Energy 36 (2011) 647e655

Contents lists available at ScienceDirect

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

Computational modeling, validation, and utilization for predicting

the performance, combustion and emission characteristics of hydrogen IC engines
Shravan K. Vudumu, Umit O. Koylu*
Department of Mechanical and Aerospace Engineering, Missouri University of Science and Technology, Rolla, 290A Toomey Hall, 400 West 13th Street, MO 65409 0050, USA

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

Article history: Hydrogen-fueled internal combustion engines are considered to be more efficient and cleaner alterna-
Received 3 May 2010 tives to their fossil-fueled counterparts. Reasonably fast and accurate predictive computational tools are
Received in revised form essential for practical design, control and optimization of hydrogen engines. To serve for this broader
10 September 2010
purpose, a computational model, which has been widely used for gasoline and diesel engines, is
Accepted 29 September 2010
investigated for its capability to simulate hydrogen engines. Specifically, fuel-specific sub-models are first
Available online 18 November 2010
incorporated by properly accounting for hydrogen’s distinct properties such as flame speed and burn
rate. The accuracy of the model is then assessed by validating it in comparison to independent experi-
Keywords:
Hydrogen
mental data. Finally, it is utilized to quantify the environmental impact of exhaust gas recirculation. With
Engine simulations these improvements, the present predictive model is shown to capture the measured engine perfor-
Emissions mance and emission data well under different operating conditions. In particular, the variations of peak
Exhaust gas recirculation in-cylinder pressure, heat release rate, brake power, brake thermal efficiency, exhaust temperature, and
NOx emissions are predicted close to the measured values. With the addition of a proportional-integral-
derivative controller to the engine model, exhaust gas recirculation level is varied, resulting in nearly an
order of magnitude reduction in NOx emissions during the present simulations.
Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction
As a part of the sustainable solutions to the increasing demands
for clean and secure energy, hydrogen can be produced from
a variety of sources while its chemical reaction in air does not
produce any greenhouse gases unlike other conventional fuels
[1e3]. Compared to gasoline, hydrogen has more energy per unit
mass, a higher flame speed, wider flammability limits, and a lower
minimum ignition energy (Table 1). These properties make
hydrogen one of the attractive alternative choices to power (IC)
internal combustion engines [4e7]. The existing engine design
methods and manufacturing plants can be fitted with minor
modifications to produce hydrogen engines in the near term while
other technologies, such as fuel cells, demand a complete re-design
of vehicles in the long term. Consequently, hydrogen engines can
also act as a transitional technology to fuel cell and hybrid vehicles
during the development and implementation of emerging energy
technologies [8e10].
* Corresponding author. Tel.: þ1 573 341 6601; fax: þ1 573 341 4607.
E-mail address: koyluu@mst.edu (U.O. Koylu).
Hydrogen IC engines have the potential for high power because
of more energy per unit mass and high flame speed, high efficiency
because of high flame speed that causes high rate of pressure rise in
the cylinder and hence near constant-volume combustion, and
near-zero emissions, except NOx at higher loads, because of the
absence of carbon in the fuel molecular structure. Hydrogen can be
operated with wide open throttle due to extremely wide flamma-
bility limits, which, unlike gasoline engines, decrease the cycle-by-
cycle variations even with very lean mixtures [4,9e11,13]. Because
of its distinct properties described above, hydrogen can also be
used as a single component fuel or in a multi-component fuel to
improve combustion of other fuels like gasoline, methane, alcohols,
LPG, biogas, and diesel [12,14e16].
Hydrogen IC engines have also technical challenges that need
to be overcome. Increasing the equivalence ratio for a higher
power demand increases NOx emissions, which are higher than
those from a regular gasoline engine for the same conditions,
limits the use of hydrogen fuel to low power density engines
[9,13,17]. Due to hydrogen’s lower minimum ignition energy, any
hot spot in the combustion chamber might become a source of
ignition, potentially resulting in pre-ignition/backfire [4,9,10]. The
high rate of combustion could cause very high rate of pressure
rise and uncontrolled abnormal combustion resulting in engine
knock [4,13].

0360-5442/$ e see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.energy.2010.09.051
648 S.K. Vudumu, U.O. Koylu / Energy 36 (2011) 647e655

shorten the development time from conceptual ideas to actual

Nomenclature products. This is especially important for non-conventional
emerging engine technologies that are in the initial stages of
Ae entrained surface area at the edge of the flame front development and commercialization. Hence, there is a crucial need
etot;s total sensible energy to develop, validate and utilize simple yet predictive models for
FMEP friction mean effective pressure hydrogen engines.
LHVi instantaneous fuel lower heating value Among many others, one of the leading quasi-dimensional
mb burned mass engine simulation software is GT-POWER by Gamma Technologies
me entrained mass of the unburned mixture Inc. [26]. After extensive development and validation for conven-
mf ;tot total fuel mass tional gasoline and diesel engines, it has become industry-standard
p pressure in the cylinder engine simulation software for relatively fast but reasonably accu-
Q heat transfer rate essential predictions. While this computational model has been
SL ; ST laminar and turbulent flame speed widely used in the literature for hydrocarbon-fueled engines, only
t time a limited number of studies are available on its application in pre-
V volume of the cylinder dicting the performance of hydrogen-fueled engines. Noteworthy is
the study by Polasek et al. [18] who compared their experiments on
Greek Symbols a hydrogen engine with a model developed in GT-POWER. However,
ru density of unburned zone they used a non-predictive model by fitting the coefficients in the
s time constant Wiebe function to their experimental data for obtaining the
l Taylor micro-scale length combustion burn rate/heat release rate. Accordingly, their approach
is engine specific and may not be applied to other hydrogen engines.
Subscripts Furthermore, many past simulations of hydrogen engines, including
tot total cylinder quantities (burned and unburned) [18], were based on sub-models that have specifically been devel-
b burned zone oped for hydrocarbon fuels. One important example is that
u unburned zone hydrogen has a much higher flame speed compared to gasoline at
a fixed equivalence ratio. Therefore, distinctive properties of
hydrogen must be accounted for using new sub-models.
2. Background and motivation
3. Research objectives
To realize maximum advantages of hydrogen with the above-
mentioned distinctive properties, detailed research is required for Based on the brief discussion above, the objective of the present
the development of fuel-specific combustion and emission models. study is to develop a hydrogen fuel-specific predictive one-
Advanced control methods and operating strategies to reduce NOx dimensional engine model based on two-zone combustion meth-
emissions at high loads are also needed. These efforts have the odology, validate it against independent experimental data for
potential of producing more efficient and lower emission hydrogen widespread implementation, and demonstrate its utilization for
engines that surpass the current fossil-fuel burning IC engines. finding operating conditions for higher performance and lower
Numerous studies exist in the literature on hydrogen-fueled emissions. Specifically, an accurate hydrogen flame speed sub-
engines; please see two recent technical reviews by White et al. [9], model is to be incorporated into the GT-POWER software so that the
Verhelst et al. [10], and references cited therein. While there are fuel-specific properties can be properly accounted for and therefore
many experimental and numerical studies on the characteristics of the computational predictions can be significantly improved.
hydrogen-fueled spark-ignition engines [4,13,16e23], very few Additionally, a predictive turbulent combustion model is to be
have used computational tools that could be extended later for adopted so that the combustion burn rate sub-model and therefore
investigating advanced combustion modes, emission control tech- the engine performance and emission characteristics do not require
nologies, and new efficiency increasing opportunities with a second experimentally-prescribed parameters [26]. This will yield
law analysis. Few recent studies have also focused on three- a computational tool that can directly predict different operating
dimensional (CFD) computational fluid dynamics tools to under- conditions, allowing it to be potentially used for any hydrogen
stand in-cylinder flow field and mixing process [24,25]. Although engine. A well-documented experimental study [13] on a spark-
such multi-dimensional models are necessary to understand the ignition port-injected IC engine fueled by gaseous hydrogen is
details of in-cylinder combustion conditions, they are computa- identified for comparing against the simulations and therefore
tionally demanding. Simpler quasi-dimensional models are desir- assessing the accuracy and suitability of the computational
able for fast computations with reasonable accuracy for practical predictions. A (PID) proportional-integral-derivative controller is
design, control and optimization purposes. Such engine simula- then added to the present model for adjusting (EGR) exhaust gas
tions also provide cost-effective technical tools that considerably recirculation and quantifying the accompanying reductions in NOx
emissions. Since GT-POWER is already established for predicting
Table 1 gasoline and diesel engines, the results presented here are expected
Combustion characteristics of hydrogen in comparisons to those of gasoline (at
to contribute to the improved design and analysis of hydrogen IC
300 K and 1 atm) [6,9,13].
engines in the automotive industry and therefore a faster and
Combustion Characteristic Hydrogen Gasoline Notes smoother transition to emerging cleaner and more efficient energy
Lower heating value 120 44 Higher energy density conversion devices.
(kJ/g) per unit mass
Stoichiometric flame 2.65e3.25 0.37e0.43 Faster burning
speed (m/s) 4. Computational methods
Flammability limits 4e75 1e7 Wider limits
in air (vol.%) Essential descriptions of the computational methods and their
Minimum ignition 0.02 0.28 Easier ignition implementation are given in the following, as further details can be
energy in air (mJ)
found in [27].
 S.K. Vudumu, U.O. Koylu / Energy 36 (2011) 647e655 649

4.1. Engine operating conditions was iterated for temperature and pressure until they satisfied the
gas density and energy that were already calculated. Numerical
The operating conditions of the hydrogen IC engine modeled convergence was ensured by considering different convergence
and simulated in this investigation were chosen similar to the criterion and checking the solution did not change. While solving
independent study by Subramanian et al. [13] because their the equations of continuity, momentum and energy, the simula-
reported test conditions and experimental data were well-docu- tions were run until these equations met the minimum conver-
mented. The specifications of the spark-ignition hydrogen engine gence criterion (relative change in  0:2%value) for each computed
from [13] used in this computational study are given in Table 2. The parameter such as temperature, pressure (in all the sub-volumes)
single-cylinder research engine was operated at wide open throttle and brake power for five consecutive engine cycles.
(no throttle restriction), and the equivalence ratio (hence the power
dm X
output) was varied by changing the amount of gaseous hydrogen ¼ _
m (1)
injected into the intake port. The simulations were optimized for dt
boundaries
(MBT) minimum advance for best torque for each case, similar to
the experimental study [13] that was used here for comparing
against the present computations. Adiabatic flame temperature of  
P   rujuj dxA 1
hydrogen (2390 K) is higher than that of gasoline (2276 K) [6] and dPA þ _  4Cf 
mu  Cp rujuj A
the rapid combustion allows very little heat loss to the surround- _
dm boundaries 2 D 2
¼ (2)
ings and hence high instantaneous local temperatures are dt dx
produced. Also, the high auto-ignition temperature of hydrogen
allows higher compression ratios to be explored in a hydrogen
engine compared to a conventional one [10]. Again following [13],
dðmeÞ dV X    
a compression ratio of 9:1 was used at 2500 rpm. ¼ P þ _
mH  hAs Tfluid  Twall (3)
dt dt
Due to the low density of hydrogen, the power densities of port- boundaries
fuel-injected hydrogen engines may be diminished relative to
gasoline-fueled engines [17]. One option is to inject hydrogen fuel 4.3. Combustion model
directly into the cylinder at very high pressures [18], but it would be
practically difficult for the injector to survive such an extreme A two-zone combustion methodology, dividing cylinder into an
thermal environment of combustion chamber over a prolonged unburned zone and a burned zone, was employed to model
operation period. An early study by Jorach et al. [21] has suggested combustion with the assumptions and details explained in the
that this could lead to higher NOx emissions relative to the non- following. At the start of combustion, all the cylinder contents
direct injection systems due to the relatively short fuel/air mixing were in the unburned zone. At each subsequent time step,
time. However, more recent studies using new concepts of mixture a mixture of fuel and air was transferred from the unburned zone
formation have shown direct injection can be used to its advantage to the burned zone. The burn rate was directly predicted from
and lower NOx emissions [6,8e10]. A port-fuel-injected system flame speed correlation (predictive combustion) instead of
would require little modifications to the combustion chamber imposing an experimentally-fitted Wiebe function (non-predictive
design during transitional period from gasoline to hydrogen combustion). If a non-predictive approach were used, the
engines [8,17]. As a result, a port-fuel-injected engine was used in prescribed burn rate would have been the same irrespective of the
[13] and therefore in the present study. cylinder conditions. While the unburned mixture of fuel and air
was entrained into the flame front at a rate proportional to the
4.2. Governing equations flame speed, the burn rate became proportional to the amount of
unburned mixture behind the flame front as shown in Equations
In the present computations, the entire system was divided into (4)e(6). The Taylor micro-scale length l and the turbulent inten-
many discrete volumes that were connected by boundaries. The sity were set to 0.1 (normalized by cylinder bore) and 1
scalar quantities such as pressure, temperature, density, internal (normalized by average piston speed) respectively. The commonly
energy, enthalpy, species concentration were assumed to be used turbulent burning velocity model originally developed by
uniform over each volume and were calculated for the center of the Blizard and Keck [28] and Hires et al. [29] is used here (more
volume. The vector variables such as mass flux, velocity, mass details and comparison with other available models can be found
fraction flux were calculated at each boundary. in [22,30]). Thus, the present predictive approach took into
Simultaneous equations of continuity, momentum and energy account the operating conditions such as cylinder geometry, spark
as shown in Equations (1)e(3) were solved with all the quantities timing, air motion, and fuel properties.
averaged across the flow direction (1-dimensional). Continuity and
dme
energy equations yielded the mass and energy for the next time ¼ ru Ae ðST þ SL Þ (4)
step, and density was calculated with a known volume. The solver dt
 
dmb me  mb
¼ (5)
Table 2 dt s
Specifications of the IC engine modeled in the present study.
l
Type Four-stoke, single-cylinder, spark-ignited,
s¼ (6)
naturally-aspirated, port-fuel-injected engine SL
Fuel Hydrogen
Number of cylinders One Once the unburned fuel and the associated air were transferred
BoreStroke 85  95 mm from the unburned zone to the burned zone, a chemical equilib-
Displacement volume 530 cm3 rium was carried out for the entire burned zone. This calculation
Compression ratio 9:1 considered all the atoms of each species (H, O, N) present in the
Engine speed 2500 rpm
burned zone at that time and calculated the equilibrium
650 S.K. Vudumu, U.O. Koylu / Energy 36 (2011) 647e655

concentration of combustion products (N2, O2, H2O, H2, NO, NO2).
The species concentrations depended on the burned zone
temperature and, to a lesser degree, pressure. With the new
composition of burned zone, the internal energies of each species
and the complete zone were computed. By energy conservation,
the new unburned and burned zone temperatures were also
obtained.
To account for the distinctive properties of hydrogen, especially
the higher flame speed, hydrogen fuel-specific flame speed model
needed to be incorporated into the computations. Changes in the
flame speed correlation were necessary because the flame speed
model offered in GT-POWER is only sufficient for several hydro-
carbons. It is worth noting that, even for methane, the equations
and constants do not accurately correlate with the data over the
entire range of temperature and pressure relevant to engine oper-
ation [31]. Consequently, a new correlation for flame speed appli-
cable for hydrogen combustion was adopted in the present study
based on the equations and the comprehensive literature review
done by Knop et al. [32]. Using this fuel-specific model, the varia-
tion of hydrogen and gasoline flame speeds with respect to
equivalence ratio (at atmospheric pressure and temperature) are
illustrated in Fig. 1, which clearly demonstrates that hydrogen has
a very high flame speed compared to gasoline.
The emissions of oxides of nitrogen (NO and NO2 ¼ NOx) were
predicted based on the three step extended Zeldovich mechanism.
In principle, nitrogen oxides (NOx) are the only harmful engine-out
emissions but the burning of lubricating oil in the combustion
chamber produces (COx) carbon oxides and (HC) hydrocarbons at
near-zero levels [9]. As reflected in the Zeldovich mechanism,
production of NOx in an engine mainly depends on the combustion
temperature and the oxygen availability.
5. Results and discussion
5.1. Comparison of hydrogen and gasoline IC engines
Before a comparison to experimental data, the computational
model for hydrogen engine was first compared to the one for
a gasoline engine with the same geometry and operating condi-
tions given in Table 2. This exercise is useful to understand the
general features of a hydrogen engine relative to a traditional
engine. Cylinder pressure variations with crank angle as well as
with relative cylinder volume are shown in Fig. 2 for an equivalence
ratio of 0.9 for both types of engines (at wide open throttle and
MBT). As expected, in comparison to the gasoline engine, hydrogen
engine has a higher rate of pressure rise and a higher maximum
pressure in the cylinder due to a significantly higher burning speed.
For the conditions considered here, the peak pressure in the
hydrogen engine was 45 bars at 14 crank angle compared to 38 bar
at 28 in the gasoline engine. The PeV diagram also demonstrates
that the heat addition process in the hydrogen engine takes place at
nearly constant volume similar to the Otto cycle due to much faster
combustion.
Table 3 compares two other relevant parameters, the ignition
delay (crank angle degrees for the first 2% of the total heat release)
50
Hydrogen
40
Gasoline
Pressure (bar)

Recirculating a portion of the exhaust gases back into the intake 30

manifold, which is called (EGR) exhaust gas recirculation, is
a convenient way to displace excess air, at the same time to increase
the specific heat capacity of the mixture in the cylinder, and hence 20
to lower in-cylinder temperatures for the same amount of heat
addition [9,17]. This will reduce NOx emissions, the possibility of
pre-ignition, knock, and backfire. Past studies have also shown that 10
EGR is an effective way to reduce NOx emissions in hydrogen-fueled
IC engines [16,17,20]. In order to vary EGR level and quantify its
effect on exhaust emissions, the present model was supplemented 0
with a (PID) proportional-integral-derivative controller, which will -90 0 90 180 270 360
be discussed more later on.
Crank Angle (degrees)

50
3 Hydrogen
Gasoline
40
2.5
Laminar Flame Speed (m/s)

Pressure (bar)

2 30

1.5 20

1 Hydrogen 10
Gasoline
0.5
0
0 0 0.2 0.4 0.6 0.8 1
0 0.5 1 1.5 2 2.5 V/Vmax
Equivalence Ratio
Fig. 2. Pressure vs. crank angle and pressure vs. volume diagrams for hydrogen- and
Fig. 1. Laminar flame speeds of hydrogen and gasoline at various equivalence ratios. gasoline-fueled spark-ignition engines.
 S.K. Vudumu, U.O. Koylu / Energy 36 (2011) 647e655 651

Table 3 8
Computed combustion properties of hydrogen in comparisons to those of gasoline
for the same engine operating conditions. 7
Fuel Ignition delay Burn duration
(for the first 2% of (for 0e90% of the 6

Brake Power (kW)

the total heat release) total heat release)
5
Hydrogen 6 crank angle 22.4 crank angle
Gasoline 13.6 crank angle 36.4 crank angle 4

3
and combustion duration (crank angle degrees for 0e90% of the
total heat release), for the hydrogen and gasoline engines under the 2
Simulations
identical operating conditions mentioned above. In terms of crank
1 Experiments
angle, the ignition delay and combustion duration of the hydrogen
engine were approximately 50% lower than those of the gasoline 0
engine. The faster ignition and shorter combustion duration were 0 0.2 0.4 0.6 0.8 1 1.2
responsible for the optimal spark timing (MBT) to be close to the
Equivalence Ratio
top dead center (8 before TDC).
Fig. 4. Comparison of predicted and measured brake powers vs. equivalence ratio.
5.2. Model validation e comparison of simulations to experiments

of the combustion, the simulations were in good agreement with

As discussed in the combustion modeling section, the burn rate
the experiments, correctly predicting the start of the combustion as
in this study was computed based on the flame speed instead of
well as the value and the timing of the peak heat release rate.
fitting a Wiebe function to the experimental data [18]. This
The variation of brake power with equivalence ratio for the
fundamental approach can therefore predict the combustion burn
hydrogen engine under investigation is shown in Fig. 4. As the
rate in any hydrogen engine because it accounts for the changes in
equivalence ratio (or the amount of fuel injected) was increased,
engine conditions. The apparent heat release rate was calculated
the power generated by the engine increased. The simulations
using Equation (7) for both the burned and unburned zones put
again captured this trend and agreed well with the experimental
together.
data as the difference was less than 8% at higher equivalence ratios
 ! and less than 15% at lower equivalence ratios. The latter observation
dVtot d mtot etot;s
p  Qtot  was attributed to the very small brake powers produced at such
dt dt extremely fuel-lean mixtures. The brake power was 7.4 kW at an
Heat release rate ¼ (7)
mf ;tot  LHVi equivalence ratio of 0.84, which was the maximum value consid-
ered during the experiments [13] due to the limitation of backfiring.
The predicted heat release rates are compared to the measured
Fig. 5 displays the change in the cylinder peak pressure with
values reported by [13] in Fig. 3 as a function of crank angle for the
brake power (or equivalence ratio) and the comparison of simula-
maximum equivalence ratio of 0.84. It should be noted that the heat
tions with measurements. The peak pressure was found to increase
release rate calculation by [13] was done using the polytropic index
almost linearly with brake power. For the maximum brake power of
method [33] by neglecting the wall heat transfer and using
7.4 kW, the peak pressure was approximately 44 bar. As mentioned
a constant polytropic index value of 1.33, which could explain some
before, such relatively high peak pressures in the cylinder are due
of the differences observed between simulations and experiments.
to the fast combustion process that causes high rate of pressure rise.
The results indicated that faster burning speed of hydrogen caused
The maximum rate of pressure rise was 2.2 bar per degree crank
relatively high rate of heat release in a small time interval. Specif-
angle. The predicted peak pressures agreed well (within 10%) with
ically, the heat release became noticeable about 3 before the TDC,
the experiments at medium to high brake powers. There were again
reached a peak value of about 84 J/degrees at 10 after TDC, and
nearly diminished around 15 . Aside from a small delay at the end
60
Simulations
100
50 Experiments
Heat Release Rate (J/deg CA).

Simulations
Peak Pressure (bar)

80 Experiments
40

60
30

40 20

20 10

0 0
-30 -15 0 15 30 45 60 0 1 2 3 4 5 6 7 8
Crank Angle (degrees) Brake Power (kW)

Fig. 3. Predicted heat release rates and comparison to experimental data for the Fig. 5. Comparison of predicted and measured peak in-cylinder pressures vs. of brake
hydrogen engine simulated in this study. power.
652 S.K. Vudumu, U.O. Koylu / Energy 36 (2011) 647e655

40 600
Simulations Simulations
Experiments 500 Experiments

Exhaust Temperature ( C)
Brake Thermal Efficiency (%)

o
30
400

20 300

200

10
100

0
0 0 1 2 3 4 5 6 7 8
0 1 2 3 4 5 6 7 8
Brake Power (kW)
Brake Power (kW)
Fig. 7. Comparison of simulated and measured exhaust gas temperatures.
Fig. 6. Comparison of simulated and measured brake thermal efficiencies.

observed differences between the predicted and measured values

some differences at the two lowest brake powers of 1e2 kW for
which the equivalence ratio of the hydrogen/air mixture in the
cylinder was less than 0.3, corresponding to extremely lean
conditions.
Brake thermal efficiency of the hydrogen engine was computed
using the Chen-Flynn friction model [31] shown in Equation (8).
The variations of brake thermal efficiency with brake power are
shown in Fig. 6 for the simulations and the experiments, both of
which resulted in an initial increase in efficiency with brake power
followed by gradual leveling off. The maximum brake thermal
efficiency was nearly 30% for the cylinder compression ratio of 9:1.
Indicated thermal efficiency would be a more relevant comparison
as this study was more concerned with the combustion process
instead of the power transmission. Nevertheless, the experimental
data of [13] were obtained for brake thermal efficiency (after fric-
tion) instead of indicated thermal efficiency (before friction).
FMEP ¼ 0:7 þ 0:008  Pmax cylinder þ 0:12  Speedmean piston
þ 0:0015  Speed2mean piston
Because the friction conditions of the actual engine were unknown,
it was necessary to adjust the constants in the friction model (as
given in the equation) for a meaningful comparison. While this did
not affect the overall computed trends, it somewhat reduced the
magnitude of brake thermal efficiency at each brake power and
resulted in good agreement between the simulations and
measurements at all operating conditions. After this consideration,
the maximum deviation of the computations from the experiments
was less than 10%.
In the present study, the standard Woschni’s equation [31] was
used to predict the heat transfer between the combustion gases and
were still satisfactory within these uncertainties.
NOx emissions from the hydrogen engine are shown in Fig. 8 as
a function of brake power, which is directly tied to the equivalence
ratio (Fig. 4). Although the experiments [13] only reported NO
while the simulations predicted both NO and NO2, the rational for
their comparison was based on the fact that a major portion (about
95%) of NOx emissions involved NO. Because of its high sensitivity
to the temperature, NOx formation was negligible below a brake
power of 4.5 kW or an equivalence ratio of 0.5. The rapid increase in
NOx after this operating condition limits the usage of hydrogen
engines to low powers/equivalence ratios. Considering that the
primary advantage of a hydrogen engine is near-zero emission, the
brake power for the hydrogen engine considered here should be
limited to 5 kW in order to satisfy the (CARB) California Air
Resources Board (LEV II) Low Emissions Vehicle II (SULEV) standard
for Super Ultra Low-Emissions Vehicles [9]. Note, however, that the
potential to expand the power band while maintaining low NOx
emissions still exists by considering other advanced methods
(8) including the possibility of stoichiometric operation with after
treatment through the use of a three-way catalyst [9].
The measurements in Fig. 8 revealed a maximum NO emission
of 7000 ppm at a brake power of about 7.4 kW (corresponding to an
equivalence ratio of 0.84) and a slight decrease after this peak. The
10000
Simulations
Experiments
8000
NOx (ppm)

the cylinder walls. Subramanian et al. [13] also reported thermo- 6000
couple measurements of exhaust gas temperatures. Unfortunately,
they did not specify the location of the thermocouple along the
length of the exhaust pipe and the pipe properties. On the other 4000
hand, the temperature of the gases at the end of the exhaust runner
was obtained during the present simulations based on reasonable
2000
assumptions about the pipe properties and thermocouple location.
The exhaust temperatures predicted in this manner relative to
those measured are illustrated in Fig. 7. The temperature of the gas 0
stream flowing in the exhaust pipe increased from about 300  C to 0 1 2 3 4 5 6 7 8
500  C with increasing brake power or equivalence ratio. In view of
Brake Power (kW)
the fact that there were some unknown experimental parameters
that could not be exactly specified during the computations, the Fig. 8. Predictions of NOx emissions as a function of brake power.
 S.K. Vudumu, U.O. Koylu / Energy 36 (2011) 647e655 653

Fig. 9. Hydrogen engine model modified with a PID controller to vary EGR level.

computations based on the extended Zeldovich mechanism fol-
lowed the data closely. For example, the predicted maximum NOx
concentration was 7300 ppm, which deviates only 4% from the
measured value. As the stoichiometric condition was approached,
some of the formed NOx dissociated due to a reduction in oxygen
amount and an increase in free radicals at the highest combustion
temperatures. The present simulations of NOx emissions were in
agreement not only with the experimental data of Subramanian
et al. [13] on a single hydrogen engine but also with the technical
review of White et al. [9], who compiled tailpipe emission data
from several different studies.
5.3. Model utilization - effect of EGR on NOx emissions
also the possibility of pre-ignition, knock and backfires. To induce
variable EGR levels, the present hydrogen engine model was
modified as shown in Fig. 9. A (PID) proportional-integral-deriva-
tive controller was used with a 15 mm-diameter throttle valve to
control the amount of EGR through the EGR circuit. An appropriate
control system was essential to supply the desired amount of
diluent back into the cylinder.
10000
Simulations
8000 Experiments

After its development and validation, the computational model

N O x (p p m )

could be utilized for investigating various aspects of hydrogen- 6000

fueled engines. One possibility is to utilize this approach to simu-
late burning of mixtures of hydrogen and another traditional fuel
4000
such as gasoline or natural gas in an IC engine [34] by using an
appropriate combustion model. Another possibility is to compu-
tationally study the reduction in NOx as a function of EGR level
2000
which will be briefly explored here.
Production of NOx depends on the combustion temperature and
the oxygen availability. Injecting a portion of the exhaust gases back 0
to the intake manifold, called (EGR) exhaust gas recirculation, 0 5 10 15 20
displaces excess air, increases the specific heat capacity of the
EGR %
mixture, and lowers in-cylinder temperatures for the same amount
of heat addition. This, in turn, reduces not only NOx emissions but Fig. 10. Variations of NOx emissions with (EGR) exhaust gas recirculation.
654 S.K. Vudumu, U.O. Koylu / Energy 36 (2011) 647e655
The effect of EGR on NOx emissions from the engine simulated in
this investigation is quantified in Fig. 10 at the maximum equiva-
lence ratio (0.84) considered. The computations revealed nearly
linear decrease in NOx concentrations from 7300 ppm to 800 ppm
when the EGR level was increased from 0 to 16%. Overall, the
predictions reasonably agreed with the experiments from another
study [35] under nearly identical engine conditions. The nearly-
linear decrease in the predicted amounts of NOx with relatively low
percentages of EGR is also consistent with the measured trends
reported by Verhelst et al. [10]. Note that if the EGR level is
increased further, it is expected that there will be less decrease in
NOx that may question implementation of EGR approximately
above 30%. While the observed nearly an order of magnitude
reduction in NOx with only 16% EGR was significant, it also
compromised the engine performance: At this maximum
percentage of EGR considered here, the brake power, maximum
pressure, and brake thermal efficiency decreased by 20%, 13%, and
13%, respectively. The application of post-combustion methods
such as (TWC) three-way catalytic converters could further reduce
the raw NOx concentrations predicted above for achieving near zero
exhaust-out emissions [10].
6. Summary and conclusions
Hydrogen is a viable fuel for use in internal combustion engines.
The unique combustion characteristics of higher flame speed,
wider flammability limits and easier ignition of hydrogen allow
cleaner and more efficient engine operations at low engine loads
but present difficulties at higher loads. In this study, engine simu-
lations were employed to study the performance, combustion and
emission characteristics of a hydrogen-fueled engine. In particular,
hydrogen fuel-specific predictive sub-models were incorporated
into the quasi-dimensional simulations. The improvements
included application of a flame speed model that is exclusively
accurate for hydrogen fuel/air mixtures and a predictive burn rate
model that can be used for any hydrogen engine. The computa-
tional predictions were then compared to independent and well-
documented experimental data from [13] in order to evaluate
accuracy and suitability for widespread implementation.
The simulations generally agreed well (typically 10% difference)
with the measurements under similar engine operational condi-
tions, validating the predictive ability of the present model. In
particular, the variations of peak in-cylinder pressure, heat release
rate, brake power, brake thermal efficiency, exhaust temperature,
and NOx emissions were predicted close to the measured values
within experimental and computational uncertainties. NOx
concentrations in the engine exhaust were negligible at lower
equivalence ratios but they sharply increased after an equivalence
ratio of 0.5, limiting the brake power of the hydrogen engine
considered here to 5 kW.
After validation, the simulations were utilized to quantify the
effect of (EGR) exhaust gas recirculation for lowering NOx emissions
by designing and adding a (PID) proportional-integral-derivative
controller to the hydrogen engine model. Similar to the gasoline
engines, EGR was found to be an effective method to reduce NOx
emissions in hydrogen engines. For example, a 16% EGR level at an
equivalence ratio of 0.84 in the present hydrogen engine resulted in
nearly an order of magnitude reduction in NOx.
The results extended the use of GT-POWER software, which is
already an industry-standard for designing gasoline and diesel
engines, to hydrogen engines by properly accounting for the
distinctive characteristics of hydrogen during simulations. This is
expected to lead to improved designs of hydrogen engines, shorten
the development time of alternative-fueled and hybrid vehicles in
the automotive industry.
Acknowledgments
The authors would like to thank the National University Trans-
portation Center at Missouri S&T for partial funding, Dr. James
A. Drallmeier of Missouri S&T for his valuable comments, and GT-
POWER team at Gamma Technologies, Inc. for their technical
support.
References
[1] Penner SS. Steps toward the hydrogen economy. Energy 2006;31:33e43.
[2] Neef HJ. International overview of hydrogen and fuel cell research. Energy
2009;34:327e33.
[3] Schafer A, Heywood JB, Weiss MA. Future fuel cell and internal combustion
engine automobile technologies: a 25-year life cycle and fleet impact
assessment. Energy 2006;31:2064e87.
[4] Negurescu N, Panna C, Popa MG, Soare D. Aspects regarding the combustion of
hydrogen in spark ignition engine. SAE Technical Paper 2006e01-0651.
[5] Li H, Karim GA. Knock in spark ignition hydrogen engines. Int J Hydrogen
Energy 2004;29:859e65.
[6] Verhelst S, Wallner T. Hydrogen-fueled internal combustion engines. Prog
Energy Combust Sci 2009;35:490e527.
[7] Berry GD, Pasternak AD, Rambach GD, Smith JR, Schock RN. Hydrogen as
a future transportation fuel. Energy 1996;21:289e303.
[8] Verhelst S, Demuynck J, Sierens R, Huyskens P. Impact of variable valve
timing on power, emissions and backfire of a bi-fuel hydrogen/gasoline
engine. Int J Hydrogen Energy 2010;35:4399e408.
[9] White CM, Steeper RR, Lutz AE. The hydrogen-fueled internal combustion
engine: a technical review. Int J Hydrogen Energy 2006;31:1292e305.
[10] Verhelst S, Sierens R, Verstraeten S. A critical review of experimental research
on hydrogen fueled SI engines. SAE Technical Paper 2006e01-0430.
[11] Verhelst S, Sierens R. Hydrogen engine-specific properties. Int J Hydrogen
Energy 2001;26:987e90.
[12] Yuksel F, Ceviz MA. Thermal balance of a four stroke SI engine operating on
hydrogen as a supplementary fuel. Energy 2003;28:1069e80.
[13] Subramanian V, Mallikarjuna JM, Ramesh A. Performance, emission and
combustion characteristics of a hydrogen fueled SI engine-an experimental
study. SAE Technical Paper 2005e26-349.
[14] Verhelst S, Maesschalck P, Rombaut N, Sierens R. Efficiency comparison between
hydrogen and gasoline on a bi-fuel hydrogen/gasoline engine. Int J Hydrogen
Energy 2009;34:2504e10.
[15] Hu E, Huang Z, Liu B, Zheng J, Gu X, Huang B. Experimental investigation on
performance and emissions of a spark-ignition engine fuelled with natural
gas-hydrogen blends combined with EGR. Int J Hydrogen Energy 2009;34:
528e39.
[16] Das LM. Near-term introduction of hydrogen engines for automotive and
agricultural application. Int J Hydrogen Energy 2002;27:479e87.
[17] Verhelst S, Landtsheere JD, Smet FD, Billiouw C, Trenson A, Sierens R. Effects of
supercharging, EGR and variable valve timing on power and emissions of
hydrogen internal combustion engines. SAE Technical Paper 2008e01-1033.
[18] Polasek M, Macek J, Takats M, Vitek O. Application of advanced simulation
methods and their combination with experiments to modeling of hydrogen
fueled engine emission potentials. SAE Technical Paper 2002e01-0373.
[19] Shudo T, Suzuki H. New heat transfer equation applicable to hydrogen-fueled
engines. Proc. ASME ICEF2002e515.
[20] Heffel JW. NOx emission reduction in a hydrogen fueled internal combustion
engine at 3000 rpm using exhaust gas recirculation. Int J Hydrogen Energy
2003;28:1285e92.
[21] Jorach R, Enderle C, Decker R. Development of a low-NOx truck hydrogen
engine with high specific power output. Int J Hydrogen Energy 1997;22:
423e7.
[22] Verhelst S, Sierens R. A quasi-dimensional model for the power cycle of
a hydrogen-fuelled ICE. Int J Hydrogen Energy 2007;32:3545e54.
[23] Errico GD, Onorati A, Ellgas S. 1D thermo-fluid dynamic modelling of an S.I.
single-cylinder H2 engine with cryogenic port injection. Int J Hydrogen
Energy 2008;33:5829e41.
[24] Rakopoulos CD, Kosmadakis GM, Pariotis EG. Evaluation of a new computa-
tional fluid dynamics model for internal combustion engines using hydrogen
under motoring conditions. Energy 2009;12:2158e66.
[25] Zhenzhong Y, Aiguo S, Fei W, Nan G. Research into the formation process of
hydrogeneair mixture in hydrogen fueled engines based on CFD. Int J Hydrogen
Energy 2010;35:3051e7.
[26] Gamma Technologies Inc. GT-POWER V6.2 user’s manual. http://www.gtisoft.
com.
[27] Vudumu SK. Experimental and computational investigations of hydrogen
safety, dispersion, and combustion for transportation applications. Ph.d.
thesis: Missouri University of Science and Technology; 2010.
[28] Blizard NC, Keck JC. Experimental and theoretical investigation of turbulent
burning model for internal combustion engines. SAE Technical Paper 740191.
[29] Hires SD, Tabaczynski RJ, Novak JM. The prediction of ignition delay and
combustion intervals for a homogeneous charge, spark ignition engine. SAE
Technical Paper 780232.
 S.K. Vudumu, U.O. Koylu / Energy 36 (2011) 647e655 655

[30] Verhelst S, Sheppard CGW. Multi-zone thermodynamic modelling of spark-
ignition engine combustion e An overview. Energy Conversion and
Management;50:1326e1335.
[31] Heywood JB. Internal combustion engine fundamentals. New York: McGraw
Hill; 1988.
[32] Knop V, Benkenida A, Jay S, Colin O. Modeling of combustion and nitrogen
oxide formation in hydrogen-fuelled internal combustion engines within a 3D
CFD code. Int J Hydrogen Energy 2008;33:5083e97.
[33] Hayes TK, Savage LD, Sorenson SC. Cylinder pressure data acquisition
and heat release analysis on a personal computer. SAE Technical Paper
860029.
[34] Bysveen M. Engine characteristics of emissions and performance using
mixtures of natural gas and hydrogen. Energy 2007;32:482e9.
[35] Subramanian V, Mallikarjuna JM, Ramesh A. Intake charge dilution effects on
control of nitric oxide emission in a hydrogen fueled SI engine. Int J Hydrogen
Energy 2007;32:2043e56.