JSAE 20119177SAE 2011-01-1838

CFD Investigation of Heat Transfer in a Diesel Engine withDiesel and PPC Combustion Modes

Helgi Fridriksson, Bengt Sunden, Shahrokh Hajireza, Martin TunerLund University, Sweden

Copyright c© 2011 Society of Automotive Engineers of Japan, Inc.

ABSTRACT

In this study, an investigation was made on a heavyduty diesel engine using both conventional dieselcombustion mode and a partially premixed combustion(PPC) mode. A segment mesh was built up andmodeled using the commercial CFD code AVL FIRE,where only the closed volume cycle, between IVCand EVO, was modeled. Both combustion modeswere validated using experimental data, before anumber of heat flux boundary conditions were applied.These conditions were used to evaluate the engineresponse in terms of engine performance and emissionlevels for the different percentage of heat rejection.The engine performance was measured in terms ofspecific fuel consumption and estimated power output,while the calculated net soot and accumulated NOxmass fractions were used for comparing the emissionlevels. The results showed improved efficiency for bothcombustion types, but only the PPC combustion modemanaged that without increasing the production of NOxemissions severely.

INTRODUCTION

The process of converting chemically bound energyof the liquid fuel, in a diesel engine, to useful workincludes a few sources for losses along the way. Themajor ones are combustion related losses, due toincomplete combustion, heat losses through walls andexhaust, gas exchange losses and friction losses [1]. Ithas commonly been stated that the energy from thefuel is divided equally into three main parts, energyconverted into useful work, energy transferred to thecoolant and energy transferred to the exhaust [2]. Forthis reason, it is of great importance to have controlof the heat losses in the system when designing aninternal combustion engine. Increased knowledge ontemperature distribution and heat losses inside andaround the engine cylinder is therefore important.

The heat transfer process in the entire engine cylinderhas not been extensively studied, but some studies inthe open literature contains some reviews of methodsfor heat transfer analysis [1, 3, 4], some of which arenot up to date for modern diesel engines. As pointedout in these reviews, the engine cylinder is in fact a

series of interacting sub-systems, each with their ownlevel of complexity for heat transfer analysis and allaffect the overall performance of the engine. Due tothe unsteady, transient behavior of the fluid motioninside the engine cylinder, the heat flux to the solidcomponents of the cylinder may vary from 0 to as muchas 10 MW/m2 during the span of a few milliseconds.Furthermore, two points on the cylinder wall, separatedonly by one centimeter, may experience the samedifference in heat flux [3, 4]. Additionally, heat transferis important to physical phenomena in the cylinder,such as droplet evaporation, autoignition and flame-wall interaction [5].

Inside the engine cylinder there are a number of mech-anisms that have proven to be quite demanding for asimulation process. Adding these processes togetheralong with their interactions creates a problem thatcannot be solved without tremendous computationalresources. Therefore, researchers have traditionallymade simplifications on their computational domainand flow descriptions in order to achieve any solutionat all. According to the literature [3, 6, 7] there arenumerous alternatives available for engine heat trans-fer simulations, depending on the level of complexitydesired in the model. Borman & Nishiwaki [3] list fivedifferent alternatives, out of which global models andmultidimensional (CFD) models are the most widelyused today.

The global heat transfer models consist of empirical orsemi-empirical correlations for the mean, crank angledependent, convective heat transfer coefficient. In[3, 6, 8] the most common correlations are listedand discussed, but currently the most widely usedheat transfer correlations are the ones derived byAnnand [9], Woschni [10] and Hohenberg [11]. Someresearchers have provided alternative correlations dur-ing recent years, such as [12–14]. These kinds ofcorrelations are frequently used in one-dimensionalgas exchange codes as shown in [15] and the resultsfrom gas exchange code simulations are often usedas boundary and initial conditions for CFD simulations,as in [16], where the engine cylinder is resolvedin a computational grid of three dimensions. Analternative option to using gas exchange codes toprovide boundary conditions for CFD simulations, is

1

to include the solid parts of the cylinder in thecomputational domain and solve for conjugate heattransfer, from the bulk gas to the cooling media, asexemplified in [17, 18].

With increasing computational power, the field ofthree-dimensional engine simulations has seen vastimprovements during recent years. Sub-models forall parts of the energy conversion process have beenadvancing and an extensive effort has been put into thedevelopment and application of improved turbulence[19], spray, combustion [20] and emission models. Notmany studies, however, have combined all of thesesub-models and evaluated the energy flow throughthe solid components of the cylinder and therebydocumenting how the heat transfer mechanism is builtup in the engine cylinder.

As stated before, it is of interest to reduce theheat losses in the system, in the attempt to achievehigher mechanical work output. The reduction ofheat losses through the walls will result in elevatedflame temperatures and even elevated exhaust temper-atures [21]. The raised temperatures might introduceincreased local emission production in the cylinder,which is not desired due to strict emission regulationsas well as efficiency penalties from exhaust gasafter-treatment devices. One way to circumvent thisproblem is to incorporate some of the newly introducedLow Temperature Combustion (LTC) strategies. Thiscategory contains a range of combustion strategies,out of which the best known is undoubtedly HCCI,or Homogeneous Charge Compression Ignition, whichdates back to 1979 [22–24]. Recently, a new conceptwithin this category has emerged which is supposed totake care of the problems related to combustion controlexperienced in HCCI, thus extending the load range.This concept is called PPC, or Partially PremixedCombustion.

The PPC concept is described in detail by Manente[25], but in short it is characterized by a fuel injectionin the compression stroke in order to ensure somefuel-air mixing before the start of combustion. Thisinjection strategy is combined with high levels of EGRin order to dilute the air-fuel mixture to ensure leancombustion. This strategy has shown to provide highefficiency and near zero emissions for a wide loadrange of a heavy duty diesel engine, due to reducedoperating temperatures [26].

Historically, some research has been done on lowheat rejection (LHR) or even adiabatic engines, bothnumerically and experimentally, as the review byJaichandar & Tamilporai [27] indicates. These havegiven quite diverged results, since the experimentalresearch usually shows little or no efficiency gains withLHR engines, while the numerical research has givenmore positive response. The research by Taymaz [2]and Yasar [21], as well as the work done by Kamoand Bryzik [28], has shown that there are potentialefficiency gains with the use of LHR engines. Tuner[29] also states that with the new LTC, the LHR enginesmight have become an attractive option.

Table 1: Scania D13 engine geometry

Displacement [L] 2.124Geom. Comp. Ratio [-] 17.3Conn. Rod Length [mm] 255Bore x Stroke [mm] 130x160

Valve-train IVC = 40 CAD ABDCEVO = 50 CAD BBDC

Spray angle [deg] 148Orifices [-] 8Orifice diameter [mm] 0.19

The aim of the work presented here was to comparethe response of the two selected combustion modesto different heat rejection levels, by monitoring theperformance and emission levels of each combustionmode and compare them. This was done by simulatingonly the closed volume cycle (from IVC to EVO),keeping the intake conditions fixed. The combustionmodes used were a traditional diesel combustion andthe newly developed partially premixed combustionmode. In the study, the same engine configurationwas used to simulate both combustion modes, usingthe same diesel fuel. Both cases were validated usingengine test bed data before being subjected to differentheat flux boundary conditions, representing differentlevels of heat rejection. The engine response for thedifferent heat rejection levels was compared for eachcase and emission levels were evaluated.

This study is a part of a ongoing research project,called D60, where simulations and experiments arecombined in order to gain increased knowledge onthe heat transfer process in diesel engines. Thename, D60, stands for The 60% efficient diesel engineand is aimed to find ways to increase the efficiencyof the diesel engine by lowering heat losses. Forother publications related to the project, please referto [8, 29–31].

PROBLEM DEFINITION

The modeled engine is a 13 L six cylinder heavy-dutydiesel engine, with the engine and injector specifica-tions for a single cylinder given in Table 1. In order toperform the heat transfer analysis on the engine, three-dimensional computational fluid dynamic simulation(CFD) was applied. Therefore, a computational mesh,shown in Fig. 1, was built up using the commercial CFDtool AVL ESEDiesel. The set of governing equationswas solved using the commercial solver AVL FIRE [32].

The computational domain consists of one sectionof the modeled engine cylinder, which was usedfor simulations between inlet valve closing (IVC) andexhaust valve opening (EVO). This means that only theclosed volume part of the engine cycle is computed. Tosimplify the geometry, a periodic condition is assumedfor each injector hole and therefore only 1/8 part ofthe cylinder is modeled, since the injector has 8 holes.This simplification will neglect possible interactions

2

Figure 1: Computational domain with compensationvolume shown at 20◦ BTDC. The grid has around113,000 and 66,000 cells at BDC and TDC,respectively.

between spray paths and all information on heattransfer mechanisms in the gas exchange process areneglected. This computational mesh was used toexamine the effects of different heat rejection levels onthe engine performance of a heavy duty diesel engine.

The average cell size in the computational meshranged from 0.6 mm to 1.2 mm and there were two celllayers placed adjacent to the wall boundary of 0.1 mmthickness. The mesh was divided into 17 angular cells,resulting in mesh densities of around 113,000 cells atBDC and 66,000 cells at TDC.

STUDIED CASES

In this study two baseline cases were produced,one for each combustion strategy. These caseswere validated against available test bed data. Theexperimental data for the diesel combustion case wasa production engine test from the engine produces.The load point used for the simulation process wasa full load case (26.2 bar IMEPg). The data for thePPC case was taken from experiments on a singleengine cylinder, performed by Manente et al. [26]. Theload point chosen from that experiment was a 75%load case (20.8 bar IMEPg). The operating conditionsfor each case are listed in Table 2, where SOI andEOI represent start of injection and end of injection,respectively.

APPLIED MODELS

Detailed in-cylinder engine modeling by CFD requiresa series of interacting models to describe the physicalphenomena occurring within the cycle.

First of all, the representation of the gas motion inthe cylinder must have a relatively accurate portrayalfor the solution to be realistic. For this a turbulencemodel is needed, of which there is a wide varietyavailable, depending on the characteristics of theflow. Today, the most commonly used turbulencemodels for engine simulations are 2-equation RANSbased models, more specifically some versions ofthe k − ε model. This model, along with other 2-equation eddy viscosity RANS models, tends to failwhen predicting flow over curved surfaces, secondary

Table 2: Operating conditions for both combustionmodes

Diesel PPC

Load (IMEPg) [bar] 26.2 20.8Engine speed [rpm] 1250 1250Swirl ratio [-] 1.7 2.095Effective comp. ratio [-] 16.8 15.5Inlet Pressure [bar] 4.15 3.66Inlet Temp. [K] 350 309EGR [%] 28.1 55.24SOI [CAD bTDC] 2.6 4EOI [CAD aTDC] 57.4 22Injection pressure [bar] 2250 2400Fuel mass [mg/cyl] 310 195Fuel [-] Diesel MK1 Diesel MK1

flows and rotational flows. For this reason, non-linear and extended eddy viscosity models have beenpresented. These models generally give better resultsin the previously mentioned problematic areas. Withinthe group of non-linear eddy viscosity models is modelpresented by Hanjalic and Popovac [19, 32, 33], the k−ζ − f model. This model has given good agreement tovalidation data for various cases that have proven to bedifficult for eddy viscosity models [19]. The model wasused along with a hybrid velocity wall treatment andstandard wall functions for the wall heat transfer. It isworth pointing out that even though the k−ζ−f modelcan provide improved results over the k − ε model, itis still a RANS based model which will only providetime averaged velocity components. That means thatinformation on instantaneous fluctuations will be lost.

The mixing of fuel and air as well as the combus-tion must be modeled correctly. There are a fewcombustion model options available and two differentmodels were applied here. For the diesel combustioncase the eddy break-up model was used [32], since ithas proven to give reasonable result for conventionaldiesel combustion. The mean reaction rate in PPCcombustion is, unlike diesel combustion, not deter-mined by the rate of dissipation of the turbulent eddies.Therefore the basis of the eddy break-up model failsfor PPC combustion. This is not an issue whenusing the ECFM-3Z combustion model presented byColin & Benkenida [20], which was used for the PPCcombustion in this work. Droplet evaporation wasmodeled by the Dukowicz model and droplet break-upby the Wave model [32].

Emission production must also be evaluated and themodels available in that section, as in other sections,are constantly being upgraded and improved. Forthis work NOx is evaluated by the extended Zeldovichmodel and soot is evaluated by the Frolov kineticmodel [32].

BOUNDARY CONDITIONS

In order to close the system of equations, a set ofboundary conditions is needed. In this study, where

3

only the closed volume is simulated, there is no intakeand no exhaust. This means that only conditions atwall boundaries need to be given. There are threewall areas in the simulation, piston, cylinder head andliner. For the fluid flow a standard no-slip condition isapplied at wall boundaries. For the baseline cases, aconstant wall temperature condition was used for bothcombustion modes. These wall temperatures wereprovided from one-dimensional simulations.

The baseline cases provided an average, crank-angledependent, heat flux to each wall boundary. This heatflux solution was scaled, for each boundary, to providethe desired conditions for the heat rejection. The heatflux levels used were the 100% heat flux (baselinecase), as well as 75%, 50%, and 25% heat fluxes.Additionally, adiabatic conditions were also simulated,i.e., zero heat flux. It is worth mentioning that whilethe heat rejection is varied during this study the intakeconditions, presented in Table 2 remain constant.

ENGINE PERFORMANCE EVALUATION

The engine performance is evaluated for all heatrejection levels by computing and comparing theindicated mean effective pressure (IMEP), the indi-cated efficiency (ηi) and the indicated specific fuelconsumption (ISFC). The CFD computations onlyprovide values for the closed part of the cycle, so thereis not enough data available from the simulations toprovide the traditional IMEP, ηi and ISFC. The valuescomputed for the purpose of comparison betweenheat rejection cases, have been obtained by usingthe symmetrical calculation duration (CD), cylinderpressure (pc) and the displaced volume (Vd). The IMEPwill then be calculated by Eq. (1)

IMEP =1Vd

∫CD

pc · dV (1)

The formulation of the indicated efficiency also includesthe total fuel mass during each cycle (mf ) and thelower heating value of the fuel (Hu). This is shown inEq. (2).

ηi =

∫CD

pc · dVmfHu

(2)

The ISFC is taken as the ratio of the indicated power(Pi) and the total fuel mass during the cycle, as shownin Eq. (3).

ISFC =Pi

mf(3)

where the indicated power is given by Eq. (4)

Pi = IMEP · VdN

nR(4)

In Eq. (4), nR is the number of crank revolutions foreach power stroke per cylinder, which is 1 for two-stroke engines and 2 for four-stroke engines.

RESULTS AND DISCUSSIONS

The results will be presented in two separate sections,one for each combustion mode. Each section will startby showing the validation of the baseline case, byviewing mean pressure traces and heat release for thesegment cycle, before moving on to the heat rejectionstudy.

DIESEL COMBUSTION

The diesel combustion case, as stated earlier, was afull load case with the gross indicated mean effectivepressure (IMEP) of 26.2 bar. Fig. 2 shows the meanpressure trace for the simulated engine comparedto the experimental data and as the figure shows,the CFD results provide an acceptable fit to theexperimental results. Figure 3 shows the apparent heatrelease characteristics for both experimental valuesand values from CFD simulations, i.e., containing fuelevaporation and heat loss. The left hand figure showsthe accumulated heat release as a function of crankangle degrees, whereas the right hand figure showsthe heat release rate as a function of crank angledegrees.

The values for the accumulated heat release donot match perfectly between the experimental andsimulation results. One possible reason for this mightbe the value of the constant wall temperature, whichwas acquired from one-dimensional simulation. Thetrends of the heat release rate are quite similar.However, even there the phasing in the diffusion part ofthe flame is not perfectly matched. This, again, mightbe traced back to the choice of values for the constantwall temperature boundary condition.

This validated case was used in the heat rejectionstudy, as previously mentioned, to investigate theeffects of heat rejection on engine performance andemission production. Firstly, the chemical heat releasewas examined as a function of crank angle degree fordifferent heat rejection levels, as shown in Fig. 4. Thisheat release is defined from the heat of formation cal-culated in each computational cell during combustion.There were no significant changes made to this heatrelease from combustion with increasing insulation.

Figure 2: Mean Pressure trace from experiments andCFD simulations for the diesel combustion case.

4

(a) Heat release (b) Heat release rate

Figure 3: Accumulated heat release and heat releaserate from experiments and CFD simulations for thediesel combustion case.

The results indicate that the combustion is not severelyaffected during this procedure. If the combustion wereto be affected, one would see a difference in thephasing of the heat release rate (RoHR). This phasingdifference is not visible here.

The performance of the engine, in terms of IMEP,indicated efficiency and indicated specific fuel con-sumption (ISFC), are shown as a function of heat fluxpercentage in Fig. 5. There, 100% heat flux representsthe validated case, while 0% presents an adiabaticcase. As expected, and shown in previous research[27], the indicated efficiency and power output of theengine is increased by introducing insulation effectson wall boundaries and specific fuel consumption isreduced. The question remains if this is at the costof elevated peak temperatures and thereby increasedemissions.

The mass fractions of NOx and soot are shown inFig. 6 as a function of crank angle degrees for differentheat rejection levels. Based on the figure, it seemsthat decreased heat rejection leads to increased tem-peratures and more NOx is produced. However, theamount of soot in the cylinder after combustion isrelatively unaffected by the decreased heat rejection,even though the peak in soot production is higher forlower heat rejection.

The average temperature in the cylinder, as a functionof crank angle, for different heat rejection levels isgiven in Fig. 7. One can see that the average

Figure 4: Heat release for the diesel engine withdifferent insulations.

Figure 5: Diesel engine performance as a function ofinsulation effects.

Figure 6: NOx and soot mass fractions from dieselcombustion for different heat rejection levels.

cylinder temperature is affected by the reduction ofheat rejection. According to Fig. 4, there is no changein the heat release, so the same amount of heat isgenerated by combustion while less heat is rejected.This will lead to a slight increase in the average cylindertemperature. The figure does not show temperatureabove 2000 K, which is often referred to as the the limitwhen thermal NOx begins to form, so it is likely that theNOx is formed due to some local temperature peaks.

The equivalence ratio, φ, and the cylinder gas temper-ature for the baseline case are shown at a mid-planeof the engine sector in Figs. 8 and 9, respectively. As

Figure 7: Average crank angle resolved cylindertemperature from diesel combustion for different heatrejections.

5

(a) TDC (b) 10 CAD ATDC (c) 20 CAD ATDC

(d) 30 CAD ATDC (e) 40 CAD ATDC (f) 50 CAD ATDC

Figure 8: Equivalence ratio for baseline case of thediesel combustion, from TDC to 50 CAD ATDC.

(a) TDC (b) 10 CAD ATDC (c) 20 CAD ATDC

(d) 30 CAD ATDC (e) 40 CAD ATDC (f) 50 CAD ATDC

Figure 9: Cylinder gas temperature for the base linecase of the diesel combustion, from TDC to 50 CADaTDC.

shown by Akihama et al. [34], the production of NOx indiesel engines begins to occur at temperatures around2000 K and equivalence ratios φ < 2. Soot, on theother hand, is formed at temperatures between 1500 Kand 2400 K for φ > 2. This range of temperature andequivalence ratios only explain the formation of NOxand soot. Andersson et al. [35] stated that soot isoxidized at lower equivalence ratios, which introducesa trade-off between NOx and soot. The red color in thetemperature field of the mid-plane section correspondsto temperature above 2200 K and the red color for theequivalence ratio corresponds to φ = 2 and above.This shows that the soot is formed around the spraypath between 10 CAD ATDC and 20 CAD ATDC andoxidized at later crank angle degrees. NOx, whichis formed at high temperatures and low equivalenceratios, is mainly formed after 20 CAD ATDC.

Figures 10 and 11 show the equivalence ratio andtemperature, respectively, for the 50% heat rejectioncase. It shows that while the equivalence ratio isrelatively unchanged between the baseline case andthe lower heat rejection, the temperature is slightlyincreased. This occurs mainly after 30 CAD andexplains the increase in NOx emissions, shown inFig. 6. The soot level is relatively unaffected due tostronger oxidation behavior at higher temperatures.

(a) TDC (b) 10 CAD ATDC (c) 20 CAD ATDC

(d) 30 CAD ATDC (e) 40 CAD ATDC (f) 50 CAD ATDC

Figure 10: Equivalence ratio for the 50% heat rejectionof the diesel combustion, from TDC to 50 CAD ATDC.

(a) TDC (b) 10 CAD ATDC (c) 20 CAD ATDC

(d) 30 CAD ATDC (e) 40 CAD ATDC (f) 50 CAD ATDC

Figure 11: Cylinder gas temperature for the 50% heatrejection of the diesel combustion, from TDC to 50 CADaTDC.

PPC COMBUSTION

Before examining the results from the PPC case, it isworth noting that not only was the load point differentfor the PPC case, compared to the diesel case, but theeffective compression ratio was also reduced from 16.8to 15.5, as shown in Table 2.

As stated before, the PPC combustion case was a highload case (20.8 bar IMEPg) from the experiments ofManente et al. [26]. The mean pressure trace is shownin Fig. 12 and shows that the agreement betweensimulation and experiment is quite good. There isa slight difference in the phasing of the combustion,which can also be seen in the rate of heat release inFig. 13. The figure also shows the accumulated heatrelease. The experimental values of the heat releasehave been adjusted to show the chemical heat release,as the simulation results show. This has been done byManente et al. according their self tuning gross heatrelease algorithm [36]. It can be noted that the heatreleased in the PPC combustion is considerably lessthan the heat released in the diesel combustion, whichis both due to the combustion mode and difference inamount of fuel used.

As for the diesel case, this validated case was used inthe heat rejection study for the PPC engine. The heatrelease was examined as a function of heat rejectionlevel, as shown in Fig. 14, and it was found that nosignificant changes were made to the heat release

6

Figure 12: Mean Pressure trace from experiments andCFD simulations.

(a) Heat release (b) Heat release rate

Figure 13: Accumulated heat release and heat releaserate from experiments and CFD simulations.

shape with increasing insulation. This, again, impliesthat the combustion is not severely affected during thisprocedure.

The performance of the engine, in terms of IMEP,indicated efficiency and ISFC, is shown as a functionof insulation effects in Fig. 15. As for the diesel engine,the indicated efficiency and power output of the engineis increased by introducing insulation effects on wallboundaries and specific fuel consumption is reduced.Compared to the diesel engine, the indicated efficiencyis higher for the PPC case, leading to lower values ofthe specific fuel consumption (ISFC).

The mass fractions of NOx and soot are shown inFig. 16 as a function of crank angle degrees fordifferent heat rejection levels. Again, it seems that

Figure 14: Heat release for the PPC engine withdifferent insulations.

Figure 15: PPC engine performance as a function ofinsulation effects.

Figure 16: NOx and soot mass fractions for differentheat rejection levels.

decreased heat rejection leads to increased temper-atures, however, the increase in NOx production isfar less than in the diesel case. Unlike for the dieselengine, the soot level at the end of simulation is notunaffected by change in insulation, i.e., decreasingheat rejection results in more soot in the cylinder.However, it is apparent that the amount of emissions forthe PPC combustion is considerably lower than in thediesel combustion, around three orders of magnitudelower.

The average temperature in the cylinder, as a functionof crank angle and for different heat rejection levels, isgiven in Fig. 17. It is clear that the PPC combustionresults in lower average cylinder temperatures thanin diesel combustion. This directly corresponds toreduced amounts of emissions, since neither soot norNOx is formed at temperatures lower than around1500 K. There are, however, zones with higher localtemperatures and equivalence ratios in the cylinderthat contribute to the amount of emissions actuallygenerated.

The equivalence ratio, φ, and the cylinder gas temper-ature for the baseline case are shown at a mid-planeof the engine sector in Figs. 18 and 19, respectively.The upper limit (red color) of the contour plots has thesame value as for the diesel case, i.e., T = 2200 K andφ = 2.

Figures 20 and 21 show the equivalence ratio and

7

Figure 17: Average crank angle resolved cylindertemperature for different heat rejections.

cylinder temperature for the 50% heat rejection case atthe same mid-plane. In the figures it can be seen thatthe temperature does not reach a high enough value forNOx to form and the areas, where the equivalence ratiois high enough for soot to form, are almost all too coldfor the soot production to start. This is in agreementwith the emissions shown in Fig. 16. The comparisonbetween Figs. 18 and 19 and Figs. 20 and 21 alsoconfirms that the change in NOx and soot productionbetween the heat rejection cases is minor, due to verylow temperature differences.

(a) TDC (b) 10 CAD ATDC (c) 20 CAD ATDC

(d) 30 CAD ATDC (e) 40 CAD ATDC (f) 50 CAD ATDC

Figure 18: Equivalence ratio for the baseline case ofthe PPC combustion, from TDC to 50 CAD ATDC.

(a) TDC (b) 10 CAD ATDC (c) 20 CAD ATDC

(d) 30 CAD ATDC (e) 40 CAD ATDC (f) 50 CAD ATDC

Figure 19: Cylinder gas temperature for the baselinecase of the PPC combustion, from TDC to 50 CADaTDC.

(a) TDC (b) 10 CAD ATDC (c) 20 CAD ATDC

(d) 30 CAD ATDC (e) 40 CAD ATDC (f) 50 CAD ATDC

Figure 20: Equivalence ratio for the 50% heat rejectionof the PPC combustion, from TDC to 50 CAD ATDC.

(a) TDC (b) 10 CAD ATDC (c) 20 CAD ATDC

(d) 30 CAD ATDC (e) 40 CAD ATDC (f) 50 CAD ATDC

Figure 21: Cylinder gas temperature for the 50% heatrejection of the PPC combustion, from TDC to 50 CADaTDC.

CONCLUSIONS

A study of engine performance, along with an estima-tion of NOx and soot emission levels, was performedon a heavy duty diesel engine for two different com-bustion modes. A specific load point was used as avalidation point for each combustion mode and exper-imental results were used for the validation process.Both baseline cases gave acceptable agreement withthe experimental data, apart from the slight differencein heat release for the diesel combustion mode.

The validated cases were used in a heat rejectionstudy, where the heat flux to the solid walls wasadjusted and used as wall boundary conditions. Forthe simulated cases, the heat release did not seem tobe affected by the change in heat rejection, indicatingthat the combustion itself was relatively unaffected.

Engine performance followed expected patterns forboth combustion modes as a function of heat rejection,i.e., efficiency increased with reduced heat rejection aswell as ISFC reduced and IMEP increased. Decreasedheat rejection caused an increase in NOx emissions,especially in the diesel combustion case. Soot emis-sions did not increase in either case, because in thediesel case the high temperature provided suitable

8

soot oxidation conditions. For the PPC case the localtemperature in the cylinder was sufficiently low to avoidexcessive soot formation.

The mass fractions of both soot and NOx proved tobe around three orders of magnitude lower in the PPCmode compared to the diesel combustion mode. Thisis due to an appropriate combination of φ and T in thePPC combustion mode.

The result from this study show that the combinationof low heat rejection and low temperature combustionis worth studying further in diesel engines. This isimportant for the task of increasing engine efficiency,while maintaining low emission levels.

ACKNOWLEDGEMENTS

This work is a part of the research project D60, whichis financially supported by the Swedish Energy Agency.The AVL FIRE code has been provided by AVL ListGMBH.

CONTACT

Current affiliation of the corresponding author:helgi.fridriksson@energy.lth.se

Lund UniversityDept. of Energy SciencesDivision of Heat TransferBox 118221 00 Lund, SWEDEN

REFERENCES

[1] Heywood, J. Internal Combustion EngineFundamentals. McGraw-Hill, New York, NY, 1988.

[2] Taymaz, I. “An experimental study of energybalance in low heat rejection diesel engine”.Energy, 31(2-3), pp. 364–371, 2006.

[3] Borman, G., and Nishiwaki, K. “Internal-combustion engine heat transfer”. Progress inEnergy and Combustion Science, 13(1), pp. 1–46, 1987.

[4] Arcoumanis, C., Cutter, P., and Whitelaw, D. S.“Heat Transfer Processes in Diesel Engines”.Chemical Engineering Research & Design, 76(2),pp. 124–132, 1998.

[5] Nishiwaki, K. “Modeling Engine Heat Transferand Flame-Wall Interaction”. JSME, pp. 35–44. The Fourth International Symposium onDiagnostics and Modeling of Combustion inInternal Combustion Engines, COMODIA 98,1998.

[6] Torregrosa, A., Olmeda, P., Romero, C., andTermicos, C. “Revising Engine Heat Transfer”.Annals of Faculty Engineering Hunedoara -International Journal of Engineering, pp. 245–265, 2008.

[7] Kleemann, A. P., Gosman, A. D., and Binder, K. B.“Heat transfer in diesel engines: A cfd evaluationstudy”. JSME, pp. 123–131. Proceedings of theFifth International Symposium on Diagnostic andModeling of Combustion in Internal CombustionEngines, Comodia, Nagoya, 2001.

[8] Fridriksson, H., Sunden, B., and Hajireza, S. “Atheoretical study on the heat transfer process indiesel engines”. WIT Transactions on EngineeringSciences, 68, pp. 177–188, 2010.

[9] Annand, W., and Ma, T. “Instantaneous heattransfer rates to the cylinder head surface of asmall compression-ignition engine”. Proceedingsof the Institution of Mechanical Engineers, 185,pp. 976–987, 1970.

[10] Woschni, G. “A universally applicable equation forthe instantaneous heat transfer coefficient in theinternal combustion engine”. SAE International(64), pp. 3065–3083. 670931, 1967.

[11] Hohenberg, G. “Advanced Approaches for HeatTransfer Calculations”. SAE Transactions, 88,pp. 61–77. SAE paper 790825, 1979.

[12] Sharief, A. “Study on Heat Transfer Correlation inIC Engines”. SAE International, 01(1816). 2008-01-1816, 2008.

[13] Depcik, C., and Assanis, D. “A Universal HeatTransfer Correlation for Intake and Exhaust Flowsin a Spark-Ignition Internal Combustion Engine”.SAE International, 111(3), pp. 734–740, 2002.

[14] Chang, J., Guralp, O., Filipi, Z., Assanis, D., Kuo,T. W., Najt, P., and Rask, R. “New Heat TransferCorrelation for an HCCI Engine Derived fromMeasurements of Instantaneous Surface HeatFlux”. SAE Paper. 2004-01-2996, 2004.

[15] Xin, Q. “Theoretical analysis of internalcombustion engine miscellaneous heat losses”.SAE Paper. 2009-01-2881, 2009.

[16] Hajireza, S. “Application of CFD Modeling inCombustion Bowl Assessment of Diesel EnginesUsing DoE Methodology”. SAE Paper. 2006-01-3330, 2006.

[17] Urip, E. “Modeling IC Engine Conjugate HeatTransfer Using the KIVA Code”. Numerical HeatTransfer, Part A: Applications, 52(1), pp. 1–23,2007.

[18] Urip, E., Liew, K. H., Yang, S. L., and Arici,O. “Numerical Investigation of Heat Conductionwith Unsteady Thermal Boundary Conditionsfor Internal Combustion Engine Application”.ASME. ASME International Mechanical Engineer-ing Congress and RD&D Expo, November, 2004.

[19] Tatschl, R., Basara, B., Schneider, J., Hanjalic,K., Popovac, M., Brohmer, A., and Mehring,J. Advanced turbulent heat transfer modelingfor ic-engine applications using avl fire. Int.Multidimensional Engine Modelling, User’s GroupMeeting, Detroit, MI, 2006.

9

[20] Colin, O., and Benkenida, A. “The 3-zonesextended coherent flame model (ECFM3Z) forcomputing premixed/diffusion combustion”. Oil &Gas Science and Technology, 59(6), pp. 593–609, 2004.

[21] Yasar, H. “First and second law analysis of lowheat rejection diesel engines”. Journal of theEnergy Institute, 81(1), pp. 48–53, 2008.

[22] Onishi, S., Jo, S., Shoda, K., Do Jo, P., andKato, S. “Active Thermo-Atmosphere Combustion(ATAC) – A New Combustion Process for InternalCombustion Engines”. SAE Paper. 790501, 1979.

[23] Christensen, M. “HCCI Combustion - EngineOperation and Emission Characteristics”. PhDThesis, Lund University, Lund, Sweden. ISBN 91-628-5424-0, 2002.

[24] Johanson, T. “Turbocharged HCCI Engine -Improving Efficiency and Operating Range”. PhDThesis, Lund University, Lund, Sweden. ISBN987-91-7473-061-6, 2010.

[25] Manente, V. “Gasoline Partially PremixedCombustion; An Advanced Internal CombustionEngine Concept Aimed to High Efficiency, LowEmissions and Low Acoustic Noise in the WholeLoad Range”. PhD Thesis, Lund University, Lund,Sweden. ISBN 987-91-628-8144-3, 2010.

[26] Manente, V., Johansson, B., Tunestal, P., Zander,C.-G., and Cannella, W. “An Advanced InternalCombustion Engine Concept for Low Emissionsand High Efficiency from Idle to Max Load UsingGasoline Partially-Premixed Combustion”. SAEPaper, p. 32. 2010-01-2198, 2010.

[27] Jaichandar, S., and Tamilporai, P. “Low HeatRejection Engines- An Overview”. SAE Paper.2003-01-0405, 2003.

[28] Kamo, R., Kamo, L., Bryzik, W., Schwarz, E.,and Mavinahally, N. “Injection characteristics thatimprove performance of ceramic-coated dieselengines”. SAE Paper. 1999-01-0972, 1999.

[29] Tuner, M. “Studying the Potential Efficiency of LowHeat Rejection HCCI Engines with a StochasticReactor Model”. SAE Paper. 2009-24-0032,2009.

[30] Tuner, M. and Frojd, K and Seidel, L. and Mauss,F. “Diesel-PPC engine: Predictive Full CycleModeling with Reduced and Detailed Chemistry”.JSAE. 20119239, SAE 2011-01-1781, 2011.

[31] Knappe, C. and Andersson, P. and Algotsson, M.and Richter, M. and Linden, J. and Alden, M.and Tuner, M. and Johansson, B. “Lase-InducedPhosphorescence and the Impact of PhosphorCoating Thickness on Crank-Angle ResolvedCylinder Wall Temperatures”. SAE Paper. 2011-01-1292, 2011.

[32] AVL. AVL FIRE v2009 user manual, 2009.

[33] Hanjalic, K., Popovac, M., and Hadziabdic, M. “Arobust near-wall elliptic-relaxation eddy-viscosityturbulence model for CFD”. International Journalof Heat and Fluid Flow, 25(6), pp. 1047–1051,2004.

[34] Akihama, K., Takatori, Y., Inagaki, K., Sasaki, S.,and Dean, A. “Mechanism of the smokeless richdiesel combustion by reducing temperature”. SAEPaper. 2001-01-0655, 2001.

[35] Andersson, O., Somhorts, J., Lindgren, R., Blom,R., and Ljungqvist, M. “Development of the Euro5 Combustion System for Volvo Cars’ 2.4 l DieselEngine”. SAE paper. 2009-01-1450, 2009.

[36] Manente, V. and Tunestal, P. and Johansson, B.“Validation of a Self Tuning Gross Heat ReleaseAlgorithm”. SAE Paper. 2008-01-1672, 2008.

10