You are on page 1of 15

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 1 8 5 1 7 e1 8 5 3 1

Available online at www.sciencedirect.com

ScienceDirect

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

On the capabilities and limitations of predictive,


multi-zone combustion models for hydrogen-
diesel dual fuel operation

Stefan Tüchler a,*, Pavlos Dimitriou b


a
Powertrain and Vehicle Research Centre (PVRC), Department of Mechanical Engineering, University of Bath,
Claverton Down, Bath, BA2 7AY, United Kingdom
b
Renewable Energy Research Center, National Institute of Advanced Industrial Science and Technology (AIST), 2-2-
9, Machiikedai, Koriyama, Fukushima, 963-0298, Japan

article info abstract

Article history: Compared with traditional hydrocarbon fuels, hydrogen provides a high-energy content
Received 19 February 2019 and carbon-free source of energy rendering it an attractive option for internal combustion
Received in revised form engines. Co-combusting hydrogen with other fuels offers significant advantages with
20 May 2019 respect to thermal efficiency and carbon emissions.
Accepted 21 May 2019 This study seeks to investigate the potential and limitations of multi-zone combustion
Available online 13 June 2019 models implemented in the GT-Power software package to predict dual fuel operation of a
hydrogen-diesel common rail compression ignition engine. Numerical results for in-
Keywords: cylinder pressure and heat release rate were compared with experimental data. A single
Hydrogen-diesel cylinder dual-fuel model was used with hydrogen being injected upstream of the intake
GT-Power manifold. During the simulations low (20 kW), medium (40 kW) and high (60 kW) load
Simulation conditions were tested with and without exhaust gas recirculation (EGR) and at a constant
Dual-fuel combustion engine speed of 1500 rpm. Both single and double diesel injection strategies were examined
Internal combustion engine with hydrogen energy share ratio being varied from 0 to 57% and 0e42 respectively. This
corresponds to a range in hydrogen air-equivalence ratios of approximately 0e0.29.
The results show that for the single-injection strategy, the model captures in-cylinder
pressure and heat release rate with good accuracy across the entire load and hydrogen
share ratio range. However, it appears that for high hydrogen content in the charge
mixture and equivalence ratios beyond the lean flammability limit, the model struggles to
accurately predict hydrogen entrainment leading to underestimated peak cylinder pres-
sures and heat release rates. For double-injection cases the model shows good agreement
for hydrogen share ratios up to 26%. However, for higher energy share ratios the issue of
erroneous hydrogen entrainment into the spray becomes more accentuated leading to
significant under-prediction of heat release rate and in-cylinder pressure.
© 2019 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.

* Corresponding author.
E-mail address: S.Tuechler@bath.ac.uk (S. Tüchler).
https://doi.org/10.1016/j.ijhydene.2019.05.172
0360-3199/© 2019 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.
18518 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 1 8 5 1 7 e1 8 5 3 1

improve thermal efficiency and reduce greenhouse gas


Nomenclature emissions of internal combustion engines. One of these
measures is to co-combust diesel fuel with gaseous fuels, such
m_ Mass flow rate in kg/s
as hydrogen.
f Mass fraction
Hydrogen is an abundant element and can be generated
xb Mole fraction
from multiple sources, ranging from natural gas reforming,
C Multiplier
gasification using biomass or coal and water electrolysis using
R Energy share Ratio
(renewable) energy. Although these measures are cost and
SL Laminar flame speed in m/s
energy intensive, hydrogen as an energy carrier is of particular
T Temperature in K
interest for sustainable power generation in internal com-
p Pressure in bar
bustion engines. It provides high energy content and is free of
F Equivalence Ratio
carbon atoms, making it attractive for reducing CO2 , CO and
Jv EGR rate
particular matter (PM) emissions. Table 1 provides a compar-
hv Volumetric efficiency
ison of the most salient features of hydrogen and diesel fuel.
AFR Air-to-fuel ratio
Compared with diesel, hydrogen provides almost thrice the
ATDC After top dead centre
energy per unit mass. However, this advantage is offset by its
CFD Computational fluid dynamics
substantially lower density compromising hydrogen storage
DoE Design of experiments
in vehicles. Hydrogen can be run over a wide range of air-to-
EGR Exhaust gas recirculation
fuel ratios (AFR) making it particularly suitable for ICE opera-
EXH Exhaust gas
tion. In conjunction with its low energy required for ignition, it
FKG Flame kernel growth
is thus possible to run on extremely lean mixtures, albeit at
HDDF Hydrogen-diesel dual-fuel
the cost of higher susceptibility to knocking. High flame speed
HRR Heat release rate
and diffusivity further ensures rapid and complete combus-
ICE Internal combustion engine
tion with the exception of low load operation where high rates
IVC Intake valve closing
of unburned hydrogen and low combustion efficiency are
LHV Lower heating value
observed [3]. Despite knock-on effects with respect to volu-
MRE Mean relative error
metric efficiency, hydrogen addition offers potential benefits
MUZ Main unburned zone
in terms of soot, CO and CO2 emissions [4], while increased
PM Particulate matter
NOx emissions can be partially compensated through exhaust
RMSE Root mean squared error
gas recirculation (EGR) [5]. Further improvements in harmful
SBZ Spray burned zone
emissions can be achieved through the application of bio-
SOC Start of combustion
diesel [6]. Nonetheless, implementation of hydrogen as an
SUZ Spray unburned zone
engine fuel is complex in compression ignition (CI) engines
TFS Turbulent flame speed
due to the high autoignition temperature and low cetane
TLS Taylor length scale
number, which requires combustion to be initiated through
TPA Three pressure analysis
the lower auto-ignition diesel fuel.
VGT Variable geometry turbine
There has been a significant amount of research efforts
df Diffusion
dedicated to the examination of hydrogen as a fuel for ICE
entr Entrainment
ranging from CI engines [3e9] to spark-ignition (SI) engines
ign Ignition
[10e13]. While the open literature features a wide range of
pm Premixed
experimental studies on co-combustion engines of hydrogen
ref Reference value
and diesel or gasoline, there are relatively few numerical
s Static
studies available. The bulk of these studies deal with three-
stoich Stoichiometric
dimensional computational fluid dynamics (CFD) simula-
t Total
tions on the in-cylinder combustion characteristics, chemical-
u Unburned
kinetics of hydrogen combustion and effects on pollutant gas
emissions [14e19]. One-dimensional models having

Introduction
Table 1 e Overview of the main properties of diesel and
Over recent decades automotive engines have come under hydrogen. Data taken from [1,2].
close scrutiny with respect to their emission levels and carbon Property Unit Diesel Hydrogen
footprint. In combination with increased consumer aware-
Carbon content (mass fraction) [%] 86 0
ness for fuel economy and sustainability, this has led to an
Density at 1 bar and 300 K [kg/m3 ] 832 0.0838
ever increasing strive for powertrain electrification and bat- Lower heating value (LHV) [MJ/kg] 43 120
tery or fuel cell electric vehicles. However, electric vehicles Autoignition temperature [K] 473 858
represent long-term solutions that require major expendi- Min. ignition energy [mJ] e 0.02
tures for charging infrastructure. Hence, it will be decades Flame velocity (F¼1) [cm/s] 30 250e350
until the internal combustion engine (ICE) will be finally Flammability limits e 0.2e7
Stoichiometric air/fuel ratio 14.5 34
phased out. Nonetheless, there are a number of options to
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 1 8 5 1 7 e1 8 5 3 1 18519

established themselves as invaluable tools for predicting en- water on a diesel engine performance. The study considered a
gine operation and pollutant gas emissions on both SI and CI hydrogen share ratio of 19% and varied engine speed as well as
engines [20,21], have not been extensively tested on hydrogen- injection timing examining the impact on emissions. The
diesel dual fuel engine operations. To the authors' knowledge study of Monemian [23] dealt with a hydrogen-diesel dual fuel
there are merely two studies that address this to present (HDDF) engine run at constant speed and considering
[22,23]. Ghazal [22] used the predictive dual-fuel model hydrogen share ratios of up to 30%. The primary objective of
implemented in the one-dimensional software GT-Power to the study focused on the calibration process of the multi-zone
investigate the effect of both port-fuel injected hydrogen and combustion model.

Fig. 1 e Overview of experimental setup including sensor locations and data acquisition system [5].
18520 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 1 8 5 1 7 e1 8 5 3 1

Both mentioned publications consider relatively low The experimental scope involves three different load con-
hydrogen share ratios and do not compare the model's pre- ditions, namely 20 kW, 40 kW and 60 kW and energy share
dictive capabilities over a wider range of hydrogen share ratios ratios as defined in Equation (1) ranging from 0 to 57%. This
and equivalence ratios. Therefore, this study seeks to address corresponds to a maximum equivalence ratio of approxi-
this gap, by testing the multi-zone predictive dual-fuel com- mately 0.29. Furthermore, EGR rate was varied from approxi-
bustion model implemented in GT-Power over an extensive mately 0 to 30%. Finally, diesel injection strategy was also
range of hydrogen share ratios taking into account the effect altered from a simple single-injection strategy to a pilot- and
of variations in load, EGR rates and diesel injection strategies. main-injection strategy. In both cases, the injection timing
To the author's knowledge this is the first time such a char- was calibrated by the engine manufacturer. Table 3 provides
acterisation is reported in the open literature. an overview of main operating conditions considered
throughout this study.

GT-power
Methodology
Simulations of HDDF operation were done using a single-
This section gives a brief introduction of the engine specifi-
cylinder model, as illustrated in Fig. 2a, within GT-Power
cations used throughout this study as well as the experi-
v2019 by Gamma Technologies, which has established it-
mental setup and scope of testing conditions are given. After
self as a reliable and accurate simulation tool for combus-
that, the GT-power model used to predict heat release rate and
tion modelling [21]. The combustion model used throughout
in-cylinder pressure is introduced followed by the approach
this study is the predictive dual fuel combustion model that
taken for the regression model that computes laminar flame
employs the DIPulse combustion model for Diesel com-
speed.
bustion and SITurb for the premixed combustion of
hydrogen. The DIPulse model on the one hand follows the
Experimental setup
injected fuel, as it evaporates, entrains surrounding gases
and eventually burns. It separates the combustion chamber
Collection of experimental data was done on a 5.2 L four cyl-
into three thermodynamic zones; the first is the main
inder heavy-duty compression-ignition engine. The engine
features intake throttling valve, high pressure EGR and a
variable geometry turbocharger (VGT). The experimental Table 2 e Engine specifications for 5.2 L heavy-duty
engine.
apparatus and instrumentation is illustrated in Fig. 1. The
setup has been extensively discussed by Dimitriou et al. [5] Engine type 4-cylinder inline
and therefore only a brief report shall be given. The engine Displacement 5.2
has been instrumented with pressure sensors in all four cyl- volume in liters
inders to obtain in-cylinder pressure and thus enable Bore  stroke in mm 115  125
computation of heat release rate. Diesel is directly injected Compression ratio 17.5:1
Crank angle at ICV 127 ATDC
into the cylinders via a common-rail system. In addition, the
Injection system Electronically controlled
engine was modified to operate in dual-fuel mode with common rail injection system
hydrogen being port-injected upstream of the intake manifold Fuel strategy Diesel (direct injection)
and downstream of the turbocharger compressor intercooler. Hydrogen (port injection)
An inline air-hydrogen mixer ensures a well-premixed
mixture and homogeneous distribution to the four cylinders.
The main engine specifications are summarised in Table 2.
Table 3 e Overview of engine test conditions for single-
All experimental test data were collected at a constant and double-injection of Diesel.
engine speed of 1500 rpm. Hydrogen mass injection was based
Single-Injection
on the amount of hydrogen energy with respect to the total
Load in kW 20 40 60
input energy (i.e. Diesel and hydrogen), referred to as Engine speed in rpm 1500 1500 1500
hydrogen share ratio and defined as Diesel rail pressure in bar 865 994 1150
EGR rate in percent 0e24 0e24 0e24
m_ H2 LVHH2 RH2 in percent 0e56 0e57 0e47
RH2 ¼ 100 (1)
m_ H2 LVHH2 þ m_ Diesel LVHDiesel FH2 in percent 0e0.21 0e0.29 0e0.23
Main injection timing (ATDC) 6 4 4
where m_ H2 and m_ D denote the mass flow rates of the two fuels
Double-Injection
and LHVH2 and LHVD denote the lower heating values of Load in kW 20 40 60
hydrogen and diesel, respectively. Furthermore, one can Engine speed in rpm 1500 1500 1500
define the equivalence ratio of the hydrogen-air mixture as Diesel rail pressure in bar 865 994 1150
EGR rate in percent 0e25 0e25 0e28
AFRstoich
H2 =air RH2 in percent 0e42 0e42 0e42
F¼ (2) FH2 in percent 0e0.10 0e0.17 0e0.18
AFRH2 =air
Pre injection timing (ATDC) 24 25 27.5
where AFRH2 =air refers to the air-to-fuel ratio of the hydrogen- (diesel)-17.3
air mixture and the subscript stoich to the stoichiometric air- (HDDF)
to-fuel ratio, given in Table 1. Main injection timing (ATDC) 0.8 0.9 1.3
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 1 8 5 1 7 e1 8 5 3 1 18521

unburned zone (MUZ), that contains all charge air at intake ratio in dual-fuel applications is significantly different
valve closing (IVC); the second is the spray unburned zone from the equivalence ratio in the unburned zone. This
(SUZ), which includes injected fuel as well as entrained necessitates the generation of a user-defined regression to
gases; finally, there is the spray burned zone (SBZ) that compute laminar flame speed based on instantaneous in-
contains products of combustion. puts from the model. In this instance, this was done via a
The SITurb model on the other hand resorts to a two- Python function that transfers the computed laminar
zone model, where unburned mass is transferred from flame speed to the SITurb, overriding the original imple-
the unburned zone into the burned zone. The model in- mented model. Details of the regression model are given
cludes mass entrainment of the unburned fuel and air in the following section.
mixtures into the flame front moving proportionally to The model was setup and calibrated following the pro-
turbulent and laminar flame speeds. The model further cess layed out in Fig. 3. Initially, experimental data from the
requires calculating laminar flame speed as a function of laboratory test were extracted including mass flow rates of
equivalence ratio, unburned gas temperature and pressure air, liquid and gaseous fuel, hydrogen/air equivalence ratio,
as well as diluent mass fraction. However, the current EGR rate, engine speed, volumetric efficiency and injection
version of GT (v2019) significantly overestimates laminar timings. After this, a three pressure analysis (TPA) was
flame speeds for hydrogen-air mixtures, even at mixtures conducted, which includes engine breathing characteristics
that are beyond the flammability limit. This takes place, as through inclusion of intake and exhaust manifolds, as well
GT-Power estimates laminar flame speed based on global as valves and ports. The model schematic of the TPA is given
equivalence ratio. While this approach works well with in Fig. 2b. The premixed hydrogen-air charge is prescribed at
pure hydrogen combustion cases, the global equivalence the manifold inlet, while Diesel is directly injected into the

Fig. 2 e (a) Three pressure analysis (TPA) model used to calibrate volumetric efficiency and initialise swirl and tumble for the
closed volume model given in (b), shown with Python user function to compute laminar flame speed for diesel-hydrogen
dual fuel operation based on equivalence ratio, unburned zone temperature, in-cylinder pressure and residual mass
fraction in the unburned zone.
18522 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 1 8 5 1 7 e1 8 5 3 1

cylinder. TPA ensures that conditions within the cylinder at Chemkin


IVC as well as swirl and tumble are accurately initialised
when running the closed volume pressure analysis with the Modelling of premixed hydrogen-air laminar flame speeds
predictive combustion model. It further allows to calibrate was done in Chemkin 18.1 using the flame speed calculation
volumetric efficiency and air mass flow rates. model, PREMIX [50], for one-dimensional, freely propagating
With turbulence parameters and engine breathing, the flames. Thermodynamic data, transport properties as well as
next step involves calibration of the predictive combustion gas-phase kinetics were evaluated using GRI-Mech 3.0 [51]
model done in a single-cylinder, closed volume model. The chemical reaction mechanism. The model uses an implicit
multi-zone combustion model features several parameters finite-difference approximation in steady-state to discretise
that can be optimised based on experimental data encom- balance equations for mass, momentum, energy and species.
passing multipliers for entrainment rate Centr , ignition delay These are then solved through pseudo time-stepping to yield
Cign , premixed combustion rate Cpm and diffusion combus- the laminar flame velocity. Throughout this study, a mixture-
tion Cdf (DIPulse) as well as turbulent flame speed CTFS , flame averaged formulation for diffusion and fluxes was chosen and
kernel growth CFKG and Taylor micro-scale length CTLS (SIT- the Soret effect included. Finally, an adaptive grid method was
urb). These parameters are optimised using a standard elitist selected controlling maximum curvature and gradient
genetic algorithm that aims at minimising the root mean ensuring zero gradients at the boundaries until convergence is
squared error between experimental and predicted heat obtained.
release rate. As a first step, laminar flame speeds are calculated for
In case of port-injected hydrogen, a Python subroutine, hydrogen-air ratios at atmospheric conditions (p0 ¼ 1bar;
which accepts values for equivalence ratio, unburned zone Tu ¼ 298K) for equivalence ratios ranging from 0.2 to 6. A
temperature and in-cylinder pressure, uses a fitted power-law comparison of the resulting flame speed calculations with
curve to determine laminar flame speed. The data used for experimental and simulation data from the literature is
curve fitting was computed using the PREMIX solver in the given in Fig. 4. The simulation data depicts trends accu-
Chemkin combustion simulation solver. rately, although flame speeds tend to be slightly

Fig. 3 e Flowchart outlining the calibration process for the one-dimensional simulation model. Experimental data, such as
air and fuel mass flow rates, engine speed, equivalence ratios, EGR rates and injection timings are extracted and input into
the three pressure analysis (TPA) model to calibrate engine breathing and initialise flow parameters within the cylinder.
The final step is then calibrating DIPulse and SITurb parameters using the single-cylinder model. One-dimensional freely
propagating combustion simulations using Chemkin-Pro are fed into a Python flame speed fit model implemented as a user
routine within GT-Power.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 1 8 5 1 7 e1 8 5 3 1 18523

Fig. 4 e Comparison of experimental and literature data for laminar flame speed of hydrogen and air at ambient conditions
p0 ¼ 1bar; Tu ¼ 298K. The data for experimental results was taken from: Aung et al. [24], Berman [25], Dong [26], Dowdy
et al. [27], Egolfopoulos and Law [28], Guenther et al. [29], Iijima and Takeno [30], Koroll et al. [31], Krejci et al. [32], Kwon and
Faeth [33], Liu and MacFarlene [34], Pareja et al. [35], Qin et al. [36], Takahasi et al. [37], Tse et al. [38], Vagelopoulos et al. [39],
Wu and Law [40], Varea et al. [41], Park et al. [42], Lamoureux et al. [43] and Das et al. [44]. Simulation results were extracted
from: Marinov et al. [45] and Alekseev et al. [46]. The dash-dotted line stems from the Chemkin PREMIX model.

Fig. 5 e Comparison of experimental and literature data for laminar flame speed of hydrogen and air at elevated
temperatures and ambient pressure p0 ¼ 1bar. The data for experimental results was taken from: Hu et al. [47], Krejci et al.
[32] and Verhelst et al. [48]. Simulation results were extracted from Alekseev et al. [49]. The dash-dotted lines denote results
from the Chemkin PREMIX model.
18524 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 1 8 5 1 7 e1 8 5 3 1

underpredicted with respect to the literature data. The 300 and 600 K and pressure from ambient conditions (1
difference to other simulation results is, however, small barA) to 6 barA. Equivalence ratio was also varied from
and discrepancies between different experiments and 0.25, to 0.325 and finally to 0.4. Subsequently, a curve fit
simulations occur as a consequence of the method used to was performed following a power-law approach, as outlined
determine flame speed and how well effects of stretch and by Heywood [52] and given in Fig. 6.
flame curvature are reduced. As HDDF operation predomi- !b2
 b1
nantly occurs in the very lean region around the lower Tu p
SL ¼ S0L
flammability limit of F ¼ 0:2, the focus of this study is Tu;ref pref
(4)
primarily on laminar flame speeds up to an equivalence with b1 ¼ 11:0669  23:239ðF  1Þ
ratio of 0.4. As a result, a piece wise regression was per-
and b2 ¼ 3:7331 þ 6:9921ðF  1Þ
formed yielding
8 Finally, the presence of burned gas in the charge as a
< 2 consequence of trapped burned gas and EGR is taken into
0:1239 þ 6:6711ðF  0:0031Þ if 0:20  F < 0:27
SL0 ¼ (3)
: 0:2472 þ 662:0649ðF  0:2575Þ2 if 0:27  F  0:40 account by fitting a curve to simulation data, as given in
Fig. 7. The data were gathered considering unburned gas
The simulations were repeated for three different temperatures from 300 to 600 K, pressures from 1 to 6
temperatures, namely at 300 K, 373 K and 447 K, while barA and equivalence ratios in the lean region from 0.25
pressure was kept constant at 1 bar. The results were to 0.4. In addition, diluent mole fractions ranging from
again compared with experimental and numerical data 0 to 0.4 were considered. The diluent mixture was
from the open literature and are given in Fig. 5. Again, the approximated using experimental data from the exhaust
simulation complies well with previously published manifold and included approximately 70% N2 , 8% O2 , 12%
simulation and laboratory data over the entire equiva- CO2 and 10% H2 O. The final fitted curve follows Equation
lence ratio range. (5).
In order to account for the effect of higher unburned gas  
temperature and pressures, as will be present within the SL ðxb Þ ¼ SL ðxb ¼ 0Þ 1  1:2406x0:6952
b (5)
combustion chamber, a design of experiments (DoE) was
conducted varying temperature in steps of 100 K between

Fig. 6 e Laminar flame speed curve fit for elevated temperatures and pressures. Regression was done at temperatures from
300 to 600 K and pressures from 1 barA to 6 barA for three equivalence ratios at 0.25 (top), 0.325 (centre) and 0.40 (top).
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 1 8 5 1 7 e1 8 5 3 1 18525

which accounts for a delay of approximately 4:5+ . At low and


Results and discussion medium load conditions, peak heat release rate tends to be
overpredicted resulting in slightly higher peak in-cylinder
In the following section, the results for in-cylinder pressure pressures. The relative error however remains within 5.5%.
and heat release rate obtained from GT-Power simulations are Towards higher loads of 60 kW start of combustion (SOC) is
compared against the experimental data from the test bed. predicted with a delay of approximately 1+ . In combination
First of all, cases with single injection are treated. These with the slightly overpredicted initial peak in heat release
comprise hydrogen share ratios from 0 to 56% and H2 -air rate, the relative error in peak pressure reduces to less than
equivalence ratios from 0 to 0.29. Thereafter, pilot injection 1%. In general, however, the predictive model shows good
and main injection operation is discussed with hydrogen agreement with both experimental in-cylinder pressure and
share ratios ranging from 0 to 42% and H2 -air equivalence heat release rate for all load conditions and both high and low
ratios from 0 to 0.18. EGR rates, giving confidence that diesel injection and com-
bustion are captured well. This is further emphasised when
Single injection computing mean relative errors (MRE) and root mean squared
errors (RMSE) between in-cylinder pressures and heat release
Diesel operation rates (HRR) respectively, as given in Table 4.
Fig. 8 depicts data for in-cylinder pressure and heat release
rate exhibiting results of both experimental and predicted
Hydrogen-diesel dual fuel operation
model. The data encompasses low (20 kW), medium (40 kW)
The next step involves considering hydrogen share ratios of
and high (60 kW) load cases and variation in EGR rate as well.
up to 28%, however at equivalence ratios beyond the lower
In addition, the annotations in the top right corner of each
flammability limit, namely up to 0.15. As a result it can be
subplot outline hydrogen energy share ratio, equivalence ratio
expected that no moving flame front occurs and that all the
and EGR rate (in the format RH2 /FH2 /Jv ) for each case. To begin
hydrogen combustion stems from entrainment of hydrogen
with, we shall first investigate how well the DIPulse model can
into the diesel spray upon injection. The flame speed model
predict diesel only operation. Therefore the test deals with 0%
takes this into account by setting laminar flame speed to zero
hydrogen share ratio. The dashed line outlines the start of
as soon as the equivalence ratio falls below the flammability
injection not including the hydraulic delay of around 500 ms,
limit of 0.2. Therefore, the characteristics in terms of pressure

Fig. 7 e Effect of burned gas dilution on laminar flame speed and corresponding curve fit.
18526 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 1 8 5 1 7 e1 8 5 3 1

Fig. 8 e Comparison between experimental (blue curve) and predicted (red curve) in-cylinder pressure and heat release rate
for (a) for diesel operation only and (b) medium hydrogen share ratio (up to 28%) and equivalence ratios beyond the
flammability limit. The left column of graphs shows low (20 kW), the centre medium (40 kW) and the right column high
(60 kW) load cases and variations with no EGR results on top and cases with EGR on the bottom. The dashed vertical line
marks the electrical signal for the main fuel injection. Annotations in the top right corner of each subplot show hydrogen
energy share ratio/equivalence ratio/EGR rate.

and heat release rate are not expected to vary dramatically. diesel-only operation for low and medium loads, it now ap-
This is confirmed when inspecting Fig. 8b in more detail. pears to be more in line or gently underpredicted, reducing
While heat release rate tended to be slightly overpredicted in relative error in peak pressure to around 1e4%. At higher loads
the offset in SOC has reduced to 0.5% and while the change
with respect to the low EGR case is minimal, the high EGR
Table 4 e Mean relative error (MRE) for in-cylinder case, as shown in the bottom right corner of Fig. 8b, experi-
pressure and root mean squared errors (RMSE) for heat ences increased underprediction of in-cylinder pressure as a
release rate between experiments and simulation for consequence of the late prediction of SOC and lower heat
single-injection diesel-only cases with and without EGR.
release rate compared to the diesel-only operation.
20 kW 40 kW 60 kW Further increasing hydrogen share ratio in turn leads to
pcyl MRE with EGR 3.1% 1.7% 2.0% higher hydrogen-air equivalence ratios that can lie within the
HRR RMSE with EGR 0.0033 0.0065 0.012 flammability limit of 0.2. This initiates flame propagation of
pcyl MRE w/o EGR 3.5% 2.1% 1.5% the hydrogen mixture from which stems an increased heat
HRR RMSE w/o EGR 0.0031 0.0062 0.0043
release rate and thus higher in-cylinder pressure. The effect of
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 1 8 5 1 7 e1 8 5 3 1 18527

this is presented in Fig. 9. The graph depicts hydrogen share entrainment and combustion of diesel and hydrogen is not
ratios of up to 57% and equivalence ratios varying between accurately captured by the dual fuel model.
0.14 and 0.29. Therefore, despite the high energy share ratio, This trend is exemplified by Fig. 10 that depicts the relative
all cases with low EGR rate (top row graphs in Fig. 9) are below error in peak pressure between experiments and simulation
the flammability limit. These shall be discussed first in more as it varies with equivalence ratio and thus hydrogen content
detail. When comparing the experimental heat release rate in the charge mixture. Initially, at zero hydrogen content
and in-cylinder pressure without EGR across Figs. 8a and 9 it within the charge mixture, peak pressures are overpredicted.
becomes apparent that both increase as hydrogen share ratio As the hydrogen share is gradually increased, the error re-
increases even though the mixture is too lean for flame duces linearly until the model starts underestimating peak
propagation (F < 0.2). This stems from the combustion of pressures and thus peak heat release rates. In sum, this leads
diesel and entrained hydrogen into the injected liquid fuel. For to relative errors in the range of approximately 5%. In light of
low to medium hydrogen share ratios of up to around 25%, the the complex, three-dimensional nature of combustion pro-
effect is still moderate and thus less obvious. However, cesses such a level or accuracy is well within the acceptable
comparing this with the simulation data shows that if range for one-dimensional modelling. However, it is impor-
hydrogen share ratio is increased further-as done in Fig. 9- the tant to notice that the model appears to struggle with the

Fig. 9 e Comparison between experimental (blue curve) and predicted (red curve) in-cylinder pressure and heat release rate
for medium hydrogen share ratio (up to 57%). Equivalence ratios vary between 0.14 and 0.29. The left column of graphs
shows low (20 kW), the centre medium (40 kW) and the right column high (60 kW) load cases and variations with no EGR
results on top and cases with EGR on the bottom. The dashed vertical line marks the electrical signal for the main fuel
injection. Annotations in the top right corner of each subplot show hydrogen energy share ratio/equivalence ratio/EGR rate.

Fig. 10 e Relative error in peak pressure plotted against hydrogen-air equivalence ratio for cases without EGR (top) and with
EGR (bottom). The numbers next to each data point denote the hydrogen share ratio in percent.
18528 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 1 8 5 1 7 e1 8 5 3 1

correct prediction of hydrogen entrainment. It is of course speed through Equation (5). The equivalence ratios are still
possible to re-calibrate the model by defining the entrainment close to the lower flammability limit yielding predicted flame
multiplier as a function of hydrogen share ratio. However, this speeds in the order of around 1e5 cm/s. Flame propagation is
measure would significantly curb the model's predictive ca- nonetheless sufficient to account for a marked increase in
pabilities. It shall be noted that similar trends illustrated in heat release rate and pressure, which is well captured by the
Fig. 10 can be witnessed when comparing heat release rates model.
between experiments and simulation. Qualitatively, this can
also be shown when comparing heat release rates between Double injection
Figs. 8b and 9. The higher the energy share ratio, the more the
predicted heat release rate falls short off the experimentally Diesel operation
determined one. Double injections cases feature a pilot-injection of a small
As soon as the hydrogen-air equivalence ratio is suffi- quantity of fuel prior to reaching top dead centre (TDC) and
ciently high to foster flame propagation, as shown in the lower the main injection taking place just after TDC. For diesel-only
row of graphs in Fig. 9, the model computes the laminar flame cases, this results in the pressure and heat release rate

Fig. 11 e Comparison between experimental (blue curve) and predicted (red curve) in-cylinder pressure and heat release rate
for (a) diesel operation and (b) medium hydrogen share ratio (up to 42%) with equivalence ratios beyond the flammability
limit. The left column of graphs shows low (20 kW), the centre medium (40 kW) and the right column high (60 kW) load cases
and variations with no EGR results on top and cases with EGR on the bottom. The left dashed vertical line marks the
electrical signal for the pilot fuel injection, while the right one denotes the signal for the main fuel injection. Annotations in
the top right corner of each subplot show hydrogen energy share ratio/equivalence ratio/EGR rate.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 1 8 5 1 7 e1 8 5 3 1 18529

distributions given in Fig. 11a. Again, the top row denote re- relative subtle feature at lower hydrogen share ratios and
sults at different load cases without EGR rate, while the lower appears to become more pronounced at higher load.
row of graphs yield results including EGR addition. In the The situation becomes more clear when hydrogen share
absence of any hydrogen in the charge mixture, the heat ratio is further increased to 42%. There is still no flame prop-
release rate is characterised by a distinct two peak shape agation expected, although equivalence ratios increase to a
following pilot and main injection. There is satisfactory range of 0.1e0.18. Again, EGR rates remain zero, with the
agreement between experimental and predicted heat release exception of the 40 kW, where a low EGR rate of approximately
rate as well as in-cylinder pressure across all load cases and 9% was included. At low load conditions, the effect is still
EGR rates. Nonetheless, the model consistently overestimates hardly noticeable, although there is a clear non-zero heat
ignition delay for the pilot-injection as well as the amount of release rate in between pilot and main-injection. However,
heat release in the process. The consequence is that the pre- this continues to have little effect on the accuracy of the
dicted peak cylinder pressure is by up to 4% larger than the predicted in-cylinder pressure. Moving towards medium and
experimentally one. A similar trend is witnessed by Piano high load cases, characteristics change substantially and sig-
et al. [21]. The calibration process seeks to minimise the error nificant discrepancies between model and experiments occur.
between measured and predicted heat release rate. It is thus The heat release profile shifts from the diesel-typical twin-
not surprising that there will be larger deviations in the heat peak shape towards a continuously increasing rate until
release rate for the pilot injection, since large deviations in the main-injection takes place. The phenomenon takes place
main injection heat release rate have a larger impact on during the compression stroke and in conjunction with the
overall error. The overprediction propagates further to the ongoing compression this accounts for considerably larger in-
main injection, which then yields pressures that are in be- cylinder pressures. As a result, one can observe a smearing of
tween 1 and 5% larger than in the experiments. This is slightly the two peaks in both heat release rate and in-cylinder pres-
exacerbated by an early SOC present for all diesel-only cases. sure. The reason for the witnessed behaviour stems presum-
Nonetheless, MRE between experimental and predicted in- ably from the higher hydrogen content in the cylinder charge
cylinder pressures range within 1.4 and 2.9%, as displayed in that is entrained within the diesel pilot-jet and is ignited along
Table 5. In addition, despite the overestimated ignition delay with the liquid fuel as the cylinder moves upwards towards
for the pilot injection RMSE for heat release rates are only TDC. While, the simulation model was able to predict com-
marginally larger compared to single Diesel injection cases. bustion performance in single-injections reasonably well, it
cannot capture the effects of double-injection at medium and
Hydrogen-diesel dual fuel operation higher loads and continues to reflect the distinct diesel two-
Initially, at low EGR rates and for hydrogen share ratios of peak profiles. Since the main effects associated with this
approximately 20e26%, the introduction of hydrogen into the occur during the compression stroke, the shortcomings of the
charge mixture does not alter the distinct two-peak heat multi-zone combustion models with respect to flow entrain-
release rate, as illustrated in Fig. 11b. All cases encompass ment become more pronounced and larger differences in the
zero EGR. Equivalence ratio is well beyond the flammability predicted characteristics transpire.
limit ranging from 0.06 to 0.08. For all load cases both heat
release rate and in-cylinder pressure are well captured.
Similar to the conventional diesel operation, the initial peak in Conclusions and outlook
heat release rate originates from combustion of a small
amount of diesel being injected during the compression stroke The focus of this study is on the capacity of widely used pre-
and hydrogen and air entrained in the spray plume. The sec- dictive multi-zone combustion models in one-dimensional
ond, larger peak stems from combustion of the main injection engine performance simulations. The study employed the
quantity and again entrains further hydrogen and air in the dual-fuel combustion model implemented in GT-Power in
process. However, the experimental data reveals that as load combination with a laminar flame speed fitting model derived
is increased the heat release rate in between the two fuel in- from one-dimensional, freely propagating flame simulations
jection pulses does not fully return to zero any more, while the in Chemkin. The performance and limitations of the simula-
simulation model maintains this trend and does not capture tion model with respect to experimental data was then
the change in behaviour as seen in the experiments. This is a examined over a range of engine loads, hydrogen share ratios,
equivalence ratios and EGR rates and for two different injec-
tion strategies. The conclusions drawn from the study can
thus be summarised as follows:

Table 5 e Mean relative error (MRE) for in-cylinder  Current version of GT-Power (v2019) necessitates a user-
pressure and root mean squared errors (RMSE) for heat defined model that computes laminar flame speed
release rate between experiments and simulation for instantaneously based on hydrogen/air equivalence ratio,
dual-injection diesel-only cases with and without EGR.
in-cylinder pressure, unburned zone temperature and re-
20 kW 40 kW 60 kW sidual mass fraction. The model proposed in this study
pcyl MRE with EGR 2.1% 2.5% 1.8% yields flame speeds in good agreement with data from the
HRR RMSE with EGR 0.0065 0.0067 0.0075 literature and proves to exhibit satisfactory performance
pcyl MRE w/o EGR 2.9% 2.5% 1.4% when applied to the predictive combustion model in GT-
HRR RMSE w/o EGR 0.0042 0.0044 0.0045
Power.
18530 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 1 8 5 1 7 e1 8 5 3 1

 For single diesel injection the model is able to capture the and emission characteristics of a four stroke-single cylinder
effects of hydrogen addition in the charge mixture well. diesel engine. Int J Mech, Aero, Ind, Mechatron and Manuf
This holds for hydrogen share ratios up to 57% and Eng 2015;9:914e20.
[9] Talibi M, Hellier P, Ladommatos N. The effect of varying EGR
hydrogen/air equivalence ratios of up to 0.29. However,
and intake air boost on hydrogen-diesel co-combustion in CI
there is a clear, discernible trend toward underprediction engines. Int J Hydrogen Energy 2017;42:6369e83.
of hydrogen entrained within the injected diesel spray [10] Li G, Yu X, Shi W, Yao C, Wang S, Shen Q. Effects of split
leading to heat release rates and in-cylinder pressures that injection proportion and the second injection timings on the
fall below what is witnessed during the laboratory tests. combustion and emissions of a dual fuel SI engine with split
 The trend indicated by the single-injection study appears hydrogen direct injection. Int J Hydrogen Energy
to be further aggravated when a double-injection strategy 2019;44:11194e204.
[11] Elsemary IMM, Attia AAA, Elnagar KH, Elsaleh MS. Spark
is taken. The model is able to reproduce the experimental
timing effect on performance of gasoline engine fueled with
data up to a hydrogen share ratio of up to 26%. However, at mixture of hydrogengasoline. Int J Hydrogen Energy
higher share ratios, the model cannot capture partial 2017;42:30813e20.
hydrogen combustion initiated by pre-injection. Under [12] Niu R, Yu X, Du Y, Xie H, Wu H, Sun Y. Effect of hydrogen
these circumstances, the model erroneously maintains a proportion on lean burn performance of a dual fuel SI engine
distinct two-peak heat release rate distribution. using hydrogen direct-injection. Fuel 2016;186:792e9.
[13] Wu H, Yu X, Du Y, Ji X, Niu R, Sun Y, Gu J. Study on cold start
 Given the performance benefit of the hydrogen-diesel dual
characteristics of dual fuel SI engine with hydrogen direct-
fuel engine and its potential to mitigate carbon footprint of
injection. Appl Therm Eng 2016;100:829e39.
internal combustion engines in the short and medium [14] Menaa A, Lounici MS, Amrouche F, Loubar K, Kessal M. CFD
term, there is a clear need for further research on model- analysis of hydrogen injection pressure and valve profile law
ling of hydrogen-diesel dual fuel combustion. effects on backfire and pre-ignition phenomena in hydrogen-
diesel dual fuel engine. Int J Hydrogen Energy
2019;44:9408e22.
[15] Ghazal OH. Combustion simulation of direct injection CI
engine operating on dual fuel using CFD. Acta Phys Pol, A
Acknowledgement 2018;134:387e90.
[16] Sharma P, Dhar A. Compression ratio influence on
The financial support of the Daiwa Anglo-Japanese Founda- combustion and emissions characteristic of hydrogen diesel
dual fuel CI engine: numerical Study. Fuel 2018;222:852e8.
tion and the Great Britain Sasakawa Foundation is highly
[17] Santoso WB, Bakar RA, Nur A. Combustion characteristics of
appreciated. The authors would also like to thank Kevin diesel-hydrogen dual fuel engine at low load, energy
Roggendorf and Jared Cromas from Gamma Technologies for procedia. In: International conference on sustainable energy
their invaluable help and suggestions during the modelling engineering and application. 32. ICSEEA); 2013. p. 3e10. 2012.
campaign. [18] Chintala V, Subramanian KA. CFD analysis on effect of
localized in-cylinder temperature on nitric oxide (NO)
emission in a compression ignition engine under hydrogen-
references diesel dual-fuel mode. Energy 2016;116:470e88.
[19] Koten H. Hydrogen effects on the diesel engine performance
and emissions. Int J Hydrogen Energy 2018;43:10511e9.
[20] Hvezda J. Multi-zone models of combustion and heat
[1] Bora BJ, Saha UK, Chatterjee S, Veer V. Effect of compression
transfer processes in SI engines. In: SAE 2014 world congress
ratio on performance, combustion and emission
exhibition. SAE International; 2014.
characteristics of a dual fuel diesel engine run on raw biogas.
[21] Piano A, Millo F, Boccardo G, Rafigh M, Gallone A, Rimondi M.
Energy Convers Manag 2014;87:1000e9.
Assessment of the predictive capabilities of a combustion
[2] Dimitriou P, Tsujimura T. A review of hydrogen as a
model for a modern common rail automotive diesel engine.
compression ignition engine fuel. Int J Hydrogen Energy
In: SAE 2016 world congress and exhibition. SAE
2017;42:24470e86.
International; 2016.
[3] Dimitriou P, Tsujimura T, Suzuki Y. Low-load hydrogen-
[22] Ghazal OH. Combustion analysis of hydrogen-diesel dual
diesel dual-fuel engine operation - a combustion efficiency
fuel engine with water injection technique. Case Stud Therm
improvement approach. Int J Hydrogen Energy
Eng 2019;13. 100380.
2019;43:13605e17.
[23] Monemian E. Simulating diesel-hydrogen combustion by GT-
[4] Nag S, Sharma P, Gupta A, Dhar A. Experimental study of
power. In: Proceedings of the FISITA world automotive
engine performance and emissions for hydrogen diesel dual
congress; 2018.
fuel engine with exhaust gas recirculation. Int J Hydrogen
[24] Aung KT, Hassan MI, Faeth GM. Flame stretch interactions of
Energy 2019;44:12163e75.
laminar premixed hydrogen/air flames at normal
[5] Dimitriou P, Kumar M, Tsujimura T, Suzuki Y. Combustion
temperature and pressure. Combust Flame 1997;109.
and emission characteristics of a hydrogen-diesel dual-fuel
[25] M. Berman, reportSandia laboratories report, technical
engine. Int J Hydrogen Energy 2018;43:13605e17.
report, SSAND84-0689, S1984.
[6] Dimitriou P, Tsujimura T, Suzuki Y. Adopting biodiesel as an
[26] Dong Y, Holley AT, Andac MG, Egolfopoulos FN, Davis SG,
indirect way to reduce the NOx emission of a hydrogen
Middha P, Wange H. Extinction of premixed H2/air flames:
fumigated dual-fuel engine. Fuel 2019;244:324e34.
chemical kinetics and molecular diffusion effects. Combust
[7] Tsujimura T, Suzuki Y. The utilization of hydrogen in
Flame 2005;142:374e87.
hydrogen/diesel dual fuel engine. Int J Hydrogen Energy
[27] Dowdy DR, Smith DB, Taylor SC, Williams A. The use of
2017;42:14019e29.
expanding spherical flames to determine burning velocities
[8] Deb M, Sastry GRK, Banerjee RS Panua R, Bose PK. Effect of
and stretch effects in hydrogenair mixtures. In: Symposium
hydrogen-diesel dual fuel combustion on the performance
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 1 8 5 1 7 e1 8 5 3 1 18531

(international) on combustion, vol. 23; 1991. p. 325e32 on combustion, vol. 20; 1985. p. 1941e9 [Twentieth
[Twenty-Third Symposium (International) on Combustion]. Symposium (International) on Combustion].
[28] Egolfopoulos FN, Law CK. An experimental and [41] Varea E, Beeckmann J, Pitsch H, Chen Z, Renou B.
computational study of the burning rates of ultra-lean to Determination of burning velocities from spherically
moderately rich H2/O2/N2 laminar flames with pressure expanding H2/air flames. Combust Flame 2015;35:711e9.
variations. In: Symposium (international) on combustion, [42] Park O, Veloo PS, Burbano H, Egolfopoulos FN. Studies of
vol. 23; 1991. p. 333e40. premixed and non-premixed hydrogen flames. Combust
[29] Günther R, Janisch G. Measurements of burning velocity in a Flame 2015;162:1078e94.
flat flame front. Combust Flame 1972;19:49e53. [43] Lamoureux N, Djebaili-Chaumeix N, Paillard CE. Laminar
[30] Iijima T, Takeno T. Effects of temperature and pressure on flame velocity determination for H2airsteam mixtures using
burning velocity. Combust Flame 1986;12:445e52. the spherical bomb method. Journal de Physique de France
[31] Koroll GW, Kumar RK, Bowles EM. Burning velocities of IV 2002;109.
hydrogenair mixtures. Combust Flame 1993;94:330e40. [44] Das AK, Kumar K, Sung CJ. Laminar flame speeds of moist
[32] Krejci MC, Mathieu O, Vissotski AJ, Ravi S, Sikes TG, syngas mixtures. Combust Flame 2011;158:345e53.
Petersen EL, Ke rmones A, Metcalfe W, Curran HJ. Laminar [45] Marinov NM, Westbrook CK, Pitz WJ. Detailed and global
flame speed and ignition delay time data for the kinetic chemical kinetics model for hydrogen. Transport
modeling of hydrogen and syngas fuel blends. J Eng Gas Phenomena in Combustion 1996;1.
Turbines Power 2012;135. [46] Alekseev VA, Christensen M, Berrocal E, Nilsson EJK,
[33] Kwon OC, Tseng L-K, Faeth GM. Laminar burning velocities Konnov AA. Laminar premixed flat non-stretched lean
and transition to unstable flames in H2/O2/N2 and C2H8/O2/ flames of hydrogen in air. Combust Flame
N2 mixtures. Combust Flame 1992;90:230e46. 2015;162:4063e74.
[34] Liu DDS, MacFarlane R. Laminar burning velocities of [47] Hu E, Huang Z, He J, Miao H. Experimental and numerical
hydrogen-air and hydrogen-air-steam flames. Combust study on laminar burning velocities and flame instabilities of
Flame 1983;49:59e71. hydrogenair mixtures at elevated pressures and
[35] Pareja J, Burbano HJ, Ogami Y. Measurements of the laminar temperatures. Int J Hydrogen Energy 2009;34:8741e55.
burning velocity of hydrogenair premixed flames. Int J [48] Verhelst S, Woolley R, Lawes M, Sierens R. Laminar and
Hydrogen Energy 2010;35:1812e8. unstable burning velocities and Markstein lengths of
[36] Aung KT, Hassan MI, Faeth GM. Flame stretch interactions of hydrogenair mixtures at engine-like conditions. In:
laminar premixed hydrogen/air flames at normal Proceedings of the combustion institute, vol. 30; 2005.
temperature and pressure. Combust Flame 1997;109. p. 209e16.
[37] Takahashi F, Mizomoto M, Ikai S. Laminar burning velocities [49] Alekseev VA, Christensen M, Konnov AA. The effect of
of hydrogen/oxygen/inert gas mixtures. Alternative Energy temperature on the adiabatic burning velocities of diluted
Sources III 1983;5:447e57. hydrogen flames: a kinetic study using an updated
[38] Tse SD, Zhu DL, Law CK. Morphology and burning rates of mechanism. Combust Flame 2015;162:1884e98.
expanding spherical flames in H2/O2/inert mixtures up to 60 [50] Kee RJ, Grcar JF, Smooke MD, Miller AJ, Meeks Ellen. PREMIX:
atmospheres. In: Proceedings of the combustion institute, a fortran program for modeling steady laminar one-
vol. 28; 2000. p. 1793e800. dimensional premixed flames. Sandia Rep 1985;143.
[39] Vagelopoulos CM, Egolfopoulos FN, Law CK. Further [51] G. P. Smith, D. M. Golden, M. Frenklach, N. W. Moriarty, B.
considerations on the determination of laminar flame Eiteneer, M. Goldenberg, C. T. Bowman, R. K. Hanson, S.
speeds with the counterflow twin-flame technique. In: Song, W. C. Gardiner, Jr, V. V. Lissianski, and Z. Qin, GRI-
Symposium (international) on combustion, vol. 25; 1994. Mech 3.0, The Gas Research Institute. URL http://www.me.
p. 1341e7. berkeley.edu/gri_mech/þ.
[40] Wu CK, K Law C. On the determination of laminar flame [52] Heywood JB. Internal combustion engine fundamentals. New
speeds from stretched flames. In: Symposium (international) York: McGraw-Hill; 1988.

You might also like