Energy 73 (2014) 325e339

Contents lists available at ScienceDirect

Energy
journal homepage: www.elsevier.com/locate/energy

Development of a new multi zone model for prediction of HCCI

(homogenous charge compression ignition) engine combustion,
performance and emission characteristics
Elaheh Neshat, Rahim Khoshbakhti Saray*
Faculty of Mechanical Engineering, Sahand University of Technology, Sahand New Town, Tabriz, Iran

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

Article history: Nowadays, computer models have contributed to a better understanding and solution of long standing
Received 11 January 2014 practical combustion problems in HCCI (homogenous charge compression ignition) engines. In this study,
Received in revised form a new multi zone model is developed for HCCI engines simulation. Model contains four different types of
13 May 2014
zones, which are core, boundary layer, zones between them, and crevice zone. Heat and mass transfer are
Accepted 7 June 2014
considered between all of zones. For accurate calculation of initial conditions at IVC, multi zone model is
Available online 1 July 2014
coupled with a single zone model, which simulates gas exchange process. The developed model is
validated by using two types of fuels, n-heptane as a fuel with low octane number and methane as a fuel
Keywords:
Multi zone model
with high octane number. Semi detailed chemical kinetics mechanisms of fuels are used for their
HCCI (homogenous charge compression combustion simulation. A new heat transfer model is used for calculating convective heat transfer, which
ignition) engine enhanced the model ability in prediction of combustion and performance characteristics of engine
Mass transfer accurately. Model results are in good agreement with experimental data in prediction of in-cylinder
Heat transfer pressure, NOx, CO and UHC emissions. Accurate mass transfer model caused to accurate prediction of
UHC (maximum error is 20.2%) and CO (maximum error is 1.8%). Near zero NOx, which is lower than 10
ppm for all of examined cases, is predicted well, too.
© 2014 Elsevier Ltd. All rights reserved.

1. Introduction engine emissions and also it over-predicts engine in-cylinder peak

pressure [2]. The multi-dimensional models utilize CFD methods
HCCI (Homogenous Charge Compression Ignition) engines have for engine performance simulation. Engine combustion, perfor-
potential for low fuel consumption and low NOx and PM emissions. mance and emission characteristics can be predicted accurately by
These engines have some disadvantages such as lack of direct using these models, if suitable chemical kinetics mechanisms are
control for combustion timing and limited engine operational used. However, these models are costly because they consume
range due to engine knock and misfire. In recent years, many more computational time [3]. The multi zone models divide the
detailed studies have been done to overcome the limitations of combustion chamber volume into multiple zones. These models are
HCCI engines. Computer models, such as single zone, multi zone, not time consuming meanwhile engine combustion, performance
and multi-dimensional, have contributed to a better understanding and emission characteristics can be predicted well using these
and solution of long standing practical combustion problems in models. This kind of models has been extensively reported in the
HCCI engines. Because of importance of chemical kinetics mecha- open literature. It can be seen that, the developed multi zone
nism in simulating HCCI combustion, it is usually utilized in HCCI models differ from each other by the considered configuration, heat
engines simulation coupled with single zone, multi zone and multi- and mass transfer between zones [4e10].
dimensional models [1]. Assanis and Fiveland [4,5] proposed a multi zone model, which
Single zone models are the simplest models for HCCI engine could predict HCCI engine performance and emission related pa-
simulation. In these models, in-cylinder temperature and compo- rameters well especially under turbocharged conditions. This
sition are considered homogenous. This model cannot predict model contained an adiabatic core zone, low temperature bound-
ary layer and crevice zones. There was no heat transfer between
core and other zones. The predicted results for UHC are in good
* Corresponding author. Tel./fax: þ98 411 3459476. agreement with corresponding experimental data but there is
E-mail address: [email protected] (R.K. Saray).

http://dx.doi.org/10.1016/j.energy.2014.06.025
0360-5442/© 2014 Elsevier Ltd. All rights reserved.
326 E. Neshat, R.K. Saray / Energy 73 (2014) 325e339

discrepancy in prediction of CO in comparison with experimental

data.
A comprehensive multi zone model was presented by Komninos
[8,9]. In this model, the combustion chamber volume is divided into
three types of zones, core zone, outer zones and crevice zones. Heat
transfer and mass exchange between zones are considered. The
results indicate that the model can predict engine performance and
emission related parameters well for the only case studied in these
works. The fuel was iso-octane and the outer zones thickness was
considered some constant values, without any description about
the selection procedure for those zones thickness.
Kongsereeparp [11] developed a segregated multi zone model.
Work and heat transfer between zones were considered but there
was no mass transfer between them. Temperature differences be-
tween the cylinder wall, piston crown and cylinder head were
considered, too. Various fuels with different octane numbers were
used as engine fuel. Results showed that the model could predict
in-cylinder pressure, accurately. But, the results were not reported
about emissions prediction.
A new multi zone model was developed by Kodavasal et al. [12].
In this model there were no mass and heat exchanges between Fig. 1. Geometry of zones.
zones. It is noted even though the zones have had constant mass
distribution, their volume fractions have changed at different time
steps. Heat has transferred from all of zones to walls. Gasoline was crevice zone and from crevice zone to the boundary layer zone.
used as HCCI engine fuel and model has had an acceptable accuracy Boundary layer zone is the nearest zone to the wall. It is a hollow
in pressure prediction. There were no reported results for cylinder and it is in contact with cylinder wall, cylinder head and
emissions. piston head. Thickness of this zone is in a direct proportion of
Guo et al. [13] developed another multi zone model, which was cylinder height near the cylinder head and piston head and the
contained boundary layer, crevice and some core zones. Heat same proportion of bore near the cylinder wall. Hence, its thickness
transfer was considered between zones but there was no mass is varying by time in axial direction and it is constant in radial di-
transfer between them. The multi zone model was able to rection. There are interfaces between this zone and all of other
reasonably capture the experimentally measured cylinder pressure zones, so mass and heat transfer from this zone to others and vice
variations over a range of operation conditions. It also provided versa. There is convective heat transfer between this zone and all
qualitative trends of CO emissions. The numerical simulation over the walls of combustion chamber. Core zone is the innermost zone
predicted the combustion completeness at extremely high air/fuel of the cylinder. This zone is a cylindrical zone, its radius is in a direct
ratios and higher engine speeds. Some differences were observed proportion of bore, and its height is varying by time and it can be
between the experimental and numerical data for NOx emissions. calculated using equation (1). Outer zones are some zones between
Notwithstanding the numerous amount of works in the multi core and boundary layer. Their height is equals to the core height
zone combustion modeling of HCCI engines, there is a need to and their thickness can be obtained using equation (2). Each of core
develop a suitable model for prediction of all engine related pa- and outer zones exchanges heat and mass with their neighboring
rameters. Therefore, the main purpose of this paper is to develop a zones. From the given brief description, it is clear that thicknesses
multi zone model for HCCI engines to predict the engine combus- of zones are constant in radial direction but they are varying by
tion, performance and emission characteristics accurately for time in axial direction which causes to work production (except for
different types of fuels in a wide range of operating conditions. crevice zone).
More attention was paid on zones configuration, accurate estima-
tion of trapped conditions and heat and mass transfer between 4Vtot
lzi ¼  2thickBL;a i ¼ 1; …; nz  2 (1)
neighboring zones. pD2

2. Model description 0:5D  thickcore  thickBL

thicki ¼ i ¼ 2; nz  2 (2)
nz  3
Closed part of engine cycle is simulated by using multi zone
model. It is notable that temperature and composition are assumed
to be uniform in each zone and pressure is assumed to be uniform
throughout the cylinder at each time step. 2.1.2. Governing equations
For each zone, first law of thermodynamics equation and
2.1. Characteristics of zones chemical kinetics equations are solved, simultaneously. Equation
(3) shows first law of thermodynamics for each zone. All terms of
2.1.1. Geometry of zones equation (3) are defined in equations (4)e(9). In this equation, Q is
Fig. 1 shows zone configuration in the developed multi zone the quantity of total heat supplied to the each zone by its sur-
model. As it is shown in this figure, there are four types of zones roundings (conduction or convection) and generated heat in each
containing crevice zone, boundary layer zone, outer zones and core zone by chemical reactions. Transferred heat to each zone consists
zone. Volume of crevice zone is constant and equal to 3% of clear- of three types; conductive heat transfer, convective heat transfer
ance volume. This zone has no heat transfer with other zones or and the thermal energy which is transferred by mass movement
wall and consequently its temperature is constant and equal to the from one zone to another. Details on these types of energy transfer
wall temperature. Mass transfers from boundary layer zone to are explained in the next sections. It is also worth noting that the
 E. Neshat, R.K. Saray / Energy 73 (2014) 325e339 327

quantity of convective heat transfer is zero for all zones except for zones. Equation (15) is used for computing conductive heat transfer
boundary layer, and conductive heat transfer and mass transfer can quantity between zones. In this equation, ktot is the total conduc-
only take place between two zones that are in direct contact with tivity for zones. Total conductivity is the sum of laminar and tur-
each other. Net production rate of each of species is also included in bulence conductivity. It is computed using Yang and Martin's
equation (10) for each zone. It should be noted that temperature of method which is already used by Komninos for HCCI engines [8,9].
zones and mass fraction of species within zones are unknowns. Equations (16)e(21) are used for calculating ktot. In these equations
a is a constant and equals to 0.06, k is Von Karman constant and
dUi dWi dQi equals to 0.41 and C, the proportionality constant, which produces
¼ þ (3)
dt dt dt characteristic velocity from engine speed, is equal to 0.32.

dUi dT Xns
dYj X ns
dmi dQi;cond vT
¼ civ mi i þ mi uj þ uj Yj (4) ¼ ktot A (15)
dt dt dt dt dt vx
j¼1 j¼1

ktot ¼ klam þ ktur (16)

dWi dV
¼P i (5)
dt dt
P rlam mtur
ktur ¼ k (17)
dQi dQi;cond dQi;conv dQi;mtran P rtur mlam lam
¼ þ þ (6)
dt dt dt dt h  i
mtur
¼ krnþ 1  exp 2akrnþ (18)
dQi;cond vT mlam
¼ kA (7)
dt vx
Zrn
 vel
dQi;conv vT 0 rnþ ¼ rdrn (19)
¼ kA  (8) mnz 1
dt vx x¼D;L 0
2 2

dQi;mtran dmin;i dmout;i S

¼ hin  hout (9) vel ¼ crpm (20)
dt dt dt 30
 h  i
dYk;i u_ k;i Mwk Pr
¼ (10) ktot ¼ klam 1 þ lam krnþ 1  exp 2akrnþ (21)
dt ri P rtur

After solving equations of first law and chemical kinetics for all of Because of cylindrical geometry of zones, the area between two
zones, in-cylinder pressure is calculated by using equations zones is varying with radius. Therefore, for estimating conductive
(11)e(14). It can be seen from equation (14) that in-cylinder pres- heat transfer between two zones (except for boundary layer),
sure can be computed by knowing characteristics of all zones such through the cylinder wall, equation (22) is used.
as volume, temperature, mass and composition. Estimated pressure
is constant throughout the combustion chamber but varies over the dQi;cond pli ðTiþ1  Ti Þ
¼ 0 1 i ¼ 1; …; nz  2 (22)
time. dt 0 1
B C
  Br þthicki C lnB riþ1 C
mi ln@ i r 2 A @r þthicki A
PVi ¼ Ru T i ¼ 1; … nz (11) i i 2
Mwi i
ki
þ kiþ1
 
V mi Contact areas between inner zones and boundary layer are
P i ¼ Ru i ¼ 1; … nz (12)
Ti Mwi constant and so conductive heat transfer can be estimated using
equation (15) by a linear approximation of temperature gradient.
X
nz Xnz All zones are in contact with boundary layer and those zones can
Vi mi
P ¼ Ru (13) exchange heat and mass with this zone.
Ti Mw i
i¼1 i¼1 The second part of heat transfer is convection. Convective heat
transfer occurs between fuel and air mixture in the boundary layer
Pnz mi
Ru zone and wall of combustion chamber. Equation (23) is used for
i¼1 Mwi
P¼ Pnz Vi
(14) estimating convective heat transfer.
i¼1 Ti

dQBL;conv pklam;BL D2 ðTw  TBL Þ

¼
dt thickBL;a
 
2.2. Heat transfer
2pklam;BL 4V pD
tot
2  2thick BL;a DðTw  TBL Þ
þ (23)
In the developed model, heat transfer belongs to two main thickBL;r
categories. First part is a mechanism similar to conduction which
occurs between zones that are in direct contact with each other. Equation (23) is a developed version of equation (8) by linear ap-
According to Fig. 1, it can be observed that each zone exchanges proximations for temperature and distance and total area of cyl-
heat with boundary layer zone and with its inner and outer zones. inder walls. Convective heat transfer for other zones is equal to zero
Fourier's law is used to calculate conductive heat transfer between because there is no contact area between them and cylinder walls.
328 E. Neshat, R.K. Saray / Energy 73 (2014) 325e339

0 1
2.3. Mass transfer X
i1
if @Dmolei  moletranðj;iÞ A < 0 then
When the above mentioned governing equations for all zones j¼1
are solved, the new in-cylinder charge temperature and
composition will be calculated for each zone. Changes in tem- 0 1
perature and composition cause to change in-cylinder charge X
i1
mtranði;iþ1Þ ¼ raðiÞMwðiÞ@Dmolei  moletranðj;iÞ A (30)
pressure which leads to mass movement from one zone to j¼1
another. A suitable mass transfer model can be used for accurate
prediction of engine exhaust emissions. New quantity for moles 0 1
of each zone is obtained by using new in-cylinder pressure and X
i1

zone temperature in equation of state of ideal gas. Difference mtranði;nz 1Þ ¼ ð1  raðiÞÞMwðiÞ@Dmolei  moletranðj;iÞ A
j¼1
between new and old mole numbers of each zone causes to mole
(mass) transfer between zones. Equations (24)e(30) are used for (31)
estimation of mole transfer between zones. At first, new moles of
each zone are calculated. Because of lacking any chemical reac- mtranði;iþ1Þ
tion during mass transfer, total moles of combustion chamber are
moletranði;iþ1Þ ¼ (32)
MwðiÞ
constant. Therefore, summation of new moles of all zones is
equal to summation of old moles of all zones. Quantity of new mtranði;nz 1Þ
mole for core zone is calculated using equation (28), and quan- moletranði;nz 1Þ ¼ (33)
MwðiÞ
tities of new moles for other zones are estimated by applying
equation (27).
Ai;iþ1
! raðiÞ ¼   (34)
mi;new Ai;iþ1 þ Ai;nz 1
PVi ¼ Ru T i ¼ 1; … nz (24)
Mwi;new i 0 1
X
i1
if @Dmolei  moletranðj;iÞ A > 0 then
Vi V1 j¼1
! ¼  (25)
mi;new m1;new
Mwi;new Ti Mw1;new T1
mtranði;iþ1Þ mtran;ði;nz 1Þ X
i1
þ ¼ Dmolei  moletranðj;iÞ (35)
Mwði þ 1Þ Mwðnz  1Þ j¼1
mi;old
molei;old ¼ (26)
Mwi;old raðiÞ
mtranði;iþ1Þ ¼ m (36)
ð1  raðiÞÞ tranði;nz 1Þ
1 T
molei;new V !
¼ T1 (27)
mole1;new i Pi1
Vi raðiÞ Dmolei  j¼1 moletranðj;iÞ
mtranði;iþ1Þ ¼ raðiÞ ð1raðiÞÞ
(37)
X X þ Mwðn
nz
mi;old nz
mi;new T X nz
Vi Mwðiþ1Þ z 1Þ
¼ ¼ mole1;new 1 (28)
i¼1
Mw i;old i¼1
Mw i;new V 1 i¼1 Ti
ð1  raðiÞÞ
mtran;ði;nz 1Þ ¼ mtranði;iþ1Þ (38)
Next step is calculating mole difference for each zone; Dmolei. raðiÞ
If Dmolei is greater than zero, it means moles from other zones
have entered to the ith zone. If it is less than zero, some moles mtranði;iþ1Þ
moletranði;iþ1Þ ¼ (39)
leave the ith zone and if it is equal to zero, there is no mass Mwði þ 1Þ
transfer between ith zone and other zones. Similar to heat
transfer, mass transfer is occurred just between zones which mtranði;nz 1Þ
are in direct contact to each other. Equations (29)e(40) explain moletranði;nz 1Þ ¼ (40)
Mwðnz  1Þ
the mechanism of mass transfer between zones. moletran (j,i) is
the transferred mole between jth and ith zones. Computational The most important advantage of mass transfer modeling is mass
procedure starts when i is equal to one, at first step total mole exchange between all zones and consequently it can affect pollut-
changes of this zone is calculated. Then transferred mass be- ants formation. By employing mass transfer between zones, the
tween core and each of neighboring zones are estimated. At the model is able to predict most important pollutants in HCCI com-
next step share of each of neighboring zones in mass transfer bustion, CO and UHC, accurately.
must be specified. Total transferred mass is divided into
two parts, first part enters (or leaves) zone two and second
part enters (or leaves) boundary layer. Quantity of each part
2.4. Initial conditions estimation
is assigned by the ratio of interface area between that zone
and core. After calculating transferred mass between these
To estimate initial values for in-cylinder charge pressure, tem-
three zones, equations are solved for zone two and for other
perature and composition, a single zone thermodynamic model is
zones.
coupled to multi zone model to simulate gas exchange process.
Equations (41)e(46) are used in gas exchange process modeling
Dmolei ¼ molei;new  molei;old (29) and are solved simultaneously.
 E. Neshat, R.K. Saray / Energy 73 (2014) 325e339 329

dT X
ns
dYj Xns
dm for solving coupled equations. The convergence error of tempera-
cv m þm uj þ uj Yj ture and mass fraction of species are considered to be 10e25. Its
dt dt dt
j¼1 j¼1 value for transferred heat and mass energies is equal to 10e12. The
dV dmin dmout dQconv flowchart of the code developed by the authors is given in Fig. 2. In
¼ P þ h  hout þ (41)
dt dt in dt dt

dYj u_ j Mwj
¼ (42)
dt r

 2 
dQconv vT pD
¼ kA ¼ kðTw  TÞ þ 2kpL (43)
dt vx L

m
P¼ Ru T (44)
Mw
0 0 11
dV B1 B dq  1
1
dqCC
¼ Vc @ ðRc  1Þ@sin q  Cr2  sin2 q
2
ðsin 2qÞ AA
dt 2 dt 2 dt

(45)
Mass flow rates through valves are calculated based on the
approach explained in Ref. [14]. These formulas are shown in
equations (46) and (47). In these equations, CD is discharge coef-
ficient, which is defined as the ratio of effective flow area to
reference flow area. CD is a function of valve lift and it is calculated
for CFR engine based on the approach explained in Ref. [11].
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 3
u
 1 u
u  g1    g
dm CD AR Pu Pd u 2g 6 6 Pd 7
7 if Pd > 2
g g 1g
¼ u 4 1  5
dt 1
ðRu Tu Þ2 Pu tg1 Pu Pu gþ1

(46)

  gþ1    g
dm CD AR Pu 2 2ðg1Þ Pd 2 1g
¼ if  (47)
dt ðRu Tu Þ
1
2 g þ 1 Pu gþ1

Gas exchange process model is used to calculate in-cylinder tem-

perature, pressure and composition at trapped condition to initiate
multi zone combustion model calculations.

2.5. Problem solving algorithm

At first time step, there is no difference between temperature

and composition of various zones and consequently there are no
mass and heat transfer between zones. Therefore, values of
conductive heat transfer energy and mass transfer energy in first
law equation are equal to zero. Then, first law equation is solved
and temperature and composition of all zones are calculated in this
step. In the next step, mass and heat transfer subroutines are called
and new composition of each zone and energy transfer due to mass
and heat transfer are calculated. Then first law equation is solved
again using new heat and mass transfer energies and this path
continues until temperature, composition, transferred heat energy
and transferred mass energy are converged. Computed mass
transfer and energy transfer are used in next time step calculations
as initial assumptions. For all time steps, this problem solving
method is used until EVO. When exhaust valve opens, calculations
are followed by gas exchange process model. In this model, inlet
and outlet mass are calculated for each time step. The composition
of outlet mass is the same as in-cylinder charge composition and
the composition of inlet mass consists of fuel, air, and EGR (exhaust
gas recirculation), introduced into the engine. DVODE solver is used Fig. 2. Algorithm for problem solving.
330 E. Neshat, R.K. Saray / Energy 73 (2014) 325e339

Fig. 3. Experimental set-up.

Table 1 pressure, then mass, enthalpy and chemical species are transferred
CFR engine specifications. between zones and the second stage is repeated until convergence.
Engine type Waukesha In present study all of equations are solved simultaneously.
Displacement Volume (cm3) 612
Bore (cm) 8.25 2.6. Experimental set-up
Stroke (cm) 11.4
Connect Rod. (cm) 24
IVO (deg aTDC) 10 Experimental data was obtained from University of Alberta
IVC (deg aTDC) 214 engine research facility. All experiments were performed using a
EVO (deg aTDC) 500 Waukesha CFR (Cooperative Fuel Research) single cylinder research
EVC (deg aTDC) 15
engine coupled to a DC motoring dynamometer. The experimental
Compression Ratio Variable
set-up is shown in Fig. 3. The basic engine specifications for the
current study are shown in Table 1. The engine was maintained at
two constant speeds of 700 and 800 rpm and ran with an open
Table 2
Experimental uncertainty. throttle. The intake system included a 2.4 kW heater with a PID
temperature controller to preheat intake air when required. Two
Parameter Uncertainty (%)
types of fuel injectors, one for the n-heptane and the other for NG
Temperature <1 (natural gas), were located upstream of the intake valve to facilitate
Crank angle position <1 proper mixing. The control module of an AFS Sparrow-II engine was
Engine speed <2
Fuel Flow rate <1
utilized to regulate the injection rate of each fuel separately or as a
Air flow rate <1 blend. External uncooled EGR line was connected to the intake after
EGR <1 the heater and before the fuel injectors and was controlled by a
IMEP <2 manual butterfly valve. Intake gas was analyzed for the EGR frac-
SOC <2
tion determination. A Vetronix PXA-1100 portable gas analyzer,
Pmax <2
MHR <2 which is capable of measuring different gases, was used in the
CA50 <2 intake system and EGR connection to the intake plenum to deter-
mine the CO2 concentration in the intake mixture. EGR was
calculated using volume concentration measurement of CO2 up-
stream and downstream of the engine. EGR was calculated as
the algorithm proposed by Komninos [8,9], it can be seen that the
calculation for each CA step is completed in two stages: First CO2:up
combustion is assumed to occur without mass transfer and an in- EGR ¼ 100  (48)
CO2:down
termediate zone temperature is determined from the first ther-
modynamic law. At the second stage, the total cylinder pressure is The lubricating oil was maintained at a constant temperature
calculated using equation of state and the current values of zone using a heater. A BEI rotary incremental encoder with a resolution
temperature, mass and volume. For each zone a new temperature is of 0.1 CAD was used to monitor engine rotational speed and coor-
estimated from the first thermodynamic law and a new mass is dinate the pressure trace with respect to crank position. Cylinder
obtained using equation of state. If the first thermodynamic law has pressure was measured with a Kistler 6043A, water-cooled pres-
not converged or the estimated new zone mass differs significantly sure transducer in combination with a Kistler 507 charge amplifier.
from the value initially assumed when calculating cylinder The pressure transducer was mounted in the cylinder wall close to

Table 3
CFR engine operating conditions for six considered cases.

Case number Fuel Rc Lambda EGR (mass base %) rpm PIVC (bar) TIVC (K) Excess air (%) Volumetric efficiency

1 n-heptane 12.7 3.87 0 700 1.527 423.076 73.78 90.11

2 n-heptane 12.7 2.32 30 700 1.512 422.692 56.30 60.66
3 n-heptane 12.7 2.46 40 700 1.525 415.216 58.81 53.15
4 Methane 21.5 3.63 0 800 1.515 441.256 68.22 80.73
5 Methane 21.5 3.17 10 800 1.501 463.826 67.02 71.71
6 Methane 21.5 2.05 40 800 1.491 482.541 48.19 42.11
 E. Neshat, R.K. Saray / Energy 73 (2014) 325e339 331

Fig. 4. Comparison of experimental and calculated pressure traces for examined cases.
332 E. Neshat, R.K. Saray / Energy 73 (2014) 325e339

Table 4
Numerical and experimental values of engine exhaust emissions.

Case Exp. CO2 Num. CO2 Exp. CO Num. CO Exp. UHC Num. UHC Exp. NOx Num. NOx
number (% by Vol.) (% by Vol.) (% by Vol.) (% by Vol.) (ppm) (ppm) (ppm) (ppm)

1 3.71 3.71 0.073 0.073 995 867 0.11 0.14

2 6.72 6.71 0.054 0.054 1261 1006 0.34 0.72
3 6.33 6.32 0.149 0.148 1705 1573 0.57 0.89
4 3.45 3.46 0.073 0.073 3919 3889 0.41 0.60
5 6.63 6.64 0.103 0.103 4293 4178 0.00 0.03
6 6.10 6.10 0.056 0.057 4575 4496 8.86 9.31

methane, as a fuel with high octane number, are used as engine

fuels. The chemical kinetics mechanism of Valeri, which contains 57
chemical species and 290 chemical reactions [17], was used for
modeling of n-heptane oxidation. GRI Mech 3.0 chemical kinetics
mechanism was used for methane combustion simulation. The
mechanism contains 53 species and 325 reactions [18]. Wide range
of operating conditions used in this investigation is presented in
Table 3. Cases are in a wide range of lambda and EGR.
Fig. 4 shows the pressure history of the examined cases both
experimentally and numerically. The model predictions and
experimental data are in good agreement as indicated in this figure.
Further analyses show that the model can predict SOC for n-hep-
tane and methane cases accurately. SOC is considered the location

Fig. 5. In-cylinder pressure curves for different numbers of zones.

the cylinder head. Experimental data were acquired using a per-

sonal computer running custom Labview software and using three
high sampling rate NI PCI-MIO-16E1 data acquisition cards.
The intake pressure at the IVC time was taken as the reference
point for the pressure trace signal. The high frequency noise was
filtered digitally from the pressure trace signal. The pressure traces
measured over 100 successive cycles were averaged to avoid cyclic
variations before they were used in calculations. Cyclic variation in
the HCCI combustion is less than that of the conventional com-
bustion, but as far as the absolute estimated error from internal and
external errors is concerned the deviation is significant. Averaging
over 100 repeated cycles, the cyclic error related to each parameter
was predicted. The uncertainty for engine parameters and com-
bustion parameters are displayed in Table 2 [15,16].

3. Results and discussion

The developed model was validated against experimental data

to examine its accuracy. Six different operating conditions are
selected from experiments done on CFR engine and the model was
used for prediction of engine combustion, performance and emis-
sion characteristics. Heptane, as a fuel with low octane number, and

Table 5
Calculated engine exhaust emissions using different zone numbers.

nz CO2 (% by Vol.) CO (% by Vol.) UHC (ppm) NOx (ppm)

4 2.51 0.051 682 0.21

7 3.23 0.063 798 0.16
11 3.71 0.073 867 0.14
20 3.71 0.073 870 0.13
Fig. 6. Zonal mass distributions for Cases 1 and 6.
 E. Neshat, R.K. Saray / Energy 73 (2014) 325e339 333

where 10% of cumulative chemical energy is released. The Engine combustion, performance and emissions characteristics
maximum amount of model uncertainty in prediction of peak can be predicted accurately by using a suitable number of zones. On
pressure value is less than 1.5% and 0.6% for methane cases and n- the other hand lower running time needs to be considered in
heptane cases, respectively. determining the number of zones. Fig. 5 shows in-cylinder pressure
Table 4 evaluates numerical and experimental engine exhaust trace for four different zone numbers. As it is shown in this figure, 4,
emissions for all of examined cases. It can be concluded from 7, 11 and 20 are used as zone numbers. The main purpose of using
Table 4 that model can predict CO and CO2 species concentration different zone numbers is predicting the aforementioned engine
accurately. UHC is rather under predicted because exhaust UHC in characteristics, which are independent of zone number. Results
addition to fuel oxidation, may be produced from lubricating oil show that using 11 zones, in-cylinder pressure can be predicted as
decomposition or other sources [19]. NOx is over predicted but it is accurate as using 20 zones. Table 5 shows calculated exhaust
worth noting that these species concentration are well predicted by emissions in comparison with corresponding experimental values
the model. Such noteworthy accuracy in prediction of species for different zone numbers. A comprehensive review on the results
concentration in exhaust gases is obtained by using suitable mass in Table 5 and Fig. 5 indicates that both 11 and 20 can be used as
transfer model employed in the developed multi zone combustion number of zones. Runtime of problem solving for 20 zones is 2
model. During compression and combustion processes, mass times higher than 11 zones. Trade of between accuracy and run-
(mixture of fuel and air) leaves all inner zones due to in-cylinder time, caused to finally 11 used as number of zones.
pressure rise, enters to the boundary layer, and then transfers Fig. 6 indicates mass distribution in different zones of combus-
from boundary layer to the crevice zone. When expansion starts, tion chamber for cases 1 and 6. As it is shown in this figure, mass of
stored mass expels out from crevice zone and enters to other zones. all zones, except for boundary layer and crevice zones, is decreased
The expelled mass is unburned. When it enters other zones, tem- during compression and combustion processes and is increased
peratures and pressure of zones are not high enough for a complete during expansion process. It causes to increase mass in crevice
combustion of new entered mass. Therefore, it is not burned or volume during compression and combustion, and it is decreased
combustion remains incomplete. Hence, UHC and CO emissions are during expansion stroke.
formed. Fig. 7 shows variation of in-cylinder CO, NOx and UHC during
closed loop of engine for Cases 1 and 6. As indicated in this figure,
the variations of these species can be interpreted by different
processes occurred in the engine cycle such as combustion and
consequently mass and energy transfer during this process and
expansion stroke. Chen et al. [20] has performed some analyses on
methane combustion. Their results show that methane and CO
variations are similar to current study during combustion process.
NOx is predicted near zero. Published researches show that HCCI
engine produces low NOx emissions [21].
Fig. 8 shows the zonal distribution of emissions for Cases 1 and
6. This figure shows that CO formation has been occurred mainly at
different temperatures for methane. It may be due to low tem-
perature reactions of n-heptane oxidation chemical kinetics
mechanism [22]. As it is shown in this figure, higher amount of CO
is observed in boundary layer at EVO. Core zone has the largest
volume around TDC, so mass of fuel and air mixture in this zone is
larger than other zones. When the mixture temperature is
increased due to compression, mixture follows partial oxidation
reactions and maximum amount of CO forms in core zone. How-
ever, because of high temperature of core zone, all of CO reacts with
oxygen and forms CO2. Therefore, there is a little amount of CO at
EVO in core zone. Mixture temperature in crevice zone is very low,
hence oxidation reactions may not take place in this zone. But
because of mass transfer, CO enters to this zone around TDC from
other zones and leaves this zone during expansion process. The
largest amount of CO is seen in boundary layer at EVO because this
zone is nearest zone to crevice zone. Therefore, UHC and CO from
crevice zone enter to the boundary layer during expansion process.
The large part of the UHC through an incomplete combustion
converts to CO and it cause to appearance of large amounts of CO at
EVO in the boundary layer. It is notable that some of CO leaves
boundary layer and enters to other zones, like zone 9, which has
largest interface shared with boundary layer zone.
Zonal temperature distribution is shown in Fig. 9 for Cases 1 and
6. As it is observed from this figure, temperatures of inner zones are
higher than outer zones during compression and expansion pro-
cesses and consequently combustion starts earlier in the inner
zones. Temperature of crevice volume is constant and equals to wall
temperature. LTHRR (low temperature heat release region) of n-
heptane oxidation is also observed in this figure. LTHRR in n-hep-
Fig. 7. In-cylinder CO, NOx, and UHC variations versus crank position. tane oxidation causes to a local jump on temperature and pressure
334 E. Neshat, R.K. Saray / Energy 73 (2014) 325e339

Fig. 8. Zonal distributions of CO and UHC versus crank position for Cases 1 and 6.
E. Neshat, R.K. Saray / Energy 73 (2014) 325e339 335

Fig. 8. (continued).
336 E. Neshat, R.K. Saray / Energy 73 (2014) 325e339

Fig. 9. Zonal temperature distributions versus crank position for Cases 1 and 6.

curves. LTHRR occurs because of reaction of fuel radical with O2 no similar region for methane oxidation due to its only high tem-
which produces alkylperoxy radical and this reaction is followed by perature combustion. Fig. 10 shows that numerical heat release rate
other reactions to produce HO2, olefins, cyclic ether, and OH. When is in good agreement with experimental data especially for n-
these molecules are produced, fuel consumption stops and heat heptane. Experimental heat release rate is calculated using the
release is decreased (NTC (negative temperature coefficient)). At modified heat release analysis method introduced by Fathi et al. for
high temperatures, olefins decompose to smaller hydrocarbons and HCCI engines [15].
their decomposition leads to CO formation [23]. Unlike the large
hydrocarbons, methane has a negligible NTC region [24]. Pnz Pns     
mj;i hj;i  mj;i hj;i
Heat release rate and cumulative heat release of Cases 1 and 6 i¼1 j¼1 CAþDCA CA
HRRðCA þ DCAÞ ¼
are indicated in Fig. 10. HRR is calculated using equation (49) for DCA
each case. LTHRR of n-heptane oxidation is clear in Fig. 10. There is (49)
 E. Neshat, R.K. Saray / Energy 73 (2014) 325e339 337

Fig. 11. Variations of pressure histories of various calculation loops.

Fig. 10. Variation of heat release rate and cumulative heat release versus crank posi-
tion for Cases 1 and 6.
previous loop. In-cylinder temperature and pressure are calculated
from energy and state equations, respectively.
Oscillating HRR for methane is because of various positions for In-cylinder mass variations of Cases 1 and 6 are shown in Fig. 12.
start of combustion at different zones. Combustion occurs at In-cylinder mass always is higher than zero, therefore it can be
different crank angles in different zones and it causes to observe concluded that there is some IEGR in the cylinder.
multiple peaks in HRR curve.
Because of simulating gas exchange process, calculation for each
case was converged after few loops. Convergence was occurred for
Case 1 after 5 loops, for Cases 2 and 3 after 6 loops and for Cases
4e6 after 4 loops. Fig. 11 illustrates in-cylinder pressure histories in
different loops for Cases 1 and 6. A convergence criterion of 1.0e-06
was employed for IVC in-cylinder charge pressure, temperature and
composition (all of species) in the model. Initial conditions of first
loop, containing pressure and composition at IVC, are obtained
from available experimental data and some approximations. It is
assumed that there is no internal EGR in the combustion chamber,
and external EGR is composed of CO2, H2O, O2, N2, and AR for the
first loop. In-cylinder temperature is calculated using state equa-
tion and experimental data of in-cylinder pressure and average
amount of introduced mass. For other loops, IVC conditions are
calculated from gas exchange process. Therefore, usually, there is
some internal EGR (IEGR) in the cylinder charge. Composition of
external EGR is taken from the composition of exhaust gases of Fig. 12. Mass variations versus crank angle for Cases 1 and 6.
338 E. Neshat, R.K. Saray / Energy 73 (2014) 325e339

4. Conclusion U internal energy (J)

u specific internal energy (J kg1)
A new multi zone model is developed for HCCI engine simula- V volume (m3)
tion. This model contains conductive heat transfer between zones, vel characteristic velocity of engine
convective heat transfer between boundary layer and cylinder wall W work (J)
and mass transfer between zones. Multi zone model is coupled x axial distance (m)
with a one dimensional thermodynamic model for gas exchange Y mass fraction
process simulation. The developed model is validated by using two
types of fuels, heptane as a fuel with low octane number and Greek symbols
methane as a fuel with high octane number. A new heat transfer a constant
model is used for calculating convective heat transfer, which t ratio of specific heat
enhanced the model ability in prediction of combustion and per- q crank angle position ( )
formance characteristics of engine accurately. Model results are in k Von Karman constant
good agreement with experimental data in prediction of in- m viscosity (kg/ms2)
cylinder pressure, NOx, CO and UHC emissions. Accurate mass r density (kg/m3)
transfer model caused to accurate prediction of UHC (maximum _u molar rate of production (mol/m3s)
error is 20.2%) and CO (maximum error is 1.8%). Near zero NOx is
predicted well, too.
Subscripts
a axial
Acknowledgment
BL boundary layer
d downstream
The authors thank gratefully Professor M. D. Checkel for
i ith zone
providing the permission to conduct experiments in Engine
J jth species
Research Laboratory of University of Alberta, Edmonton, Canada. In
lam laminar
addition, the authors wish to thank gratefully Dr. V. Hosseini from
tran transferred
Sharif University of Technology, due to his ongoing scientific
tur turbulent
support.
U upstream
Nomenclature
References
A area (m2)
[1] Zhao F, Assanis DN, Asmus TN, Dec JE, Eng JA, Najt PM. Homogeneous charge
AR effective flow area (m2) compression ignition (HCCI) engines. USA: SAE; 2003.
CA crank angle [2] Amjad AK, Saray RK, Mahmoudi SMS, Rahimi A. Availability analysis of n-
Cr connecting road heptane and natural gas blends combustion in HCCI engines. Energy 2011;36:
6900e9.
cv specific heat constant at constant volume (J kg1 K1) [3] Kong SC, Marriot CD, Rutland CJ, Reitz RD. Experiments and CFD modeling of
CD discharge coefficient direct injection gasoline HCCI engine combustion. SAE; 2002.
CD combustion duration [4] Fiveland SB, Assanis DN. Development of a two-zone HCCI combustion model
accounting for boundary layer effects. SAE; 2001.
D cylinder diameter (m) [5] Fiveland SB, Assanis DN. Development and validation of a quasi-dimensional
EGR exhaust gas recirculation model for HCCI engine performance and emissions studies under turbo-
EVC exhaust valve closing angle charged conditions. SAE; 2002.
[6] Easley WL, Agarwal A, Lavoie GA. Modelling of HCCI combustion and emis-
EVO exhaust valve opening angle
sions using detailed chemistry. SAE; 2001.
h enthalpy (J kg1) [7] Ogink R, Golovitchev V. Gasoline HCCI modeling: an engine cycle simulation
IVC inlet valve closing angle code with a multi-zone combustion model. SAE; 2002.
[8] Komninos NP, Hountalas DT, Kouremenos DA. Description of in-cylinder
IVO inlet valve opening angle
combustion processes in HCCI engines using a multi-zone model. SAE; 2005.
k thermal conductivity (W m1 K1) [9] Komninos NP. Assessing the effect of mass transfer on the formation of HC and
L length of combustion chamber (m) CO emissions in HCCI engines, using a multi-zone model. Energy Convers
lz length of zone (m) Manage 2009;50:1192e201.
[10] Tzanetakis T. Multi zone modeling of a primary reference fuel HCCI engine.
m mass (kg) University of Toronto; 2006.
MW molecular weight [11] Kongsereeparp P. Chemical kinetic based simulation for an HCCI engine and
nz number of zones its combustion canada. University of Alberta; 2008.
[12] Kodavasal J, McNenly MJ, Babajimopoulos A, Aceves SM, Assanis DN,
ns number of species Havstad MA, et al. An accelerated multi-zone model for engine cycle simu-
P pressure (Pa) lation of homogeneous charge compression ignition combustion. Int J Engine
Q heat (J) Res 2013;14:416e33.
[13] Guo H, Li H, Neill WS. A study on the performance of combustion in a HCCI
r radius (m) engine using n-heptane by a multi-zone model. In: ASME 2009 internal
ra area fraction combustion engine division fall technical conference. Switzerland; 2009.
Rc compression ratio [14] Heywood JB. Internal combustion engine fundamentals. Singapore: McGraw
Hill Inc; 1998.
rpm engine speed (revolution per min) [15] Fathi M, Saray RK, Checkel MD. Detailed approach for apparent heat release
rn normal distance from cylinder wall (m) analysis in HCCI engines. Fuel 2010;89:2323e30.
Ru universal ideal gas constant (J mol1 K1) [16] Bahlouli K, Atikol U, Saray RK, Mohammadi V. A reduced mechanism for
predicting the ignition timing of a fuel blend of natural-gas and n-heptane in
S stroke (m)
HCCI engine. Energy Convers Manage 2014;79:85e96.
SOC start of combustion [17] Golovitchev VI, Atarashiya K, Tanaka K, Yamada S. Towards universal EDC-
T temperature (K) based combustion model for compression ignited engine simulation. SAE;
t time 2003.
[18] Frenklach M, Wang H, Goldenberg M, Smith GP, Golden DM, Bowman CT,
TDC top dead center et al. GRI-Mech e an optimized detailed chemical reaction mechanism for
thick thickness (m) methane combustion. GRI-Mech; 1995.
 E. Neshat, R.K. Saray / Energy 73 (2014) 325e339 339

[19] Vressner A. HCCI engine using in-cylinder pressure, ion current and optical [22] Li Y, Jia M, Chang Y, Liu Y, Xie M, Wang T, et al. Parametric study and opti-
diagnostics. Sweden: Lund University; 2007. mization of a RCCI (reactivity controlled compression ignition) engine fueled
[20] Chen G-B, Li Y-H, Cheng T-S, Chao Y-C. Chemical effect of hydrogen peroxide with methanol and diesel. Energy 2014;65:319e32.
addition on characteristics of methane-air combustion. Energy 2013;55: [23] Curran HJ, Gaffuri P, Pitz WJ, Westbrook CK. A comprehensive modeling study
564e70. of n-heptane oxidation. Combust Flame 1998;114:149e77.
[21] Lu X, Qian Y, Yang Z, Han D, Ji J, Zhou X, et al. Experimental study on com- [24] Deshmukh SR, Vlachos D. A reduced mechanism for methane and one-step
pound HCCI (homogenous charge compression ignition) combustion fueled rate expressions for fuel-lean catalytic combustion of small alkanes on no-
with gasoline and diesel blends. Energy 2014;64:707e18. ble metals. Combust Flame 2007;149:366e83.