Parametric study of instantaneous heat transfer based on multidimensional model in direct-injection hydrogen-fueled engine

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 3 8 ( 2 0 1 3 ) 1 2 4 6 5 e1 2 4 8 0

Available online at www.sciencedirect.com

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

Parametric study of instantaneous heat transfer based on multidimensional model in directinjection hydrogen-fueled engine Khalaf I. Hamada a,d, M.M. Rahman a,b,*, A. Rashid A. Aziz c a

Faculty of Mechanical Engineering, University of Malaysia Pahang, 26600 Pekan, Pahang, Malaysia Automotive Engineering Centre, University of Malaysia Pahang, 26600 Pekan, Pahang, Malaysia c Center for Automotive Research, University of Technology Petronas, Bandar Seri Iskandar, Tronoh, Perak, Malaysia d Department of Mechanical Engineering, College of Engineering, University of Tikrit, Tikrit, Iraq b

article info

abstract

Article history:

This paper presents a parametric study on instantaneous heat transfer of a direct-injection

Received 2 February 2013

hydrogen-fueled engine using a multidimensional model. A simplified single-step mech-

Received in revised form

anism was considered for estimating the reaction rate of hydrogen oxidation. The modified

9 July 2013

wall-function was used for resolving the near-wall transport. An arbitrary Lagrangian

Accepted 12 July 2013

eEulerian algorithm was adopted for solving the governing equations. Experimental

Available online 6 August 2013

measurements were implemented to verify the developed model. They show that the instantaneous heat-transfer model is sufficiently accurate. The influence of engine speed, equivalence ratio, and the start of injection timing were investigated. The flow fields

Keywords: Heat transfer correlation Hydrogen Direct injection Injection timing Equivalence ratio

appeared to have greater size vectors and coarser distribution with an increase of engine speed. A heterogeneous distribution was obtained for an ultra-lean mixture condition (4
0.5), which decreased with an increase of equivalence ratio. There was no pronounced influence of the start of injection on the flow field pattern and mixture homogeneity. Thermal field analysis was used to demonstrate trends in the instantaneous heat transfer. It was observed that there was a crucial distinction between the lean and ultra-lean mixtures as well as the engine speed. Furthermore, a non-uniform behavior was found for the impact of the equivalence ratio on temperature distribution. It is clear that the developed models are powerful tools for estimating the heat transfer of the hydrogenfueled engine. The developed predictive correlation is highly accurate in predicting the heat transfer of the hydrogen-fueled engine, focusing on the equivalence ratio as a governing variable. Copyright ª 2013, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.

1.

Introduction

The growth of alternatives to the fossil-fueled internal-combustion engines (ICEs) for personal transportation offers

significant prospects. They promise to reduce the dependence of the world on fossil fuels (depleted resources) and their adverse environmental effects. The challenge is to find highly efficient ways to produce, deliver, and use the energy that

* Corresponding author. Faculty of Mechanical Engineering, University of Malaysia Pahang, 26600 Pekan, Pahang, Malaysia. Tel.: þ60 94246239; fax: þ60 94246222. E-mail addresses: [email protected], [email protected] (M.M. Rahman). 0360-3199/$ e see front matter Copyright ª 2013, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ijhydene.2013.07.051

12466

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 3 8 ( 2 0 1 3 ) 1 2 4 6 5 e1 2 4 8 0

enhances the quality of life. The environment and climate are under threat. The energy carrier hydrogen is an alternative to fossil fuels which have the potential to achieve the required goals [1,2]. Development of internal combustion engines (ICEs) with hydrogen as an alternative fuel has made considerable progress [3e11]; however, many challenges still remain. Therefore, much work is under way in an attempt to overcome these hurdles [3,11e13]. An excellent technical and historical survey of the literature on H2ICE research has been accomplished by several researchers [3,14,15]. These reviews deal with the major effects, difficulties, technical problems, efficiency improvement and economy of utilizing hydrogen as fuel for ICEs. Within this field, many studies have been presented which characterize the combustion process, overall performance and emissions of hydrogen-fueled engines [16e20]. There are few, however, which investigate heat transfer [11]. Heat transfer is one of the noteworthy issues in the study of ICEs. It has a direct effect on the main distinguishing parameters for the engine, such as in-cylinder pressure and temperature. Effort has been devoted to developing a highly efficient global heat transfer model for ICE applications [21]. An overview of the state of the art shows an abundance of correlations for estimation of heat transfer in ICEs [22]. Since 2000, the heat-transfer issue has arisen as one of the crucial features in the modeling of the hydrogen-fueled engine (H2ICE). The importance of investigating heat transfer for H2ICE can be ascribed to the vast differences in its properties compared with hydrocarbon fuels [23,24]. The differences relate not only to the quantity of the heat liberated from the combustion process but also the unique behavior of hydrogen combustion [25]. Several efforts have been directed towards developing realistic thermodynamic models of the engine cycle. These types of model enable a cheap and fast optimization of engine settings for operation with hydrogen fuel [26]. Several submodels are required to solve the equations governing the conservation of energy and mass, including the combustion, turbulence, and heat transfer models. Numerous correlations in the literature have been developed to model the heat transfer in fossil fuel engines [27e29]. Existing models are based on the empirical formulation for heat transfer in hydrocarbon-fueled ICEs. However, the phenomenon of heat transfer in hydrogen and fossil-fuelled ICEs is different in the rate and behavior of heat release, as well as in the heat loss due to the different properties [25,26,30e32]. A new heat transfer model applicable to H2ICE was proposed by Shudo and Suzuki [33] based on Woschni’s model. It contains two calibration parameters that depend on the ignition timing and equivalence ratio (4). Therefore, dependency of heat transfer correlation on the equivalence ratio is stated to be the subject of further studies [11,26,34]. It is not possible to predict accurately the trends with a variation in the equivalence ratio using a recalibration of Annand and Woschni’s correlations [26]. Applicability of a quasi-dimensional heat transfer model for a hydrogen-fueled ICE (H2ICE) was presented by Nefischer et al. [35]. This model is based on the realistic description of the characteristic velocity for the turbulent flow field proposed by Schubert et al. [36]. The predicted heat transfer was in reasonable agreement with experiments for port injection;

however, the direct-injection effect was not predictable. Recent developments in the analysis of heat transfer in a H2ICE have focused only on the instantaneous spatially averaged. This is due to the importance of the heat transfer correlation for modeling the engine cycle [26,37]. Multidimensional modeling represents the approach for modern computational fluid dynamics (CFD) codes. It provides very valuable and complete information compared with experiments of the in-cylinder flow pattern. It is difficult to simulate adequately the heat fluxes through the cylinder walls due to the unsteady and highly inhomogeneous temperature distribution of the cylinder charge, as well as the uncertainty that exists in the determination of the initial and boundary conditions of the problem [38]. CFD codes become an efficient tool for the calculation of heat transfer in ICE applications. Most current CFD codes utilize either a wallfunction or a near-wall modeling approach to describe the flow conditions near the wall in the heat transfer calculations. An abundance of literature has proposed various heat transfer models, which are based on the wall-function hypothesis. The basic formula for all available models was developed by Launder and Spalding [39]. This approach is extensively considered in multidimensional simulations, because it provides a robust technique of computing the thermal boundary layer without increasing the computational nodes that are placed within this layer [40]. Of course, various formulations can be found where the number of nodes in the boundary layer is increased and there, the law-of-the-wall is applied with satisfactorily results [38]. A critical evaluation of current heat transfer models used in CFD in-cylinder engine simulations reveals that the establishment of a comprehensive wallfunction formulation for estimation of the heat fluxes is included. It was assessed which wall-function formulation is more suitable for each engine type and under what operating conditions. The model of Launder and Spalding [39] revealed its weaknesses in all cases considered, as has also been demonstrated by many other researchers [40]. However, its improved version achieves the best estimation of wall heat loss [41,42]. Parametric investigations become more achievable due to the inclusion of several sub-models, which describe each phenomenon. The three dimensional compressible averaged NaviereStoke’s equations are solved on a moving mesh. Turbulent fluxes are modeled by an eddy viscosity concept using ke3 model [43]. The present study attempts to address the influence of the equivalence ratio on multidimensional instantaneous heat transfer (IHT). IHT is devoted to weighting the influence of operating parameters on heat transfer. Therefore, the parametric study of heat transfer is essential to investigate the trends of IHT. The outcome of the present study is expected to establish a technical contribution for the automotive sectors.

2.

Multidimensional engine model

A multidimensional model is used to characterize the instantaneous heat transfer for a direct-injection hydrogenfueled internal combustion engine (DIH2ICE). This model is based on a computational fluid dynamics approach using the finite volume technique. Fig. 1(a) shows the physical domain

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 3 8 ( 2 0 1 3 ) 1 2 4 6 5 e1 2 4 8 0

12467

Fig. 1 e Physical domain of the multidimensional engine model.

illustrating some of the basic elements of the model used in operation of the engine. This model has a combustion chamber fitted with a piston and intake and exhaust ports, each of which has valves named as exhaust valves and intake valves. The engine possesses a pair of intake and exhaust valves for every cylinder, and these have circular shapes on them for runners. The combustion chamber possesses a number of pistons, each with a cup-shaped molding at the crown that encloses its roof. This model also shows alterations, such as the introduction of new fuel inlets. For the meshing process, the physical domain is divided into three main regions: the intake, exhaust, and the combustion chamber. Each region is composed of several blocks and each block is sub-divided into many smaller volumes. The current model is used to simulate the supply of fuel into the combustion chamber. The snapper technique is used for simulating the motion of the piston and valves. These are considered as solid objects moving through the meshing domain. The computational grid domain is displayed in Fig. 1(b). This computational grid is related to the Proton Campro single cylinder research engine. The current model was used to simulate the fuel supply in the combustion chamber based on a new program namely “SETVELIN”. This program was developed and integrated with the main KIVA code. The adopted algorithm of the newly developed program to simulate the hydrogen fuel injection is presented in a flowchart in Fig. 2. Improved wall function is used to solve transport equations (continuity, momentum and energy) near to the wall. This improved wall function gives the best estimation of wall heat loss [41,42]. The logarithmic law of the wall is employed to define the velocity and temperature profiles in the near-wall regions. qw ¼

r  Cp  nl  F  ðT  Tw Þ Prl  y

(1)

F¼

8 > > <

1:0

for R0
11:05

R0 Prl RPR

for R0 > 11:05

> > : 1kln R0 þBþ11:05ðPrl RPR1Þ

where

Fig. 2 e Flowchart for the injection program SETVELIN.

12468

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 3 8 ( 2 0 1 3 ) 1 2 4 6 5 e1 2 4 8 0

R0 ¼ c0:25  k1=2  y=nl ; m r is the density of the in-cylinder gaseous; Cp is the specific heat for in-cylinder gaseous; nl is the laminar kinematic viscosity for the in-cylinder gaseous; y is the distance from the wall in the normal direction; Prl is the laminar Prandtl number; RPR is the reciprocal of the turbulent Prandtl number; T is the gaseous temperature; Tw is the wall temperature; B, cm and k are the Karmann’s model’s constants; k is the kinetic energy.

2.1.

Governing equations

weak coupling between the species and turbulence equations with the flow field solution. Determination of the time step Dt in phase B is based on the accuracy rather than stability conditions. This is due to the implicit difference for the diffusion terms as well as the sub-cycling of the convective terms. Hence, there are no stability restrictions on (Dt). The flow field is frosted during phase C, as the computational mesh moves to a new position. Therefore, the convective transport associated with the moving mesh relative to the fluid is calculated. The convection time step is derived using the Courant stability condition, as in Eq. (3): ur  Dt <1 Dx

(3)

where The governing equations are discretized within space and time domains. The discretization process involves a combination of the arbitrary LagrangianeEulerian (ALE) and variably implicit time-discretization methods [44]. The time derivatives are approximated by the first-order difference. Accordingly, any quantity (J) within governing equations is formulated as Eq. (2): vJ Jnþ1  Jn y vt Dtn

(2)

where Dtn ¼ tnþ1  tn is the time step; n ¼ 0, 1, 2. is the number of iteration cycle. Each computational cycle is executed in three separated phases: A, B, and C. Phase A is a Lagrangian method, where the mesh and control volumes are not moved. The impact of the fuel injection and chemical reactions on gas quantities is calculated within phase A. Phase B is also a Lagrangian method, but the control volume moves with the fluid. Fluid diffusion computations are involved within phase B. Therefore, the governing equations are solved by Lagrangian method using a finite volume scheme. Phase C is a rezoning/ remapping phase in which the fluid field is static and the mesh moves to a new position. The Eulerian method or the rezoning process is considered within phase C. It includes the movement of the grid to new positions as well as the exchange of fluxes of mass, momentum, energy, and turbulence quantities. The combination of these three phases is equivalent to the process whereby the mesh moves to a new position in a time step (dt) and the fluid also moves. The adopted procedure for solving the governing equations is the SIMPLE technique [45]. This approach solves the individual equations utilizing the conjugate residual method [46]. The SIMPLE algorithm adopts an iterative routine. It consists of two steps: the pressure field frosts and solving the other flow quantities. The acquired terms are frosted and solve the implicit finite difference equations for the pressure correction term. Furthermore, it compares the predicted and corrected pressure fields as a condition for convergence criteria. The Lagrangian equations for phase B quantities, namely: species density, velocity, temperature, pressure, turbulent kinetic energy, and turbulent dissipation rate are solved using the conjugate residual method [46]. The species and turbulence equations are solved independently. This is because of the

ur is the fluid velocity relative to the grid velocity; Dt is the size of time step; Dx is the discretization length for the space domains. The updating of the cell properties is carried out through employing the state equations, and the individual species densities are summed to estimate the density. The QuasiSecond-Order Upwind (QSOU) scheme is used for approximating the difference equations. It is evident that this scheme achieves greater accuracy [47]. The conservation equations are discretized by the finite volume technique on an arbitrary hexahedral mesh applying the ALE approach [48,49]. The spatial differences are created on a mesh, which sub-divides the computational field into a number of small hexahedrons. The vertices are moved with the fluid (Lagrangian), which is kept fixed (Eulerian). In the ALE method, no interpolation is required for determining vertex motion in the Lagrangian phase of the calculation. A staggered grid mode is adopted in the multidimensional model. That means, thermodynamic ( p, r, T, I, rm) and turbulent (k, 3 ) quantities are stored at the cell centers while velocity is stored at vertexes at the cell corners. Quantities needed at points where they are not fundamentally located are obtained by averaging neighboring values. Spatial difference is carried out by integrating the differential term over the volume of a standard cell.

2.2.

Boundary and initial conditions

The CFD computations consider the compression stroke, including the hydrogen injection process and the combustion/ expansion phase. Initial and boundary conditions of the computational model are derived from experimental results of engine measurements, as well as from a preliminary cycle simulation. Turbulent law-of-wall velocity conditions with fixed temperature walls are used in this study. The normal gas velocity is equaled to the normal wall velocity while the tangential components are determined by matching to a logarithmic profile. Constant wall temperature is used as the condition at the solid surfaces throughout the computation. The heat flux to the walls is calculated using a modified Reynolds analogy formula. Uniform values are assumed for the in-cylinder pressure and temperature, species concentration, turbulent kinetic energy, and turbulence length scale at the time of the start of the computations. A preliminary cycle

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 3 8 ( 2 0 1 3 ) 1 2 4 6 5 e1 2 4 8 0

simulation has been executed to generate the initial conditions for pressure, temperature, and species concentration. The initial value of turbulent kinetic energy (k) is assumed to be 10% of the total kinetic energy based on mean piston speed. The initial value of dissipation rate (3 ) is calculated using standard procedures [50]. In addition, the injection parameters (injection velocity, mass of hydrogen fuel, and injection duration) are used as boundary conditions at the injection region, as shown in Fig. 1. A multidimensional model is executed for engine speed 2000
rpm
5000 with an interval of 1000 rpm, and equivalence ratio 0.3
4
0.9 with an interval of 0.2, and start of injection 70 deg BTDC (before top dead center)
SOI (start of ignition)
130 deg BTDC with an interval of 30 deg. The same operating parameters are considered in the one dimension model.

3.

Experimental details

To carry out an experiment that could provide trustworthy data, the most appropriate engine and test cell setup are required. The experimental tests are conducted based on the Proton Campro research engine, which has a single cylinder, spark ignition, four strokes, and a water-cooled engine. The cylinder is made of aluminum, and is equipped with a pentroof that acts as a combustion chamber, with a spark plug located at the center. Moreover, the structure of this cylinder involves dual intake and exhaust valves that are activated with the help of a double overhead camshaft (DOHC). The overall geometry of the cylinder depends upon the Proton Campro 0.4 L engine, as shown in Table 1. This cylinder is fed with the fuel via the injector located at its top, which ultimately is fed by the fuel system comprising hydrogen gas bottles maintained at a pressure of 200 bar, a pressure regulator, pressure gages, and micro-motion flow meter. The pressure of the injected fuel inside the cylinder is also carefully regulated by the piezoelectric transducer located at the top of the cylinder and is bound to constantly measure the internal pressure of the hydrogen gas. This transducer type is the water-cooled Kistler Thermo Comp. Fig. 3(a) illustrates the experimental setup of the integrated fueling systems. The engine cooling system was modified in order to carry out the experiment. The heat generated by combustion is transferred to the open cooling circuit, which is

Table 1 e Proton Campro engine specifications. Parameter

Value

Bore Stroke Displacement Geometric compression ratio Connecting rod length Intake valve open Exhaust valve open Intake valve close Exhaust valve close Intake valve diameter Exhaust valve diameter

76.0 88.0 399.25 14.0 131.0 12 45 48 10 30.0 27.0

Unit mm mm cm3

12469

fed by tap water. A specified heat exchanger is used to transfer the heat. The coolant of the engine is circulated through an external pump (0.18 kW), which is driven by an electric motor. The coolant temperature is maintained at 70  5  C. The oil circulation system comprises the oil pump, filter, and connection pipes with a specified heat exchanger. The oil temperature is controlled by utilizing the same external cooling circuit to be within a range of 90  5  C. To ensure the accuracy of the data obtained in this experiment and to make the working of the model clearer, a crank angle encoder is also used, which determines the pressures via identifying the physical location and angle of the engine. This encoder is provided with a combustion analyzer unit that feeds the pressure data into the crank angle domain. Then, to organize the functioning of the engine, a programmable engine control unit (ECU) (Orbital Inc.) is used, which gives details on: the timings of the injection, its duration, the spark timings, air fuel ratio, and the throttle position. This is connected to the Electronic Remote Interface (ERI) installed in the computer that is used in this experiment to obtain the required data, and this overall system of ECU is powered with a 12 V DC source. Devices to control and measure the speed and power absorption and the amount of fuel utilized were also introduced in this model. A David McClure DC30 eddy current dynamometer was used to control the speed of the engine and absorb the power; its capacity is about 30 kW and its maximum rated speed is about 6000 rpm (Fig. 3(b)). This device is capable of braking and motoring, through setting the speed of the engine to a desired level and letting the torque be variable. GASMET was used for measurements of the fuel utilized; it is a device based on a Zirconia cell measurement. This device gave results showing that the voltage was directly proportional to the oxygen concentration and thus, the measurement of the concentration of the oxygen was helpful in determining an estimation of the equivalence ratio.

4.

Model validation

The collected database from the experimental test rig was used to validate the baseline engine model. The baseline engine model was performed for a test rig of a single-cylinder Proton Campro engine. Grid sensitivity analysis was carried out for a multidimensional model based on the motoring condition at an engine speed of 2000 rpm. The developed model was validated experimentally based on the in-cylinder pressure traces. This was performed for three operating parameters: the engine speed, equivalence ratio, and start of ignition.

4.1.

Grid sensitivity analysis

e mm deg BTDC deg BBDC deg ABDC deg ATDC mm mm

The grid sensitivity is analyzed to eliminate the effects of grid dependency on the model’s prediction. This is performed through reducing the grid element size. The singlecylinder baseline engine was operated by the David McClure dynamometer at a motoring condition of 2000 rpm. The number of cells is defined as a coarse to fine mesh from 33,291 to 52,897. The sensitivity of the results to the spatial

12470

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 3 8 ( 2 0 1 3 ) 1 2 4 6 5 e1 2 4 8 0

Fig. 3 e The test-cell (a) the tested engine coupled with the eddy current dynamometer with control console (b) the fueling system.

resolution is assessed through individually varying the resolution in the cylinder region. The intake and exhaust regions are excluded for the grid sensitivity analysis because these two regions are not relevant during the combustion phase. Fig. 4 illustrates the computational domain with the hexahedron elements for the multidimensional engine model at the TDC position. The developed model is executed using an adaptive time stepping scheme. The time step for the successive cycle is computed considering the solution of the previous cycle in view of various accuracy-related constraints. The maximum time step is selected to be equal to 105 s. The time interval is converted to be equivalent to a unit of crank angle degree. According to the accuracy constraint, the time step is reduced automatically as the mesh is refined.

Fig. 4 e Computational domain with hexahedral element for multidimensional engine model at TDC position.

12471

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 3 8 ( 2 0 1 3 ) 1 2 4 6 5 e1 2 4 8 0

Fig. 5 presents the results of the grid sensitivity analysis. The in-cylinder pressure traces for the experimental and model prediction results at 2000 rpm under motoring conditions are considered to investigate the sensitivity of the mesh size. A significant sensitivity is observed in predictions of peak pressure for a minor difference in the corresponding crank angle. The prediction of in-cylinder pressure traces is approached from the experimental results as the grid size is reduced. The final fine grid distribution (52,897 cells) is the closest to the experimental results. Further refinement in the spatial domain leads to an increase in computational time with a negligible benefit regarding the accuracy of the results. This reveals that the fine mesh size is the best compromise between time and accuracy. Hence, the fine mesh distribution (52,897 cells) is selected for the present study. The requirement for spatial resolution over the temporal resolution is justified by using the fine mesh distribution with the accuracy criterion.

4.2.

Experimental validation tests

The developed model is validated experimentally in terms of the in-cylinder pressure traces. It is performed for various engine speed and equivalence ratio at SOI ¼ 130 deg BTDC under the wide open throttle (WOT) and full-load condition. This injection timing is chosen to simulate the direct-injection case. Further retardation for SOI is unachievable experimentally due to abnormal combustion. Table 2 lists the different experimental conditions used for the model validations. These conditions include both low and high engine speeds, as well as ultra-lean and lean mixture strengths. The engine is tested to yield sufficient torque (misfire limit) and avoid knocking noise. Fig. 6 illustrates the comparisons between the experimental and predicted (simulation) results of the incylinder pressure for different engine speeds and various

Table 2 e The experimental operating points of the cases utilized for validating of the multidimensional model. Case Case Case Case Case

1 2 3 4

Engine speed 1800 1800 3000 3000

rpm rpm rpm rpm

4 0.65 0.97 0.78 0.41

SOI 130 130 130 130

deg deg deg deg

BTDC BTDC BTDC BTDC

Load condition Full Full Full Full

load load load load

mixture strengths under WOT and full-load conditions. It can be seen that there is good agreement between the predicted (simulation) and experimental results for all tested conditions. However, there are some deviations in terms of the incylinder pressure (about 5%) and the locations of peak pressure (around 10%). The deviation of the in-cylinder pressure traces appears during the end of the compression stroke and combustion period. The predicted results are slightly overestimated during the end of the compression stroke. This is due to the uncertainty of the compression ratio [19]. Although the compression ratio for the simulation as well as the experiment is considered constant, the real compression ratio for the experiment appears to fluctuate. Therefore, this causes some uncertainty in the predicted results during this interval. It can be seen that there is an increase in the discrepancy between the predicted and experimental results with initiation of combustion until the expansion stroke. It can also be seen that the predicted results show too slow combustion at the beginning and a rapid combustion near the peak incylinder pressure. Consequently, the initiation timing of the predicted results is retarded compared with the experimental. This is because a simplified single-step oxidation was used that converts the hydrogen and oxygen to water in the multidimensional modeling.

5.

Results and discussion

Multidimensional analysis of IHT for the in-cylinder fourstroke DIH2ICE model is performed for three operational parameters: the engine speed, equivalence ratio, and injection timing. The engine speed is varied from 2000 to 5000 rpm with change steps of 1000 rpm. Equivalence ratio is varied within the range 0.3
4
0.9 with change steps of 0.2. The injection timing is selected within the range 70 deg BTDC
SOI
130 deg BTDC with change steps of 30 deg. The parametric analysis was performed in terms of: the flow field, the thermal field, trends of heat release rate, trends of heat loss, and trends of the heat transfer coefficients.

5.1.

Fig. 5 e Effect of mesh size on the comparison between simulated and experimental in-cylinder pressure under motoring condition at 2000 rpm.

Flow field

The in-cylinder flow field for the DIH2ICE is presented to clarify the fuel-injection process in a gaseous state, and its interaction based on the modified KIVA code. The pattern of the flow field, combined with the history of hydrogen concentration inside the combustion chamber during the injection phase, is used to capture hydrogen injection. Fig. 7(a) presents the flow field of hydrogen concentration during the injection process for engine speeds of 3000 and 5000 rpm,

12472

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 3 8 ( 2 0 1 3 ) 1 2 4 6 5 e1 2 4 8 0

Fig. 6 e Comparison between the experiment and predicted results of the in-cylinder pressure traces for different speeds and mixture strengths under WOT and full-load conditions.

4 ¼ 0.7, and SOI ¼ 100 deg BTDC at 80 deg BTDC. The proper instants are selected to reveal the hydrogen injection based on the boundary conditions of the injection region (SOI ¼ 100 deg BTDC and injection duration of 50 deg). Therefore, it is selected during the opening period of the gas injector. An identical distribution of the hydrogen concentration is observed because the hydrogen fuel in the gaseous state has very low density and remarkably high diffusivity [51,52]. Additionally, the hydrogen fuel is injected within the combustion chamber during the closing period of the valves. There is no flow entering the combustion chamber during this period. Accordingly, the mixing process between the fuel and

fresh air within the combustion chamber is mainly dependent on the physical properties of the fuel. Furthermore, it can be noticed that the velocity vectors have greater size as well as a coarser distribution near to the piston surface at higher engine speeds. This is because of increasing turbulence levels with an increase of engine speed [53,54]. Fig. 7(b) presents the flow field of hydrogen concentration during the injection process for different instants: 3000 rpm, 4 ¼ 0.5 and SOI ¼ 100 deg BTDC. Two instants (70 and 80 deg BTDC) are considered to illustrate the progress of the fuelinjection process with piston movement. It can be seen that adverse behavior occurs as the crank angle is advanced where

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 3 8 ( 2 0 1 3 ) 1 2 4 6 5 e1 2 4 8 0

12473

Fig. 7 e The flow field of hydrogen concentration during the injection process for different engine speed, equivalence ratios, and SOI [ 100 deg BTDC (isometric view).

the gradient of concentration of the hydrogen fuel increases progressively. The maximum hydrogen concentration is increased from 3.2  105 to 3.8  105 as the crank angle is advanced from 80 to 70 deg BTDC, respectively, whereas the minimum concentration remains constant at 1.0  105. This means that the heterogeneity in hydrogen distribution is increased within the computational domain. This is because

of increasing the injected amount of hydrogen fuel as the piston level gradually increases. Even the total amount of hydrogen fuel is increased within the combustion chamber, but it is still not enough to acquire a homogenous distribution. Fig. 7(c) presents the flow field of hydrogen concentration during the injection process for different equivalence ratios of 4 ¼ 0.5 and 0.7, 3000 rpm and SOI ¼ 100 deg BTDC at 80 deg

12474

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 3 8 ( 2 0 1 3 ) 1 2 4 6 5 e1 2 4 8 0

BTDC. The maximum hydrogen concentration 4  104 is obtained at higher mixture strengths 4 ¼ 0.7. It can be seen that the mixture becomes more homogeneous with an increase of mixture strength due to increasing the amount of hydrogen. Therefore, the total amount of hydrogen injected inside the combustion chamber increases per unit time.

5.2.

Thermal field

The thermal field represents a local distribution of the temperature within the computational domain. Fig. 8 presents the description of the temperature contours at the TDC position, which is the ignition timing point. It reveals the essential elements from the combustion chamber, including: the spark plug position, intake ports, exhaust ports, and fuel injection regions. Fig. 9 demonstrates the temperature contours for different engine speeds (2000 and 4000 rpm) and instants (5, 10, 15, and 20 deg ATDC) at 4 ¼ 0.7 and SOI ¼ 100 deg BTDC. These four instants are selected to differentiate the progress of the combustion events as well as to analyze the heat transfer phenomenon. The first instant (5 deg ATDC) is immediately after the start of the ignition process, which reveals a hot spot at the spark plug position. At this stage, the temperature distributions are almost identical for the different engine speeds. This is because the flame propagation is in the initiation stage. At the second instant (10 deg ATDC), the growth of the flame becomes more developed and covers a wider area of the combustion chamber. The temperature levels are also increased as the crank angle advances due to the increase in the amount of burning mixture. These growths and development continuously increase over the next two instants (15 and 20 deg ATDC). It can also be seen that the growth of the flame to cover a wider area rapidly increases with an increase of the engine speed. In addition, the temperature levels increase with an increase of the engine speed. The differences are clearer in terms of the instants at 10 and 15 deg ATDC. The maximum local temperature obtained is 2400 and 2600 K for 4000 rpm at the instants of 10 and 15 deg ATDC, respectively. This is because of the increase in the reaction

Fig. 8 e The temperature contours at TDC position (isometric view).

rate with an increase of the engine speed, which causes a higher level of turbulence [53,54]. This difference in the temperature distributions for both engine speeds continues with the further progress of the piston movement. Also, the difference still exists even in the temperature levels. Additionally, there is a delay in completing the combustion process in the case of 2000 rpm. Again, the direct effect of the engine speed on the turbulence level, which leads to a direct impact on the flame propagation and reaction rate, represents the main reason for this delay. This behavior for the local distribution of the temperature within the combustion chamber produces the same trends for spatial heat transfer. This is because of a direct relationship between the distribution of temperature and the nature of the process of heat transfer. These results are achieved according to the improved wallfunction for the definitions of the velocity and temperature within the region in the vicinity of the walls. Fig. 10 demonstrates the temperature contours for different equivalence ratios (4 ¼ 0.3, 0.5, 0.7 and 0.9) and instants (10 and 15 deg ATDC) at an engine speed of 3000 rpm and SOI ¼ 100 deg BTDC. These two instants are selected based on the observed behavior in Fig. 9. The results here can be classified into the ultra-lean (4 ¼ 0.3 and 0.5) and lean (4 ¼ 0.7 and 0.9) mixture strength ranges. There is a considerable difference in the distribution of the temperature levels and the space of the burnt mixture for the ultra-lean range. It can be seen that the higher local temperature distribution, as well as a wider burning space, is observed at 4 ¼ 0.5. The maximum temperature is obtained (2000 K) at the stronger mixture strength (4 ¼ 0.5) and second instant (15 deg ATDC). This is due to a direct relationship between the flame speed and mixture strength because the hydrogen has a significantly higher dependency for the laminar flame speed on mixture strength condition [17]. Therefore, the combustion rate becomes faster for the stronger mixture condition. At the first instant with the lean mixture strength (4 ¼ 0.7 and 0.9), the temperature contours are almost identical. However, the temperature contours are differentiated for the second instant. The maximum temperature (2600 K) is obtained in the case of the richer mixture strength (4 ¼ 0.9). This is because the laminar flame speed for hydrogen fuel increases dramatically with an increase of the equivalence ratio [17]. Therefore, the richer mixture strength produces a faster burning rate, which means a quicker flame development and thus, higher heat release. Accordingly, the heat transfer rate has to be higher with the richer mixture strength. It is seen that the trend of temperature distribution is non-uniform for the ultralean and lean mixture strengths. This is because of the nature of the combustion model in the multidimensional modeling. It is based on the kinetic controlled and equilibrium reaction mechanisms. These mechanisms for combustion modeling depend on some properties of fuel that are employed within the model formulation. The forward pre-exponential factor (kfr) and the stoichiometric coefficients (am and bm) are the most important properties [55]. These characteristics are usually evaluated empirically to be appropriate for any application. An excessive rate of heat release is produced when the values chosen for these characteristics are too high. However, combustion is not sustained due to insufficient heat release when the

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 3 8 ( 2 0 1 3 ) 1 2 4 6 5 e1 2 4 8 0

12475

Fig. 9 e The temperature contours for different engine speeds and instants at 4 [ 0.7, IT [ TDC, and SOI [ 100 deg BTDC.

considered values are low [55]. The values of these characteristics are considered in the present study according to previous reported works [56,57]. However, these values were revealed to have a degree of uncertainty in the prediction of a laminar flame speed. The predicted flame speed represented the experimental data well only for the 0.55e1.1 equivalence ratio range. Predictions by the global reaction mechanism were considered poor outside this range of equivalence ratio. The poor prediction is attributed to chemical and thermal structural changes in the flame as the stoichiometry varies, which could not be properly accounted for in the global reaction model. Therefore, it is desirable to

clarify the impact of injection timing on the temperature distribution. Fig. 11 demonstrates the temperature contours for different injection timings (SOI ¼ 70 and 130 deg BTDC) and instants (5, 10, 15, and 20 deg ATDC) at an engine speed of 3000 rpm, 4 ¼ 0.7, and IT ¼ TDC. It can be seen that there is no pronounced difference for the temperature distribution between the different injection timings, even with an advanced stage of the combustion process. This is because the injection process is completed with sufficient time prior to the ignition event, even for the retarded injection timing (SOI ¼ 70 deg BTDC). Therefore, there is no effect on heat transfer.

12476

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 3 8 ( 2 0 1 3 ) 1 2 4 6 5 e1 2 4 8 0

Fig. 10 e The temperature contours for different equivalence ratios and instants at engine speed of 3000 rpm, IT [ TDC, and SOI [ 100 deg BTDC.

5.3.

Heat release rate and cumulative heat loss

Heat release and cumulative heat loss during combustion are used to characterize the heat transfer of ICEs. The heat release rate is determined based on the kinetic controlled and equilibrium reactions for multidimensional modeling. The cumulative heat loss is a global integral for the rate of

instantaneous heat loss, which is based on the definition of the modified wall function. Fig. 12(a) reveals the trends in heat release rate and the cumulative heat loss against crank angle for different engine speeds of 2000 and 5000 rpm at (4 ¼ 0.7), IT ¼ TDC and SOI ¼ 100 deg BTDC. The influence of engine speed can be clearly seen during and after the combustion process. The heat release rate is increased with an increase in

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 3 8 ( 2 0 1 3 ) 1 2 4 6 5 e1 2 4 8 0

12477

Fig. 11 e The temperature contours for different SOI and instants at engine speed of 3000 rpm, 4 [ 0.7, and IT [ TDC.

engine speed. The peak rates of heat release were around 65 and 140 J/deg at 2000 and 5000 rpm, respectively. Moreover, the combustion process takes place in a shorter period with an increase in engine speed. The cumulative heat loss appeared to have a sharper gradient during the combustion process as well, as it presented greater values at higher engine speed. At a crank angle of 130 deg ATDC, the cumulative heat loss increased about 10% with an increase in the engine speed from 2000 to 5000 rpm. This is because of enhanced convection heat transfer to the cylinder walls with an increase in engine speed, which boosts the turbulence level and reduces the boundary-layer thickness [53,54]. Fig. 12(b) demonstrates the trends in heat release rate and the cumulative heat loss against crank angle for different equivalence ratios (4 ¼ 0.5 and 4 ¼ 0.7) at an engine speed of 3000 rpm, IT ¼ TDC and SOI ¼ 100 deg BTDC. The selected

values for the equivalence ratio (4) are based on the range of the finest engine operation for DIH2ICE [58]. The peak rates of heat release were around 62 and 116 J/deg at 4 ¼ 0.5 and 4 ¼ 0.7, respectively. It can also be seen that the combustion process takes place in a shorter period with an increase in the equivalence ratio. Moreover, combustion duration is decreased from 18 to 12 deg as the equivalence ratio increases from (4 ¼ 0.5) to (4 ¼ 0.7), due to the slower combustion for the leaner mixture. This is because of increasing flame development due to increasing laminar flame speed [17]. Therefore, the mixture is burned faster because the burning rate depends mainly on the flame development [59,60]. The cumulative heat losses increase with an increase in the equivalence ratio. This is due to the higher heat rejection from the combustion chamber. Richer mixtures produce higher temperatures inside the cylinder (as shown in Fig. 10) and faster flame speeds.

12478

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 3 8 ( 2 0 1 3 ) 1 2 4 6 5 e1 2 4 8 0

was shown that the neglected parameter 4 has a vital impact on the heat transfer for the H2ICE. An integrated program for simulating hydrogen injection in the gaseous state in the DIH2ICE was developed. This program was implemented in the multidimensional model. The flow pattern of a typical early injection was revealed for the DIH2ICE. The flow fields appeared to have greater vector size and a coarser distribution with increasing engine speed. A heterogeneous distribution was obtained for the ultra-lean mixture condition (4
0.5) due to an insufficient amount of injected hydrogen. The heterogeneity decreased with increasing 4 because of the increase of the total amount of hydrogen injected inside the combustion chamber per unit time. There was no pronounced influence of SOI on the flow field pattern and mixture homogeneity because only the early period of injection was considered. In terms of the thermal field analysis, it was shown that the neglected parameter, 4, has a vital impact on the heat transfer for H2ICE. Therefore, including 4 as a parameter in the development of heat-transfer correlation is a powerful approach for estimating the heat transfer in H2ICE. The multidimensional model with the integrated injection program is a practical strategy for alternative-fuel development in the automotive sector. It is promising for performing optimization of the mixture formation and combustion analysis by 3D-CFD simulation.

Acknowledgments The authors would like to thank Universiti Malaysia Pahang for financial support under project no. RDU110332. Furthermore, the authors would like to express appreciation to Universiti Teknologi PETRONAS for providing laboratory facilities and the experimental database.

references Fig. 12 e Variation in heat release rate and cumulative heat loss against crank angle for different engine speeds and equivalence ratios at IT [ TDC and SOI [ 100 deg BTDC.

The temperature gradient of heat transfer, as well as the approach velocity of the hot plume to the wall, is increased. Therefore, the cumulative heat loss increases with an increase in the equivalence ratio due to the higher heat rejection. The cumulative heat loss increased around 35% with an increase in the equivalence ratio from 4 ¼ 0.5 to 4 ¼ 0.7 at a crank angle of 130 deg ATDC.

6.

Conclusions

A multidimensional engine model based on the finite volume approach was executed for various operating conditions, including the engine speed, equivalence ratio, and SOI. A parametric analysis was performed for qualifying and quantifying the influence of the operating parameters on IHT. It

[1] Dresselhaus M, Crabtree G, Buchanan M, editors. Basic research needs for the hydrogen economy. Washington, DC: Office of Basic Energy Sciences, Department of Energy; 2003www.sc.doe.gov/bes/reports/abstracts.html#NHE [accessed 11.04.10]. [2] Crabtree G, Dresselhaus M, Buchanan M. The hydrogen economy. Phys Today 2004;57(12):39e44. [3] White CM, Steeper RR, Lutz AE. The hydrogen-fueled internal combustion engine: a technical review. Int J Hydrogen Energy 2006;31(10):1292e305. [4] Rahman MM, Hamada KI, Noor MM, Kadirgama K, Maleque MA, Bakar RA. Heat transfer characteristic of intake port for spark ignition engine: a comparative study. J Appl Sci 2010;10(18):2019e26. [5] Kamil M, Rahman MM, Bakar RA. Modeling of SI engine for duel fuels of hydrogen, gasoline and methane with port injection feeding system. Energy Educ Sci Tech 2012;29(2):1399e416. [6] Kamil M, Rahman MM, Bakar RA. Performance evaluation of external mixture formulation strategy in hydrogen fuelled engine. J Mech Eng Sci 2011;1:87e98. [7] Rahman MM, Kamil M, Bakar RA. Engine performance and optimum injection timing for 4-cylinder direct injection hydrogen fueled engine. Simul Model Pract Theory 2011;19(2):734e51.

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 3 8 ( 2 0 1 3 ) 1 2 4 6 5 e1 2 4 8 0

[8] Das L. Hydrogeneoxygen reaction mechanism and its implication to hydrogen engine combustion. Int J Hydrogen Energy 1996;21(8):703e15. [9] Hamada KI, Rahman MM, Rashid AAA. Influence of engine speed and mixture strength on instantaneous heat transfer for direct injection hydrogen fuelled engine. Energy Edu Sci Tech 2013;30(2):805e24. [10] Abdullah MA, Rashid AAA. An experimental investigation of hydrogen combustion in a direct injection spark ignition natural gas engine. SAE Paper 2009. Technical Paper No. 2009-01-1499. [11] Verhelst S, Wallner T. Hydrogen-fueled internal combustion engines. Prog Energy Combust Sci 2009;35:490e527. [12] Rahman MM, Hamada KI, Kadirgama K, Bakar RA. Cycle analysis of in-cylinder heat transfer characteristics for hydrogen fueled engine. Sci Res Essays 2012;7(8):891e902. [13] Rahman MM, Hamada KI, Kadirgama K. Heat transfer of intake port for hydrogen fueled port injection engine: a steady state approach. Int J Phys Sci 2011;6(16):4036e43. [14] Das LM. Hydrogen engines: a view of the past and look into the future. Int J Hydrogen Energy 1990;15(6):425e46. [15] Ball M, Wietschell M. The future of hydrogen e opportunities and challenges. Int J Hydrogen Energy 2008;34(2):615e27. [16] Yi HS, Lee SJ, Kim ES. Performance evaluation and emission characteristics of in-cylinder injection type hydrogen fueled engine. Int J Hydrogen Energy 1996;21(7):617e24. [17] Natkin RJ, Tang X, Boyer B, Oltmans B, Denlinger A, Heffel JW. Hydrogen IC engine boosting performance and NOx study. SAE Paper 2003. Technical Paper No. 2003-01-0631. [18] Subramanian V, Mallikarjuna JM, Ramesh A. Performance, emission and combustion characteristics of a hydrogen fuelled SI engine: an experimental study. SAE Paper 2005. Technical Paper No. 2005-26-349. [19] Knop V, Benkenida A, Jay S, Colin O. Modelling of combustion and nitrogen oxide formation in hydrogen-fuelled internal combustion engines within a 3D CFD code. Int J Hydrogen Energy 2008;33(19):5083e97. [20] Ma F, He Y, Deng J, Jiang L, Naeve N, Wang M, et al. Idle characteristics of a hydrogen fueled SI engine. Int J Hydrogen Energy 2011;36(7):4454e60. [21] Bohac SV, Baker DM, Assanis DN. A global model for steadystate and transient SI engine heat transfer studies. SAE Paper 1996. Technical Paper No. 960073. [22] Sanli A, Sayin C, Gumus M, Kilicaslan I, Canakci M. Numerical evaluation by models of load and spark timing effects on the in-cylinder heat transfer of a SI engine. Numer Heat Transf 2009;56(5):444e58. [23] Wei SW, Kim YY, Kim HJ, Lee JT. A study on transient heat transfer coefficient of in cylinder gas in the hydrogen fueled engine. J Korean Hydrogen Energy Soc 2001;12(4):267e75. [24] Demuynck J, Pauwels S, Verhelst S, De Paepe M, Sierens R. Experimental research on the heat transfer inside a hydrogen combustion engine: evaluation and construction of measurement methods. In: Proc FISITA 2008 world auto cong 2008. Paper No. F2008-SC-037. [25] Shudo T, Suzuki H. Applicability of heat transfer equations to hydrogen combustion. JSAE Rev 2002;23(3):303e8. [26] Demuynck J, De Paepe M, Huisseune H, Sierens R, Vancoillie J, Verhelst S. On the applicability of empirical heat transfer models for hydrogen combustion engines. Int J Hydrogen Energy 2011;36(1):975e84. [27] Eichelberg G. Some new investigations on old combustion engine problems. Engineering 1939;148:463e547. [28] Annand WJ. Heat transfer in the cylinders of reciprocating internal combustion engines. P I Mech Eng 1963;177(36):973e90. [29] Woschni G. A universally applicable equation for the instantaneous heat transfer coefficient in the internal

[30]

[31]

[32]

[33]

[34]

[35]

[36]

[37]

[38]

[39]

[40]

[41]

[42]

[43]

[44] [45] [46]

[47]

[48]

[49]

12479

combustion engine. SAE Paper 1967. Technical Paper No. 670931. Hamada KI, Rahman MM, Aziz ARA. Time-averaged heat transfer correlation for direct injection hydrogen fueled engine. Int J Hydrogen Energy 2012;37(24): 19146e57. Rahman MM, Hamada KI, Noor MM, Bakar RA, Kadirgama K, Maleque MA. In-cylinder heat transfer characteristics of hydrogen fueled engine: a steady state approach. Am J Environ Sci 2010;6(2):124e9. Rahman MM, Hamada KI, Noor MM, Kadirgama K, Bakar RA, Rahim MFA. Heat transfer characteristics in exhaust port for hydrogen fueled port injection engine: a transient approach. Adv Mater Res 2011;152e153:1909e14. Shudo T, Suzuki H. New heat transfer equation applicable to hydrogen-fuelled engines. In: ASME fall tech conf 2002. Paper No. ICEF2002-515. Safari H, Jazayeri SA, Ebrahimi R. Potentials of NOx emission reduction methods in SI hydrogen engines e simulation study. Int J Hydrogen Energy 2009;34(2):1015e25. Nefischer A, Hallmannsegger M, Wimmer A, Priker G. Application of a flow field based heat transfer model to hydrogen internal combustion engines. SAE Paper 2009. Technical Paper No. 2009-01-1423. Schubert C, Wimmer A, Chmela F. Advanced heat transfer model for CI engines. SAE Paper 2005. Technical Paper No.2005-01-0695. Verhelst S, Sierens R. A quasi-dimensional model for the power cycle of a hydrogen-fuelled ICE. Int J Hydrogen Energy 2007;32(15):3545e54. Rakopoulos CD, Kosmadakis GM, Pariotis EG. Critical evaluation of current heat transfer models used in CFD incylinder engine simulations and establishment of a comprehensive wall-function formulation. Appl Energy 2010;87(5):1612e30. Launder BE, Spalding DB. The numerical computation of turbulent flows. Comput Methods Appl Mech Eng 1974;3(2):269e89. Han Z, Reitz RD. A temperature wall function formulation for variable-density turbulent flows with application to engine convective heat transfer modeling. Int J Heat Mass Tranf 1997;40(3):613e25. Amsden AA. KIVA-3V: a block-structured KIVA program for engines with vertical or canted valves. Los Alamos Nat Lab Rep No. LA-13313-MS; 1997. Xin J, Shih S, Itano E, Maeda Y. Integration of 3D combustion simulations and conjugate heat transfer analysis to quantitatively evaluate component temperatures. SAE Paper 2003. Technical Paper No. 2003-01-3128. Torregrosa AJ, Olmeda PC, Romero CA. Revising engine heat transfer. Annal Fac Eng HUNEDOARA e Int J Eng 2008;6(3):245e65. Holst MJ. Notes on the KIVA-II software and chemically reactive fluid mechanics. Lawrence Livermore Nat Lab; 1992. Patankar SV. Numerical heat transfer and fluid flow. New York: McGraw-Hill; 1980. O’Rourke PJ, Amsden AA. Implementation of a conjugate residual iteration in the KIVA computer program. Los Alamos Nat Lab Rep No. LA-10849-MS; 1986. Van Leer B. Towards the ultimate conservative difference scheme V. A second-order sequel to Godunov’s method. J Comput Phys 1979;32(1):101e36. Hirt CW, Amsden AA, Cook JL. Arbitrary LagrangianeEulerian computing method for all flow speeds. J Comput Phys 1974;14(3):227e53. Pracht WE. Calculating three-dimensional fluid flows at all speeds with an EulerianeLagrangian computing mesh. J Comput Phys 1975;17(2):132e59.

12480

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 3 8 ( 2 0 1 3 ) 1 2 4 6 5 e1 2 4 8 0

[50] Das S, Dent JC. Simulation of the mean flow in the cylinder of a motored 4-valved spark ignition engine. SAE Paper 1995. Technical Paper No. 952384. [51] Verhelst S, Sierens R. Hydrogen engine-specific properties. Int J Hydrogen Energy 2001;26(9):377e82. [52] Shudo T. Thermophysical properties of working fluid and heat transfer in hydrogen. JSME Trans B 2005;71(702):730e6. [53] Heywood JB. Internal combustion engine fundamentals. New York: McGraw-Hill; 1988. [54] Ferguson CR, Kirkpatrick AT. Internal combustion engine: applied thermosciences. 2nd ed. New York: John Wiley and Sons Inc.; 2001. [55] Amsden AA. KIVA-3: a KIVA program with block-structured mesh for complex geometries. Los Alamos Nat Lab Rep No. LA-12503-MS; 1993. [56] Kumar P, Franchek M, Grigoriadis K, Balakotaiah V. Fundamentals-based low-dimensional combustion

[57]

[58]

[59]

[60]

modeling of spark-ignited internal combustion engines. AIChE J 2011;57(9):2472e92. Rahman MM, Hamada KI, Aziz ARA. Multidimensional computational modeling of direct injection for hydrogen fueled engine. Adv Sci Lett 2012;13:317e21. Mohammadi A, Shioji M, Nakai Y, Ishikura W, Tabo E. Performance and combustion characteristics of a direct injection SI hydrogen engine. Int J Hydrogen Energy 2007;32(2):296e304. Verhelst S, Sierens R. A laminar burning velocity correlation for hydrogen/air mixtures valid at spark-ignition engine conditions. In: Spring technical conference of the ASME internal combustion engine division 2003. p. 35e43. Paper No. ICES2003-555, Salzburg, Austria. Stefaan V, Roger S, Sebastian V. A high speed single cylinder hydrogen fuelled internal combustion engine. In: World automotive congress, 23e27 May, Barcelona, Spain 2005. p. 1e10.