Applied Mathematical Modelling 37 (2013) 2961–2982

Contents lists available at SciVerse ScienceDirect

Applied Mathematical Modelling

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

Numerical simulation of two-stage combustion in SI engine with

prechamber
Arkadiusz Jamrozik a, Wojciech Tutak a, Arkadiusz Kociszewski a, Marcin Sosnowski b
a
Czestochowa University of Technology, Al. Armii Krajowej 21, Czestochowa, Poland
b
Jan Dlugosz University in Czestochowa, Al. Armii Krajowej 13/15, Czestochowa, Poland

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

Article history: Burning lean mixture in spark ignition (SI) engine leads to decrease in temperature of com-
Received 14 April 2011 bustion process and is one of the methods of limiting nitric oxide emission and increasing
Received in revised form 5 July 2012 the engine efficiency. The two-stage combustion system of stratified mixture (engine with
Accepted 20 July 2012
prechamber) can be an effective method of lean mixture combustion. The paper presents
Available online 16 August 2012
the results of three-dimensional modeling of fuel mixture preparation and combustion
in SI engine with sectional combustion chamber powered by liquefied fuel. Three dimen-
Keywords:
sional modeling was performed in KIVA-3V code. The modeling results were compared
Spark ignition (SI) engine
Two-stage combustion system
with results obtained from the analysis of experimental measurements of two-stage com-
Prechamber bustion test engine operating at the Institute of Internal Combustion Engines and Control
Excess air factor Engineering (Czestochowa University of Technology). The performed simulations of the
Turbulent kinetic energy combustion process provided data concerning the spatial and temporal distributions of
Heat release rate turbulent kinetic energy, pressure, temperature and nitric oxides concentration in the com-
Nitric oxide bustion chambers of the engine. The engine model with two-stage combustion system
properly represents the real processes which occur in the combustion chambers of the test
engine. Pressure and temperature courses in function of CA obtained from the experiment
and modeling were in good qualitative and quantitative consistence. Comparison of mod-
eled and measured nitric oxide emissions revealed relatively significant discrepancies. In
case of k = 1.4, the measured values of NOx concentration were 1.75 times higher than
the modeled values. In case of k = 2.0, the modeled and measured values were close to each
other and were within the range of measurement error.
Ó 2012 Elsevier Inc. All rights reserved.

1. Introduction

The problem of atmospheric air pollution caused by the exhausts of piston engines, particularly in highly motorized coun-
tries, is presently one of the most important aspects of struggle for the protection of the environment. The design and devel-
opment of the combustion engines results from the necessity of limiting the toxic components concentration in exhaust gas
and reducing fuel consumption. Limiting emission of toxic components in exhaust gas of piston engines can be achieved by
proper organization of the combustion process, fuel additives and the neutralization of exhausts and afterburning in systems
outside the engine [1,2].
The emission of nitric oxides (NOx) is the most difficult to be limited among numerous harmful exhaust gas components.
In the exhaust system and then in the atmosphere, the NO further transforms into nitric dioxide (NO2). The sum of these

E-mail addresses: jamrozik@imtits.pcz.czest.pl (A. Jamrozik), tutak@imc.pcz.czest.pl (W. Tutak), kocisz@imtits.pcz.czest.pl (A. Kociszewski),
m.sosnowski@ajd.czest.pl (M. Sosnowski)

0307-904X/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.apm.2012.07.040
2962 A. Jamrozik et al. / Applied Mathematical Modelling 37 (2013) 2961–2982

Nomenclature

Symbols
cp specific heat at constant pressure, J/kg K
CA crank angle, deg
CO carbon monoxide
D molecular diffusion coefficient, m2/s
~
F rate of momentum gain per volume unit due to spray, kg/m2 s2
fcoll droplet collision source term
fbu droplet breakup source term
G thermal conductivity, J/m s K
~
g specific body force, m/s2
h specific enthalpy, J/kg
HC hydrocarbon
I internal energy, J/kg
~
J heat flux vector, J/m2 s
k turbulent kinetic energy, J/m3
M molecular weight, kg
NOx nitric oxides
p pressure, Pa
Pr turbulent Prandtl number
Q heat source term, J/m3
Qcyl heat supplied to the engine cylinder, J
Qp heat supplied to the prechamber, J
Qtot total heat supplied to the engine, J
R universal gas constant, J/kmol K
r droplet radius, m
Sc Schmidt number
SI spark ignition
t time, s
T temperature, K
TDC top dead centre
~
u velocity vector, m/s
ux, uy, uz velocity fluctuations in the direction of x, y, z, m/s
V volume, m3

Greek symbols
d Dirac delta function
e dissipation rate of turbulence, m/s3
j isentrope ratio
k excess air factor averaged for the main chamber and the prechamber
kp excess air factor in prechamber
~
h droplet acceleration, m/s2
l dynamic viscosity, kg/m s
m turbulent viscosity coefficient, kg/m s
n kinetic reaction rate coefficient, m3/mol s
q mass density, kg/m3
r viscous stress tensor, Pa
u droplet sphericity, m
x_ kinetic reaction rate, 1/s

Superscripts
c chemical reaction
e equilibrium reaction
s spray
Subscripts
b backward
f forward
i species i
r kinetic reaction
 A. Jamrozik et al. / Applied Mathematical Modelling 37 (2013) 2961–2982 2963

components is commonly called NOx. Nitric oxides creation rate increases as the temperature in the combustion chamber
rises, especially above 1600 K.
Lean mixture burning leads to decrease in temperature of combustion process and is one of the methods of limiting nitric
oxides emission. Increasing the excess air factor excessively results in the decrease of engine performance [3]. Conventional
spark ignition engines operate properly only in a narrow range of excess air factor. Exceeding this range towards the richer
mixtures results in the phenomenon of knock and increase in NOx emission. On the other hand, exceeding the range towards
the leaner mixtures results in increase of the non-repeatability of engine cycles and misfire as well as increase in HC and CO
emission [4,5].
Therefore the effective method of lean mixture burning can be the two-stage system of stratified mixture combustion in
the engine with prechamber. In such system the combustion chamber consists of two parts: the main chamber in cylinder of
the engine and the prechamber in the engine head connected with the main chamber by connecting duct(s). Very lean mix-
ture prepared in the engine inlet system (k = 1.5  3.0) is aspirated to the cylinder. However rich mixture of k between 0.5
and 0.9 is delivered to the prechamber. Due to pressure increase in prechamber resulting from combustion, the burning con-
tent of the prechamber is forced by the connecting duct to the main combustion chamber in the cylinder where many mov-
ing ignition kernels develop. In consequence, the lean mixture in the main chamber (which could not be ignited by single
spark discharge) ignites in many regions. The combustion is fast enough to provide high engine cycle efficiency and avoid
the disadvantages connected with combustion during expansion stroke. Slight amounts of NOx are produced and particles
of CO and HC are burnt at this stage of combustion [6].
One of the first attempts to study and analyze the effectiveness of lean mixtures ignition and combustion in the engine
with prechamber is so called the pilot flame torch ignition system. This system was developed and patented in 1963 and
1966 by Gussak [7,8].
Two-stage combustion system in the engine with a small ignition chamber was the subject of studies conducted since
1978 at University of California, Berkeley by a team headed by prof. Oppenheim [9,10]. The research result was the devel-
opment of controlled burning system, where lean mixture ignition was generated by the pilot flame generator.
Studies of stratified mixture burning using the sectional combustion chamber were conducted in 1990 in the UK [11]. The
research consisted of numerical modeling of the mixture preparation process using PHOENICS code and experimental stud-
ies. The measurements of toxic components concentration in exhaust gases revealed that nitric oxides emission was reduced
to very low level in comparison with a conventional engine.
The APIR (Auto-inflammation Pilotée par Injection de Radiucaux) lean mixture combustion system was developed at the
French University of Orleans in 1999. A spark plug placed in a small prechamber was used in this system instead of the tra-
ditional spark plug. The prechamber volume was 1% of the total volume of the combustion chamber of the engine. The pre-
chamber was supplied with rich mixture and the prechamber was connected with the main chamber by four ducts of
diameter equal one millimeter. The research results showed positive effect of the two-stage combustion on the stability
of the engine operation, repeatability of the successive cycles, resistance to knock and toxic components emission in com-
parison with the conventional spark ignition engine [12].
Research concerning lean gaseous mixture combustion carried out in 2001 in Japan showed that the improvement of en-
gine thermal efficiency by changing the compression ratio can be obtained by stratified mixtures combustion in two-stage
system with prechamber [13]. The results showed that the increase in compression ratio and engine efficiency can be ob-
tained by generating turbulence due to the geometry of the chamber.
The studies on improving the combustion process by moving the ignition point from the main combustion chamber of
conventional engine to the small prechamber were carried out in the Swiss Institute of Technology in Lausanne [14,15].
The experimental researches were the continuation of numerical studies performed in KIVA-3V code. The aim of the research
was to determine the possibility of limiting the exhaust gas toxic components emission to the level of Swiss regulations by
applying the two-stage system of lean mixture (k = 1.6  1.7) combustion. The mostly investigated issues were the dimen-
sions and geometry of the prechamber as well as the number, location and diameter of the ducts connecting the main cham-
ber and the prechamber. The study showed that the most favorable conditions for ignition and propagation of the flame front
can be achieved by using prechamber with a capacity equal 3% of the total volume above the piston at the BDC. The number
of ducts should be 4 and a small total area of the ducts (9.37 mm2).
Almost all of the major global automotive companies conducted or maintained the work on a two-stage system of strat-
ified mixture combustion. In some cases, the research resulted in the implementation of the new engine design into mass
production.
The most known system of this type is the CVCC (Compound Vortex Controlled Combustion) [3] developed in the seven-
ties of the XX century by Honda. Companies such as Ford, General Motors, Volkswagen, Walker, Eaton, Heintz, Nilov, Porsche,
Toyota, Mitsubishi [16,17] designed their own systems.
The preparation and combustion of stratified mixtures in engines equipped with prechamber (Honda CVCC) was aban-
doned and the fuel direct injection to the single combustion chamber (GDI) was introduced [18].
At present times the two stage system of stratified mixtures combustion (with prechamber) is applied mostly in station-
ary supercharged gaseous engines of medium and high power operating at constant rotational speed.
The two-stage lean mixture combustion system with prechamber was applied in the SI stationary gaseous engines of high
power and cylinder diameter of above 200 mm by manufactures such as the Austrian company Jenbacher AG [19], Danish
2964 A. Jamrozik et al. / Applied Mathematical Modelling 37 (2013) 2961–2982

German MAN B&W Holeby [20], Finnish Wärtsilä NSD Corporation [21] and by the US companies Waukesha Engine Dresser
[22] and Caterpillar Inc. [23].
During recent years, the research concerning numerical modeling using more and more advanced mathematical models
has been intensively developing. The development of numerical modeling is boosted by increasing computational power that
allows modeling not only flow processes but also combustion in 3D. One of the more advanced numerical models used for
combustion process in piston engines modeling is KIVA-3V developed in Los Alamos National Laboratory in the United States
of America [24,25] and [26].

2. The computational code

The KIVA-3V [24,25] and [26] code allows calculation of three-dimensional flow and combustion in the engine chambers
of any geometry including the effects of turbulence and heat exchange with walls. KIVA-3V is an example of a three-dimen-
sional field model. Combustion process is the result of solving the basic conservation equations and other equations that
determine the flow field in the region of the flame front, the course of chemical reactions of combustion and instantaneous
thermodynamic state of the medium. The model is based on solving the equations of conservation of mass, momentum, en-
ergy and quantities of ingredients, which describe the unsteady, three-dimensional flow field of chemical reaction (combus-
tion). These equations are the three-dimensional Navier–Stokes equations [24] for the compressible fluid mixture.
Continuity equation for total fluid density:
@q
uÞ ¼ q_ s ;
þ r  ðq~ ð1Þ
@t
u is the fluid velocity vector, q_ s is source term due to spray.
where q is the total mass density, ~
Three-dimensional fluid velocity vector:

u ¼ uðx; y; z; tÞ~i þ v ðx; y; z; tÞ~j þ wðx; y; z; tÞ~

~ k:
Vector differential operator:
@ ~@ ~ @
r ¼ ~i þj þk :
@x @y @z
Momentum equation for the fluid mixture [24]:
 
@ðq~uÞ 1 2
þ r  ðq~
u~ g þ~
uÞ ¼  2 rp  A0 r qk þ r  r þ q~ Fs; ð2Þ
@t a 3
where a is the dimensionless quantity characterized the type of flow, p is the fluid pressure, A0 is the value connected with
the model of turbulence (A0 = 0 for laminar flows, A0 = 1 for turbulent flows), k is the turbulent kinetic energy, r is the viscous
stress tensor (Newtonian fluid), ~g is the specific body force (assumed constant), ~ F s is the rate of momentum gain per volume
unit due to spray.
Turbulent kinetic energy:
1  2 
k¼ q ux þ u2y þ u2z ; ð3Þ
2
where: q is the density and ux, uy, uz are the velocity fluctuations in the direction of x, y, z.
Viscous stress tensor in Newtonian form [24]:

r ¼ l½r~ uÞT  þ mr  ~
u þ ðr~ uH; ð4Þ
where l is the gas dynamic viscosity, m is the viscosity coefficients of turbulent field, superscript T is transpose, H is the unit
dyadic.
Internal energy equation [24]:
@ðqIÞ
þ r  ðq~
uIÞ ¼ pr  ~ J þ A0 qe þ Q_ c þ Q_ s ;
u  r ~
u þ ð1  A0 Þr : r~ ð5Þ
@t
where I is the specific internal energy exclusive of chemical energy, ~ J is the heat flux vector and, e is the dissipation rate of
turbulence, Q_ c is the source term due to chemical heat release, Q_ s is the source term due to spray interactions.
Heat flux vector
X  
~ q
J ¼ GrT  qD hi r i ; ð6Þ
i1
q

lc p
G¼ ; ð7Þ
Pr
l
D¼ ; ð8Þ
qSc
 A. Jamrozik et al. / Applied Mathematical Modelling 37 (2013) 2961–2982 2965

where G is the thermal conductivity, l is the viscosity coefficient, cp is the specific heat at constant pressure, Pr is the tur-
bulent Prandtl number, T is the fluid temperature, D is the molecular diffusion coefficient, Sc is the Schmidt number, hi is the
specific enthalpy of species i, qi is the density of species i.
The process of turbulence can be modeled using one of the three submodels: SGS (Sub-Grid Scale), k–e or RNG k–e (ReNor-
malisation Group) [24]. The k–e model was used in case of turbulence modeling in which the kinematic viscosity lt is depen-
dent on the turbulent kinetic energy k and the rate of turbulence kinetic energy dissipation e.
Viscous stress tensor for the k–e model is a function of turbulent viscosity:
2
k
r!l¼ : ð9Þ
qe
Two additional transport equations: turbulence kinetic energy k and dissipation rate of turbulence kinetic energy e [24]:
  
@ qk 2 l
þ r  ðq~ u þ r : r~
ukÞ ¼  qkr  ~ uþr rk  q2 e; ð10Þ
@t 3 Prk
  
@ qe 2 l e
ueÞ ¼  qer  ~
þ r  ðq~ uþr u  q2 eÞ;
re þ ðr : r~ ð11Þ
@t 3 Pre k
where Pre = 1, 3 is the turbulence Prandtl number for e, Prk = 1, 0 is the turbulence Prandtl number for k.
Continuity equation for species i:
  
@ qi q
uÞ ¼ r  qDr i
þ r  ðqi~ þ q_ ci þ q_ s di ; ð12Þ
@t q
where q_ ci is the source term due to chemistry, di is Dirac delta function.
These conservation equations are supplemented with the following additional equations for the model of ideal gas [4]:
X q 
i
p ¼ RT ; ð13Þ
i
Mi

Xq  
i
IðTÞ ¼ Ii ðTÞ ; ð14Þ
i
q

Xq  
i
cp ðTÞ ¼ cpi ðTÞ ; ð15Þ
i
q

RT
hi ðTÞ ¼ Ii ðTÞ þ ; ð16Þ
Mi
where R is universal gas constant, Mi is the molecular weight of species i, cpi(T) and hi(T) are the specific heat and specific
enthalpy of species i taken from JANAF tables.
General equation of the chemical reactions [4]:
X X
ai X i () bi X i ; ð17Þ
i¼1 i¼1

where Xi represents one mole of species i and, ai and bi are integer stoichiometric coefficients for (kinetic/equilibrium) reaction r.
Four kinetic reactions and six equilibrium reactions are included in KIVA-3V in the submodel of chemical reactions of the
combustion process of liquid fuel. One global reaction (C8H17) with extended Zeldovich mechanisms [27] for NO formation:
4C8 H17 þ 49O2 ) 32CO2 þ 34H2 O; ð18Þ
8
< O þ N2 () N þ NO
>
O2 þ N () O þ NO : ð19Þ
>
:
N þ OH () H þ NO
The equilibrium reactions are:
8
> H2 () 2H
>
>
>
> O2 () 2O
>
>
>
< N () 2N
2
: ð20Þ
>
> O 2 þ H2 () 2OH
>
>
>
> O2 þ 2H2 O () 4OH
>
>
:
O2 þ 2CO () 2CO2
2966 A. Jamrozik et al. / Applied Mathematical Modelling 37 (2013) 2961–2982

The chemical heat release term is given by

X
Q_ c ¼ Q rx
_ r; ð21Þ
r

where Qr is the negative of the heat of reaction at absolute zero, x

_ r is the kinetic reaction rate.
X c
Qr ¼ ðai  bi ÞðDh Þi ; ð22Þ
i

where (Dhc)i is the heat of formation of species i at absolute zero.

The chemical source term in Eq. (12) is given by:
X
q_ ci ¼ Mi ðbi  ai Þx
_ r: ð23Þ
r

The kinetic reaction r proceeds at a rate of x

_ r given by

Y q ai Y q bi
i i
x_ r ¼ nrf  nrb ; ð24Þ
i
Mi i
Mi

where nrf, nrb are the forward and backward rate coefficients for reaction r and are assumed to be of a generalized Arrhenius
form (kinetic reactions) [24]:
nrf ¼ Af exp ðEf =TÞ; ð25Þ

nrb ¼ Ab exp ðEb =TÞ; ð26Þ

where Af, Ab are values constant, Ef, Eb are activation temperatures (Kelvin).
The rate of equilibrium reaction is implicitly determined by the condition:
Y q ðbi ai Þ
i
¼ K e ðTÞ; ð27Þ
i
Mi

where Ke(T) depends only on temperature and is the concentration equilibrium constant for equilibrium reaction r and is
assumed to be of the form [24]:
 
K e ðTÞ ¼ exp A ln T n þ B=T n þ C þ DT n þ ET 2n ; ð28Þ

T
Tn ¼ : ð29Þ
1000 K
The temperature is determined by successive iterations:
P
xr
rQ r
T n ¼ T n1 þ ; ð30Þ
qcp
xr
x_ r ¼ : ð31Þ
Dt
The ignition in KIVA-3V software is realized by delivering specific amount of energy to a cell or block of cells. The energy
flux delivered to the mixture during ignition can be modified by changing the value of xignit parameter but it is not possible
to modify the energy flux during individual phases of spark discharge. The delivering of energy to the earlier specified region
is terminated if the temperature of the region exceeds 1600 K before the end of ignition window. The front of the flame is
assumed to have enough energy to sustain its development and supply of external energy is no longer needed after exceed-
ing the temperature of 1600 K.
The fuel injection submodel is described by the equation containing ten independent variables and time. The independent
variables are: three droplet position components ~ x, three velocity components ~
u, equilibrium radius r, temperature Td, dis-
tortion from sphericity u and the time rate of change of u (u _ ¼ ddtu).

f ð~
x; ~ udrdT d dudu
_ tÞd~
u; r; T d ; y; y; _: ð32Þ
The time evolution of f is obtained by solving a form of the spray equation [24]:
@f @ @ @ @
uÞ þ r~u  ðf ~
þ r~x  ðf ~ hÞ þ ðf r_ Þ þ ðf T_ d Þ þ ðf u
_Þþ ðf u
€ Þ ¼ fcoll þ fbu ; ð33Þ
@t @r @T d @u @u_
where ~h, r_ , T_ d , and u € are the time rate of change, following an individual drop, of its velocity, radius, temperature, and oscil-
lation velocity u _ . These functions determine the trajectories of individual droplets. The terms fcoll and fbu are sources due to
droplet collisions and breakups. The submodel takes into account the phenomenon of droplet collisions and oscillations,
 A. Jamrozik et al. / Applied Mathematical Modelling 37 (2013) 2961–2982 2967

breakups, coalescences and effects of evaporation. The heat exchange with the wall was modeled using the heat transfer sub-
model based on the so-called turbulent wall law (law-of-the-wall).
Solving the differential equations in KIVA-3V is done using a numerical method based on finite difference method. Im-
plicit Finite Difference Scheme is used for continuity equation and momentum equation. Explicit Finite Difference Scheme
is used for internal energy equation. The computational algorithm based on the Arbitrary Lagrangian–Eulerian Method
(ALE) in which the grid is made up of arbitrary cuboids. The location of grid nodes may be arbitrary and depends on time.
The grid of any shape moves with the fluid (Lagrangian scheme) and is fixed in space (Eulerian scheme). In this method it is
possible to conduct calculations in two phases. In the first phase, grid nodes move with the instantaneous local fluid velocity
without analyzing the phenomenon of convection through the walls of cells.
In the second phase, the flow field with the convection through the walls of cells is calculated for the frozen flow field and
new location of the grid. The movements of the grid can simulate the actual change of piston position in the cylinder,
increasing or decreasing its volume in accordance with changes in geometry of the workspace.
The KIVA-3V code has a three-module structure and consists of: k3prep – preprocessor (geometric mesh generation); ki-
va3v – main program (solver); k3post – postprocessor (visualization of modeling results). The preprocessor input data (geo-
metric mesh) is prepared in file called iprep. The preprocessor generates otape11 and otape17 files, which are input data to
the main program after renaming respectively to itape11 and itape17 along with itape5, itape8 and itape7 files. The main pro-
gram generates plotgmv files with data for visualization in postprocessor (native or third-party) and data concerning in-cyl-
inder conditions.
Within the confines of the research the three-dimensional model of air–fuel mixture creation and combustion process in
SI engine with the sectional combustion chamber powered by liquid fuel was designed and analyzed in KIVA-3V code. The
results of numerical modeling were compared with measurements of two-stage combustion test engine built and developed
in the Institute of Internal Combustion Engines and Control Engineering (Czestochowa University of Technology).

3. The test engine

The test engine has been constructed on the basis of a four-stroke compression-ignition engine manufactured by ‘‘AND-
ORIA’’ Diesel Engine Manufacturers of Andrychow, which (after some constructional changes) is designed for the combustion
of gaseous fuel as a spark-ignition engine owing to a new fuel supply system and an ignition installation. The engine is a
stationary, two-valve unit of a horizontal cylinder configuration. The engine block is made of cast iron and is integrated with
the crankcase [28].
The engine cooling system is based on the evaporation of the water jacket. Within the confines of the research changes in
the base engine design were done in order to enable the implementation of two-stage combustion system with the sectional
combustion chamber [28].
The main engine element that underwent modernization was the head (Fig. 1). The changes in its design allowed an addi-
tional combustion chamber (prechamber) to be installed in the previously existing head of the S320 ER engine by setting the
compression ratio at 8.6.
The additional combustion chamber (prechamber) was made of an alloy of enhanced strength at high temperatures called
Nimonic 90. In the prechamber (its volume makes up 4.5% of the total combustion volume) an M141.25 spark plug and an
M70.75 piezo-quartz sensor were installed. The prechamber can be connected with the cylinder by a duct of 3, 4, 6 or 9 mm
diameter.
The fuel mixture in the prechamber is enriched at the end of the compression stroke with gaseous fuel supplied through
the additional fuel supply system. The main elements of the additional fuel supply system are a set of non-return valves and
a solenoid inlet valve. The water-cooled valves are placed in a special water jacket.
The main parameters of the test engine are shown in the Table 1.

4. The influence of prechamber geometry on cycle parameters of SI engine with two-stage combustion system

The first part of the numerical analysis was to investigate the influence of the connecting duct diameter on chosen engine
cycle parameters. Within the confines of calculations performed in KIVA-3V code, the modeling of processes occurring in
combustion chambers of the engine with prechamber powered by gasoline for four diameters of the duct connecting the pre-
chamber with cylinder (3, 4, 6 and 9 mm) were conducted. The courses of kinetic energy of turbulence, pressure and tem-
perature variations in engine combustion chambers in function of crank angle were estimated. Moreover the distribution of
above mentioned quantities in the combustion chambers was calculated for chosen crank angles. Characteristics of heat re-
lease rate and the total heat released during combustion in function of CA were presented.

4.1. The model of the test engine with two-stage combustion system

The workspace of the combustion chambers of analyzed model was built in accordance with the test engine geometry and
is depicted in Fig. 2.
2968 A. Jamrozik et al. / Applied Mathematical Modelling 37 (2013) 2961–2982

Fig. 1. Test engine head with prechamber. 1 – valve rocker, 2 – inlet valve, 3 – injector, 4 – prechamber head, 5 – retaining cover, 6 – spark plug, 7 – sealing
ring, 8 – prechamber body, 9 – piston, 10 – pressure sensor [28].

The total volume of those chambers is 237 cm3. The prechamber volume is approximately 4.5% of total volume above the
piston at TDC and it is located asymmetrically regarding the cylinder axis. It is connected with the main chamber by cylin-
drical duct of diameter equal 3, 4, 6 or 9 mm (Fig. 3).
The computational domain discretization was performed using the KIVA-3V preprocessor (k3prep), which generates fully
structured mesh. Several meshes of different resolution were generated in order to assure the mesh-independent solution.
Finally the mesh built of 19 blocks and 24,500 hexahedral cells was chosen. The time step couldn’t be precisely defined as the
solver calculates it automatically. The subsequent validation revealed the satisfactory consistence of the numerical and

Table 1
Main engine parameters.

Displacement volume 1810 cm3

Number of cylinders 1
Cylinder configuration Horizontal
Cylinder bore 120 mm
Connecting-rod length 275 mm
Piston stroke 160 mm
Compression ratio 8.6
Engine speed 1000 rpm

Fig. 2. The computational domain.

 A. Jamrozik et al. / Applied Mathematical Modelling 37 (2013) 2961–2982 2969

experimental results. Table 2 presents the chosen input parameters of modeled process with their values corrected according
to the literature [4,29] and experimental research at the test stand [28].
The kinetic reaction rate equations and rate coefficients for kinetic reactions were adopted on the basis of the literature
[25,30], and [31].
4C8 H17 þ 49O2 ) 32CO2 þ 34H2 O;
  
15106 qC8 H17 0:25
nrf ¼ 4:6  1011 exp  nrb ¼ 0 x_ r ¼ nrf ½O2 1:5 ;
T MC8 H17

O þ N2 () N þ NO;
 
38000
nrf ¼ 7:6  1013 exp  nrb ¼ 1:6  1013 x_ r ¼ nrf ½N2 ½O  nrb ½N½NO;
T

O2 þ N () O þ NO;
   
3150 19500
nrf ¼ 6:4  109 T exp  nrb ¼ 1:5  109 T exp  ;
T T

x_ r ¼ nrf ½O2 ½N  nrb ½O½NO;

N þ OH () H þ NO;
 
23650
nrf ¼ 4:1  1013 nrb ¼ 2:0  1014 exp  x_ r ¼ nrf ½N½OH  nrb ½H½NO:
T
The equilibrium constants were adopted on the basis of [24,25].
H2 () 2H;

½H2 ð01Þ ½Hð20Þ ¼ K e1 ;

K e1 ¼ exp ð0:990207 ln T  51:7916=T þ 0:993074  0:343428T þ 0:0111668T 2 Þ;

O2 () 2O;

½O2 ð01Þ ½Oð20Þ ¼ K e2 ;

K e2 ¼ exp ð0:43131 ln T  59:6554=T þ 3:50335  0:340016T þ 0:0158715T 2 Þ;

N2 () 2N;

½N2 ð01Þ ½Nð20Þ ¼ K e3 ;

K e3 ¼ exp ð0:794709 ln T  113:208=T þ 3:16837  0:443814T þ 0:0269699T 2 Þ;

O2 þ H2 () 2OH;

Fig. 3. The geometric mesh with connecting duct of diameter equal: 3, 4, 6, 9 mm.
2970 A. Jamrozik et al. / Applied Mathematical Modelling 37 (2013) 2961–2982

Table 2
The chosen input parameters.

Quantity KIVA name Dimension Value

Rotational speed rpm rpm 1000
Cylinder bore bore cm 12.0
Piston stroke stroke cm 16.0
Squish squish cm 1.38
Fuel gasoline – C8H17
Initial pressure (180° before TDC) presi MPa 0.1
Initial temperature (180° before TDC) tempi K 365.0
Kinetic energy of turbulence (180° before TDC) tkei J/m3 0.17
Turbulence scale (180° before TDC) scli cm 2.2
Ignition advance angle (before TDC) calign deg 12
Ignition energy factor xignit – 1.0e+4

½O2 ð01Þ ½H2 ð01Þ ½OHð20Þ ¼ K e4 ;

K e4 ¼ exp ð0:652939 ln T  9:8232=T þ 3:93033 þ 0:16349T  0:0142865T 2 Þ;

O2 þ 2H2 O () 4OH;

½O2 ð01Þ ½H2 Oð02Þ ½OHð40Þ ¼ K e5 ;

K e5 ¼ exp ð1:15882 ln T  76:8472=T þ 8:532155  0:86832T þ 0:0463471T 2 Þ;

O2 þ 2CO () 2CO2 ;

½O2 ð01Þ ½COð02Þ ½CO2 ð20Þ ¼ K e6 ;

K e6 ¼ exp ð0:980875 ln T þ 68:4453=T  10:5938 þ 0:57426T  0:041457T 2 Þ;

4.2. The course of calculation

The calculations started at BDC at the beginning of compression stroke and lasted for 360° CA until the end of engine
power stroke. Lean and homogeneous mixture prepared in the inlet system was located in engine combustion chambers
at the time of the beginning of compression stroke. Forty-five degree before TDC the mixture in prechamber was enriched
to stoichiometric composition by injecting additional dose of fuel, which was approximately 2.5% of energetic share of all the
fuel delivered to the engine.
The ignition by spark discharge 12 deg before TDC occurred in the prechamber. As a result of combustion and therefore
pressure increase in prechamber, the burning mixture was forced by the connecting duct to the main chamber in the cylinder
and ignited the lean mixture. The research was conducted for four values of excess air factor equal 1.4, 1.6, 1.8 and 2.0 aver-
aged for the main chamber and the prechamber.
The configurations with different diameters of the connecting duct (3, 4, 6 and 9 mm) were analyzed during the
simulations.
The injection of enriching charge into the prechamber caused that the mixture composition in the prechamber was in the
range of 0.85 for the average excess air factor k = 1.4 and 1.1 for the average excess air factor k = 2.0 (Table 3).
Fig. 4 shows the distribution of fuel mixture in the combustion chambers of the engine model at the time of ignition –
12 deg before TDC at excess air factor of 2.0 (averaged for the entire combustion volume) an excess air factor in prechamber
equal 1.1.
In the prechamber, the mixture prepared during the compression stroke is not homogeneous and the value of excess air
factor of the mixture varies from 0.715 to 2.03. Fig. 5 shows the density distribution of the concentration of the mixture in
the prechamber depending on the initial composition of the mixture at the average excess air factor in the prechamber equal
kp = 1.1.
The largest part of the prechamber of around 32% is occupied by the mixture of excess air factor equal 0.715–0.85 and the
smallest part of the prechamber (around 3%) is occupied by the mixture of excess air factor equal 1.75–1.9.
 A. Jamrozik et al. / Applied Mathematical Modelling 37 (2013) 2961–2982 2971

Table 3
The values of excess air factor in prechamber of engine model
at the time of ignition.

k kp
1.4 0.85
1.6 0.95
1.8 1.00
2.0 1.10

Fig. 4. Air–fuel mixture distribution in combustion chambers of engine model at the time of ignition for average k = 2.0 and kp = 1.1.

4.3. Modeling results

The primary result of mathematical modeling and experimental research was to determine the courses of pressure and
temperature in the cylinder of the SI engine with two-stage combustion system. These courses can be used to verify the re-
sults of three-dimensional modeling using KIVA-3V. Comparison of the results obtained from KIVA-3V with the results of the
analysis of the test engine cycle showed some differences between them. The results were used to validate the model. The
comparison was performed for the case in which the energetic share of the enriching charge was approximately 2.5% of ener-
getic share of all the fuel delivered to the engine, the ignition advance angle 12 deg before TDC and for duct of diameter equal
6 mm (Figs. 6 and 7).
The values of maximum pressure in the cylinder of the test engine designated in the indication process, depending on the
value of k, were in the range of 3.2 MPa for k = 2.0 to 4.5 MPa for k = 1.4. These values differed from the average pressure in
the combustion chambers of engine model. In case of the engine model at k = 2.0 the maximum pressure was at the level of
3.1 MPa and in case of k = 1.4  4.7 MPa.
The temperature in the test engine was calculated on the basis of pressure. The maximum temperature calculated in
KIVA-3V code (averaged for the cylinder and prechamber) was equal 2140 K for k = 1.4 and 1550 K for k = 2.0. In case of
the test engine, the temperature was in the range of 1520 K for k = 2.0 to 1990 K for k = 1.4.

Erro de 7,5 % para lambda igual a 1,4

p e erro de 1,9 % para lambda igual a 2,0
100
∑ Vi / V p , %

80

60

40
Vi / Vp ,%

20

0
0.70 0.85 1.00 1.15 1.30 1.45 1.60 1.75 1.90 2.05
λp

Fig. 5. Density distributions of the concentration of air–fuel mixture in the prechamber at the time of ignition for the average excess air factor in the
prechamber equal kp = 1.1.
2972 A. Jamrozik et al. / Applied Mathematical Modelling 37 (2013) 2961–2982

4 λ = 1.4
λ = 1.6

pressure, MPa
3 λ = 1.8
λ = 2.0
2

0
240 280 320 360 400 440 480
crank angle, deg

2500

2000 λ = 1.4
temperature, K

λ = 1.6
1500
λ = 1.8
1000 λ = 2.0

500

0
240 280 320 360 400 440 480

crank angle, deg

Fig. 6. Courses of pressure and temperature in KIVA-3V model.

4 λ = 1.4
pressure, MPa

λ = 1.6
3 λ = 1.8
λ = 2.0
2

0
240 280 320 360 400 440 480
crank angle, deg

2500

2000 λ = 1.4
temperature, K

λ = 1.6
1500 λ = 1.8
λ = 2.0
1000

500

0
240 280 320 360 400 440 480

crank angle, deg

Fig. 7. Courses of pressure and temperature in the test engine.

One of the most important issues in combustion engine is fast air-fuel mixture combustion [5]. The combustion process in
piston engine is elongated in case of lean mixtures combustion due to heat transferred to the walls. The combustion speed is
dependent on the quality of fuel and air mixing before the ignition. The mixing takes place during the compression stroke as
a result of fresh charge swirl caused by the geometry of combustion chambers. The swirl in cylinder is accompanied by the
increase of mixture turbulence. One of the main parameters describing the turbulence in the flow is the turbulent kinetic
energy k [30]. High but not excessive turbulence level before the ignition particularly in the area of spark plug is desired
in case of SI engine with charge prepared outside the working volume.
 A. Jamrozik et al. / Applied Mathematical Modelling 37 (2013) 2961–2982 2973

Fig. 8 depicts the turbulent kinetic energy in function of CA, averaged for both the main chamber and the prechamber
during the compression stroke for four connecting duct diameters equal: 3, 4, 6 and 9 mm. The difference in the average ki-
netic energy for various ducts diameters results from the fact that the average velocity of charge flowing in the duct of smal-
ler diameter is significantly higher in comparison with the cases of greater duct diameters. The higher average velocity leads
to higher turbulent kinetic energy in the combustion chamber.
Fig. 9 depicts the turbulent kinetic energy in function of CA in prechamber only. The highest turbulent kinetic energy va-
lue averaged for the main chamber and prechamber and for the prechamber only was obtained in case of the engine
equipped with the connecting duct of 3 mm diameter. The lowest turbulent kinetic energy value was obtained in case of
the connecting duct diameter equal 9 mm. At the time of spark discharge 12 deg before TDC the k value in the prechamber
was: 1760 J/m3 – 3 mm connecting duct diameter, 855 J/m3 – 4 mm connecting duct diameter, 254 J/m3 – 6 mm connecting
duct diameter and 112 J/m3 – 9 mm connecting duct diameter. The modeling results have been confirmed experimentally in
the case of a duct diameter of 6 mm [32].
Fig. 10 shows the distribution of turbulent kinetic energy in combustion chambers of the engine with the prechamber at
the time of the spark discharge 12 deg before TDC.
Lean mixture combustion in traditional SI engine with single combustion chamber leads to deterioration of main engine
cycle parameters such as pressure and temperature and as a result it causes the deterioration of engine efficiency revealed by
decreased values of maximal torque and maximal indicated pressure. Engines with sectional combustion chamber provide
increased combustion speed and in consequence allow significantly greater mixture impoverishment in comparison with
conventional engines. Figs. 11–14 depict courses of pressure and temperature averaged for the main chamber and precham-
ber of engine model in function of CA at analyzed range of mixture composition k = 1.4  2.0.
The charts show that in case of the engine with two stage combustion system equipped with the connecting duct of 3 mm
diameter the obtained values of pressure and temperature complied with the compression process. The combustion took
place only in the prechamber as the connecting duct of such small diameter did not allow the flame to propagate to the main
chamber and ignite the charge in the cylinder. The engine equipped with connecting duct of diameter equal 4 mm and 6 mm
was characterized by similar values of maximal pressure and temperature while combusting mixture of k = 1.2  1.8. Signif-
icant differences in obtained values of pressure and temperature occurred while combusting the leanest mixture of k = 2.0. In
case of the leanest mixture and the connecting duct of diameter equal 4 mm, the combustion process in cylinder was delayed
until the end of expansion stroke due to choke effect in the connecting duct. Therefore obtaining high values of pressure and
temperature was impossible. The engine equipped with the connecting duct of the diameter equal 9 mm and operating at
k = 1.2  1.8 obtained the lowest values of pressure and temperature. At k = 2.0, the analyzed parameters of engine cycle val-
ues were lower than those obtained in the engine equipped with duct of 6 mm diameter and higher than values obtained in
case of the engine with the connecting duct of 4 mm diameter.
Temperature distribution in the working areas of engine model with prechamber while combusting mixture of k = 2.0 at
selected CA and for four analyzed connecting duct diameters are depicted in Figs. 15–17.
Differences in combustion speed in reference to the diameter of the duct connecting the main chamber and the precham-
ber can be revealed while analyzing the temperature distribution in combustion chambers of the engine model. The most
favorable configuration was the one with connecting duct of diameter equal 6 mm. In this case the combustion process
was the most dynamic one, and the time in which flame propagates into the volume of combustion chamber was the short-
est. In the configuration of engine equipped with 4 mm connecting duct, too small diameter of the duct did not allow the
flame to propagate fast enough from the prechamber to the main chamber in the cylinder and prevented the mixture in
the main chamber to ignite at the desired time. In case of the engine with 9 mm duct, the deterioration of flame propagation
speed and the decrease of maximal values of engine cycle parameters is revealed in comparison to the engine with 6 mm
duct even though the combustion process was proper. The combustion in case of engine with the connecting duct of
3 mm diameter occurred only in the prechamber as the connecting duct of such small diameter did not allow the flame
to propagate into the cylinder. The deterioration of maximal engine work cycle parameters is connected with low velocity

120
3 mm
4 mm 100
6 mm
9 mm 80
3
k, J/m

60

40

20

0
260 270 280 290 300 310 320 330 340 350 360
crank angle, deg

Fig. 8. Course of turbulent kinetic energy averaged in the main chamber and the prechamber of the engine with two-stage combustion system.
2974 A. Jamrozik et al. / Applied Mathematical Modelling 37 (2013) 2961–2982

2100
3 mm
1800
4 mm
6 mm 1500
9 mm

3
1200

k, J/m
900

600

300

0
260 270 280 290 300 310 320 330 340 350 360
crank angle, deg

Fig. 9. Course of turbulent kinetic energy in the prechamber of the engine with two-stage combustion system.

Fig. 10. Distribution of turbulent kinetic energy in combustion chambers of the engine with prechamber at the time of spark discharge 12 deg before TDC.

7 3150
3 mm 3 mm
6 2700
4 mm 4 mm
6 mm
temperature, K
pressure, MPa

5 2250 6 mm
9 mm 9 mm
4 1800

3 1350

2 900

1 450

0 0
240 270 300 330 360 390 420 450 480 240 270 300 330 360 390 420 450 480
crank angle, deg crank angle, deg

Fig. 11. Courses of pressure and temperature at k = 1.4.

of flame propagation and insufficient heat produced during the first phase of lean and homogeneous mixture combustion.
Figs. 18–21 depict heat release rate during combustion and amount of heat released during combustion in function of CA
of engine model equipped with connecting duct of 3, 4, 6 and 9 mm diameter for k = 1.4  2.0. These values were calculated
using the first thermodynamics law for ideal gas on the basis of pressures and temperatures averaged for the whole volume
of combustion obtained during numerical modeling.
 A. Jamrozik et al. / Applied Mathematical Modelling 37 (2013) 2961–2982 2975

6.3 2800
3 mm 2400 3 mm
5.4
4 mm 4 mm

temperature, K
2000 6 mm
pressure, MPa
4.5 6 mm
9 mm 9 mm
3.6 1600

2.7 1200

1.8 800

0.9 400

0 0
240 270 300 330 360 390 420 450 480 240 270 300 330 360 390 420 450 480
crank angle, deg crank angle, deg

Fig. 12. Courses of pressure and temperature at k = 1.6.

4.9 2450
3 mm 3 mm
4.2 2100
4 mm 4 mm
6 mm

temperature, K
pressure, MPa

3.5 1750 6 mm
9 mm 9 mm
2.8 1400

2.1 1050

1.4 700

0.7 350

0 0
240 270 300 330 360 390 420 450 480 240 270 300 330 360 390 420 450 480
crank angle, deg crank angle, deg

Fig. 13. Courses of pressure and temperature at k = 1.8.

4.2 2100
3 mm 1800 3 mm
3.6
4 mm 4 mm
temperature, K

6 mm 1500 6 mm
pressure, MPa

3
9 mm 9 mm
2.4 1200

1.8 900

1.2 600

0.6 300

0 0
240 270 300 330 360 390 420 450 480 240 270 300 330 360 390 420 450 480
crank angle, deg crank angle, deg

Fig. 14. Courses of pressure and temperature at k = 2.0.

Heat release rate:

dQ dI dV
¼ þp ; ð34Þ
da da da
 
dQ 1 dV dp
¼ jp þV ; ð35Þ
da j  1 da da

j ¼ 1:392  8:13105 T ½33; ð36Þ

where j is the isentrope ratio.
The analysis of depicted graphs confirms the fact that the most favorable case was the one of engine equipped with the
connecting duct of 6 mm diameter. Maximal values of pressure and temperature were obtained for the mixture of
k = 1.4  2.0 in engine of 6 mm connecting duct. In this configuration heat release rate and the heat released during combus-
tion for the whole analyzed mixture composition range are the most favorable and prove the proper curse of combustion.
2976 A. Jamrozik et al. / Applied Mathematical Modelling 37 (2013) 2961–2982

Fig. 15. Temperature distribution in combustion chambers of the engine model for k = 2.0 at 5 deg before TDC.

Fig. 16. Temperature distribution in combustion chambers of the engine model for k = 2.0 at 5 deg after TDC.

Even though the mixture was leant to the level of k = 2.0, the combustion process in this engine was fast enough to pro-
vide high values of indicated pressure and in consequence proper engine efficiency.
Fig. 22 shows the courses of pressure and temperature in the prechamber and the main chamber of engine model
equipped with the connecting duct of 6 mm diameter at k = 2.0.
The pressure courses obtained from the modeling (Fig. 22) were experimentally confirmed by Jamrozik, Tutak and Kow-
alewicz [3,34]. The engine was equipped with two pressure sensors, one in the prechamber and the other in the main
chamber.

4.4. Modeling of nitric oxide creation process in SI engine with sectional combustion chamber

KIVA-3V code allows specifying the level of toxic components emission in exhaust gases. Among many harmful exhaust
components, NOx emission is the most difficult to be limited in the spark-ignition engine.
 A. Jamrozik et al. / Applied Mathematical Modelling 37 (2013) 2961–2982 2977

Fig. 17. Temperature distribution in combustion chambers of the engine model for k = 2.0 at 30 deg after TDC.

0.78 4.5

0.65 3 mm 3.6
4 mm
0.52 6 mm
dQ/dα, kJ/deg

2.7
9 mm
0.39
Q, kJ

1.8 3 mm
0.26 4 mm
6 mm
0.9
0.13 9 mm

0
0

-0.13 -0.9
300 320 340 360 380 400 420 440 460 480 300 320 340 360 380 400 420 440 460 480
crank angle, deg crank angle, deg

Fig. 18. Heat release rate and amount of heat released during combustion in function of CA for k = 1.4.

0.36 4

0.3 3 mm 3.2
4 mm
0.24 6 mm
2.4
dQ/dα, kJ/deg

9 mm
0.18
Q, kJ

1.6
3 mm
0.12
4 mm
0.8 6 mm
0.06 9 mm

0 0

-0.06 -0.8
300 320 340 360 380 400 420 440 460 480 300 320 340 360 380 400 420 440 460 480
crank angle, deg crank angle, deg

Fig. 19. Heat release rate and amount of heat released during combustion in function of CA for k = 1.6.
2978 A. Jamrozik et al. / Applied Mathematical Modelling 37 (2013) 2961–2982

0.18 3.5

0.15 2.8
3 mm
4 mm
0.12
6 mm
dQ/dα, kJ/deg

2.1
9 mm
0.09

Q, kJ
1.4
3 mm
0.06 4 mm
0.7 6 mm
0.03 9 mm

0 0

-0.03 -0.7
300 320 340 360 380 400 420 440 460 480 300 320 340 360 380 400 420 440 460 480
crank angle, deg crank angle, deg

Fig. 20. Heat release rate and amount of heat released during combustion in function of CA for k = 1.8.

0.12 3

0.1 3 mm 2.4
4 mm 3 mm
0.08 6 mm 4 mm
1.8
dQ/d α, kJ/deg

9 mm 6 mm
0.06 9 mm
Q, kJ

1.2
0.04
0.6
0.02

0
0

-0.02 -0.6
300 320 340 360 380 400 420 440 460 480 300 320 340 360 380 400 420 440 460 480
crank angle, deg crank angle, deg

Fig. 21. Heat release rate and amount of heat released during combustion in function of CA for k = 2.0.

5 3000

prechamber 2500
prechamber
4
cylinder
temperature, K

cylinder
pressure, MPa

2000
3
1500
2
1000
1 500

0 0
345 360 375 390 405 420 435 450 345 360 375 390 405 420 435 450
crank angle, deg crank angle, deg

Fig. 22. Courses of pressure and temperature in the prechamber and the main chamber of engine model equipped with the connecting duct of 6 mm
diameter at k = 2.0.

The main component of NOx in exhaust gases is nitric oxide. NO is created in the engine combustion chambers at a high
temperature above 1000 °C as a result of nitrogen and oxygen molecule reaction. During the exhaust of gases from the en-
gine, the greater amount of available oxygen and lower temperatures lead to creation of more stable nitrogen dioxide NO2.
Finally the ratio of NO2 to NO in the exhaust gas is generally from 0.05 to 0.1.
NO concentration in combustion chambers of the engine model for the whole range of the analyzed excess air factor of
1.4–2.0 is shown in Figs. 23–26.
 A. Jamrozik et al. / Applied Mathematical Modelling 37 (2013) 2961–2982 2979

Fig. 23. NO concentration in combustion chamber of engine model at selected CA for k = 2.0 and 1.8.

300 400

240 320
prechamber prechamber
NO, ppm
NO, ppm

180 cylinder 240 cylinder

average average
120 160

60 80

0 0
345 360 375 390 405 420 435 450 345 360 375 390 405 420 435 450
crank angle, deg crank angle, deg

Fig. 24. Variations of NO concentration in prechamber as well as in cylinder and average concentration at the prechamber and cylinder for k = 2.0 and 1.8.

Results of modeling in KVA-3V showed that the concentration of nitric oxide in prechamber of SI engine with two stage
combustion system powered by lean mixtures of k from 1.8 to 2.0 is much bigger than the concentration of this component
in the main chamber.
The final NO emission in SI engine with two stage combustion system operating within the analyzed k range from 1.4 to
2.0 is determined by the content of nitric oxide in the cylinder. The share of NO produced in the prechamber is negligibly
small.
Nitric oxide (NO) emission of two-stage combustion system was obtained as a result of modeling in KIVA-3V code
whereas the experimental research delivered data concerning NOx emission. The NOx consists mostly of NO (concentration
in the exhaust gas is about 95% of NOx concentration) created in the combustion chambers and NO2 created during the ex-
haust of combustion products.
The comparison of nitric oxide (NO) emission of engine model and NOx emission of the test engine for Qin/Qtot = 2.5% is
shown in Fig. 27.
The figure shows that for both engines the leaner fuel mixture causes the decrease in the concentration of nitric oxide.
The most favorable conditions for the formation of nitric oxide in the combustion chambers occurred in case of the richest
2980 A. Jamrozik et al. / Applied Mathematical Modelling 37 (2013) 2961–2982

Fig. 25. NO concentration in combustion chamber of medaled engine at selected CA for k = 1.6 and 1.4.

600 1000

480 800
NO, ppm
NO, ppm

360 600
prechamber
prechamber
cylinder
240 cylinder 400
average
average
120 200

0 0
345 360 375 390 405 420 435 450 345 360 375 390 405 420 435 450
crank angle, deg crank angle, deg

Fig. 26. Variations of NO concentration in prechamber as well as in cylinder and average concentration at the prechamber and cylinder for k = 1.6 and 1.4.

2000 2000

1600 experimental (NOx) 1600

NOx, ppm

KIVA-3V (NO)
NO, ppm

1200 1200

800 800

400 400

0 0
1.3 1.5 1.7 1.9 2.1
λ

Fig. 27. Comparison of nitric oxide emission of engine model and the test engine.
 A. Jamrozik et al. / Applied Mathematical Modelling 37 (2013) 2961–2982 2981

mixture of the average excess air factor k = 1.4. In this case, the NOx emission in the test engine was equal 1450 ppm and for
engine model the emission was equal 860 ppm. The lowest concentration occurred for the highest excess air factor (2.0) in
case of the test engine – 64 ppm and in case of the engine model – 29 ppm. The difference between above mentioned results
can be neglected as the analyzer measurement error in this measurement range was equal ±32 ppm of NOx [34].

5. Summary

The KIVA-3V code used for numerical calculations allows generating 3D geometric mesh of combustion chambers of the
engine with prechamber (two stage combustion system) and allowed to perform numerical calculations of processes occur-
ring in this engine. Performed simulations of two-stage combustion process delivered information concerning spatial and
time-dependent turbulent kinetic energy, pressure, temperature and NO distribution in combustion chambers of the engine
model. The data would be extremely difficult to obtain by experimental means. Calculations were conducted for four values
of excess air factor: 1.4, 1.6, 1.8, 2.0 (averaged for the main chamber and the prechamber). The pressure, temperature, heat
release rate and amount of heat released during combustion showed that despite the substantial depletion of fuel mixture
(up to k = 2.0), the combustion process can be fast enough to provide high values of indicated pressure and proper engine
efficiency in consequence.
Results of modeling the concentration of nitric oxide in the combustion chambers of the engine with two-stage combus-
tion system showed that even though NO concentration in the prechamber was significantly higher than in the cylinder, the
crucial influence on the final NO emission of the engine with two stage combustion process for k from 1.4 to 2.0 has the NO
concentration in the main chamber. It is caused by the fact that the participation of NO created in the prechamber is neg-
ligibly small.
Numerical modeling results reveal that the mathematical model of SI engine with two-stage combustion system properly
reflects the processes which occur in working spaces of the test engine. Satisfying compatibility of numerical modeling and
experimental research results was obtained. Moreover, the usefulness of numerical modeling of flow processes as a tool,
which assists during designing and developing combustion systems, was proved one more time. Comparison of results ob-
tained using the KIVA-3V code with the results of the analysis of the test engine cycle shows slight quantitative differences in
the values of maximum pressures and temperatures in the combustion chambers of the engine. In case of 2.5% of energetic
share of the enriching charge, the maximal pressure values and the CA at which they appear are similar for the modeling and
experiment. The maximum cycle temperature averaged for the main chamber and the prechamber for the entire analyzed
range of k, calculated in KIVA-3V is about 150 K (from 7.5% to 10%) higher than the temperature calculated on the basis of the
test engine indication. Comparison of nitrogen oxide emission of KIVA-3V model with the emission of the test engine,
showed rather large discrepancies. For Qin/Qtot = 2.5% at k = 1.4, the measured values of NOx were 1.75 times higher than
the modeled values, and at k = 2.0 the measured values were close to each other, and the values are within the range of mea-
surement error.

References

[1] M. Bernhardt, J. Michałowska, S. Radzimirski, Automotive Air Pollution (in Polish), WKŁ, Warszawa, 1976.
[2] K. Cupiał, A. Jamrozik, A. Kociszewski, Vergleich von Gasmotoren, die mit verschiedenen Verbrennungssystemen arbeiten, Erste Internationale
Fachthemenkonferenz Gasmotoren, Motortech GmbH, Celle, 2003. pp. 7–13.
[3] A. Kowalewicz, Combustion Systems of the High Speed Internal Combustion Engines (in Polish), WKŁ, Warszawa, 1980.
[4] J.B. Heywood, Internal Combustion Engine Fundamentals, McGraw – Hill Book Company, 1988.
_
[5] A. Kowalewicz, S. Luft, A. Rózycki, M. Gola, Chosen aspects of feeding in SI engines with poor gasoline–air mixtures (in Polish), J. Kones 6 (3–4) (1999)
156–162.
[6] A. Jamrozik, Creation and Combustion of Heterogeneous Burn Mixtures in Spark Ignition Engines, Ph.D. Thesis, Czestochowa University of Technology,
2004 (in Polish).
[7] L.A. Gussak, G.V. Evart, D.A. Ribiński, Carburetor Type Internal Combustion Engine with Prechamber, US Patent 3092,088, 1963.
[8] L.A. Gussak, Method of Prechamber Torch Ignition in Internal Combustion Engines, US Patent 3230,939, 1966.
[9] A.K. Oppenheim, K. Teichman, K. Hom, H.E. Stewart, Jet Ignition of an Ultra-Lean Mixture, SAE Paper 780637, vol. 87, 1978.
[10] A.K. Oppenheim, H.E. Stewart, K. Hom, Pulsed Jet Combustion Generator for Premixed Charge Engines, US Patent 4926,818, 1990.
[11] S.J. Charlton, D.J. Jager, M. Wilson, Computer Modelling and Experimental Investigation of a Lean Burn Natural Gas Engine, SAE Paper 900228, 1990.
[12] C. Robinet, P. Higelin, B. Moreau, O. Pajot, J. Andrzejewski, Anew Firing Concept for Internal Combustion Engines: ‘‘I’APIR’’, SAE Paper 1999-01-0621.
[13] H. Endo, K. Tanaka, Y. Kakuhama, Y. Goda, T. Fujiwaka, M. Nishigaki, Development of the lean burn Miller cycle gas engine, The Fifth International on
Diagnostics and Modeling of Combustion in Internal Combustion Engines – COMODIA, 2001.
[14] R.P. Roethlisberger, D. Favrat, Investigation of the prechamber geometrical configuration of a natural gas spark ignition engine for cogeneration: part I,
numerical simulation, Int. J. Therm. Sci. 42 (3) (2003) 223–237.
[15] R.P. Roethlisberger, D. Favrat, Investigation of the prechamber geometrical configuration of a natural gas spark ignition engine for cogeneration: part II,
experimentation, Int. J. Therm. Sci. 42 (3) (2003) 239–253.
[16] C.C.J. French, Alternative Engines – Interest or a Real Alternative (in Polish), No. 10, 1990, pp. 5–10.
[17] M. Noguchi, S. Sanda, N. Nakamura, Development of Toyota lean burn engine, SAE Paper 760757, 1976.
[18] B. Sendyka, A. Sochan, S. Kudzia, J. Soczówka, Determination of the total efficiency in the engine with direct injection of the petrol, J. Kones Int.
Combust. Engines 10 (3–4) (2003).
[19] Diesel & Gas Turbine, Worldwide Catalog, 68th Annual Product & Buyer’s Guide for Engine, Power Markets, 2003.
[20] P. Janicki, Production and development of gas engines in the H. Cegielski – FSA (in Polish), V International Scientific Conference Gas Engines 2000,
2000, pp. 179–193.
[21] T. Stenhede, Combined Heat and Power Solutions of Wärtsilä, VI International Conference Gas Engines, 2003, pp. 645–656.
[22] M.P. Jeffery, Design and development of the Waukesha AT25GL series gas engine, Energy Sources Technology Conference and Exhibition New Orleans,
1988.
2982 A. Jamrozik et al. / Applied Mathematical Modelling 37 (2013) 2961–2982

[23] D. Mooser, CAT Gas Engines, KEC Kiel Engine Center, Caterpillar medium speed gasmotor, Entwicklung & Betriebserfahrung, Erste Internationale
Fachthemenkonferenz Gasmotoren, Motortech GmbH, Celle, 2003.
[24] A.A. Amsden, P.J. O’Rourke, Butler T.D., KIVA-II, A Computer Program For Chemically Reactive Flows with Spray, Los Alamos National Laboratory, LA-
UR-11560-MS, 1989.
[25] A.A. Amsden, KIVA-3, A KIVA Program with Block-Structured Mesh for Complex Geometries, Los Alamos National Laboratory, LA-12503-MS, 1993.
[26] A.A. Amsden, KIVA-3V, A block-structured KIVA program for engines with vertical or canted valve, Los Alamos National Laboratory, LA-UR-97-689,
1997.
[27] Y.B. Zeldovich, The oxidation of nitrogen in combustion explosions, Acta Physicochim. USSR 21 (1946) 577–628.
[28] K. Cupiał, A. Jamrozik, SI engine with the sectional combustion chamber, J. Kones 9 (3–4) (2002) 62–66.
[29] W. Tutak, Modeling of fresh charge swirl in the combustion chamber of a piston engine, Ph.D. Thesis, Czestochowa University of Technology, 2002 (in
polish).
[30] I. Andersson, Cylinder Pressure and Ionization Current Modeling for Spark Ignited Engines, Linköping Studies in Science and Technology, Thesis No.
962, 2002.
[31] P. Belardini, C. Bertoli, Multi-Dimensional Modeling of Combustion and Pollutants Formation of New Technology Light Duty Diesel Engines, Oil & Gas
Science and Technology – Rev. IFP, vol. 54 (No. 2) (1999), pp. 251–257.
[32] W. Tutak, A. Jamrozik, A. Kociszewski, Flow processes in the gas engine prechamber, in: Proceedings of International Conference Simulation in
automotive engineering, Kazimierz Dolny, 2005, pp. 457–464 (in polish).
[33] J.A. Gatowski, E.N. Balles, M. Chun, F.E. Nelson, J.A. Ekchian, J.B. Heywood, Heat Release Analysis of Engine Pressure Data, SAE Paper 841359, 1984.
[34] A. Jamrozik, W. Tutak, A study of performance and emissions of SI engine with a two-stage combustion system, Chem. Process Eng. 32 (4) (2011) 453–
471.