You are on page 1of 30

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/229475866

Development and validation of a comprehensive two‐zone model for


combustion and emissions formation in a DI diesel engine

Article  in  International Journal of Energy Research · November 2003


DOI: 10.1002/er.939

CITATIONS READS
59 187

3 authors, including:

Dimitrios C. Rakopoulos Dimitrios C. Kyritsis


National Technical University of Athens Khalifa University of Science & Technology
73 PUBLICATIONS   4,153 CITATIONS    118 PUBLICATIONS   2,449 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Application of Nanotechnology and Photocatalysis for Water Treatment View project

Thermodynamic Simulations of Engine Combustion Processes View project

All content following this page was uploaded by Dimitrios C. Kyritsis on 24 November 2017.

The user has requested enhancement of the downloaded file.


INTERNATIONAL JOURNAL OF ENERGY RESEARCH
Int. J. Energy Res. 2003; 27:1221–1249 (DOI: 10.1002/er.939)

Development and validation of a comprehensive


two-zone model for combustion and emissions
formation in a DI diesel engine

C.D. Rakopoulos1,n,y,z,}, D.C. Rakopoulos1,} and D.C. Kyritsis2,k


1
Internal Combustion Engines Laboratory, Thermal Engineering Section, Mechanical Engineering Department, National
Technical University of Athens, 9 Heroon Polytechniou Str., Zografou Campus, Athens 15780, Greece
2
Department of Mechanical and Industrial Engineering, University of Illinois at Urbana-Champaign,
1206 West Green Str., Urbana, IL 61801, U.S.A.

SUMMARY
A two-zone model for the calculation of the closed cycle of a direct injection (DI) diesel engine is presented.
The cylinder contents are taken to comprise a non-burning zone of air and another homogeneous zone in
which fuel is continuously supplied from the injector holes during injection and burned with entrained air
from the air zone. The growth of the fuel spray zone, consisting of a number of fuel–air conical jets equal to
the injector nozzle holes, is carefully modelled by incorporating jet mixing to determine the amount of
oxygen available for combustion. Application of the mass, energy and state equations in each one of the
two zones yields local temperatures and cylinder pressure histories. For calculating the concentration of
constituents in the exhaust gases, a chemical equilibrium scheme is adopted for the C–H–O system of the 11
species considered, together with chemical rate equations for the calculation of nitric oxide (NO). A model
for the evaluation of soot formation and oxidation rates is incorporated. A comparison is made between
the theoretical results from the computer program implementing the analysis, with experimental results
from a vast experimental investigation conducted on a fully automated test bed, direct injection, standard
‘Hydra’, diesel engine located at the authors’ laboratory, with very good results, following a multi-
parametric study of the constants incorporated in the various sub-models. Pressure indicator diagrams and
plots of temperature, NO, soot density and of other interesting quantities are presented as a function of
crank angle, for various loads and injection timings, elucidating the physical mechanisms governing
combustion and pollutants formation. Copyright # 2003 John Wiley & Sons, Ltd.

KEY WORDS: diesel engine; two-zone model; combustion; emissions formation

n
Correspondence to: Professor C. D. Rakopoulos, Thermal Engineering Section, Mechanical Engineering Department,
NTUA, 9 Heroon Polytechniou Str., Zografou Campus, Athens 15780, Greece.
y
E-mail: cdrakops@central.ntua.gr
z
Professor Internal Combustion Engines.
}
Director of I.C. Engines Laboratory.
}
Research assistant.
k
Assistant professor of Combustion.

Received 27 September 2002


Copyright # 2003 John Wiley & Sons, Ltd. Accepted 6 March 2003
1222 C. D. RAKOPOULOS, D. C. RAKOPOULOS AND D. C. KYRITSIS

1. INTRODUCTION

Public concern for maintaining a clean environment has motivated extensive research into the
sources of pollution and ways to reduce it. One such major polluting contributor is the internal
combustion engine, either in the form of spark-ignition (Otto) or Diesel version. In parallel to
the environmental threat posed by these engines (Ferguson, 1986; Rakopoulos, 1991), their
main source of fuel, the crude oil, is being depleted at increasing rates, so that the development
of less polluting and more efficient engines is today of paramount importance for engine
manufacturers (Watson and Janota, 1982; Heywood, 1988; Rakopoulos, 1992). Also, to this end
corroborates the fact of increasing threat posed by the rivals of the internal combustion engine,
for smaller size engines, such as the electric motors, the hybrid engines, the fuel cells and the like
(Ferguson, 1986; Rakopoulos and Kyritsis, 2001).
Experimental work aimed at fuel economy and low pollutants emissions from Diesel (as well
as Otto) engines involves successive changes of each of the many parameters concerned, which is
both very demanding in terms of money and time. With the advent of digital computers, the
obvious alternative is today the simulation of the engine performance by a mathematical model,
where the effect of design and operation changes can be quickly and cheaply estimated by the
use of computers, provided that the main mechanisms are recognized and correctly modelled
(Benson and Whitehouse, 1979; Horlock and Winterbone, 1986; Ferguson, 1986; Heywood,
1988).
The process of diesel combustion is inherently very complex due to the transient and
heterogeneous character of combustion, which is mainly controlled by turbulent mixing of fuel
and air in the fuel jets issuing from the nozzle holes. High speed photography studies and in-
cylinder sampling have revealed some interesting features of combustion (Heywood, 1988). The
first attempts to simulate the diesel engine cycle substituted the ‘internal combustion’ by
‘‘external heat addition’’ (Ferguson, 1986). Apparent heat release rates were empirically
correlated to fuel injection rates and used in a thermodynamic cycle calculation to obtain
cylinder pressure, in a uniform mixture (Austen and Lyn, 1960–61). Models based on droplet
evaporation and combustion, while still in a mono-zone mixture, can only take partially into
account the heterogeneous character of diesel combustion (Whitehouse and Way, 1969–70).
The need for accurate predictions of exhaust emissions pollutants forced the researchers to
attempt developing two-zone combustion models (Khan et al., 1971; Whitehouse and Sareen,
1974; Kouremenos et al., 1989). Eventually, some multi-zone combustion models have appeared
(Shahed et al., 1975; Hodgetts and Shroff, 1975; Hiroyasu et al., 1983; Kouremenos et al., 1987),
carrying the expected drawbacks of the first attempts, where the detailed analysis of fuel–air
distribution permits the calculation of exhaust gas composition with reasonable accuracy, albeit
under the rising of computing time (cost) when compared to lower zones diesel combustion
models.
At this point it should be mentioned that multi-dimensional models are proving useful in
examining problems characterized by the need for detailed spatial information and complex
interactions of many phenomena simultaneously (Butler et al., 1981; Gosman and Harvey,
1982). However, these are limited by the relative inadequacy of sub-models for turbulence,
combustion chemistry and by computer size and cost of operation, to crude approximations to
the real flow and combustion problems (Hiroyasu et al., 1982; Heywood, 1988).
Therefore, it is felt that a reasonable choice seems to be a two-zone model, which includes
the effect of changes in engine design and operation on the details of the combustion

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Energy Res. 2003; 27:1221–1249
DEVELOPMENT AND VALIDATION OF A COMPREHENSIVE TWO-ZONE MODEL 1223

process, through a phenomenological model where the geometric details are fairly well
approximated by detailed modelling of the various mechanisms involved, along the lines of
recent multi-zone models philosophies (Rakopoulos et al., 1995; Rakopoulos and Hountalas,
1998, 2000). This is going to have the advantage of relative simplicity and very reasonable
computer time cost.
Thus, the object of the present work is the development of a comprehensive two-zone model,
applied for a direct injection (DI) diesel engine, similar in broad outline to others, but with
several differences that one should expect from an independent research group. The present
paper gives a detailed description of such a model, which contains upgraded jet mixing, heat
transfer and chemistry sub-models, incorporating a multi-parametric study of the constants in
the various sub-models needed for increased accuracy. It keeps as simple as possible the
numerical analysis treatment of the governing differential and algebraic equations, thus leading
to good solution convergence with reduced computer time cost. Furthermore, the present work
is extensively verified using data from a vast experimental investigation, conducted at the
authors’ laboratory on a fully automated test bed, four-stroke, water cooled, standard ‘Hydra’,
direct injection, high speed, diesel engine, which offers the ability to have its operating
conditions varied very easily. Plots of pressure, temperatures in the two zones, nitric oxide (NO),
soot density and of other interesting quantities are presented as a function of crank angle (time),
for various loads and injection timings, providing insight into the physical mechanisms
governing combustion and pollutants formation.

2. GENERAL DESCRIPTION OF THE MODEL

The model deals with a bowl-in-piston combustion chamber (DI) into which fuel is injected, in a
predominant radial direction, from a number of holes (four in the present application) drilled
symmetrically on the injector tip. The model includes only those processes occurring during
the portions of compression and expansion strokes, when all valves are closed (closed cycle).
It incorporates the main processes taking place in the cylinder, i.e. the in-cylinder air motion,
the fuel spray development and mixing, the spray impingement on the walls, the turbulent
heat transfer and the chemistry of combustion. The droplets evaporation and the fuel
ignition delay are implicitly taken into account through the combustion sub-model. The fuel
considered is dodecane (C12H26), representing a common fuel for commercial diesel engines.
The main calculation procedure is based on the integration of the first law of thermodynamics
and the perfect gas state equation combined with the various sub-models, for each zone
separately.

2.1. Conservation and state equations


During compression only one zone (of pure air) exists. Then, the first law of thermo-
dynamics for a closed system is applied together with the perfect gas state equation (Heywood,
1988), i.e.
dQ ¼ dE þ p dV ð1Þ

p  V ¼ mðRmol =Mch ÞT ð2Þ

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Energy Res. 2003; 27:1221–1249
1224 C. D. RAKOPOULOS, D. C. RAKOPOULOS AND D. C. KYRITSIS

where dQ is the heat loss to the chamber walls, Mch the charge molecular weight and V the
instantaneous cylinder volume given by,
V ¼ Vc‘ þ ðpD2 =4Þr ½1 þ l1  cos j  ðl2  sin2 jÞ1=2  ð3Þ

where Vc‘ is the cylinder clearance volume and l the crank radius to piston rod length ratio.
During combustion and expansion, depending on the number of injector nozzle holes, an
equal number of zones constitute the entire burning section. In this case, apart from the perfect
gas state equation, the first law of thermodynamics for an open system is applied for each zone.
For the surrounding air zone, which only loses mass (air) to the burning zone, the first law is
written as (Benson and Whitehouse, 1979)
dQ ¼ dE þ p dV þ ha dma ð4Þ

while for the burning zone, which gains mass (air) from the air zone and also an enthalpic flow
from the fuel ready to be burned in the time step, the first law becomes
dQ ¼ dE þ p dV  ha dma  hf dmf ð5Þ

In the above equations, the internal energy E of the mixture is computed by knowing the
instantaneous composition and the specific internal energies of the constituents. The latter ones
are given as fourth order polynomial expressions of the absolute temperature T, including the
enthalpy of formation at absolute zero (Benson and Whitehouse, 1979). Similar expressions are
then derived for specific enthalpies and heat capacities, by applying the simple thermodynamic
relations connecting these quantities for a perfect gas.

2.2. Heat transfer model


Heat transfer between cylinder trapped mass and surrounding walls is calculated using the
formula of Annand (1963):
Q’ lg
q’ ¼ ¼ a ðReÞb ðTw  Tg Þ þ c  ðTw4  Tg4 Þ ð6Þ
F D
where Tg is the absolute zone temperature, Re=rDup/mg is the Reynolds number, up=2NS/60 is
the mean piston speed with S the piston stroke and lg, mg the gas thermal conductivity and
dynamic viscosity, respectively, expressed as polynomial functions of temperature Tg. The
constants a,b,c in the above equation are evaluated as described in the following section.

2.3. Fuel spray penetration and air entrainment rate


The correlation of Hiroyasu et al. (1980) is used for the spray tip location as a function of time,
based on relevant experimental data and turbulent gas jet theory. The fuel injected inside the
combustion chamber breaks up into globules, forming a cone-shaped spray corresponding to
each one of the ‘z’ nozzle holes. If the kmol of air trapped in the cylinder (in one cycle) are
watot ¼ matot =Ma ; then the phenomenon will continue until each spray penetration reaches a value
of (D/2+pD/z), or until it entrains a maximum quantity of air equal to watot =z: It is noted that
given the global AFR, the total quantity of fuel to be injected in the cycle wf tot ¼ mf tot =Mf is
known.

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Energy Res. 2003; 27:1221–1249
DEVELOPMENT AND VALIDATION OF A COMPREHENSIVE TWO-ZONE MODEL 1225

At first, the parameters related to the point of jet break-up are computed. A mean value of
fuel injection rate is used:
m%’ f inj ¼ mf tot =ðz  Djinj Þ ð7Þ
j

where index ‘j’ denotes each one of the z sprays and Djinj is the total duration of fuel injection in
degrees crank angle. Then, the mean spray velocity is given by
%’ f inj 6N =ðr‘  F Þ
u%inj ¼ m ð8Þ
j

where F ¼ pD2N =4: Therefore, the mean pressure drop in the nozzle is
Dp ¼ 0:5r‘ ðu%inj =cD Þ2 ð9Þ
where the discharge coefficient cD is taken to be equal to 0.39.
The break-up time tbr is found now by equating the two expressions of Hiroyasu et al. (1980)
correlation before and after tbr, corresponding to the break-up length x=xbr,
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
x ¼ 0:39 2Dp=r‘  t for 05t  tbr ð10Þ
pffiffiffiffiffiffiffiffiffiffiffiffi
x ¼ 2:95ðDp=ra Þ0:25 DN  t for t  tbr ð11Þ
so that qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
tbr ¼ 28:61r‘  DN = ra  Dp ð12Þ
where ra is the density of air inside the cylinder just before combustion begins and the break-up
length is, xbr ¼ u% inj  tbr :
At this point we take into account the air swirl by modifying the above correlations, by
introducing the air swirl ratio RS=NS/N with NS the cylinder charge rotational speed. Then, the
break-up length with swirl is,
xbrs ¼ xbr ð1 þ p  Rs  N  xbr =30  u% inj Þ1 ð13Þ
and the corresponding break-up time
tbrs ¼ xbrs =u%inj ¼ ðxbrs =xbr Þ  tbr ð14Þ
We are now in a position to start our main calculations. At each crank angle step, the fuel
injection rate and the cumulative fuel injected in each spray are, respectively,
m’ f injj ¼ m’ f inj =z and mf injj ¼ mf inj =z ð15Þ
Similarly, the spray tip velocity and the pressure drop across the nozzle are, respectively,
uinj ¼ m’ f injj 6N =ðr‘  F Þ and Dp ¼ 0:5r‘ ðuinj =cD Þ2 ð16Þ
Now we discriminate the following cases, according to the value of time t, for the cases of
no-swirl and of presence of swirl (index s).
For t  tbrs:
x ¼ uinj  t and xs ¼ xð1 þ pRs N  x=30uinj Þ1 ð17Þ
The spray angles are y=0 and ys=0, while the fuel inside the spray is effectively mfj=0.
For t>tbrs:
pffiffiffiffiffiffiffi
x ¼ 2:95ðDp=ra Þ0:25 DN  tb and xs ¼ xð1 þ pRs N  x=30uinj Þ1 ð18Þ

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Energy Res. 2003; 27:1221–1249
1226 C. D. RAKOPOULOS, D. C. RAKOPOULOS AND D. C. KYRITSIS

In the above relation, the exponent of time b=0.50 if xs  D=2; that is before wall
impingement, while it is taken equal to 0.48 if xs  D/2, that is after wall impingement. It is seen
that the same correlations are opted to be used both before and after wall impingement, using
different values for b. Thus, the complexities involved due to uncertainties in modelling the real
physical mechanisms existing in the spray picture after wall impingement are avoided (Khan
et al., 1971; Whitehouse and Sareen, 1974).
The spray angles (in rad) are
rffiffiffiffiffi pffiffiffi!  2
1 ra 3 x
y ¼ 2 arctan 0 4p and ys ¼ y  ð19Þ
A r‘ 6 xs

where constant A0 is given by the empirical relation (LN is the nozzle hole length),
A0 =3+0.28(LN/DN).
The mass of entrained air in the spray (of conical shape) is, respectively,
 
p y 2
maj ¼ tan ra  ðx  xbr Þ3 ð20Þ
3 2
!2
p ys
majs ¼ tan  ra  ðxs  xbrs Þ3 ð21Þ
3 2

where in the above relations it is considered that the volume taken by the fuel is negligible
against that of the air, as well as the paraboloid part at the base of the cone against that of the
pure cone.
The above relations for the mass entrained apply only in the case where tbrs 5t 
ðDjinj =6N Þ þ tbrs : In the case where t > ðDjinj =6N Þ þ tbrs ; one has to subtract from the spray
volume a conical part at the tail (index t) of the spray. Then, the above equations become,
respectively,
   
p y 2 3 p yt 2
maj ¼ ra tan ðx  xbr Þ  ra tan ðxt  xbr Þ3 ð22Þ
3 2 3 2

   
p ys 2 p yts 2
majs ¼ ra tan ðxs  xbrs Þ3  ra tan ðxts  xbrs Þ3 ð23Þ
3 2 3 2
where
pffiffiffiffiffiffiffi
xt ¼ 2:95ðDp=ra Þ0:25 DN ½t  ðDjinj =6N Þ  tbr bt ð24Þ
is the cone length to be subtracted in the case of no-swirl (for the exponent bt apply the same as
for the exponent b above), while
xts ¼ xt ð1 þ pRs N  x=30uinj Þ1 ð25Þ
is the corresponding cone length to be subtracted when swirl exists.
The corresponding spray angles for the tail part of the cone to be subtracted are, respectively,
yt ¼ y and yts ¼ yt ðxt =xts Þ2 ð26Þ

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Energy Res. 2003; 27:1221–1249
DEVELOPMENT AND VALIDATION OF A COMPREHENSIVE TWO-ZONE MODEL 1227

For the fuel mass which is effectively inside the spray (the fuel starts to be considered inside
the spray only after the break-up length), we have:
mf j ¼ mf injj ðt  tbrs Þ=t for t  Dfinj =6N ð27Þ

mf injj
mf j ¼ ðt  tbrs Þ for Dfinj =6N 5t  ðDfinj =6N Þ þ tbrs ð28Þ
ðDfinj =6N Þ

mf j ¼ mf injj for t > ðDfinj =6N Þ þ tbrs ð29Þ

Lastly, the air–fuel ratio (by mass) is, respectively,


AFRspr ¼ maj =mf j and AFRsprs ¼ majs =mf j ð30Þ

2.4. Combustion model


The semi-empirical model of Whitehouse and Way (1969–70) is used for calculating the rate of
combustion. The injected fuel in the burning zone mixes up with the air entrained from the air
zone via a mixing and diffusion process, while the burning rate of the fuel is effected by an
Arrhenious-type expression.
For the preparation rate P (in kg 8CA1):
P ¼ Kmf1x
inj
mxf up pO
m
2
ð31Þ

For the reaction rate R (in kg 8CA1):


Z j
K 0 pO
R ¼ pffiffiffi2ffi eEred =T ðP  RÞ dj ð32Þ
N0 T o

In the above relations, mf inj is the total mass of injected fuel (cumulative) up to the present
crank angle j (time t), which is given by the following expression:
Z j
dmf inj
mf inj ¼ dj ð33Þ
o dj
where ðdmf inj =djÞ is the fuel mass injection rate. Furthermore,
Z j
mf up ¼ mf inj  P dj ð34Þ
o

where mf up is the cumulative mass of not yet prepared fuel, N0 is the engine speed (in rps) and
pO2 is the partial pressure of oxygen (in bar) in the zone. Constants K, K0 , x, m and Ered are
evaluated as described in the following section.
Finally, the combustion rate is given by the relations:
dmf b =dj ¼ R if R5P and dmf b =dj ¼ P if R>P ð35Þ

2.5. Chemistry of combustion


Combustion products are defined by dissociation considerations. For the C–H–O–N system
the complete chemical equilibrium scheme proposed by Way (1977) is used. The following 11

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Energy Res. 2003; 27:1221–1249
1228 C. D. RAKOPOULOS, D. C. RAKOPOULOS AND D. C. KYRITSIS

species are included in the calculations:


ð6Þ N2 ; ð5Þ O2 ; ð1Þ CO2 ; ð3Þ H2 O; ð2Þ CO; ð4Þ H2 ;
ð8Þ NO; ð9Þ OH; ð10Þ N; ð11Þ H and ð12Þ O
where each species ‘i’ is referred to by the number in parenthesis in front of its name, with
number ‘7’ reserved for the fuel.
For the burning zone, given its volume, temperature, mass of fuel burned and mass of air
entrained, the concentration of each one of the above species can be calculated by solving a
system of 11 equations consisting of four atom balance equations (one for each element) and
seven equilibrium equations. The chemical reactions considered to be in equilibrium are:
ð1Þe N2 þ O2 Ð 2NO ð2Þe 2H2 O þ O2 Ð 4OH
ð3Þe 2CO þ O2 Ð 2CO2 ð4Þe 2H2 þ O2 Ð 2H2 O
ð5Þe N2 Ð 2N ð6Þe H2 Ð 2H
ð7Þe O2 Ð 2O
The equilibrium constants for the above seven reactions are as follows:
2 4 2
pNO pOH pCO
Kp1 ¼ ; Kp2 ¼ 2
; Kp3 ¼ 2
2

pN 2  pO 2 pH 2 O  pO 2 pCO  pO2
2 2 2 2
pH 2O
pN pH pO
Kp4 ¼ 2 p
; Kp5 ¼ ; Kp6 ¼ ; K p7 ¼ ð36Þ
pH 2 O2 pN 2 pH 2 pO 2
where pi are the partial pressures of the species made non-dimensional with respect to the
standard atmospheric pressure patm ¼ 1 atm:

2.6. Nitric oxide formation model


As the consideration of chemical equilibrium cannot predict correctly the NO concentration, the
generally accepted kinetics formation scheme proposed by Lavoie et al. (1970) is used. The
equations which describe the above model, together with their forward reaction rate constants
kif (in m3 kmol1 s1), are:
N þ NO Ð N2 þ O; k1f ¼ 3:1  1010 expð160=T Þ
N þ O2 Ð NO þ O; k2f ¼ 6:4  106 T expð3125=T Þ
N þ OH Ð NO þ H; k3f ¼ 4:2  1010
The change of (NO) concentration (in kmol m3) is expressed as follows:
1 dððNOÞV Þ R1
¼ 2ð1  a2 Þ ð37Þ
V dt 1 þ aR1 =ðR2 þ R3 Þ
where Ri is the one-way equilibrium rate for reaction i, defined as
R1 ¼ k1f ðNÞe ðNOÞe ; R2 ¼ k2f ðNÞe ðO2 Þe ; R3 ¼ k3f ðNÞe ðOHÞe
with index ‘e’ denoting equilibrium concentration and term a=(NO)/(NO)e.

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Energy Res. 2003; 27:1221–1249
DEVELOPMENT AND VALIDATION OF A COMPREHENSIVE TWO-ZONE MODEL 1229

2.7. Net soot formation model


The net soot formation rate is calculated by using the model proposed by Hiroyasu et al. (1983),
as modified by Lipkea and DeJoode (1994).
According to this model, there is a soot formation (index sf) and a soot oxidation (index sc)
rate formulated as follows:
dmsf
¼ Asf dm0:8 0:5
f p expðEsf =ðRmol T ÞÞ ð38Þ
dt

dmsc
¼ Asc msn ðpO2 =pÞp n expðEsc =ðRmol T ÞÞ ð39Þ
dt
where pressures are expressed in bar and masses in kg, dmf is the unburned fuel vapour mass to
be burned in the time step dt and pO2 is the partial pressure of oxygen in the zone.
Therefore, the net soot formation is expressed as follows:
dmsn dmsf dmsc
¼  ð40Þ
dt dt dt
The exponent n, the constants Asf, Asc and the activation energies Esf, Esc in the above
equations are evaluated as described in the following section.

3. STEP-BY-STEP NUMERICAL SOLUTION PROCEDURE AND COMPUTER


PROGRAM STRUCTURE

3.1. General description


The equations of the model exposed in the previous section are solved numerically using a
marching technique, with a time step size of 18 crank angle. The steps used are described in
detail in the following sub-sections. Before the start of calculations, the design characteristics of
the engine in hand are provided, as well as the operating data at the start of the cycle, i.e. at IVC
event. The calculations stop at EVO event. The corresponding program is written in Fortran V
language and executed on a Pentium-III Personal Computer. The computer program structure
follows very closely the step-by-step procedure described below.

3.2. Computation of compression phase


(1.1) Introduce the data at IVC event, i.e. j1, p1, T1, trapped composition (air with no fuel)
and compute V1 from engine geometry. From the perfect gas state equation calculate the
trapped kmol of air watot ; so that given the global AFR the total kmol of fuel to be injected in the
cycle wf tot are known. Select crank angle step size Dj equal, here, to 18.
(1.2) Calculate the initial internal energy E1 using its T1 relation and similarly for the heat
capacities cp and cv.
(1.3) For new crank angle j2=j1+Dj, compute V2 from engine geometry.
(1.4) Estimate temperature T2 assuming isentropic change:
T2 ¼ T1 ðV1 =V2 Þg1 ¼ T1 ðV1 =V2 ÞRmol =cv ; with g=cp(T1)/cv(T1)=constant.

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Energy Res. 2003; 27:1221–1249
1230 C. D. RAKOPOULOS, D. C. RAKOPOULOS AND D. C. KYRITSIS

Then, find pressure p2 from the perfect gas state equation:

p2 ¼ ðV1 =V2 ÞðT2 =T1 Þp1

(1.5) Calculate the internal energy E2 using its T2 relation.


(1.6) Calculate the work in the step: dW ¼ p  dV ¼ 0:5ðp1 þ p2 ÞðV2  V1 Þ:
(1.7) Calculate dQ from the heat loss sub-model of Annand (1963).
(1.8) Apply the first law of thermodynamics for a closed system:

f ðEÞ ¼ E2  E1 þ dW  dQ ¼ 0

Solve this equation with respect to T2 using the Newton–Raphson numerical method
(McCracken and Dorn, 1964), so that a better estimate of T2 is found.
(1.9) Calculate p2 for the revised value of T2, using the gas state equation at time moments
1 and 2.
(1.10) Repeat steps 5–9 until the error f(E) in the first law equation is negligible.
(1.11) Set conditions at the end of time step (index 2) as initial conditions for the next time
step (new state, index 1) and repeat all steps from 1.2 to 1.10 . Carry on this way, until j2 equals
the value at the start of fuel injection jsoinj.
(1.12) During the jet break-up period there is no combustion, so continue with the
compression phase calculations keeping also track of the fuel injection rate m’ f inj and the
cumulative injected fuel mf inj :

3.3. Computation of combustion and expansion phases for both zones


(2.1) Connect the forming two-zone system with the previous single-zone one:

T2u ¼ T2 ; T2b ¼ T2 ; V2u ¼ V2 ; V2b ¼ 0; biu ¼ bi ; bib ¼ 0

where biu and bib (i=1, . . . , 12) are the number of kmol of constituents (including fuel) per kmol
of wf tot ; at the beginning of the time step.
(2.2) Set conditions at the end of previous time step (old state, index 2) as initial conditions for
the current time step (new state, index 1) for both zones.
(2.3) For crank angle j2=j1+Dj, compute V2 from engine geometry.
(2.4) Calculate the fuel injection rate m’ f inj and cumulative injected fuel mf inj :
(2.5) Calculate the fuel inside each spray mf j and the corresponding entrained air majs :
(2.6) Estimate pressure p2 at the end of the time step, to be checked later on, by assuming
isentropic change (with g=1.35): p2 ¼ p1 ðV1 =V2 Þg :
(2.7) Define a mean temperature at the beginning and end of time step as:
P  P  P  P 
biu  T1u þ bib  T1b aiu  T2u þ aib  T2b
Tbulk1 ¼ P P ; Tbulk2 ¼ P P ð41Þ
biu þ bib aiu þ aib

where aiu and aib (i =1 , . . . , 12) are the number of kmol of constituents, per kmol of wf tot ;
at the end of the time step. Then, calculate total heat loss dQ for both zones from the
heat loss sub-model of Annand (1963) at conditions: (p1+p2)/2, (Tbulk1+Tbulk2)/2 and
(V1+V2)/2.

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Energy Res. 2003; 27:1221–1249
DEVELOPMENT AND VALIDATION OF A COMPREHENSIVE TWO-ZONE MODEL 1231

(2.8) Distribute total dQ in the two zones in proportion to their kmol and absolute
temperatures (Shahed et al., 1975):
P P
ð biu Þ  T1u þ ð aiu Þ  T2u
dQu ¼ P P P P  dQ ð42Þ
ð biu Þ  T1u þ ð aiu Þ  T2u þ ð bib Þ  T1b þ ð aib Þ  T2b
P P
ð bib Þ  T1b þ ð aib Þ  T2b
dQb ¼ P P P P  dQ ð43Þ
ð biu Þ  T1u þ ð aiu Þ  T2u þ ð bib Þ  T1b þ ð aib Þ  T2b
(2.9) Make computations in the unburned zone, as described in a following sub-section.
(2.10) Compute the isenthalpic mixing of the newly incoming air in the burning zone, as
described in a following sub-section.
(2.11) Make computations in the burning zone, as described in a following sub-section.
(2.12) Check temperatures of the two zones. If T2u0, T2b0 are the temperatures calculated up
to the present iteration step and T2u, T2b the same temperatures as found by executing steps
2.9–2.11, then relative errors eu=|(T2u0T2u)/ T2u0| and eb=|(T2b0T2b)/T2b0| must be less than
a specified accuracy, otherwise repeat steps 2.7–2.11 until temperatures convergence.
(2.13) Check volumes of the two zones and change pressure p2 if necessary, since the volume
constraint V2=V2b+V2u must be satisfied, where V2u and V2b were found from steps 2.9 and
2.11, respectively. If relative error ev=|(V2V2bV2u)/V2| is greater than a specified accuracy,
then set a new value in p2 as
p2 ¼ p2 ðV2b þ V2u Þ=V2

and repeat steps 2.7–2.12 until satisfaction of the volume constraint relation.
(2.14) Calculate various useful quantities, such as the gross and net heat release rates as well
as their cumulative values at each step (Heywood, 1988).
(2.15) Repeat steps 2.2–2.14 and carry on this way, until j reaches the EVO event, i.e. the end
of the closed cycle.

3.4. Calculations in the unburned (air) zone


The calculations in this zone are similar to the ones during the compression phase.
(3.1) Calculate the initial internal energy E1 using its T1u relation and similarly for the
enthalpy h1 and heat capacity cv(T1u) of air.
(3.2) Calculate V1u from the gas state equation,
V1u ¼ wf tot  w1u  Rmol  T1u =p1
P
where w1u ¼ biu are the total kmol in the unburned zone at state 1 per kmol of wf tot :
(3.3) Calculate the kmol of the various constituents. Total kmol in the unburned zone, at
state 2, per kmol of wf tot :
w2u ¼ ½watot  zðmajs =Ma Þ=wf tot

where the air loss due to entrainment by the fuel sprays is, wloss ¼ w1u  w2u :
Thus, the mixture composition is:
aiu ¼ biu ; for i ¼ 1; . . . ; 12 except 5 and 6

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Energy Res. 2003; 27:1221–1249
1232 C. D. RAKOPOULOS, D. C. RAKOPOULOS AND D. C. KYRITSIS

with
a5u ¼ b5u  0:21  wloss and a6u ¼ b6u  0:79  wloss
(3.4) Make a first estimate of T2u considering isentropic change.
(3.5) Calculate the internal energy E2 using its T2u relation and similarly for the heat capacity
cv(T2u) of air.
(3.6) Calculate V2u from the gas state equation,
V2u ¼ wf tot  w2u  Rmol  T2u =p2
(3.7) Calculate the work in the step, dW ¼ 0:5ðp1 þ p2 ÞðV2u  V1u Þ
(3.8) Apply the first law of thermodynamics for an open system:
f ðEÞ ¼ E2  E1 þ dW  dQu þ wf tot  wloss  ‘1 ¼ 0
Solve this equation with respect to T2u using the Newton–Raphson numerical method, so that
a better estimate of T2u is found.
(3.9) Repeat steps 3.5–3.8 until the error f(E) in the first law equation is negligible.

3.5. Calculations for mixing air and fuel in the spray zone
Before proceeding with calculations in the burning zone, it is necessary to compute the
isenthalpic mixing of the newly incoming air into the burning zone (Horlock and Winterbone,
1986).
(4.1) Find the new composition which results from mixing wgain (where wgain=wloss) kmol of
entrained air per kmol of wf tot ; taking also into account the addition of dmf b kg of fuel to be
burned in the time step:
bibmix ¼ bib ; for i ¼ 1; . . . ; 12 except 5; 6 and 7
with
b5bmix ¼ b5b þ 0:21  wgain ; b6bmix ¼ b6b þ 0:79  wgain
and
b7bmix ¼ b7b þ ðdmf b =Mf Þ=wf tot
(4.2) Calculate the partial pressure of oxygen at state 1 after mixing, which is needed in the
Whitehouse and Way (1969–70) burning model,
X
pO21 ¼ p1 ðb5bmix =w1bmix Þ with w1bmix ¼ bibmix
the total kmol in the burning zone at state 1 after mixing per kmol of wf tot
(4.3) Calculate enthalpy H1b for the content of the burning zone before the current input of
air, from its relation at the initial temperature T1b.
(4.4) Similarly, calculate enthalpy H1u of the newly entered air of temperature T1u.
(4.5) Estimate temperature T1bmix after mixing, in relation to the kmol:
T1bmix ¼ ðw1b  T1b þ wgain  T1u Þ=ðw1b þ wgain Þ
(4.6) Calculate enthalpy H1bmix and heat capacity cp ðT1bmix Þ of the burning zone after mixing,
from their T1bmix relations.

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Energy Res. 2003; 27:1221–1249
DEVELOPMENT AND VALIDATION OF A COMPREHENSIVE TWO-ZONE MODEL 1233

(4.7) Since isenthalpic mixing is assumed, it holds for the first law:
f ðH Þ ¼ H1bmix  H1b  H1u ¼ 0
Solve this equation with respect to T1bmix using the Newton–Raphson numerical method, so
that a better estimate of T1bmix is found.
(4.8) Repeat steps 4.6 and 4.7 until the error f(H) in the first law equation is negligible.
(4.9) Calculate the volume of the burning zone after mixing, using the perfect gas state
equation: V1bmix ¼ wftot  w1bmix  Rmol  T1bmix =p1

3.6. Calculations in the burning zone


Now we are in a position to effect combustion calculations in the burning zone assuming that
fuel still exists, otherwise we have entered the ‘pure’ expansion phase. In the analysis the fuel is
neglected volumetrically and thermodynamically, apart from the quantity of fuel dmf b to be
burned in the time step, which is in the vapour phase.
(5.1) Calculate internal energy of the burning zone at state 1 after mixing, E1bmix ; from its T1bmix
relation.
(5.2) Estimate temperature T2b at the end of the time step (state 2), i.e. after combustion,
considering isentropic change:
T2b ¼ T1bmix ðp2 =p1 Þðg1Þ=g  dmf b  qvs =ðwf tot  w1bmix  cv Þ
where qvs is the lower heat of combustion of the fuel (J kg1) and cv ¼ Rmol =ðg  1Þ; with
g ¼ 1:35:
(5.3) Estimate the volume of the burning zone, using the gas state equation:
V2b ¼ wf tot  w1bmix  Rmol  T2b =p2
(5.4) Calculate the mixture composition in the burning zone, considering chemical
equilibrium for all species apart from NO which is kinetically controlled. Solve by the method
of Way (1977), as described in the previous section.
(5.5) Calculate the internal energy E2b using its T2b relation and similarly for the heat capacity
cv(T2b) in state 2.
(5.6) Calculate the volume in state 2, using the gas state equation
V2b ¼ wf tot  w2b  Rmol  T2b =p2
P
where w2b ¼ aib are the total kmol in the burning zone in state 2, per kmol of wf tot :
(5.7) Calculate the work dW1 in the time step, for a change from state 1 after mixing to state 2:
dW1 ¼ 0:5ðp1 þ p2 ÞðV2b  V1bmix Þ:
(5.8) Apply the first law of thermodynamics for a change from state 1 after mixing to state 2 (a
closed system is considered, since the addition of air has already been taken into account):
f ðEÞ ¼ ðE2b  E1bmix Þ þ dW1  dQb ¼ 0:
Solve this equation with respect to T2b using the Newton–Raphson numerical method, so that
a better estimate of T2b is found.
(5.9) Repeat steps 5.4–5.8 until the error f(E) in the first law equation is negligible.
(5.10) Calculate partial pressure of oxygen in state 2: pO22 ¼ p2 ða5b =w2b Þ:
(5.11) For a mean partial pressure of oxygen equal to 0:5ðpO21 þ pO22 Þ and for a mean
temperature of 0:5ðT1bmix þ T2b Þ apply the combustion sub-model and repeat steps 5.4–5.10 until
the error f(E) in the first law equation is negligible.

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Energy Res. 2003; 27:1221–1249
1234 C. D. RAKOPOULOS, D. C. RAKOPOULOS AND D. C. KYRITSIS

(5.12) Compute the quantity of NO in the burning zone through the chemical kinetics scheme
described in a previous section, by solving the relevant differential equation for NO with a
fourth-order Runge–Kutta numerical scheme (Conte, 1965), using typically a sub-step of 1/20th
the main time step.
(5.13) Compute the net soot formed in the burning zone through the relevant differential
equations scheme described in a previous section, using a simple Euler scheme (McCracken and
Dorn, 1964; Conte, 1965).
(5.14) Finally, calculate the work for the total change from state 1 (before mixing) to state 2
since this is the real work produced: dW ¼ 0:5ðp1 þ p2 ÞðV2b  V1b Þ:

3.7. Calculation of the mean state of the two zones


Having calculated the states in the two zones at every time step, proceed to calculate the mean
state of these assuming isenthalpic mixing. Some quantities are directly calculated, such as the
heat loss dQ ¼ dQu þ dQb ; the work dW ¼ 0:5ðp1 þ p2 ÞðV2  V1 Þ and the mixture composition
X
ai ¼ aiu þ aib ; i ¼ 1; . . . ; 12 and w2 ¼ ai ¼ w2u þ w2b
Specifically, for calculating the mean temperature, follow the steps below:
(6.1) Calculate the enthalpy of the unburned zone H2u ; using its T2u relation. Similarly for the
enthalpy of the burning zone H2b :
(6.2) Make a first estimate of temperature T2 ; in relation to the kmol of each zone: T2 ¼
ðT2u  w2u þ T2b  w2b Þ=w2 :
(6.3) Calculate enthalpy H2 and heat capacity cp(T2) for the mean state, using their T2
relations.
(6.4) The first law for isenthalpic mixing reads: f ðH Þ ¼ H2  H2u  H2b ¼ 0
Solve this equation with respect to T2 using the Newton–Raphson numerical method, so that
a better estimate of T2 is found.
(6.5) Repeat steps 6.3 and 6.4 until the error f(H) in the first law equation is negligible.

4. EXPERIMENTAL TEST FACILITY

4.1. Engine description


Facilities to monitor and control engine variables such as engine speed, engine load, water and
lube-oil temperatures, fuel and air flows, etc., are installed on a fully automated test bed, single
cylinder, water cooled, Ricardo–Cussons, ‘‘Hydra’’, experimental standard engine, located at
the first author’s laboratory. This engine has the ability to operate on the Otto (spark-ignition)
or direct injection (DI) diesel or indirect injection (IDI) diesel, four-stroke principle, by
changing various parts of the crank gear mechanism, cylinder and head.
For the present investigation it is used as a naturally aspirated DI diesel engine, operating in a
rotational speed range of 1000–4500 rpm.
The engine main design characteristics are summarized below:

Cylinder bore : 0.08026 m


Piston stroke : 0.08890 m
Compression ratio : 19.81:1

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Energy Res. 2003; 27:1221–1249
DEVELOPMENT AND VALIDATION OF A COMPREHENSIVE TWO-ZONE MODEL 1235

Inlet valve opening : 88 BTDC


Inlet valve closing : 428 ABDC
Exhaust valve opening : 608 BBDC
Exhaust valve closing : 128 ATDC
Swept volume : 0.450 l

The engine has a re-entrant bowl-in-piston combustion chamber, with an injector nozzle
having four holes, 0.25 mm diameter each, drilled symmetrically around its tip, forming a spray
cone angle of 1608. The diesel fuel injection pump is fitted with an 11 mm diameter plunger. A
‘‘Bosch’’ injector body is used, which opens at a pressure of 250 bar. A range of 0–408 advance
(static) is provided.
The diesel fuel used has a specific gravity at 158C of 0.834, cetane number of 53, kinematic
viscosity at 208C of 2.95 mm2 s1 and lower calorific value of 43 100 kJ kg1.

4.2. Test installation description


The engine is mounted on a fully automated test bed and coupled to a ‘‘McClure’’ DC motoring
dynamometer, having load absorbing and motoring capabilities. There is one electric sensor for
speed and one for load (torque), the signal both of which are fed to indicators on the control
panel and to the controller. Via knobs on the control panel, the operator can set
the dynamometer to control speed or load. There is also a capability of setting automatically
the static injection timing from a switch on the control panel.
The coolant (water) and lube-oil circulation is achieved by electrically driven pumps, the
temperature being controlled by water fed heat exchangers. Heaters are used to maintain oil and
coolant temperatures during warm up and light load conditions. Thermocouples are located at
strategic points of the engine, with their indications shown on a multi-point electronic
temperature indicator. The engine exhaust system is connected to a shop made silencer system.
A shop made tank and flowmeter system is used for fuel consumption measurements. A viscous
type, laminar flowmeter is used for air flow measurements.
The exhaust gas analysis consisted of a group of analysers for measuring soot (smoke), nitric
oxide (NO), total unburned hydrocarbons and carbon monoxide. The nitric oxide concentration
(in ppm) in the exhaust was measured by a ‘‘Signal’’ chemiluminescent analyser and was fitted
with a thermostatically controlled heated line. The smoke level in the exhaust was measured
with a ‘‘Bosch’’ smoke opacimeter, the readings of which are provided as Bosch smoke units or
equivalent smoke density (mgrams of soot per cubic meter of exhaust gases).

4.3. Transducers measuring set-up and data acquisition system


For measuring the pressure in the cylinder, a ‘‘Kistler’’ miniature piezoelectric transducer is
used, flush mounted to the cylinder head, and connected to a ‘‘Kistler’’ charge amplifier.
Another ‘‘Kistler’’ transducer is connected on the injector side of the pipe connecting injection
pump and injector, to provide the fuel pressure signal. The signal from a ‘‘Tektronix’’ TDC
magnetic pick-up marker is used for time reference (Rakopoulos et al., 1998).
The output signals from the magnetic pick-up and the two piezoelectric transducers signals,
while they are continuously monitored on a dual beam ‘‘Tektronix’’ storage oscilloscope, are
connected to the input of a ‘‘Keithley’’ DAS-1801ST A/D board installed on an IBM
compatible Pentium III PC. This board can acquire input data at a total throughput rate of

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Energy Res. 2003; 27:1221–1249
1236 C. D. RAKOPOULOS, D. C. RAKOPOULOS AND D. C. KYRITSIS

312.5 ksamples s1 from up to eight differential analogue inputs, also utilizing dual-channel
Direct Memory Access (DMA) operation. It is then appropriate for the recording of high
frequency engine signals, with the added advantage of high signal resolution (much less than
18C A, depending also on the number of simultaneous recorded signals). This high acquisition
rate capability also secures the nearly simultaneous sampling of all input signals.
Control of this high speed acquisition system is achieved by developing a computer code
based on the ‘‘TestPoint’’ control software (Rakopoulos and Mavropoulos, 1999). The resulted
program, specially designed for transient internal combustion engine applications, is an object-
oriented functional code. It allows the direct on-screen selection of the desired values of all
acquisition parameters, namely: engine type (two-stroke or four-stroke); engine speed (in rpm);
desired number of recorded cycles; selection of the acquired channels; sampling interval
(in 8C A).
After the selection of the above values, the resulting sampling rate per channel is displayed on
screen. When acquisition starts, data are initially stored in the computer’s memory. This secures
the continuous gap-free acquisition for the large amount of multi-channel input data. When the
desired amount of engine cycles is recorded, the data are transferred from the memory to a
corresponding file in the hard disk and directly displayed on-screen (as X–Y graphs). Special
warning signs and other secondary options are also available.

5. THEORETICAL AND EXPERIMENTAL RESULTS COMPARISON-DISCUSSION

A comparison is made in this section between the measured values derived from the
experimental investigation and the ones calculated by the model, so that the model could
be tested performance- and emissionswise. To effect comparison it was necessary to calibrate the
model, following a multi-parametric analysis of the various sub-model constants applied for all
engine operating conditions examined. The tactics followed consisted of the three steps stated
below:
(i) Determine the values of heat transfer correlation constants so that the theoretical and
experimental compression curves best match. The constants b and c are set at their most usual
values b=0.75 and c=3.4  108 W m2 K1 and then a is found as a=0.375. The temperature
of the cylinder walls Tw is taken equal to 450 K.
(ii) Determine the values of the combustion sub-model constants so that maximum
combustion pressure, indicated mean effective pressure and NO concentration in the exhaust,
theoretical and experimental values, best match. The constants x, m, K0 and Ered are set at their
most usual values x=0.33, m=0.4, K0 =1.2  1010 K0.5 bar1deg1 and Ered=16 500 K and then
constant K is found as K=0.035 barm deg1.
(iii) Determine the constants in the net soot formation correlation so that exhaust soot
density, theoretical and experimental values, best match. The following values are found:
Asf=Asc=0.8 and the exponent of pressure in the soot formation equation n=2.5 is found to
better fit the data, instead of the value 1.8 suggested in the original paper. The corresponding
activation energies are set at their usual values, i.e. Esf=30 000 and Esc=50000 kJ kmol1.
The effect of the two major parameters of load and injection timing are examined in this
work, concerning both performance and exhaust emissions. Results are presented at an engine
speed of 2500 rpm, three loads corresponding to 40, 60 and 80% of full load and two static
injection timings of 15 and 208 crank angle before TDC.

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Energy Res. 2003; 27:1221–1249
DEVELOPMENT AND VALIDATION OF A COMPREHENSIVE TWO-ZONE MODEL 1237

Figures 1 and 2 show the calculated and experimental pressure (indicator) diagrams at 80% of
full load, at static injection timings of 15 and 208C A BTDC, respectively. In both cases, a very
good coincidence is observed between calculated and experimental values. As expected, the
pressure values during combustion are higher for the higher injection timing case (Heywood,
1988).
Figures 3 and 4 show the calculated temperature crank angle diagrams at 80% of full load, at
static injection timings of 15 and 208C A BTDC, respectively, as provided by the two-zone
model. Each of these diagrams gives the temperature history of the unburned (air) zone, the
burning (spray) zone and the mean state. It is observed a sudden increase in the temperature
of the burning zone with the initiation of combustion, while the temperature of the unburned
zone changes slightly. As expected, the temperature values in the burning zone during
combustion are higher for the higher injection timing case. The high temperatures of the
burning zone are mainly responsible for the formation of NO and soot (Horlock and
Winterbone, 1986).
Figure 5 shows the comparison of calculated and experimental loads, for the three loads and
the two static injection timings examined. A very satisfactory coincidence exists, with all points
lying very close to the bisectrix of the diagram. It is to be noted that in this and some figures to
follow, the three loads examined of 40, 60 and 80% of full load are expressed as indicated mean
effective pressures, i.m.e.p. (in bar). The calculated i.m.e.p. are provided directly by the two-
zone model, while the available experimental brake torque values are firstly converted into
brake mean effective pressures, b.m.e.p., and then expressed as i.m.e.p. values by addition of the
corresponding friction mean effective pressures, f.m.e.p. (Heywood, 1988), which are provided
by the engine manufacturer.

80.00

SPEED= 2500 rpm


LOAD= 80 %
ST. INJ.= -15 deg.
60.00
experimental
calculated
PRESSURE (bar)

40.00

20.00

0.00

-200.00 -100.00 0.00 100.00 200.00


CRANK ANGLE (deg.)

Figure 1. Calculated and experimental pressure (indicator) diagram, at 80% of full load and a static
injection timing of 158C A BTDC.

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Energy Res. 2003; 27:1221–1249
1238 C. D. RAKOPOULOS, D. C. RAKOPOULOS AND D. C. KYRITSIS

100.00

SPEED = 2500 rpm


LOAD= 80%
80.00 ST. INJ.= -20 deg.
experimental
calculated
PRESSURE (bar)

60.00

40.00

20.00

0.00

-200.00 -100.00 0.00 100.00 200.00


CRANK ANGLE (deg.)

Figure 2. Calculated and experimental pressure (indicator) diagram, at 80% of full load and a static
injection timing of 208C A BTDC.

3000.00

SPEED = 2500 rpm burning zone


LOAD = 80 %
unburned zone
ST. INJ. = -15 deg.
mean state
TEMPERATURE (K)

2000.00

1000.00

0.00

-200.00 -100.00 0.00 100.00 200.00

CRANK ANGLE (deg.)

Figure 3. Calculated temperature of the unburned zone, the burning zone and the mean state versus crank
angle, at 80% of full load and a static injection timing of 158C A BTDC.

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Energy Res. 2003; 27:1221–1249
DEVELOPMENT AND VALIDATION OF A COMPREHENSIVE TWO-ZONE MODEL 1239

3000.00

SPEED = 2500 rpm burning zone


LOAD = 80 %
unburned zone
ST. INJ. = -20 deg.
TEMPERATURE (K) mean state

2000.00

1000.00

0.00

-200.00 -100.00 0.00 100.00 200.00


CRANK ANGLE (deg.)

Figure 4. Calculated temperature of the unburned zone, the burning zone and the mean state versus crank
angle, at 80% of full load and a static injection timing of 208C A BTDC.

10.00

SPEED= 2500 rpm

st. inj. -15 deg.


CALCULATED I.M.E.P. (bar)

st. inj. -20 deg.


8.00

6.00

4.00

4.00 6.00 8.00 10.00


EXPERIMENTAL I.M.E.P. (bar)

Figure 5. Comparison of calculated and experimental indicated mean effective pressures, i.m.e.p.,
for the three loads examined of 40, 60 and 80% of full load and the two static injection timings of 15 and
208C A BTDC.

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Energy Res. 2003; 27:1221–1249
1240 C. D. RAKOPOULOS, D. C. RAKOPOULOS AND D. C. KYRITSIS

Figure 6 shows the calculated and experimental peak combustion pressures versus load, for
the two static injection timings examined. It can be observed a very good coincidence between
calculated and experimental values. As is well known for this kind of engines, the peak pressure
increases with either the increase in load or the injection timing (Watson and Janota, 1982).
Figure 7 shows the calculated and experimental indicated efficiency versus load, for the two
static injection timings examined. It is to be noted that the calculated indicated efficiency values
are provided directly by the two-zone model, while the available experimental (brake) fuel
consumption values are firstly converted into brake efficiency values (by also knowing the
experimental brake power and the fuel calorific value) and then expressed as indicated efficiency
from the direct analogy between brake and indicated values for load and efficiency (Ferguson,
1986). It can be observed a very good coincidence between calculated and experimental values.
As is well known, the indicated efficiency increases with decreasing load.
Figure 8 shows the NO concentration history (in ppm) calculated when considering chemical
equilibrium or chemical kinetics conditions, at 80% of full load and a static injection timing of
158C A BTDC. It is seen that when considering chemical kinetics, which is the correct case, the
maximum NO value is lower than the chemical equilibrium case and furthermore that it is
frozen at nearly this value during the expansion phase when temperature drops (Lavoie et al.,
1971).
Figure 9 shows the calculated NO concentration history (in ppm) for various loads and a
static injection timing of 158C A BTDC. It can be seen that NO formation values increase with
increasing load. This is expected since NO formation is favoured for near stoichiometric
conditions towards the lean side, which are approached by the higher loads in a diesel engine,

110.00
SPEED = 2500 rpm

calculated , st. inj. -15 deg.

experimental , st. inj. -15 deg.


PEAK PRESSURE (bar)

100.00
calculated , st. inj. -20 deg.
experimental , st. inj. -20 deg.

90.00

80.00

70.00

5.00 6.00 7.00 8.00 9.00


I.M.E.P. (bar)

Figure 6. Calculated and experimental peak combustion pressures versus load (expressed as i.m.e.p.), for
the two static injection timings examined.

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Energy Res. 2003; 27:1221–1249
DEVELOPMENT AND VALIDATION OF A COMPREHENSIVE TWO-ZONE MODEL 1241

60.00

SPEED = 2500 rpm

calculated , st. inj. -15 deg.

INDICATED EFFICIENCY (%)


56.00
experimental , st. inj. -15 deg.
calculated , st. inj. -20 deg.
experimental , st. inj. -20 deg.

52.00

48.00

44.00

5.00 6.00 7.00 8.00 9.00


I.M.E.P. (bar)

Figure 7. Calculated and experimental indicated efficiency versus load (expressed as i.m.e.p.), for the two
static injection timings examined.

2000.00

SPEED = 2500 rpm chemical kinetics


LOAD = 80 %
1600.00 ST. INJ. = -15 deg. chemical equilibrium

1200.00
NO (ppm)

800.00

400.00

0.00

-200.00 -100.00 0.00 100.00 200.00


CRANK ANGLE (deg.)

Figure 8. Nitric oxide (NO) concentration history calculated when considering chemical equilibrium or
chemical kinetics conditions, at 80% of full load and a static injection timing of 158C A BTDC.

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Energy Res. 2003; 27:1221–1249
1242 C. D. RAKOPOULOS, D. C. RAKOPOULOS AND D. C. KYRITSIS

1000. 00

SPEED = 2500 rpm


ST. INJ. = -15 deg.

800.00
LOAD = 40 %
LOAD = 60 %

600.00 LOAD = 80 %
NO (ppm)

400.00

200.00

0.00

-200.00 -100.00 0.00 100.00 200.00


CRANK ANGLE (deg.)

Figure 9. Calculated nitric oxide (NO) concentration history, for the three loads of 40, 60 and 80% of full
load and a static injection timing of 158C A BTDC.

and high temperatures (Benson and Whitehouse, 1979). The latter ones are higher the higher the
load.
Figure 10 shows the comparison of calculated and experimental nitric oxide concentration
values at the exhaust, for the three loads and the two static injection timings examined. A very
satisfactory coincidence is observed between calculated and experimental values. As expected,
the NO values increase with increasing load, due to approaching stoichiometric conditions and
getting higher temperatures. By the same token, the NO values are higher for the higher
injection timing case, since then temperatures in the cylinder are higher as seen in Figures 3
and 4 (Khan et al., 1971).
Figure 11 shows the calculated soot density history (in mg m3) for various loads and a static
injection timing of 158C A BTDC. It can be seen that soot density values increase with
increasing load. This is expected since soot formation is favoured under pyrolysis conditions, i.e.
at rich conditions and high temperatures which are approached by the higher loads (Khan et al.,
1972). During the expansion phase, the soot density values fall sharply due to the end of soot
formation and the increase of the total cylinder volume.
Figure 12 shows the comparison of calculated and experimental soot density values at the
exhaust, for the three loads and the two static injection timings examined. A very satisfactory
coincidence is observed between calculated and experimental values. As expected, the values
increase with increasing load, due to approaching richer conditions and getting higher
temperatures. The soot values are lower for the higher injection timing case, since then the
higher temperatures involved in the cylinder have a higher impact on the soot oxidation than
formation (Watson and Janota, 1982).

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Energy Res. 2003; 27:1221–1249
DEVELOPMENT AND VALIDATION OF A COMPREHENSIVE TWO-ZONE MODEL 1243

3000.00
SPEED = 2500 rpm

calculated , st. inj. -15 deg.

experimental , st. inj. -15 deg.


calculated , st. inj. -15 deg.
2000.00
experimental , st. inj. -15 deg.
NO (ppm)

1000.00

0.00

5.00 6.00 7.00 8.00 9.00


I.M.E.P. (bar)

Figure 10. Calculated and experimental exhaust nitric oxide (NO) concentration values versus load
(expressed as i.m.e.p.), for the two static injection timings examined.

1200. 00

SPEED = 2500 rpm LOAD = 40 %


ST. INJ. = -15 deg.
LOAD = 60 %
LOAD = 80 %

800.00
SOOT (mg/m**3)

400.00

0.00

-200.00 -100.00 0.00 100.00 200.00


CRANK ANGLE (deg.)

Figure 11. Calculated soot density history, for the three loads of 40, 60 and 80% of full load and a static
injection timing of 158C A BTDC.

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Energy Res. 2003; 27:1221–1249
1244 C. D. RAKOPOULOS, D. C. RAKOPOULOS AND D. C. KYRITSIS

160.00

SPEED = 2500 rpm


calculated , st. inj. -15 deg.
120.00
SOOT (mg/m**3 ) experimental , st. inj. -15 deg.
calculated , st. inj. -20 deg.
experimental , st. inj. -20 deg.
80.00

40.00

0.00

5.00 6.00 7.00 8.00 9.00


I.M.E.P. (bar)

Figure 12. Calculated and experimental exhaust soot density values versus load (expressed as i.m.e.p.), for
the two static injection timings examined.

5000. 00
HEAT TRANSFER COEFFICIENT (W/m**2/K)

SPEED = 2500 rpm LOAD= 40 %


ST. INJ.= -15 deg.
LOAD= 60 %
4000. 00
LOAD= 80 %

3000. 00

2000. 00

1000. 00

0.00

-200.00 -100.00 0.00 100.00 200.00


CRANK ANGLE (deg.)

Figure 13. Calculated gas heat transfer coefficient versus crank angle, for the three loads of 40, 60 and 80%
of full load and a static injection timing of 158C A BTDC.

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Energy Res. 2003; 27:1221–1249
DEVELOPMENT AND VALIDATION OF A COMPREHENSIVE TWO-ZONE MODEL 1245

20.00

SPEED = 2500 rpm


LOAD = 80%
ST. INJ. = -15 deg.
16.00

CUMULATIVE FUEL MASS (mg)

12.00

injected
8.00 prepared
reacted
burned
4.00

0.00

- 40.00 -20.00 0.00 20.00 40.00 60.00


CRANK ANGLE (deg.)

Figure 14. Calculated cumulative fuel injected, prepared, reacted and burned versus crank angle, at 80%
of full load and a static injection timing of 158C A BTDC.

Figure 13 shows the variation with crank angle of calculated gas heat transfer coefficient
in the cylinder, for various loads and a static injection timing of 158C A BTDC. As
expected, heat transfer coefficients become very high during the combustion period. They
increase with increasing load, due to higher temperatures in the cylinder (Benson and
Whitehouse, 1979).
Figure 14 gives a sample example of the calculated cumulative fuel injected, prepared, reacted
and burned as a function of crank angle, at 80% of full load and static injection timing of
158C A BTDC. This picture follows closely the description of the fuel burning sub-model
(Whitehouse and Way, 1969–70) exposed in a previous section.

6. CONCLUSIONS

A comprehensive, two-zone, thermodynamic model of direct injection diesel engine cycles is


developed. Emphasis is given to including most of the processes taking place in the cylinder
during the closed cycle and to describing as many as possible of these processes by correct
modelling of physical laws.
An extended experimental investigation is carried out on a fully automated test bed, Ricardo-
Hydra standard, DI diesel engine, located at the authors’ laboratory, in order to generate a
performance and emissions data base that can be used for validation tests of this model.
The two-zone model proposed proves to be very effective in predicting the engine
performance and exhaust emissions. Specifically, the effect of the two major operating

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Energy Res. 2003; 27:1221–1249
1246 C. D. RAKOPOULOS, D. C. RAKOPOULOS AND D. C. KYRITSIS

parameters of load and injection timing upon cylinder pressures, efficiency, NO and soot
concentration values is well verified for a wide spectrum of variation of these parameters.
It is important to note that the same values of the set of calibration constants of the various
sub-models are used here for all conditions examined, performance- and emissionwise, following
a judicious multi-parametric analysis for their optimum choice.
It is concluded that the analysis model used here predicts the performance and emission
aspects of a DI diesel engine with success and in a theoretically sound way, thus forming a
serious rival to its multi-zone counterpart due to the much lower computer time cost and effort
needed by it.

NOMENCLATURE

ai =number of kmol of species i, at the end of time step, per kmol of fuel to be
injected totally in the cycle
AFR =air–fuel ratio (by mass)
bi =number of kmol of species i, at the beginning of time step, per kmol of fuel to
be injected totally in the cycle
cp =molar specific heat capacity under constant pressure (J kmol1 K1)
cv =molar specific heat capacity under constant volume (J kmol1 K1)
D =cylinder diameter (m)
DN =injector nozzle hole diameter (m)
E =internal energy, (J), or activation energy (J kmol1)
Ered =reduced activation energy (K)
F =surface (m2)
h =specific enthalpy (J kg1)
H =enthalpy (J)
kf =forward reaction rate constant
Kp =chemical equilibrium constant
m =mass (kg)
M =molecular weight (kg kmol1)
N =engine rotational speed (rpm)
p =pressure (Pa)
Q =heat (J)
Ri =one-way equilibrium rate for reaction i
Rmol =universal gas constant, 8314.3 (J kmol1 K1)
Rs =swirl ratio
t =time (s)
T =absolute temperature (K)
u =spray velocity (ms1)
V =volume (m3)
w =number of kmol of cylinder charge or of the contents of each zone, per kmol
of fuel to be injected totally in the cycle
watot =total kmol of air trapped in the cylinder
wf tot =total kmol of fuel to be injected in the cycle
W =work (J)

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Energy Res. 2003; 27:1221–1249
DEVELOPMENT AND VALIDATION OF A COMPREHENSIVE TWO-ZONE MODEL 1247

x =spray penetration (m)


z =number of injector nozzle holes

Greek letters

g =specific heat capacities ratio


Dp =pressure drop (Pa)
Dj =crank angle computational step (deg)
Djinj =duration of fuel injection (8C A)
y =spray angle (rad)
r =density (kg m3)
j =crank angle (deg)

Superscripts

} =mean value
 =time derivative

Subscripts

a =air
b =burning zone
br =spray break-up
e =chemical equilibrium
f =fuel
g =gas
i =species number
inj =injected
j =spray number
‘ =liquid fuel
mix =state at the beginning of time step after mixing
O2 =oxygen
s =swirl
sc =soot oxidized
sf =soot formed
sn =soot net
spr =spray
t =spray tail
tot =total
u =unburned zone
up =unprepared
w =wall
0 =initial value
1 =state at the beginning of time step
2 =state at the end of time step

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Energy Res. 2003; 27:1221–1249
1248 C. D. RAKOPOULOS, D. C. RAKOPOULOS AND D. C. KYRITSIS

Abbreviations

A =after
B =before
BDC =bottom dead centre
EVO =exhaust valve opening
IVC =inlet valve closing
i.m.e.p. =indicated mean effective pressure
rpm =revolutions per minute
TDC =top dead centre
8C A =degree of crank angle

REFERENCES
Annand WJD. 1963. Heat transfer in the cylinders of reciprocating internal combustion engines. Proceedings of the
Institute of Mechanical Engineers 177:973–990.
Austen AEW, Lyn WT. 1960–61. Relation between fuel injection and heat release in a direct injection engine and the
nature of the combustion processes. Proceedings of the Institute of Mechincal Engineers (AD) 1:47–62.
Benson RS, Whitehouse ND. 1979. Internal Combustion Engines. Pergamon Press: Oxford.
Butler TD, Cloutman LD, Dukovicz JK, Ramshaw JD. 1981. Multidimensional numerical simulation of reactive flow in
internal combustion engines. Progress Energy Combustion Science 7:293–315.
Conte SD. 1965. Elementary Numerical Analysis. McGraw-Hill: New York.
Ferguson CR. 1986. Internal Combustion Engines. Wiley: New York.
Gosman AD, Harvey PS. 1982. Computer analysis of fuel–air mixing and combustion in axisymmetric D.I. diesel engine.
SAE paper 820036.
Heywood JB. 1988. Internal Combustion Engine Fundamentals. McGraw- Hill: New York.
Hiroyasu H, Kadota T, Arai M. 1980. Supplementary comments: fuel spray characterization in diesel engines. In
Combustion Modeling in Reciprocating Engines, Mattari JN, Amann CA (eds). Plenum Press: New York; 369–408.
Hiroyasu H, Kadota T, Arai M. 1983. Development and use of a spray combustion modeling to predict diesel engine
efficiency and pollutant emı́ssions. Bulletin JSME 26(214):569–576.
Hiroyasu H, Yoshimatsu A, Arai M. 1982. Mathematical model for predicting the rate of heat release and exhaust
emissions in IDI diesel engines. In Diesel Engines for Passenger Cars and Light Duty Vehicles, Institute of Mechanical
Engineers, paper C102/82; 207–213.
Hodgetts D, Shroff HD. 1975. More on the formation of nitric oxide in a diesel engine. In Combustion in Engines,
Institute of Mechanical Engineers, paper C95/75; 129–138.
Horlock JH, Winterbone DE. 1986. The Thermodynamics and Gas Dynamics of Internal Combustion Engines. Clarendon
Press: Oxford.
Khan IM, Greeves G, Probert DM. 1971. Prediction of soot and nitric oxide concentrations in diesel engine exhaust. In
Air Pollution Control in Transport Engines, Institute of Mechanical Engineers, paper C142/71; 205–217.
Khan IM, Wang CHT, Langridge BE. 1972. Effect of air swirl on smoke and gaseous emissions from direct-injection
diesel engines. SAE paper 720102.
Kouremenos DA, Rakopoulos CD, Karvounis E. 1987. Thermodynamic analysis of direct injection diesel engines by
multi-zone modelling. ASME-WA Meeting, Proceedings of the AES, Boston MA, vol. 3(3):67–77.
Kouremenos DA, Rakopoulos CD, Hountalas DT. 1989. Computer simulat!ıon with experimental validation of the
exhaust nitric oxide and soot emissions in divided chamber diesel engines. ASME-WA Meeting, Proceedings of the
AES, San Francisco CA, vol. 10, No. 1:15–28.
Lavoie GA, Heywood JB, Keck JC. 1970. Experimental and theoretical study of nitric oxide formation in internal
combustion engines. Combustion Science and Technology 1:313–326.
Lipkea WH, DeJoode AD. 1994. Direct injection diesel engine soot modeling: formulation and results. SAE paper
940670.
McCracken DD, Dorn WS. 1964. Numerical Methods and Fortran Programming. Wiley: New York.
Rakopoulos CD. 1991. Influence of ambient temperature and humidity on the performance and emissions of nitric oxide
and smoke of high speed diesel engines in the Athens/Greece region. Energy Conversion and Management 31:447–458.
Rakopoulos CD. 1992. Olive oil as a fuel supplement in DI and IDI diesel engines. International Energy Journal 17:787–
790.

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Energy Res. 2003; 27:1221–1249
DEVELOPMENT AND VALIDATION OF A COMPREHENSIVE TWO-ZONE MODEL 1249

Rakopoulos CD, Giakoumis EG, Hountalas DT. 1998. Experimental and simulation analysis of the transient operation
of a turbocharged multi-cylinder IDI diesel engine. Energy Research 22:317–331.
Rakopoulos CD, Hountalas DT. 1998. Development and validation of a 3-D multi-zone combustion model for the
prediction of DI diesel engines performance and pollutants emissions. Transactions of SAE, Journal of Engines,
107:1413–1429, SAE paper 981021.
Rakopoulos CD, Hountalas DT. 2000. Development of a new 3-D multi-zone combustion model for indirect injection
diesel engines with a swirl type prechamber. Transactions of SAE, Journal of Engines, 109:718–733, SAE paper 2000-
01-0587.
Rakopoulos CD, Kyritsis DC. 2001. Comparative second law analysis of internal combustion engine operation for
methane, methanol and dodecane fuels. Energy}The International Journal, 26:705–722.
Rakopoulos CD, Mavropoulos GC. 1999. Experimental instantaneous heat fluxes in the cylinder head and exhaust
manifold of an air-cooled diesel engine. Energy Conversion and Management 41:1265–1281.
Rakopoulos CD, Hountalas DT, Taklis GN, Tzanos EI. 1995. Analysis of combustion and pollutants formation in a
direct injection diesel engine using a multi-zone model. Energy Research 19:63–88.
Shahed SM, Chiu WS, Lyn WT. 1975. A mathematical model of diesel combustion. In Combustion in Engines, Institute
of Mechanical Engineers, paper C94/75;119–128.
Watson N, Janota MS. 1982. Turbocharging the Internal Combustion Engine. MacMillan Press: London.
Way RJB. 1977. Methods for determination of composition and thermodynamic properties of combustion products for
internal combustion engine calculations. Proceedings of the Institute of Mechanical Engineers 190(60/76):687–697.
Whitehouse NO, Sareen BK. 1974. Prediction of heat release in a quiescent chamber diesel engine allowing for fuel/air
mixing. SAE paper 740084.
Whitehouse ND, Way RJB. 1969–70. Rate of heat release in diesel engines and its correlation with fuel injection data.
Proceedings of the Institute of Mechanical Engineers 184:17–29.

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Energy Res. 2003; 27:1221–1249

View publication stats

You might also like