Table 1Emission standards, g/kW h [5].

Year Standard CO HC NOx PM

1996 Euro2 4 1.1 7 0.152000 Euro3 2.1 0.66 5 0.12005 Euro4 1.5 0.46 3.5 0.022008 Euro5 1.5 0.46 2 0.02

in the 1970s for the automobile industry to reduce the gasoline en-gine emissions. The TWC is capable of reducing the three emissionsat the same time and it is much less expensive than the SCR devicesused in lean burn engines. However, in order for the TWC to oper-ate efficiently, the engine must operate at near stoichiometric fuel–air ratio (i.e. without excess air). When the engine operates nearthe stoichiometric mixture, the in-cylinder temperature increases,and consequently, the thermal stresses and the knocking tendencyincrease. This would lead to some restrictions on the use of turbo-charging, high compression ratio, and maximum brake torque(MBT) spark advance timing. As a result, the engine would operateless efficiently than a similar lean burn engine.

In order to reduce the in-cylinder temperature, an inlet chargedilution must be employed. One of the methods used to dilutethe inlet charge is to recycle some of the exhaust gases back intothe cylinder intake with the inlet mixture. This method is calledexhaust gas recirculation (EGR). Using EGR with the stoichiometricinlet mixture will lead to a decrease in the in-cylinder temperatureand a decrease in knocking tendency and could permit the engineto use turbocharging, relatively higher compression ratio, and MBTspark advance timing to achieve a relatively higher thermal effi-ciency compared to non diluted stoichiometric mixture operation.In addition, adding EGR to the inlet mixture will reduce the oxygenpartial pressure in the inlet mixture, and consequently the in-cyl-inder NOx production will decrease. Furthermore, as EGR will beadded to a stoichiometric mixture, the use of a TWC for necessaryemission reductions is also possible.

Although the use of EGR with a TWC technique is expected toeconomically produce lower emissions than lean-burn strategy,the effect of using EGR compared to lean burn on some of engineperformance parameters such as engine fuel consumption is stillnot quite certain. Some conflicting results were found in the liter-ature review regarding to this issue. For instance, Corbo andcoworkers [7] converted a heavy duty turbocharged diesel engineto work on natural gas fuel. They employed both lean burn andstoichiometric mixture with EGR and a TWC approaches after con-version and compared the engine performance and emissions forboth cases. The authors concluded that the use of both EGR andlean burn techniques led to a similar maximum thermal efficiencyof 34%. Nellen and Boulouchos [8] used a stoichiometric mixturewith cooled EGR and a TWC in a turbocharged natural gas SI engineused for cogeneration applications. The authors aimed to achievelow emissions and high efficiency by using this concept. Theauthors optimized the same engine for lean burn operation withan oxidation catalyst. They concluded that the EGR concept re-sulted in a more superior engine performance and emissions com-pared to the lean burn technique. The engine achieved a thermal

A. Ibrahim, S. Bari / Energy Conversion and Management 50 (2009) 3129–3139 3131

efficiency of 40% at bmep of 12 bar by using EGR technique com-pared to 38% at the same bmep for the lean burn operation. Saa-num and Bysveen [9] compared the use of lean burn andstoichiometric mixture with EGR strategies in a natural gas SI en-gine. They found that the maximum brake thermal efficiency washigher for lean burn operation than for EGR operation. The authorsconcluded that a penalty in engine efficiency must be acceptedwhen EGR is used as an alternative to excess air. Perhaps theseconflicting results are due to the differences in the optimized oper-ating conditions and engine design parameters used for each tech-nique. For example the lean burn technique would give higherefficiency if the spark timing was optimized for maximum braketorque (MBT) condition rather than low NOx emission condition.In the current study, both EGR and lean-burn strategies are com-pared and assessed regarding to their effect on a natural gas engineperformance and NO emission at similar operating conditions. Thiswould help to assess the use of EGR as an alternative to excess airand identify some of the operating conditions and engine designparameters that can be optimized for better engine performancewhen EGR technique is employed. For this purpose, a computersimulation of the four-stroke spark-ignition natural gas enginewas developed. A two-zone combustion model was constructedto simulate the in-cylinder conditions during combustion. The sim-ulation has been validated by experimental results and a goodagreement between the results was found.

2. Model description

The following assumptions and approximations are consideredfor simplification:

1. The contents of the cylinder are fully mixed and spatially homo-geneous in terms of composition and properties during intake,compression, expansion, and exhaust processes. Thus, the ther-modynamic properties vary only with time (or crank angle).

2. For the combustion process, two zones (each is spatially homo-geneous) are used. The two zones are the unburned and theburned zones. The two zones are separated from each otherby the flame front (see Fig. 1).

3. The intake and exhaust manifolds are assumed to be infiniteplenums containing gases at constant temperature andpressure. The exhaust pressure was set at a value of102 kPa, which is slightly higher than the atmosphericpressure.

Fig. 1. Schematic of the two-zone combustion modeling.

4. All gases are considered to be ideal gases during the enginethermodynamic cycle.

5. All crevice effects are ignored, and the blow-by is assumed to bezero.

6. The cylinder wall temperature is assumed to be constant(400 K) and the heat transfer is determined using Woschni cor-relation [10].

7. The engine is in steady state such that the thermodynamic stateat the beginning of each thermodynamic cycle (two crankshaftrevolutions) is the same as the end state of the cycle.

The flow rates in both the intake and exhaust processes weredetermined from quasi-steady one-dimensional compressible flowrate equations [10]:

_m ¼ CdARpoffiffiffiffiffiffiffiffiRTop pt

po

� �1=cffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

2cc� 1

1� pt

po

� �c�1c

" #vuut ð1Þ

where _m is the mass flow rate through intake and exhaust valves,Cd is the discharge coefficient (assumed to be 0.7), AR is the refer-ence area which was selected to be equivalent to the curtain areaas suggested by Heywood [10] (AR = p dvlv(t), where dv is valvediameter, lv(t) is valve lift as a function of time (or crank angle)),To and po are stagnation temperature and pressure upstream ofthe valve respectively, pt is static pressure down stream of thevalve, and finally c and R are specific heat ratio and gas constantof the mixture flowing through the valve respectively. For flow intothe cylinder through an intake valve, po is the intake manifold pres-sure, and pt is the cylinder pressure. For flow out of the cylinderthrough an exhaust valve, po is the cylinder pressure, and pt isthe exhaust pressure.

When the flow through the valve is choked, i.e. ptpo6

2cþ1

� � cc�1

, themass flow rate is calculated from the following equation [10]:

_m ¼ CdARpoffiffiffiffiffiffiffiffiRTop

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffic

2cþ 1

� �cþ1c�1

vuut ð2Þ

In the present model, the thermodynamic cycle simulationstarts with assumed guesses of the values of pressure and temper-ature of the contents within the cylinder at the instant the intakevalve opens. After two crankshaft revolutions (720 crank angle de-grees), the calculated values of pressure and temperature are com-pared to the initial guesses. If the calculated values are not withinan acceptable tolerance to the initial guesses, the simulation is re-peated using the final calculated values as initial guesses.

2.1. The combustion process

The following assumptions are assumed during combustion:

1. The flame front thickness is assumed to be negligible.2. The cylinder pressure is assumed to be the same in the burned

and unburned zones.3. Only the convective heat transfer mode, between the cylinder

contents and the cylinder wall, is considered.4. The heat transfer between the two zones is neglected.5. For the burned zone, ten species (CO2, H2O, CO, N2, O2, OH, NO,

H, O, and H2) are considered in chemical equilibrium duringcombustion and expansion.

6. The combustion chamber wall area in contact with the burnedgases is assumed to be proportional to the square root of theburned mass fraction to account for the greater volume filledby burned gases against the unburned volume as suggestedby Ferguson [11].

Table 2Engine specifications.

Number of cylinders 1Bore, mm 76.2Stroke, mm 111.125Capacity, cc 507Maximum speed, rpm 3000Max. cylinder pressure, bar 150Inlet valve opens, deg BTDC 9Inlet valve closes, deg ABDC 34Exhaust valve opens, deg BBDC 43Exhaust valve closes, deg ATDC 8

3132 A. Ibrahim, S. Bari / Energy Conversion and Management 50 (2009) 3129–3139

2.1.1. The thermodynamic formulationsFig. 1 is a schematic of the engine cylinder during combustion,

which shows the cylinder heat transfer from both the unburned (u)and burned (b) zones, and the piston work. The basic relations usedin the development of the present simulation are the first law ofthermodynamics, the conservation of mass law, and the ideal gaslaw.

These three principles were applied to both the unburned andburned control volumes in order to derive expressions for the time(or crank angle) derivative of the unburned and burned gas tem-peratures and volumes in addition to the cylinder pressure duringcombustion. These expressions are expressed in terms of enginedesign parameters and operating conditions. The Euler numericalsolution technique as described by Caton [12] was used to solvethe differential equations to determine the in-cylinder pressureand temperature.

2.1.2. The burning rateThe S-shaped mass fraction burned profile, the Wiebe function,

was used to determine the burning rate:

Xb ¼ 1� exp �ah� ho

Dh

� �mþ1" #

ð3Þ

where h is the crank angle, ho is the crank angle at the start of com-bustion, Dh is the total combustion duration (from Xb = 0 to Xb = 1),and a and m are adjustable parameters which fix the shape of thecurve. Actual mass fraction burned curves have been fitted with-a � 5, and m � 2 as suggested by Heywood [10]. The empirical rulefor relating the mass burning profile to crank angle at maximumbrake torque (MBT) spark timing is used in this model. With opti-mum spark timing, half the charge is burned at about 10 crank an-gle degrees after top dead centre [10]. Thus, referring to Eq. (3),putting Xb = 0.5 at h = 370 degrees enables ho to be determined ata specified combustion duration. In the current study, all resultswere obtained at the MBT spark timing condition.

2.1.3. The combustion angleA turbulent flame propagation model developed by Tabaczynski

and coworkers [13] was used by Hires and coworkers [14] to ob-tain explicit relations for the flame development angle, Dhd, andthe rapid burning angle, Dhb, as function of engine design andoperating variables:

Dhd ¼ Cð�SpmÞ1=3 LSl

� �2=3

ð4Þ

Dhb ¼ C 0BL�

� �qi

q�u

� �10=9

ð�Spm�Þ1=3 Li

S�l

� �2=3

ð5Þ

where m is the kinematic viscosity, L is the distance between cylin-der head and piston, Sl is the laminar burning speed, Sp is the meanpiston speed, q is the density, and B is the cylinder bore. The sub-script i denotes the value at ignition and the subscript u refers tothe unburned mixture, whereas the superscript (�) denotes the va-lue at cylinder conditions where Xb = 0.5. C and C

0are constants,

which depend on engine geometry. The empirical correlation ofthe laminar burning speed of the natural gas-air-EGR mixture wasdetermined from Ref. [15]:

Sl ¼ Sl0Tu

300

� �a p100

� �b

ð3:4259 x2EGR � 3:6993 xEGR þ 1:002Þ ð6Þ

Sl0 ¼ �177:43 U3 þ 340:77 U2 � 123:66 U� 0:2297 ð7Þ

a ¼ 5:75 U2 � 12:15 Uþ 7:98 ð8Þ

b ¼ �0:925 U2 þ 2 U� 1:473 ð9Þ

where Sl0 is the reference burning velocity, cm/s, Tu is the unburnedmixture temperature, K, p is the cylinder pressure, kPa, a and b arefitting coefficients, xEGR is the volumetric fraction of EGR in the un-burned mixture, and U is the equivalence ratio. The empirical lam-inar flame speed correlation was validated for equivalence ratiorange of 0.49–1.43, pressure range from 50 to 1000 kPa, EGR ratiorange from 0 to 0.43, and the tested temperature ranged from 300to 400 K [15].

Since the dynamic viscosity of hydrocarbon-air combustionproducts differs little from that of air as demonstrated by Heywood[10], the cylinder content dynamic viscosity could be expressedusing the air dynamic viscosity correlation which has the followingform [10]:

l ¼ 3:3� 10�7 T0:7 ð10Þ

where l is the dynamic viscosity in kg/ms, and T is the temperaturein K. Hence, the kinematic viscosity can be determined using thewell known correlation: m ¼ l

q .Both Eqs. (4) and (5) are used in the present model in order to

calculate the combustion duration (Dh = Dhd + Dhb) at differentoperating conditions. The combustion duration is then used todetermine the burned mass fraction using the Wiebe function.

2.2. NO formation kinetic model

The extended Zeldovich mechanism [10] is used to determinethe rate of change of NO mole fraction during combustion andexpansion processes as follows:

dZNO

dt¼

2r1 1� ZNO=ZNO;eð Þ2� �

1þ ZNO=ZNO;eÞð r1=ðr2 þ r3Þð11Þ

where:

r1 ¼ kþ1p

�RTbZO;eZN2;e ð12Þ

r2 ¼ k�2p

�RTbZNO;eZO;e ð13Þ

r3 ¼ k�3p

�RTbZNO;eZH;e ð14Þ

where Z is the mole fraction, p is the cylinder pressure, �R is the uni-versal gas constant, and Tb is the burned gas temperature. The sub-script e refers to equilibrium value. The rate constants (k), in units ofm3/kmol s, were calculated from Ref. [10]. The prescribed combus-tion model with the integration of both NO formation and knockingsub-models was previously used to optimize a natural gas engineperformance under EGR operation in two different studies; moredetails can be found in Ref. [16,17].

A. Ibrahim, S. Bari / Energy Conversion and Management 50 (2009) 3129–3139 3133

3. Experimental set-up and model validation

3.1. Experimental set-up

In order to validate the prescribed model, experimental datawas collected from a 507 cc single cylinder variable compressionratio Ricardo E6 SI engine. Table 2 shows the Ricardo engine spec-ifications. The engine used to run on petrol before it was convertedto run on natural gas. Fig. 2 shows a schematic of the experimentalset-up. The recycled exhaust gas was taken from a hole located onthe exhaust pipe with the help of a small vacuum pump. The hotexhaust gas was cooled by passing it through a water-cooled heatexchanger. A regulating valve was used to control the amount ofthe recycled exhaust gas while a 6 mm square orifice flow meterwas used to measure the flow rate of the exhaust gases recycledback to engine intake as follows:

_m ¼ CdAffiffiffiffiffiffiffiffiffiffiffiffi2qDp

pð15Þ

where _m is the mass flow rate across the orifice meter, Cd is the dis-charge coefficient which is equivalent to 0.6, A is the orifice holearea, q is the density of the gas downstream of the orifice meter,and Dp is the pressure difference across the orifice meter measuredby a u-tube manometer with an accuracy of ±0.01 kPa.

A supercharger was installed in order to provide the enginewith inlet charge at high pressure. The supercharger was drivenby an electrical motor via a belt. Air, natural gas, and cooled ex-haust gas were mixed in the supercharger intake before they weredelivered at high pressure to an intercooler. The intercooler cooleddown the air–fuel–exhaust gas mixture before it was delivered to

Fig. 2. Schematic of the e

the engine intake. A pressure gauge was used to measure the inletpressure with an accuracy of ±2 kPa. The natural gas was suppliedfrom a pipe line which supplied a continuous flow of natural gas ata pressure of slightly higher than the atmospheric pressure. Thenatural gas flow rate, and hence the air–fuel ratio, was controlledby a regulating valve.

The natural gas flow rate was measured using Dwyer RMC flowmeter with an accuracy of ±0.045 m3/h while the air flow rate wasmeasured using Alcock viscous flow meter with an accuracy of±0.09 m3/h. The pressure drop across the air flow meter, whichwas measured using an inclined water manometer, was used withmeter calibration constant to calculate the air flow as follows:

_Va ¼ ch ð16Þ

where _Va is the air flow down stream the meter cell in m3/hr, c isthe calibration constant (c = 4.57), and h is the pressure differenceacross the meter in cm water.

The engine load was applied by an electrical direct currentdynamometer, which is capable to run as a variable speed motorin order to determine the engine friction power. The engine loadwas measured with an accuracy of ±0.4 Nm by balancing the tor-que exerting on the dynamometer casing with an equivalent oppo-site torque by hanging weights on the torque arm which isattached to the dynamometer casing. A spring balance of maxi-mum force of 50 N was connected to the arm torque to providean upward force for fine balancing adjustment. The reading was ta-ken when the torque arm was horizontal and the dynamometercasing was floated. Toque was calculated as follows:

T ¼ F � R ð17Þ

xperimental set-up.

3134 A. Ibrahim, S. Bari / Energy Conversion and Management 50 (2009) 3129–3139

where F is the net force on the torque arm, and R is the radius of thetorque arm which is equivalent to 45.7 cm.

The engine speed was measured by a mechanical tachometerwith an accuracy of ±10 rpm. The in-cylinder pressure was mea-sured using water-cooled piezoelectric Kistler type 7061B pressuresensor, TDC position sensor, and shaft encoder. A charge amplifiertype Kister 5007 was used to convert the output electrical chargefrom the piezoelectric pressure sensor into DC voltage. The outputsignals from the charge amplifier, TDC position sensor, and shaftencoder were received by an analogue digital converter in orderto convert the continuous signals into digital numbers which aresuitable to be handled by a laptop via a data acquisition card asindicated in Fig. 2. Engine NO emissions were measured usingthe chemiluminescence technique with an accuracy of ±30 ppm.

Fig. 4. A comparison between measured and calculated brake power at WOT,stoichiometric fuel–air mixture, rc = 8, MBT spark timing, and different speedconditions.

Fig. 5. A comparison between measured and calculated brake specific fuelconsumption at WOT, stoichiometric fuel–air mixture, rc = 8, MBT spark timing,

3.2. Model validity

Fig. 3 shows the p–V diagram as predicted by the model at aninlet pressure of 98 kPa (wide open throttle, WOT, condition), stoi-chiometric mixture, engine speed of 1000 rpm, engine compres-sion ratio of 8, MBT spark timing, and no EGR inlet condition.The p–V relationship was integrated in order to calculate the indi-cated work and consequently the indicated power. The engine fric-tion power was determined experimentally using the motoringtest and then it was used to calculate the engine brake power. Bothbrake power and brake specific fuel consumption were calculatedat different engine speeds and compared with experimental resultsas shown in Figs. 4 and 5 respectively.

In addition, both measured and calculated in-cylinder pressuredata were compared at different operating conditions. Fig. 6 com-pares the measured and calculated in-cylinder pressure at percent-age EGR dilution in the inlet mixture of 10%, inlet pressure of113 kPa, compression ratio of 8, and 1500 rpm operating condi-tions. Fig. 7 compares measured and calculated in-cylinder pres-sure at excess air factor (k) of 1.2, atmospheric inlet conditions,compression ratio of 8, and 1500 rpm operating conditions.

Also, both measured and calculated NO emissions were com-pared at different operating conditions. Fig. 8 compares measuredand calculated NO emissions at an inlet pressure of 113 kPa, com-pression ratio of 8, 1500 rpm, and different %EGR dilution condi-tions. Fig. 9 compares measured and calculated NO emissions atatmospheric inlet condition, compression ratio of 12, 1500 rpm,and different %EGR dilution conditions.

Fig. 3. p–V diagram at inlet pressure of 98 kPa, stoichiometric mixture, rc = 8, MBTspark timing, and 1000 rpm.

and different speed conditions.

Figs. 4–9 show that there is a good agreement between theexperimental and the computer model results at various operatingconditions, which gives the confidence that the computer modelhas been well constructed.

4. Results and discussion

4.1. The effect of EGR vs. excess air on power

Fig. 10 shows the effect of the increase of excess air factor onbrake power at an inlet pressure of 130 kPa (which simulates tur-bocharged inlet pressure conditions), 1500 rpm, and MBT sparktiming condition. The excess air factor, k, is defined as the ratio be-tween the actual air–fuel-ratio to the stoichiometric air–fuel–ratio:

k ¼ ðA=FÞact

ðA=FÞstð18Þ

Similarly, Fig. 11 shows the effect of the increase of the percent-age of the mass of recycled exhaust gases, %EGR, on brake power at

Fig. 6. A comparison between measured and calculated in-cylinder pressureat.%EGR of 10%, inlet pressure of 113 kPa, rc = 8, and 1500 rpm.

Fig. 7. A comparison between measured and calculated in-cylinder pressure atk = 1.2, atmospheric inlet condition, rc = 8, and 1500 rpm.

Fig. 8. A comparison between measured and calculated NO emissions at inletpressure of 113 kPa, rc = 8, 1500 rpm, and different %EGR dilution conditions.

Fig. 9. A comparison between measured and calculated NO emissions at atmo-spheric inlet pressure, rc = 12, 1500 rpm, and different %EGR dilution conditions.

Fig. 10. Variations of brake power with excess air factor at inlet pressure of130 kPa, 1500 rpm, rc = 10, and MBT spark timing condition.

Fig. 11. Variations of brake power with EGR dilution at inlet pressure of 130 kPa,1500 rpm, rc = 10, and MBT spark timing condition.

A. Ibrahim, S. Bari / Energy Conversion and Management 50 (2009) 3129–3139 3135

Fig. 12. A comparison between the effect of both air and EGR dilution on brakepower at inlet pressure of 130 kPa, 1500 rpm, rc = 10, and MBT spark timingcondition.

Fig. 13. A comparison between the effect of both air and EGR dilution oncombustion duration at 1500 rpm, rc = 10, and MBT spark timing condition.

3136 A. Ibrahim, S. Bari / Energy Conversion and Management 50 (2009) 3129–3139

an inlet pressure of 130 kPa, 1500 rpm, and MBT spark timing con-dition. %EGR is defined as the ratio between the mass of the recy-cled exhaust gases to the total inlet mass (i.e. the mass of air, fuel,and recycled exhaust gases):

%EGR ¼ mEGR

ma þmf þmEGR� 100 ð19Þ

The engine brake power decreases with the increase of both ex-cess air and EGR as they both replace some of the inlet fuel massbecause they were added at a constant inlet pressure. In order tocompare the effect of both excess air and EGR on engine perfor-mance using the same basis of charge dilution, the air dilution isdefined in a similar way like EGR dilution. Air dilution can be de-fined as the ratio between the mass of the excess air to the total in-let mass. The mass of the excess air can be calculated as thedifference between the actual mass of air used for combustionand the stoichiometric mass of air used for stoichiometriccombustion:

%Air dilution ¼ ma;act �ma;st

ma;act þmf� 100

¼ ðA=FÞact � ðA=FÞst

ðA=FÞact þ 1� 100 ¼ k� 1

kþ ðF=AÞst� 100 ð20Þ

Knowing that the stoichiometric fuel–air ratio for natural gas isabout 1/15.8, then Eq. (20) expresses the percentage of air dilutionas a function of excess air factor, k only. Air dilution is equivalent tozero for stoichiometric mixture and increases with the increase of kas shown in Table 3.

Fig. 12 compares the effect of both air and EGR dilution on en-gine power at an inlet pressure of 130 kPa, 1500 rpm, and MBTspark timing condition. Adding EGR to the stoichiometric mixturedecreased power more rapidly than excess air. That was mainly be-cause of the more significant effect of EGR dilution on combustionduration compared to air dilution. Adding EGR to the inlet chargedecreased the in-cylinder oxygen concentration and slowed downthe combustion rate significantly. The extended combustion dura-tion resulted in burning most of the fuel away from the top deadcentre which led to a loss in engine power. In addition, the higherinlet charge temperature in the presence of EGR (assumed to be333 K) compared to the inlet charge temperature in the presenceof air (assumed to be 310 K) reduced the inlet density and hencethe power was reduced.

The use of EGR dilution at a rate of 20% and inlet pressure of130 kPa reduced engine power by about 20% compared to usingair at the same dilution rate and inlet pressure conditions. How-ever, the loss of engine power which occurred due to the use ofEGR instead of excess air was recovered by increasing the inletpressure. It was found that when inlet pressure increased from130 kPa at no EGR condition to about 150 kPa at 20% EGR dilution,brake power increased and became equivalent to the correspond-ing air dilution power results.

Fig. 13 shows the increase of combustion duration with boththe addition of air at constant inlet pressure of 130 kPa and theaddition of EGR at constant and variable inlet pressure. The inletpressure was varied from 130 to 150 kPa to obtain power equiva-

Table 3Variations of % air dilution with k.

k Air dilution (%)

1 01.1 8.61.2 15.81.3 221.4 27.3

lent to air dilution power results as it was mentioned earlier. Theincrease in inlet pressure increased the inlet density and slightlydecreased the combustion duration as shown in Fig. 13. Fig. 13indicates that the increase of EGR dilution has a stronger effecton combustion duration than the increase of excess air dilution.These results agree well with different experimental results suchas the results obtained by Einewall and coworkers [18]. Einewalland coworkers [18] compared the effect of lean burn and EGR onthe heat release rate using a natural gas engine having a fast-burn-ing combustion chamber with high turbulence. The authors foundthat EGR has a stronger influence on the flame development anglethan excess air since the laminar flame speed was more reducedcompared to lean burn operation. The authors concluded that ex-cess air leads to much shorter duration during early combustioncompared to EGR for high dilution levels.

4.2. The effect of EGR vs. excess air on bsfc

Fig. 14 shows the effect of both air and EGR dilution on brakespecific fuel consumption at 1500 rpm and MBT spark timing

Fig. 14. A comparison between the effect of both air and EGR dilution on brakespecific fuel consumption at 1500 rpm, rc = 10, and MBT spark timing condition.

A. Ibrahim, S. Bari / Energy Conversion and Management 50 (2009) 3129–3139 3137

condition. The addition of excess air was studied at constant inletpressure of 130 kPa whereas the addition of EGR was studied fortwo different inlet pressure conditions, which were the constant130 kPa inlet pressure condition and the variable inlet pressurecondition that led to brake power results similar to air dilutionbrake power results at an inlet pressure of 130 kPa. The fuel con-sumption was slightly improved with the increase of air dilutionup to about 15%. The increase of air dilution decreased the maxi-mum cylinder burned gas temperature which led to a decrease inthe dissociation losses near top dead centre which resulted intoan increase in the fraction of sensible energy that was convertedto work. However, the fuel consumption started to increase athigher air dilution conditions due to the significant increase ofthe combustion duration at higher air dilution conditions as shownin Fig. 13. On the other hand, engine fuel consumption increasedwith EGR dilution which had a more significant effect on combus-tion duration than excess air. The extended combustion durationled to most of fuel to be burned earlier during compression strokeand later during expansion stroke which resulted in a decrease inpower and an increase in the fraction of fuel energy that was dis-

Fig. 15. A comparison between the effect of both air and EGR dilution on engineheat loss as a percentage from fuel power at 1500 rpm, rc = 10, and MBT sparktiming condition.

sipated as heat loss. Fig. 15 shows the effect of air and EGR dilutionon engine heat loss as a percentage from fuel power. The percent-age heat loss was calculated as follows:

%heat loss ¼_Q

_mf � LHV� 100 ð21Þ

where _Q is the heat transfer, _mf is the fuel mass flow rate, and LHVis the lower heating value.

The use of EGR dilution at a rate of 20% and inlet pressure of130 kPa increased engine fuel consumption by about 18% com-pared to using air at the same dilution rate and inlet pressure con-ditions. On the other hand, when 20% EGR dilution was added at aninlet pressure of 150 kPa in order to obtain the same amount ofpower as when 20% air dilution was employed at 130 kPa, the fuelconsumption increased by only 10%.

4.3. The effect of EGR vs. air on NO emission

Fig. 16 shows the effect of using both EGR and air dilution onengine NO emission at 1500 rpm and MBT spark timing condition.NO emission increases with the increase of air dilution until itreaches to its peak value at about 9% air dilution which is equiva-lent to an excess air factor of about 1.1. This is due to NO formationmechanism depends on both temperature and oxygen concentra-tion. Although burned gas temperature is higher for stoichiometricmixture (zero dilution) as shown in Fig. 17, the availability of oxy-gen at leaner mixtures results in the maximum NO emission to beoccurred at mixtures slightly leaner than stoichiometric. Theburned gas temperature and consequently NO emission decreasewhen air dilution is increased above 9%.

On the other hand, the use of EGR dilution decreased NO emis-sion significantly as shown in Fig. 16. The use of EGR dilution de-creases both temperature and oxygen concentration and leads tolower NO emission compared to the use of air dilution. The useof 20% EGR dilution decreased NO emission by about 70% com-pared to using the same percentage of air dilution. Furthermore,employing EGR dilution technique would allow using a threeway catalyst, which usually has a conversion efficiency of about90%, for exhaust gas after-treatment. That makes the EGR dilutiontechnique capable of producing extremely lower NO emissioncompared to using air dilution technique.

These results agree well with several experimental results doneby different researchers. For example, Bhargava and coworkers

Fig. 16. A comparison between the effect of both air and EGR dilution on NOemission at 1500 rpm, rc = 10, and MBT spark timing condition.

Fig. 17. A comparison between the effect of both air and EGR dilution on maximumcylinder burned gas temperature at 1500 rpm, rc = 10, and MBT spark timingcondition.

Fig. 18. A comparison between the effect of both air and EGR dilution on start ofcombustion angle optimized for MBT condition at 1500 rpm, rc = 10, and MBT sparktiming condition.

3138 A. Ibrahim, S. Bari / Energy Conversion and Management 50 (2009) 3129–3139

[19] studied the effect of replacing air with EGR on natural gas en-gine emissions at constant load and speed. The authors found thatNOx emissions were reduced from 4 to about 1 g/bhp-h with a per-centage reduction of about 75% when EGR replaced excess air atthe same amount of charge dilution of about 22%. Corbo andcoworkers [7] compared engine emissions using both EGR and leanburn techniques for a converted natural gas engine. The authors re-ported that the use of lean burn technique led to NOx emissions of4.7 g/kW h whereas the use of EGR technique with a three way cat-alyst resulted in only 0.1 g/kW h NOx emissions. Although theauthors did not report the value of NOx emission before the cata-lyst, this could be calculated assuming the catalyst conversion effi-ciency to be comparable to 90%. This would lead NOx emissionbefore the catalyst to be about 1 g/kW h with a percentage reduc-tion of about 78% compared to using lean burn technique. Saanumand Bysveen [9] compared lean burn and EGR operation of an en-gine fueled with natural gas and hydrogen enriched natural gas.The authors studied the effect of both lean burn and EGR on NOx

emissions at various engine loads. It was demonstrated that NOx

emission increases with the increase of excess air until it reachesits maximum value at an excess air factor ranged from 1.1 to 1.2depending on engine load and then decreases with the increaseof excess air. On the other hand, the use of EGR reduced NOx emis-sion efficiently with a near linear decrease with the increase of EGRdilution. The authors concluded that EGR technique resulted inmuch lower NOx emissions compared to using excess air at thesame level of dilution.

Fig. 19. A comparison between the effect of both air and EGR dilution on maximumcylinder pressure at 1500 rpm, rc = 10, and MBT spark timing condition.

4.4. The effect of EGR vs. excess air on in-cylinder maximumtemperature and pressure

Both in-cylinder pressure and temperature are important vari-ables which affect the in-cylinder mechanical and thermal stressesin addition to the formation of NO emission. Fig. 17 shows the vari-ations of maximum burned gas temperature with both EGR and airdilution at 1500 rpm and MBT spark timing condition. The use ofair dilution up to 15% resulted in a slight increase in the maximumburned gas temperature compared to using EGR dilution. This isbecause EGR contains water vapor which has higher specific heatcapacity than air. However, when air dilution increased above15%, the maximum burned gas temperature decreased comparedto using EGR dilution. This is because the use of EGR at higher rates

of dilution increased the combustion duration significantly asshown in Fig. 13. That led to an excessive advance for the start ofcombustion timing which was optimized for MBT condition asshown in Fig. 18. The increased advance of the start of combustiontiming led to initiating the combustion earlier in the compressionstroke which increased the maximum temperature when EGR dilu-tion was used at rates higher than 15% compared to air dilution.

Fig. 19 shows the effect of both EGR and air dilution on maxi-mum cylinder pressure. When both air and EGR dilution werestudied at constant inlet pressure of 130 kPa, the use of EGR sloweddown the combustion rate more significantly and consequently themaximum cylinder pressure was lower than the correspondingmaximum pressure for air dilution strategy. On the other hand,when EGR was added at higher inlet pressure conditions whichvaried from 130 kPa at zero dilution to about 150 kPa at 20% EGRdilution in order to obtain similar power as for air dilution strategy,the maximum cylinder pressure increased compared to using airdilution at an inlet pressure of 130 kPa. That indicates the signifi-cant dependence of the maximum cylinder pressure on inletpressure.

Fig. 20. A comparison between the effect of both air and EGR dilution on averageexhaust gas temperature at 1500 rpm, rc = 10, and MBT spark timing condition.

A. Ibrahim, S. Bari / Energy Conversion and Management 50 (2009) 3129–3139 3139

4.5. The effect of EGR vs. air on average exhaust temperature

The average exhaust gas temperature can indicate the amountof available thermal energy in the exhaust gases which can be usedfor a turbocharger in order to increase the inlet pressure. Also, theaverage exhaust temperature is an important quantity for deter-mining the performance of catalytic converters which are used toachieve further emission reduction. The exhaust temperature wascalculated as an enthalpy-averaged temperature using the follow-ing integration as suggested by Heywood [10]:

Tex ¼Z

_mcpTdtZ

_mcpdt�

ð22Þ

where _m is the mass flow rate exiting through the exhaust valve, cp

is the exhaust gas specific heat, and T is the instantaneous exhaustgas temperature. The integration is performed from the time the ex-haust valve opens to the closing time. Fig. 20 shows the variationsof the average exhaust temperature with both air and EGR dilutionat 1500 rpm and MBT spark timing condition. The more significanteffect of EGR dilution on combustion duration compared to air dilu-tion led to terminating the combustion process later in the expan-sion stroke and increased the exhaust gas temperature comparedto using air dilution strategy.

5. Conclusions

The effect of both EGR and lean burn techniques on engine per-formance and NO emission was compared numerically using atwo-zone combustion model. The following conclusions wereobtained:

� The use of EGR slowed down the combustion rate and increasedthe combustion duration significantly compared to using air atthe same amount of dilution. The longer combustion durationwhich was achieved when EGR was used resulted in moreadvance in start of combustion timing which was optimizedfor MBT condition.

� The addition of EGR to the inlet charge decreased engine powermore rapidly compared to using excess air at the same inletpressure of 130 kPa; however, the loss of power was recoveredby increasing inlet pressure from 130 kPa at zero dilution to150 kPa at 20% EGR dilution.

� The fuel consumption increased by 18% when the inlet chargewas diluted at a rate of 20% using EGR instead of excess air atthe same inlet pressure of 130 kPa; however, the fuel consump-tion increased by only 10% when EGR was added at higher inletpressure of 150 kPa.

� The use of EGR is capable of achieving extremely low NO emis-sion compared to excess air. NO emission decreased by about70% when the inlet charge was diluted at a rate of 20% usingEGR instead of air.

References

[1] Baldassarri LT, Battistelli CL, Conti L, Crebelli R, De Berardis B, Iamiceli AL, et al.Evaluation of emission toxicity of urban bus engines: compressed natural gasand comparison with liquid fuels. Sci Total Environ 2006;355(1–3):64–77.

[2] The Environmental Protection Agency (EPA), USA. Clean alternative fuels:compressed natural gas. Fact sheet, <http://www.epa.gov/oms/consumer/fuels/altfuels/420f00033.pdf>; 2002 [accessed August 2006].

[3] Catania AE, D’Ambrosio S, Mittica A, Spessa E. Experimental investigation offuel consumption and exhaust emissions of a 16V pent-roof engine fueled bygasoline and CNG. Society of Automotive Engineers, SAE paper no. 2001-01-1191; 2001.

[4] The Engine Manufacturers Association, USA. The use of exhaust gasrecirculation systems in stationary natural gas engines. <www.enginemanufacturers.org/admin/library/upload/830.pdf>; 2004 [accessed August2006].

[5] Rabl A. Environmental benefits of natural gas for buses. J Transport Res, Part D2002;7:391–405.

[6] Environmental protection agency handbook. Control technologies forhazardous air pollutants. EPA 625/6-91-014; 1991.

[7] Corbo P, Gambino M, Iannaccone S, Unich A. Comparison between lean-burnand stoichiometric technologies for CNG heavy-duty engines. Society ofAutomotive Engineers, SAE paper no. 950057; 1995.

[8] Nellen C, Boulouchos K. Natural gas engines for cogeneration: highestefficiency and near-zero-emissions through turbocharging, EGR and 3-waycatalytic converter. Society of Automotive Engineers, SAE paper no. 2000-01-2825; 2000.

[9] Saanum I, Bysveen M, Tunestal P, Johansson B. Lean burn versus stoichiometricoperation with EGR and 3-way catalyst of an engine fueled with natural gasand hydrogen enriched natural gas. Society of Automotive Engineers, SAEpaper no. 2007-01-0015; 2007.

[10] Heywood JB. Internal combustion engine fundamentals. New York: McGraw-Hill Book Company; 1988.

[11] Ferguson CR. Internal combustion engines. John Wiley & sons; 1986.[12] Caton JA. Comparisons of instructional and complete versions of

thermodynamic engine cycle simulations for SI engines. Int J Mech Eng Educ2000;29(4).

[13] Tabaczynski RJ, Ferguson, CR, Radhakrishnan, K. A turbulent entrainmentmodel for spark-ignition engine combustion. SAE transactions, vol. 86, SAEpaper no. 770647; 1977.

[14] Hires SD, Tabaczynski RJ, Novak JM. The prediction of ignition delay andcombustion intervals for a homogeneous charge spark ignition engine. SAEtransactions, vol. 87, SAE paper no. 780232; 1978.

[15] Liao SY, Jiang DM, Huang ZH, Cheng Q, Gao J, Hu Y. Approximation offlammability region for natural gas – air – diluent mixture. J Hazard Mater2005;A125:23–8.

[16] Ibrahim A, Bari S. Optimization of a natural gas SI engine employing EGRstrategy using a two-zone combustion model. J Fuel 2008;87(10–11):1824–34.

[17] Ibrahim A, Bari S, Bruno F. A Study on EGR utilization in natural gas SI enginesusing a two-zone combustion model. SAE paper no. 2007-01-2041; 2007.

[18] Einewall P, Tunestal P, Johansson B. Lean burn natural gas operation vs.stoichiometric operation with EGR and a three way catalyst. Society ofautomotive engineers, SAE paper no. 2005-01-0250; 2005.

[19] Bhargava S, Clark NN, Hildebrand MW. Exhaust gas recirculation in a lean burnnatural gas engine. Society of automotive engineers, SAE paper no. 981395;1998.