Sie sind auf Seite 1von 9

Applied Thermal Engineering 37 (2012) 344e352

Contents lists available at SciVerse ScienceDirect

Applied Thermal Engineering

journal homepage:

Validation of a zero-dimensional model for prediction of NOx and engine

performance for electronically controlled marine two-stroke diesel engines
Fabio Scappin a, Sigurur H. Stefansson a, Fredrik Haglind a, *, Anders Andreasen b, Ulrik Larsen a

Technical University of Denmark, Department of Mechanical Engineering, DK-2800 Kgs. Lyngby, Denmark
MAN Diesel & Turbo, Engine Process Research, Process Development, Marine Low Speed Development, Teglholmsgade 41, DK-2450 Copenhagen SV, Denmark

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 24 May 2011
Accepted 22 November 2011
Available online 30 November 2011

The aim of this paper is to derive a methodology suitable for energy system analysis for predicting the
performance and NOx emissions of marine low speed diesel engines. The paper describes a zerodimensional model, evaluating the engine performance by means of an energy balance and a two
zone combustion model using ideal gas law equations over a complete crank cycle. The combustion
process is divided into intervals, and the product composition and ame temperature are calculated in
each interval. The NOx emissions are predicted using the extended Zeldovich mechanism. The model is
validated using experimental data from two MAN B&W engines; one case being data subject to engine
parameter changes corresponding to simulating an electronically controlled engine; the second case
providing data covering almost all model input and output parameters. The rst case of validation
suggests that the model can predict specic fuel oil consumption and NOx emissions within the 95%
condence intervals given by the experimental measurements. The second validation conrms the
capability of the model to match measured engine output parameters based on measured engine input
parameters with a maximum 5% deviation.
2011 Elsevier Ltd. All rights reserved.

Diesel engine
Engine tuning

1. Introduction
The development of marine low speed diesel engines with lower
emissions is primarily driven by the MARPOL Annex VI regulation
[1] adopted by the International Maritime Organization (IMO).
MARPOL Annex VI contains regulations on NOx, SOx and Particulate
Matter emissions. On July 1, 2010 the revised MARPOL Annex VI
entered in to force and it contains new limits for NOx emissions for
both new and existing ships as well as reduced SOx and PM emissions for all ships.
Allowable NOx emissions are reduced to 14.4 g/kWh for large
marine low speed engines (130 rpm) installed on ships constructed from January 1, 2011 and onwards, according to the Tier II
standard, and to 3.4 g/kWh for engines installed on ships constructed from January 1, 2016 and onwards, according to the Tier III
standard in designated emission control areas.
The widely used low speed two-stroke diesel engine can be
combined with a waste heat recovery unit and potentially offer
uniquely high fuel efciency and low specic emissions for diesel
engine ship propulsion. In designing and optimizing such

* Corresponding author. Tel.: 45 45 25 41 13; fax: 45 45 93 52 15.

E-mail address: (F. Haglind).
1359-4311/$ e see front matter 2011 Elsevier Ltd. All rights reserved.

a combined energy system, the focus is on the interaction among

components in the system rather than on component behavior.
Furthermore the newly adapted Engine Efciency Design Index
(EEDI), expected to enter into force on January 1, 2013 [2], may
impose further constraints on the layout of the engine, auxiliaries,
etc. Thus, a fast, yet thermodynamically realistic engine model
which can be integrated with an energy system analysis software is
highly desired for optimizing engine performance in combination
with waste heat recovery for minimal NOx as well as Green House
Gas (GHG) emissions.
The literature offers various modeling methodologies for the
internal combustion engine. Scope of the application, accuracy and
calculation time demand, are the determining parameters for the
modeling approach [3e8]. Zero-dimensional models [9e13] with
a few combustion zones are effective tools for providing reasonable
estimations of NOx emissions and/or engine performance with low
computational effort. Multi-zone combustion models [14e20]
seem to offer more accurate NOx predictions, a more realistic
modeling of the fuel spray, as well as provide the modeling of other
emissions such as soot. These advantages come with the cost of
increased computing time, possibly without providing additional
information essential for energy system analysis.
Analytical models for cycle simulation of four-stroke engines
and HCCI engines have been described with some validation

F. Scappin et al. / Applied Thermal Engineering 37 (2012) 344e352


Abbreviations and chemical formulas

Bottom Dead Center
Exhaust Valve Closing
Exhaust Valve Opening
Homogeneous Charge Compression Ignition
Specic fuel consumption
Start of injection
Variable turbine geometry
molar concentration
surface area (m2)
crank radius (m)
air-to-fuel ratio ()
adiabatic Flame Temperature (K)
bore (m)
rod length (m)
c1, c2
heat transfer coefcients
specic enthalpy (kJ/kg)
gas thermal conductivity (kW m1 K1)
correlation factors for scavenging efciency (e)
combustion shape factor (e)
mass (kg)
mass ow (kg/s)
a particular interval in which the combustion is
divided (e)
number of intervals in which the combustion is divided
absolute pressure (kPa)
heat (kJ)
universal gas constant (8.314 4621 kJ kmol1 K1)
compression ratio (e)
scavenging efciency on volumetric basis (e)
instantaneous piston speed (m/s)
scavenging ratio (e)
time (s)
temperature (K)

[24e26]. Zannis and Rakopoulos [27,28] developed a multi-zone

combustion model for predicting performance and emissions in
large-scale two-stroke engines with comparisons to experimental
data and Kowalski [29,30] proposed a simple method for simulating a two-stroke ship engine with some NOx emission validation.
Most recently, Payri et al. [31] published a zero-dimensional thermodynamic model of a four-stroke direct injection diesel engine
with similarities with the model described in the present paper,
although not including emission modeling.
While the development and methodologies found in these
papers offer clear descriptions of the models with some specic
validations to experimental data, few or no non-dimensional
models for electronically controlled marine two-stroke diesel
engines have been developed and validated using experimental
results for engine tuning adjustments (e.g. changed injection
timing, timing of exhaust valve closing and scavenging pressure;
see e.g. [32e35]). Being able to predict with a reasonable accuracy
the effects of engine tuning on performance and NOx emissions is of
particular importance for modern marine diesel engines.
The aim of this paper is to derive a methodology suitable for
energy system analysis and for predicting the performance and NOx




internal energy (kJ)

volume (m3)
instantaneous piston speed (m/s)
mechanical work (kJ)
residual gas fraction (e)
heat transfer scaling factor
air excess ratio
duration of combustion period (rad)
crank angle (rad)
gas kinematic viscosity (m2/s)
StefaneBoltzmann constant (5.670373$108 Wm2 K4)
specic heat ratio
angular velocity (rad/s)

heat release
reference state
charge (air fuel)
diffusive combustion
fuel; forward
inlet; initial
ignition delay
fuel injection
opening scavenge valve
pre-mixed combustion
reference, residual

emissions of low speed, two-stroke diesel engines and the effect of

engine tuning. Additionally the methodology is to be validated
against actual measurements on two engines.
The description of the diesel engine model is covered in section
2; the description of the experimental data used as a basis for
validating the model is outlined in section 3; the results and their
analysis are discussed in section 4. Lastly, in section 5, the conclusions are outlined.
2. Model description
The overall sequence of running the sub models corresponds to
the stages in a two-stroke engine cycle, i.e. it runs from compression to scavenging. Scavenging is modeled in order to calculate the
content of remaining exhaust gasses in the cylinder at the start of
compression, so the model has to be run typically three or four
cycles to converge. In this section the equations and methods used
in the model are described. Firstly the energy balance and heat loss
models, then the methodology for modeling the combustion event
and later the models for NOx formation, exhaust gas temperature
and scavenging are presented.


F. Scappin et al. / Applied Thermal Engineering 37 (2012) 344e352

2.1. Energy balance

The engine energy balance for a control volume consisting of the
cylinder gasses is solved over a complete crank cycle:

_ o ho
_ i hi  m



where U is the internal energy, W is the external work, QHR is the

fuel heat release and QW is the heat loss through the cylinder wall,
_ o ho are the inlet and outlet enthalpy ows,
_ i hi and m
while m
respectively. The rate of work rate is calculated as follows:



where P is the pressure inside the cylinder and the trapped volume
V is given as a function of the crank angle q:


 c 2
1 rc  1 1  cosq 
sin2 q


VC is the compression volume, c/a the piston rod length/crank

radius and rc the engine compression ratio. Burn fraction and heat
release rate in internal combustion engines are mostly governed by
functions based on the law of normal distribution of continuous
random variables. Best known of these functions is the Wiebe
function, [36,37]. While the Wiebe function by no means describes
the complex fuel air mixing in the diesel combustion process, it can
provide valuable thermodynamic input for the model in terms of
a realistic (shape of the) heat release. The variant used in the
present model is the double Wiebe function following the Miyamoto model [38] (see also [39]):

 q Mp
q Mp 1
1 dQHR
exp  6:9
Mp 1
u dt

M d

q Md 1
6:9 d Md 1
exp  6:9





Dqp and Dqd are the duration of the pre-mixed and diffusive
combustion periods, while Mp and Md are combustion shape
factors. Qp is the heat release due to the burning of the fuel during
the pre-mixed combustion, while Qd is the remaining heat release.
Following Miyamoto [38], Mp and Dqp are given the values 3 and
7, respectively, while half of the fuel injected during the ignition
delay is supposed to burn during the pre-mixed combustion. The
remaining parameters, Md and Dqd, are subject to model calibration
to match the maximum pressure, power output and SFC as discussed in section 4. The ignition delay tid is described by the
Arrhenius formula [40]:
tid 1:8  105



Here, AF is the air-to-fuel ratio in the cylinder and Tcyl is the

temperature in the cylinder when the ignition starts. Several
methodologies have been proposed on gas-to-wall heat transfer in
compression ignition engines. Among those is the widely used
Woschni correlation [41] which is implemented in the model, in all
the stages of the cycle:

Acyl as B0:2 Pt0:8 Tt0:55 vt0:8 *Tt  Tw


as is a scaling factor used for calibration to match a specic engine

geometry. The correlation is based on a simple Nusselt and Reynolds number correlation by using the cylinder bore as the

characteristic distance, and a velocity factor that is dependent on

pressure rise caused by combustion and mean piston speed. Thus,
the inuence of combustion on the heat transfer is included and the
heat transfer process is divided throughout the engine cycle into
the relevant stages [8]. Generally the correlation is appropriate if
experimental data are available for calibration of the engine specic
heat transfer characteristics [6]. The instantaneous characteristic
velocity v is dened as follows:

v c1 Sp c2

Vs Tr
P  Pmot
Pr Vr


where Pmot Pr(Vr/V)g, and g is the specic heat ratio. The

constants c1 and c2 need to be calibrated according to each specic
engine and are varying with each stage of the cycle. Typical values
for the gas exchange process are c1 6.18 and c2 0 [5] during the
compression stroke, c1 2.28 and c2 0 during combustion and in
the expansion stage c1 2.28 and c2 3.24 e-3. Vs is the swept
volume, Vr, Tr, and Pr are the volume, temperature and pressure
evaluated at any reference condition, in this study dened at the
time of exhaust valve closing (EVC) [5].
The cylinder scavenging entails a rapid outgoing mass ow
when the exhaust valve opens. The pressure dependency on the
crank angle q is set by the following equation, which describes
a sudden pressure drop, from the pressure at exhaust valve opening
(PEVO) to the outgoing exhaust pressure Po:

PEVO  Po expq qEVO  1

q  qEVO 1


The equation cannot be described as realistic and serves only the

purpose of avoiding problems with discontinuities in the modeled
engine cycle. The initial pressure for this equation is dened as
earlier described and the nal pressure is an input value for the
model (e.g. from measurements). This approach is suitable for
predicting the engine performance with reasonable accuracy in the
case of lacking knowledge about the exhaust valve geometry and
The gas is considered ideal, and the specic heat and enthalpy
are calculated as a function of the temperature and composition,
using correlations suggested by Gyftopoulos and Beretta [42].
2.2. Combustion and mixing
The combustion and NOx formation models are nested in a loop
with step size equal to the time duration of the entire combustion
event divided by a chosen number of steps. 30e90 steps have been
found to provide good accuracy and speed. Before the loop starts,
combustion parameters are calculated; total heat release, global AF,
ignition delay, Wiebe parameters and initial guess on AFT. In the
loop two zones are considered; a zone with combustion and a zone
containing the remaining cylinder gasses. The loop starts by solving
a system of ordinary differential equations by integration. This
system includes equations for heat losses by Woschni, energy and
mass balances, ideal gas law and heat release by Wiebe. The solution provides values of T, P, V, mass and HR over the time interval of
the present step. The mass of burnt fuel is now given and the mass
of oxygen involved is found by multiplying a local lambda factor
with the mass of burnt fuel divided by the concentration (at this
step) of oxygen present in the cylinder gasses. This way the mass of
gasses in the combustion zone is scaled to keep the oxygen
concentration constant. The local lambda is constant for the 360
engine cycle and is correlated with the load. This correlation is
subject to further investigation at the present time, but is showing
promise in the future work of validating the models part load

F. Scappin et al. / Applied Thermal Engineering 37 (2012) 344e352

The evaluation of the AFT requires knowledge of the combustion

products composition. Thus the combustion products are now
determined, based on an initial guess of AFT, using the 11 species
chemical equilibrium scheme described in Rakopoulos [43], except
for the formation of NO. Then the calculation of the exact AFT in the
burning zone is carried out [47] and the combustion products are
re-calculated using the calculated AFT. Next NOx formation is
modeled (see Section 2.3) and lastly the gasses from the two zones
are mixed to form the new composition of the cylinder gasses. This
is done by adding the components (by mass weighted) of the two
zones together and dividing by the total mass of gas.
2.3. NOx formation
The emission of nitrogen oxides is mainly in the form of NO and
reactions that involve NO2 are negligible concerning NOx formation in
low speed diesel engines [44]. Four contributions determine the total
NO emission due to combustion; thermal NO, prompt NO, fuel NO and
NO arising from the N2O mechanism. Previous studies conrm the
relevance of the thermal mechanism in the NO formation, considering the contribution of the prompt mechanism negligible [21,45].
The N2O mechanism becomes important in fuel-lean pre-mixed
combustion, which pertains more to gas turbines [46] and smaller
four-stroke engines with larger pre-mixed burn fraction [47]. Fuel NO
contribution becomes relevant in the case the engine is operating
with heavy fuel oil [21]. In fuel-lean combustion, all the fuel nitrogen
is converted to NO [48] and the fuel NO contribution to the total NO
emission depends only on the fuel and not on the engine operation. A
5e10% contribution from fuel-bound nitrogen to the total NOx
emission may be expected in the case of heavy fuel oil [49].
In the present model only the thermal mechanism is considered,
since no simple and applicable model is available and the effect is
isolated to the type of fuel used, not the engine process and the
interplay with auxiliaries. The extended Zeldovich mechanism,
extensively described [21e23] and apparently an effective instrument, which couples accuracy and low computing resources for the
calculation of thermal NOx, is implemented in the model:







The forward and backward reaction rate constants that dene

reactions K1eK3 (Eqs. (9e11)) depend only on the temperature at
which the reactions are occurring, here assumed to be the adiabatic
ame temperature (AFT) of the burning zone. With the temperature T in Kelvin, the reaction rate constants, in cm3/(gmol s), used in
reactions K1eK3 are [16]:

k1f 7$1013 exp37800=T

k1b 1:55$10



k2f 1:33$1010 $T$exp3600=T


k2b 3:2$109 $T$exp19700=T


k3f 7:1$1013 exp450=T


k3b 1:7$1014 exp24560=T



The NO formation rate in the interval (in the combustion calculation) is calculated, as a function of composition and temperature, by
integrating the following equation [5]:

K1f N2 OK1b NONK2f O2 NK22b NOO


where all the concentrations (denoted [ ]), with the exception of

NO, are obtained from the chemical equilibrium calculation in the
burning zone. The numerical method used is described in ref. [5]
As described in Section 2.2, the combustion is divided into
intervals and complete mixing is supposed at the end of each nth
interval between the burning and the non-reacting zone. The
hypothesis of obtaining a complete mixing of the two zones at the
end of each interval is not fully realistic, and neither would be the
assumption of a zero rate mixing modeleconditions that represents
extremes in the mixing process in the combustion chamber [50].
Moreover, it is expected that the innite-rate mixing model would
cause under-prediction of the NO formation, because of the lower
oxygen concentration in the reactants and of the lower temperature of the products, after they have been mixed with the nonreacted air. Conversely, a zero rate mixing model is expected to
over-predict the NO emissions.
2.4. Exhaust gas temperature
The exhaust gas temperature is calculated by an overall energy
balance of the engine over a complete cycle:


_  Q_
_ intake air CPairi Ti m
_ f hf m
_ f LHV  W
mexh cpexh


LHV is the lower heating value of the fuel. Qw is the total of heat
losses found via the Woschni correlation in each of the stages of the
2.5. Scavenging efciency and determination of initial temperature
at EVC
The initial temperature at EVC is inuenced by the amount of
trapped hot residual gasses and the heat transfer from hot surfaces
to the gas. By evaluating the amount of trapped residual gasses after
each cycle, determined by residual gas fraction Xr, the initial
temperature can be evaluated using a simple initial temperature
model [25,51].
The scavenging efciency is used to specify the connection
between the trapped residual gas and the fresh air at the start of the
compression stroke [52]. The relationship between the volumetric
scavenge efciency, SEV, and the residual gas fraction is dened as:

Xr 1  SEV


The scavenge efciency on volumetric basis is evaluated using

a numerical t to acquire the correlation between theory and
experimental data conducted for different scavenging characteristics in two-stroke engines, Blair [52]:

SEV 1  exp k0 k1 SRV k2 SR2V


where SRV is the scavenge ratio, or the ratio between the delivered
fresh air and the swept volume of the combustion chamber. The
correlation factors k0, k1 and k2 are scavenging coefcients for
ported uniow scavenging in two-stroke engines, from Blair [52].
Once the scavenging ratio and the scavenging efciency are known
measures, the modied temperature at the time of EVC (Tmix) can

Fig. 1. Response surface of NOx as a function of EVC and SOI. All other independent
variables are kept at their center levels.

be determined by mixing of the fresh air charge with trapped

residual gas from the previous cycle:

*Ti i Xr *
*Texh i  1
Tmix i 1  Xr *


where i is the cycle number index and subscripts r, ch and t

represents the residual gas, the charge (air fuel), and the total incylinder gas respectively at EVC [25]. Assuming identical specic
heats for the intake air and exhaust gases, the expression for initial
temperature is simplied [5]:

Tmix i 1  Xr *Ti i Xr *Texh i  1


Hence, the trapped cylinder mass for the ith cycle at EVC is
determined using the ideal gas law:

mc i

RTmix i


3. Experimental and response surface model validation data

In order to validate the developed engine model, experimental
ndings are used from Goldsworthy [21,58] and Mayer [54] based
on measurements on two direct injection, turbocharged, uniow
scavenged, low speed two-stroke marine MAN B&W diesel engines,
models 4T50 ME-X and 7L70 MC.
The data from the 4T50 ME-X engine is comprehensive and
consists of 50 data points, and it contains simultaneous variations
in several pivotal parameters severely affecting both engine
performance and emissions. The parameters are: Exhaust valve
closing timing (EVC), start of injection (SOI), injection pressure
(Pinj), injection prole (InjProf), injection nozzle hole size (Nozz),
and turbine area (VTG), controlling mainly (but not only)
compression pressure ratio to the scavenging pressure, pressure
rise, injection length and intensity, as well as scavenging pressure.
The number of tests has been kept at a minimum by employing
principles from the theory of design of experiments [56] with the
test plan laid out according to a central composite face centered
design (response surface methodology), which contains three
levels for all six variables (factors), with several repetitions of the
center point. Since many parameters are varied at a time, the
experimental data are used to provide response surface models [57]
of responses SFC, Pmax, Pcomp, Pscav, and NOx used for validating the

Fig. 2. aed. Predicted versus measured responses. Pcomp has been left out, since it is
similar to Pmax in terms of R2.

F. Scappin et al. / Applied Thermal Engineering 37 (2012) 344e352


Table 1
Engine operative conditions.

Fig. 3. Heat release versus crank angle at 75% engine load.

present model. The response surface is illustrated for NOx in Fig. 1.

The agreement with the experimental ndings is excellent and the
model error is minimal as visualized in Fig. 2aed.
The sensitivity of NOx and SFC on compression and maximum
pressures has been calculated effectively by changing SOI but
simultaneously allowing EVC to adjust in order to have unchanged
compression pressure for varying maximum pressure, and to have
constant maximum pressure for varying compression pressure.
Effectively scavenge pressure is changed by adjusting VTG, but
again simultaneously allowing EVC and SOI to adjust in order to
have unchanged maximum and compression pressure. The sensitivity of NOx and SFC on pressure levels hereby provided is used for
validating the present model. These sensitivities are referred to as
(surrogate) experimental, since the response surface models used
are close to a 1:1 match of the measured data.
The experimental data for the 7L70 MC engine is obtained from
a paper by Goldsworthy [58].

Engine model

4T50 ME-X

7L70 MC

Bore [mm]
Stroke [mm]
Compression volume [m3]
Compression ratio
Fuel LHV [kJ/kg]
Specic humidity of the charge air (ISO)
Speed [Revs/min]
Maximum pressure [MPa]
Compression pressure [MPa]
Scavenge pressure [MPa]
Fuel per cylinder [kg/s]
Air per cylinder [kg/s]
Scavenge temperature [ C]
Cooling load [kW]
Injection angle [degrees after BDC]

42 700

12.6 (effective)
42 700

the heat release is considered acceptable. The deviation at the peak

of heat release rate shows that the model seems to predict more
energy released during early stages of the combustion as seen in
Fig. 3. However, when looking at the cumulative heat release (burn
fraction) in Fig. 4, there is good agreement with the timing of the
energy release.
In order to validate further the engine model, a simulation
model of the 7L70 MC mk6 MAN B&W engine is built according to
the data provided in two articles by Goldsworthy [21,58]; see
Table 1. The articles provide a large part of the model input
parameters and results, and this data is thus suitable for validation
purposes because it puts constraints on the model, i.e. there are
only a few parameters available for calibration adjustments to help
match the measurements. Table 2 shows how the measured values
and the model outputs are in good agreement, i.e. within 5% of
measured data values.

4. Results and discussions

Firstly the model validation results using 4T50 ME-X engine
data are shown and secondly validation results using the 7L70 MC
engine data.
4.1. General validation by 4T50 ME-X and L70MC experimental
Data obtained from MAN Diesel & Turbo providing the
measured heat release rate is used to calibrate the injection timing
and the combustion shape parameters, see Figs. 3 and 4. The
agreement between the model results and the measured values for

Fig. 4. Burn fraction versus crank angle at 75% engine load.

Table 2
The 7L70 MC engine model results compared to experimental data.

Maximum pressure (MPa abs)

Compression pressure (MPa abs)
Specic fuel consumption (g/kWh)
Power output per cylinder (MW)
NOx emissions (g/kWh)


Engine model





Fig. 5. Normalized SFC versus maximum pressure by varying injection timing.


F. Scappin et al. / Applied Thermal Engineering 37 (2012) 344e352

Fig. 6. NOx emissions versus maximum pressure by varying injection timing.

Since the model is using the ideal gas equation of state (EOS),
somewhat lower calculated pressures, compared to the measurements, are to be expected. Investigating a case of adiabatic
compression using the ideal gas and the real gas EOS developed by
Lemmon et al. [59], respectively, resulted in about 5% lower
compression pressure, when assuming ideal gas at the relevant
conditions as compared to the real gas EOS. This explains the
deviation in compression pressure as well as maximum pressure
shown in Table 2. As also seen from the table, the modeled NOx
emission and SFC are in excellent agreement with the experimental
As mentioned only a few parameters are available for calibration
since the measured parameter values provided are extensive.
Provided parameters relevant for the presented methodology are:
Engine geometry, effective compression ratio, engine speed, intake
air composition and fuel mass ows, intake air temperature and
pressure, pressure loss due to scavenging and start of injection.
Parameters available for calibration are: Timing of end of injection,
timing of opening of the exhaust valve, and heat release function
parameters in the diffusive combustion phase (Eq. (4)). In the
following sections the model is validated against parameter variations from 4T50 ME-X.

Fig. 8. NOx emissions versus compression pressure by varying EVC timing keeping
maximum pressure constant.

its effect on the maximum cylinder pressure and AFT. Advancing

the injection timing increases the pre-mixed combustion temperature which results in higher NOx formation [55]. Here, the injection timing is used as a control parameter to vary the maximum
pressure and the effect on SFC and NOx emissions are evaluated. As
seen in Figs. 5 and 6 results suggest that the model is able to predict
the effect of maximum pressure on SFC and NOx emissions all
within the 95% condence intervals (/2s) of the measured data,
indicated by error bars to the mean value in the result in Figs. 5e10
4.3. Effect of compression pressure on SFC and NOx emissions

The injection timing is an important factor in the combustion

process, and SFC and NOx emissions are strongly inuenced due to

Generally, when increasing the compression pressure, more

work is needed for the compression stroke, resulting in higher SFC
and specic NOx emissions [14]. Increasing the compression pressure results in higher maximum pressure which could exceed the
structural limits of the engine [45]; thus when calibrating the
compression pressure by the timing of EVC, the maximum pressure
is kept constant at a reference structural limit of 160 bar, by
adjusting the injection timing accordingly. Figs. 7 and 8 illustrate
how SFC and NOx emissions both increase with increasing
compression pressure. A comparison between model results and
measured data suggests that the model is also able to predict the
SFC and NOx within the 95% condence although the model seems
to over-predict NOx emissions at relatively low compression

Fig. 7. Normalized SFC versus compression pressure by varying EVC timing keeping
maximum pressure constant.

Fig. 9. Normalized SFC versus scavenge air pressure while keeping maximum pressure
and compression pressure constant.

4.2. Effect of maximum pressure on SFC and NOx emissions

F. Scappin et al. / Applied Thermal Engineering 37 (2012) 344e352


due to the use of the ideal gas law, in the range of less than 5%
deviation. It is recommended though that future work should
include implementation of more accurate equations of state in the
methodology and faster optimization methods.
In conclusion, the model derived here features the desired
rapidity of execution and provides a sufcient accuracy, with
regards to essential engine performance characteristics, for being
used for future energy system analysis. The model has demonstrated capability of responding well to tuning engine parameters
regarding SFC and NOx emissions, thus enabling the possibility to
use the model to explore different scenarios, e.g. SFC minimization
within the IMO NOx emission limits either as a stand-alone engine
or combined with a waste heat recovery system.
Fig. 10. NOx emissions versus scavenge air pressure keeping maximum pressure and
compression pressure constant.

pressures. This could be explained by the fact that no detailed

information about the scavenging process is involved in the model.
Therefore, there might be some differences in the actual cylinder
properties at the time of EVC compared to calculated properties.
Another factor possibly causing the deviations is the use of the ideal
gas law which underestimates the actual pressure at the end of the
compression stroke. This might be improved by implementing real
gas behavior.
4.4. Effect of scavenging pressure on SFC and NOx emissions
Figs. 9 and 10 illustrate how increasing scavenge pressures
result in low specic NOx emissions and SFC [7,54]. Here, the
compression pressure and the ring ratio (Pmax/Pcomp) were kept
constant by adjusting the timing of EVC and the injection timing.
The results suggest that the model also predicts this effect within
the 95% condence interval limits of the measured data.
5. Conclusions
A zero-dimensional model for electronically controlled twostroke low speed diesel engines was derived. A two zone
combustion model, an energy balance and the ideal gas law
equations were applied and coupled with a combustion and NOx
formation calculation subroutine using the extended Zeldovich
mechanism. Initial properties of the trapped gas in the cylinder at
the start of compression were determined from the mixture of fresh
air introduced in the cylinder and trapped residual gasses evaluated
from the scavenging efciency and residual gas fraction. The
modeled heat release was calibrated to match measured data, by
selecting the correct timing of fuel injection and adjusting
combustion shape parameters.
The effects of engine tuning parameters on SFC and NOx emission
at 75% engine load were validated by data from experimental tests
on the MAN B&W 4T50 ME-X test engine. The results indicate that
the trends on SFC and NOx emissions of varied maximum pressure,
attained by changed injection timing, can be captured within the
95% condence interval of the measurements. Also the effects on SFC
and NOx emissions of adjusting the compression pressure, by
changing the timing of EVC, and of scavenging pressure, respectively, were predicted within limits of the measure values.
The model was also validated using data from the MAN B&W
7L70 engine comprising both engine operating parameters and
measured engine outputs. There was good agreement between
model and measurements, although modeled pressures were low

The authors would like to thank Brian Elmegaard, Michael Vincent Jensen and Kim Rene Hansen from Technical University of
Denmark, Department of Mechanical Engineering for the advice and
discussions during the work. A special acknowledgement is given to
Prof. Spencer C. Sorenson from Technical University of Denmark,
Department of Mechanical Engineering who provided many helpful
suggestions during the model development and for the manuscript.
[1] Revised MARPOL Annex VI, Regulations for the Prevention of Air Pollution
from Ships and NOx Technical Code 2008, International Maritime Organization, London, UK, 2009.
[2] Mandatory energy efciency measures for international shipping adopted at
IMO environment meeting, Marine Environment Protection Committee
(MEPC) e 62nd session: 11 to 15 July 2011.
[3] E. Pedersen, H. Valland, H. Engja, Modelling and simulation of Diesel engine
processes, Paper 34, CIMAC Congress, Interlaken, Switzerland, 1995.
[4] H. Grimmelius, E. Mesbahi, P. Schulten, D. Stapersma, The use of diesel engine
simulation models in ship propulsion plant design and operation, Paper 227,
CIMAC conference, Wien, Austria, 21e24 May 2007.
[5] S.C. Sorenson, Engine Principles and Vehicles, Technical University of Denmark,
Lyngby, Denmark, 2008.
[6] J.B. Heywood, E. Sher, The Two-Stroke Cycle Engine, Taylor & Francis, London,
UK, 1999.
[7] J.B. Heywood, Internal Combustion Engine Fundamentals, McGraw-Hill, New
York, USA, 1988.
[8] S.H. Chan, J. Zhu, Modelling of engine in-cylinder thermodynamics under high
values of ignition retard, International Journal of Thermal Sciences 40 (2001)
[9] G. P. Merker, B. Hohlbaum, M. Rauscher, Two-zone model for calculation of
Nitrogen-Oxide formation in direct-injection diesel engines, SAE paper
932454, 1993.
[10] R. J. Asay, K. I. Svensson, D. R. Tree, An Empirical, Mixing-Limited, ZeroDimensional Model for Diesel Combustion, SAE paper 2004-01-0924, 2004.
[11] D.T. Hountalas, Prediction of marine diesel engine performance under fault
conditions, Applied Thermal Engineering 20 (2000) 1753e1783.
[12] D. Descieux, M. Feidt, One zone thermodynamic model simulation of an ignition
compression engine, Applied Thermal Engineering 27 (2007) 1457e1466.
[13] X. Tauzia, A. Maiboom, P. Chesse, N. Thouvenel, A new phenomenological heat
release model for thermodynamical simulation of modern turbocharged
heavy duty, Diesel Engines 26 (2006) 1851e1857.
[14] G. Stiesh, G.P. Merker, A Phenomenological Heat Release Model for Direct
Injection Diesel Engines, 22nd CIMAC Congress, Copenhagen, 1998.
[15] A. Maiboom, X. Tauzia, S.R. Shah, J.F. Hetet, New phenomenological six-zone
combustion model for direct injection diesel engines, Energy Fuels 23
(2009) 690e703.
[16] O. Durgun, Z. Sahin, Theoretical investigation on heat balance in DI diesel
engines for neat diesel fuel and gasoline fumigation, Energy Convers. Manage.
50 (2009) 43e51.
[17] A.M. Kulkarni, G.M. Shaver, S.S. Popuri, T.R. Frazier, D.W. Stanton, Computationally efcient whole-engine model of a Cummins 2007 turbocharged diesel
engine, J. Eng. Gas Turbines Power 132 (2010) 0228031.
[18] M. Andersson, B. Johansson, A. Hultquist, C. Nhre, A Real Time NOx Model for
Conventional and Partially Premixed Diesel Combustion, SAE paper 2006-010195, 2006.
[19] D. Jumg, D. A. Assanis, Multi-Zone DI Diesel Spray Combustion Model for Cycle
Simulation Studies of Engine Performance and Emissions, SAE paper 2001-011246, 2001.


F. Scappin et al. / Applied Thermal Engineering 37 (2012) 344e352

[20] I. Arsie, F. Di Genova, A. Mogavero, C. Pianese, G. Rizzo, A. Caraceni, P. Ciof, G.

Flauti, Multi-Zone Predictive Modeling of Common Rail Multi-Injection Diesel
Engines, SAE paper 2006-01-1384, 2006.
[21] L. Goldsworthy, Reduced kinetic schemes for oxides of nitrogen emissions
from a slow speed marine diesel engine, Energy Fuels 17 (2003) 450e456.
[22] T. Chikahisa, M. Konno, T. Murayama, T. Kumagai, Analysis of NO formation
characteristic and its control concepts in diesel engines from NO reaction
kinetics, JSAE Review 15 (1994) 297e303.
[23] K. Kikuta, T. Chikahisa, Y. Hishinuma, Study on predicting combustion and
NOx formation in diesel engines from scale model experiments, Transactions
of the Japan Society of Mechanical Engineers 43 (2000) 89e96.
[24] L. Eriksson, I. Andersson, An analytical Model for Cylinder Pressure in a Four
Stroke SI engine, SAE paper2002-01-0371, 2002.
[25] M. Shahbakhti, C.R. Koch, Physics based control oriented model for HCCI
combustion timing, Journal of Dynamic Systems, Measurement and Control
132 (2010) 021010.
[26] D.J. Rausen, A.G. Stefanopoulou, A mean-value model for control of homogeneous charge compression ignition (HCCI) engines, Journal of Dynamic
Systems, Measurement, and Control 17 (2005) 355e362.
[27] T. C. Zannis, V. T. Lamaris, D. T. Hountalas, S. E. Glaros, Development and
validation of a multi-zone combustion model for predicting performance
characteristics and NOx emission in large scale two-stroke diesel engines,
ASME, 2009e11382, 2009.
[28] C.D. Rakopoulos, K.A. Antonopoulos, D.C. Rakopoulos, Development and
application of multi-zone model for combustion and pollutants formation in
direct injection diesel engine running with vegetable oil or its bio-diesel,
Energy Conversion and Management 48 (2007) 1881e1901.
[29] J. Kowalski, W. Tarelko, NOx emission from a two-stroke ship engine. Part 1:
modeling aspect, Applied Thermal Engineering 29 (2009) 2153e2159.
[30] J. Kowalski, W. Tarelko, NOx emission from a two-stroke ship engine. Part 2:
laboratory test, Applied Thermal Engineering 29 (2009) 2160e2165.
[31] F. Payri, P. Olmeda, J. Martn, A. Garca, A complete 0D thermodynamic
predictive model for direct injection diesel engines, Applied Energy 88 (2011)
4632e4641. doi:10.1016/j.physletb.2003.10.071.
lu, A. Kolip, The effects of injection timing on
[32] A. Parlak, H. Yasar, C. Hasimog
NOx emissions of a low heat rejection indirect diesel injection engine, Applied
Thermal Engineering 25 (2005) 3042e3052.
[33] Cenk Sayin, Metin Gumus, Impact of compression ratio and injection
parameters on the performance and emissions of a DI diesel engine fueled
with biodiesel-blended diesel fuel, Applied Thermal Engineering 31 (16)
(November 2011) 3182e3188.
[34] J. Benajes, S. Molina, J. Martn, R. Novella, Effect of advancing the closing angle
of the intake valves on diffusion-controlled combustion in a HD diesel engine,
Applied Thermal Engineering 29 (2009) 1947e1954.
[35] J.M. Desantes, J. Benajes, S. Molina, C.A. Gonzlez, The modication of the fuel
injection rate in heavy-duty diesel engines. Part 1: effects on engine performance and missions, Applied Thermal Engineering 24 (2004) 2701e2714.
[36] J.I. Ghojel, Review of the development and applications of the Wiebe function:
a tribute to the contribution of Ivan Wiebe to engine research, International
Journal of Engine Research 11 (2010) 297e312.
[37] J. Galindo, H. Climent, B. Pl, V.D. Jimnez, Correlations for Wiebe function
parameters for combustion simulation in two-stroke small engines, Applied
Thermal Engineering 31 (2011) 1190e1199.

[38] N. Miyamoto, T. Chikahisa, T. Murayama, R. Sawyer, Description and Analysis

of a Diesel Engine Rate of Combustion and Performance Using Wiebes
functions, SAE Paper 850107, SAE Trans. vol. 94, 1985.
[39] H. Yasar, H.S. Soyhan, H. Walmsley, B. Head, C. Sorusbay, Double-Wiebe
function: an approach for single-zone HCCI engine modelling, Applied
Thermal Engineering 28 (2008) 1284e1290.
[40] G.P. Merker, C. Shwarz, G. Stiesh, F. Otto, Simulating Combustion, SpringerVerlag, Berlin Heidelberg, Germany, 2006, p. 108.
[41] G. Woschni, A Universally Applicable Eqution for the Instantaneous Heat
Transfer Coefcient in the Internal Combustion Engine, SAE Paper 670931,
[42] E.P. Gyftopoulos, G.P. Beretta, Thermodynamics: Foundations and Applications, Macmillan, New York, 1991.
[43] C.D. Rakopoulos, D.T. Hountalas, E.I. Tzanos, G.N. Taklis, A fast algorithm for
calculating the composition of diesel combustion products using 11 species
chemical equilibrium scheme, Adv. Eng. Software 19 (1994) 109e119.
[44] E.C. Zabetta, P. Kilpinen, Improved NOx submodels for in-cylinder CFD simulation of low e and medium speed compression ignition engines, Energy Fuels
15 (2001) 1425e1433.
[45] J. E. Dec, R. E. Canaan, PLIF imaging of NO formation in a DI diesel engine, SAE
Technical Paper 980147, 1998.
[46] S.R. Turns, An Introduction to Combustion, second ed. McGraw-Hill, Singapore, 2000.
[47] S.M. Correa, A review of NOx formation under gas-turbine combustion
conditions, Combust. Sci. Technol 87 (1993) 329e362.
[48] Z. Bazari, DI diesel combustion and emission predictive capability for use in
cycle simulation, SAE Technical Paper 920462, 1992.
[49] J.W. Miller, H. Agrawal, W.A. Welch, Criteria Emissions from the Main Propulsion Engine of a Post-Panamax Class Container Vessel Using Distillate and
Residual Fuels (CARB Report), University of California, USA, 2009.
[50] R. C. Peterson, K. J. Wu, The Effect of Operating Conditions on Flame
Temperature in a Diesel Engine, SAE Technical Paper 861565, 1986.
[51] L. Eriksson, I. Andersson, An Analytical Model for Cylinder Pressure in a Four
Stroke SI Engine, SAE Technical Paper 2002-01-0371, 2002.
[52] G.P. Blair, Design and Simulation of Two-Stroke Engines, Society of Automotive Engineers, Inc., Warrendale, USA, 1996.
[53] M. Finch Pedersen, A. Andreasen, S. Mayer, Two Stroke Engine Emission
Reduction Technology: State-of-the-Art, Paper 85, CIMAC Congress, Bergen,
Norway, 14e17 June 2010.
[54] S. Mayer, T. A. E. Tuner, A. Andreasen, Design of experiments analysis of the
NOx-SFOC trade-off in two-stroke marine engine, Paper 38, CIMAC Congress,
Bergen, Norway, 2010.
[55] A. Sarvi, R. Zevenhoven, Large-scale diesel engine emission control parameters, Energy 35 (2010) 1139e1145.
[56] G.E.P. Box, W.G. Hunter, J.S. Hunter, Statistics for Experimenters, John Wiley &
Sons, Hoboken, New Jersey, USA, 1978.
[57] R.H. Myers, D.C. Montgomery, C.M. Anderson-Cook, Response Surface Methodology, third ed. John Wiley & Sons, Hoboken, New Jersey, USA, 2009.
[58] L. Goldsworthy, Real time model for oxides of nitrogen emissions from a slow
speed marine diesel, Journal of Marine Engineering and Technology (2002) No. A2.
[59] E.W. Lemmon, R.T. Jacobsen, S.G. Penoncello, Daniel Friend, Thermodynamic
properties of air and mixtures of nitrogen, argon, and oxygen from 60 to 2000
K at pressures to 2000 MPa, J. Phys. Chem. Ref. Data 29 (2000) 331e385.