Effects of the Injector Spray Angle on Combustion and Emissions of a 4-Stroke Natural Gas-Diesel DF Marine Engine

Abstract

This work studied the effect of the injector spray angle (SA) on the combustion and emissions of a 4-stroke port-injection natural gas-diesel dual-fuel (NG-Diesel DF) marine engine to determine the optimal SA for the fuel injector, aiming to reduce exhaust gas emissions while keeping the engine performance. Three-dimensional (3D) simulations of the combustion process and emission formations occurring in the engine cylinder in both diesel and DF modes were carried out using the AVL FIRE R2018a code. The engine’s in-cylinder temperature, pressure, and emission characteristics were analyzed. To clarify the effect of the injector SA on the combustion and emission characteristics of the engine, only the injector SA has been varied from 145 to 160°. Meanwhile, all other boundary conditions for the simulations and operating conditions of the engine have remained unchanged. The simulation results have been compared and showed a good agreement with the engine experimental results. The study has successfully investigated the effects of the injector SA on the combustion and emission characteristics of the engine. A better SA for the fuel injector, to reduce the NO emissions (145°) or soot and CO₂ emissions (150°), while keeping the engine power almost unchanged, without the use of any exhaust gas post-treatment equipment, has also been suggested.

1. Introduction

Emissions from ships have negative impacts on human health and the Earth’s environment. The International Maritime Organization (IMO) released many regulations on marine emissions to limit these impacts. Marine emission regulations are getting stricter and stricter nowadays. According to the International Convention for the Prevention of Pollution from Ships (MARPOL) Annex VI, since 1 January 2020, all ships around the World must comply with the usage of fuels containing a maximum of 0.5% sulfur. The carbon (C) intensity per marine transport mean must be reduced by at least 40% by 2030 in comparison to 2008, with a reduction target of at least 70% by 2050. Greenhouse gases (GHGs) from ships must be reduced by at least 50% by 2050, compared to 2008. Since 1 January 2016, NOx emissions from ships have been limited to 3.4 g/kWh for engines with speeds less than or equal to 130 rpm (revolution per minute). This limitation gradually decreases with the increase of the engine speed and reaches only 2 g/kWh for engines with a rate higher than or equal to 2000 rpm [1]. Therefore, the emission reduction technologies must be applied to both new-building and existing ships to meet stricter emission regulations [2,3,4]. Technically, many solutions can be applied to reduce emissions from marine engines, however, each solution has its own advantages and disadvantages. Choosing an appropriate solution for an actual engine and application thus becomes a technical issue that needs to be considered carefully. This paper focuses on two effective solutions to reduce emissions in the marine sector: using alternative fuels and optimizing fuel injectors.

Among the various alternative fuels, natural gas (NG) has been widely used for marine engines. It is expected to continue to be used for a long time in the future because it has many advantages. NG produces low exhaust gas emissions (EGEs), requires no processing, has low prices, abundant reserves, etc. Especially, if NG is used as a primary fuel in dual-fuel (DF) engines, pure diesel existing engines can be modified to NG–diesel DF engines very easily with only a low cost [5,6,7]. However, NG has a high auto-ignition temperature (low cetane number). Therefore, it always needs external energy for ignition, such as spark plugs in spark-ignition engines (SIEs) or pilot fuels (usually diesel oil) in DF engines. The detailed properties of NG and its effects on the combustion and emission formations of a NG–diesel DF marine engine have been presented and can be found in our previous studies [8,9].

Methane (CH₄) occupies approximately 96% of NG by volume [10,11] and has been employed to represent NG in many previous studies by many researchers [5,12,13,14,15,16]. These studies have demonstrated the ability of CH₄ to represent NG with a high accuracy in simulation results. Methane was thus used to represent NG in this study.

Internal combustion engines (ICEs) convert thermal energy exported from the chemical energy of fuels through combustions to the mechanical energy on engine shafts. Combustion is a process of fuel oxidation by oxygen (O₂) in the air. Ideally, all injected fuels should be in contact with all of the O₂ molecules available in the engine’s combustion chamber so that the fuel can be burned as completely as possible. In the direct injection engines (DIEs), fuel atomization and injection characteristics play very important roles in the combustion quality of engines. They are also the major factors in reducing EGEs, while keeping or even increasing the engine power [17]. The atomization of the fuel is not only influenced by fuel injector properties but also by the interaction between the injected fuel droplets and the air existing in the combustion chambers of engines [18]. Regarding the fuel injection characteristics, the injection method (port- or direct-injection), injection strategy (single- or multiple-injection), injection timing, and SA play important roles in the combustion and emission characteristics of DIEs [5,6]. The effectiveness of the multiple-injection strategy in reducing EGEs applied to a heavy-duty diesel marine engine has been demonstrated and is available to be referred to in our previous research [19]. On the remaining aspect, the injector SA strongly affects the combustion and emission characteristics of engines, because it determines the targeting point of the injection into the engine’s combustion chamber [20]. The targeting point of the fuel injection, together with the shape of the piston surface strongly affects the turbulence of the fluid flow inside engine cylinders. Additionally, the distance between the injection targeting point and the piston surface determines the liquid fuel film formed on the piston surface, due to the wall-wetting phenomena [21,22]. These factors strongly affect the combustion quality and emission formation in ICEs, especially soot. Therefore, the study on the injector SA is very important and has practical implications for both engine designers and operating engineers.

Wei et al. [17] numerically studied the effects of the SA in a swirl chamber combustion system of DIEs by using the AVL-FIRE code. The research concluded that the SA of the injector has a strong influence on the fuel-air equivalence ratio and the temperature distribution inside the combustion chamber. It also changes the combustion and emission characteristics of the engine. The research found that an injector SA of 150° produces the best emission performance. Kim et al. [20] experimentally investigated the impact of the fuel SAs and injection timing on the combustion and emissions of a diesel engine. The research found that the fuel SA has a significant influence on the combustion, performance, and emissions of the engine. They concluded that smaller SAs (narrow SAs) produce a higher maximum combustion pressure and heat release rate than wider SAs. Yoon et al. revealed the same conclusion in their study [23]. Mobasheri and Peng [24] revealed in their publication, that an appropriate injector SA can reduce NOx and soot emissions without affecting the engine fuel consumption because of the improvement of the air-fuel mixture quality.

An optimization for the fuel injector of a heavy-duty 2-stroke direct injection DF marine engine was carried out and can be referred to in our previous study [25]. The present study numerically studied the effects of the injector SA on the combustion process and emission formations of a 4-stroke port-injection NG-diesel DF marine engine by using the CFD method. The combustion of the fuel and emission formations inside the engine cylinder were simulated by the AVL FIRE code. The final goal of this study is to specify the optimal injector SA for the engine, with respect to reducing EGEs while retaining the engine power. The CFD models were verified by the measured results reported in the technical measurement data of the engine. The study successfully pointed out the optimal injector SA for the engine to achieve certain emission reductions.

2. Numerical Analysis

2.1. Specifications of the Researched Engine

The principle schematic of the researched engine in this work is presented in Figure 1. The piston surface shape of the engine is a ω-type. The diesel pilot fuel nozzle for the pilot injection has twelve identical holes with an originally designed SA of 155°. The specifications of the researched engine in this study are presented in

Table 1. Engine specifications.

Parameter	Value	Unit
Engine type	4-Stroke DF Engine
No. of cylinders	6
Gas supplying method	Port-Injection
Ignition	Diesel Direct Injection
Compression ratio	13.5:1
Cylinder bore $\times$ Stroke	350 $\times$ 400	mm
Rated power	2880	kW
Rated speed	720	rpm
IMEP	20	Bar

.

The engine can run smoothly in two working modes: diesel mode and DF mode. In the diesel mode, the engine operates with pure diesel, similar to conventional compression ignition (CI) diesel engines. In the DF mode, diesel oil serves as a pilot fuel to provide an ignition source while the NG plays a role as a primary fuel. In this mode, NG is injected into the intake port of the engine during the intake stroke by a gas nozzle. The injected gas, therefore, mixes with the charge fresh air to form a gas/air premixed mixture prior to being supplied to the cylinder during the intake stroke of the engine. In contrast, the diesel pilot fuel is injected directly into the engine cylinder at the end of the compression stroke by a pilot fuel nozzle installed in the center of the cylinder head.

In this research, C₁₃H₂₃ was utilized to represent diesel oil. It serves as a pilot fuel to provide an ignition source to ignite the premixed NG-air mixture. Meanwhile, CH₄ was employed to represent NG and played a role as a primary fuel in the DF mode. The properties of fuels in this study were presented and can be found in our previous studies [8,9]. All properties of the fuel were temperature-dependent functions.

2.2. Computational Mesh, Initial and Boundary Conditions

The AVL FIRE R2018a ESE-diesel platform was employed to build a three-dimensional (3D) model of the combustion chamber of the engine and a computational mesh for the CFD analysis. Because the combustion chamber of the engine is axially symmetric; the pilot fuel nozzle has twelve identical holes; and for the calculation time reduction purpose, only one-twelfth of the entire combustion chamber was used to create the computational domain. The calculation began from the IVC (intake valve closing) to the EVO (exhaust valve opening) and was carried out in series using a twelve-core Intel Xeon processor. It took approximately 36 h of CPU time.

The 3D movable computational mesh was generated using the dynamic layering method. The movement of the piston surface was controlled by Equation (1).

$${\mathrm{P}}_{\mathrm{l}}=\sqrt{{\left({\mathrm{r}}_{\mathrm{c}}+\mathrm{L}\right)}^2-{\mathrm{x}}_{\mathrm{offset}}^2}-{\mathrm{r}}_{\mathrm{c}}\cos \left({\mathsf{\theta }}_{\mathrm{c}}\right)-\sqrt{{\mathrm{L}}^2-({\mathrm{r}}_{\mathrm{c}}\sin \left({\mathsf{\theta }}_{\mathrm{c}}\right)+{ \mathrm{x}}_{\mathrm{offset}}){ }^2}$$

(1)

where, ${\mathrm{P}}_{\mathrm{l}}$ is the piston location, ${\mathrm{r}}_{\mathrm{c}}$ is the crank radius, $\mathrm{L}$ is the length of the connecting rod, ${\mathrm{x}}_{\mathrm{offset}}$ is the piston pin offset, while ${\mathsf{\theta }}_{\mathrm{c}}$ is the current CAD.

The computational mesh with the boundary conditions (BCs) when the piston is at 40 CADs (crank angle degrees) ATDC (after the top dead center) is presented in Figure 2. The combustion chamber of the engine is built by the cylinder head, cylinder liner, and the piston surfaces. This engine is designed to run smoothly in both the diesel and DF modes without the compression ratio (CR) adjusting requirements. That is, the CR was the same in both the diesel and DF combustion simulations. To achieve the correct calculated CR in the CFD simulation, which is the same as the actual CR in the experiments, the “compensation volume” option was active to create a compensation volume on the top of the piston, to compensate for the lacking volume part of the swept volume when building the computational mesh for the simulations.

For the CFD analysis, the BCs at the cylinder liner, cylinder head (cover), and piston surfaces were assigned as impermeable walls. Owing to the axially symmetric characteristic of the combustion chamber geometry, the cutting surfaces of the computational domain were assigned as periodic (cyclic) BCs. The simulation started from the IVC of 35 CADs after the bottom dead center (ABDC) to the EVO of 62 CADs before the bottom dead center (BBDC) of the engine. In the diesel mode, the injection of the diesel fuel started at 12 CADs BTDC and took place in 32 CADs (corresponding to 7.5 milliseconds). In the DF mode, the diesel pilot fuel was injected into the engine cylinder at 12 CADs BTDC as well, but took place in 12 CADs only (corresponding to 2.35 milliseconds). In addition, the combustion simulations had been carried out, based on the assumption that the engine was running in stable thermal conditions. In which, the thermal condition on the combustion chamber wall was stable. With this assumption, the temperatures of the cylinder head and piston surfaces were set to 297 °C while the temperature on the cylinder liner was set to 197 °C, based on the reference to similar studies [8,9] and the suggestion in the literature [26]. The temperatures on the cylinder head and piston surfaces were higher than that on the cylinder liner surface because of the cooling effect of the cylinder liner’s freshwater cooling system of the engine. This engine is supplied with the charge air by a turbocharger. The pressure and temperature inside the engine cylinder at the start of the simulation (i.e., at the IVC) are 3.5 bar and 47 °C, respectively, the same as those recorded in the experiments. In addition, all of the properties of the air were set as temperature-dependent to ensure the response of air to the change of temperature. Regarding the initial condition of the flow inside the engine cylinder, the in-cylinder flow of the ICEs is characterized by swirl, tumble, and squish. They influence the mixing quality between fuel and air. Their intensities are mainly influenced by the engine (piston) speed, piston bowl geometry, inlet port geometry, inlet valve, and inlet velocity of air or fuel-air mixture. Since squish flows are mainly formed when the piston reaches TDC, the initial flow characteristics inside the engine cylinder are characterized only by swirl and tumble. Since squish flows are mainly formed when the piston reaches TDC, the initial flow characteristics inside the engine cylinder are characterized only by swirl and tumble. Based on the structure of the piston bowl geometry ($\omega$-type), engine speed (720 rpm), and intake air pressure (3.5 bar) which causes the intake air velocity, as well as the guidance in the literature [26], a value of 2880 ($1/\min$) has been set for the swirl and tumble intensities as the initial condition of the in-cylinder flow for the simulations in this study. The BCs and initial conditions for the CFD analysis were selected from the technical report of the engine and are listed in

Table 2. Boundary and initial conditions.

Boundary Condition	Boundary Type/Specific Condition
Cylinder head	Fixed wall/Temp./297 °C
Cylinder liner	Layering wall/Temp./197 °C
Piston surface	Mesh movement/Temp./297 °C
Segment-cut surfaces	Periodic
Initial Conditions	Values
Temperature at IVC	47 °C
Pressure at IVC	3.5 bar
Swirl/Tumble	$2880 1/\min$
IVC	35 CADs ABDC
EVO	62 CADs BBDC
PSOI	12 CADs BTDC
Pilot injection duration	7.5 milliseconds (Diesel mode)
	2.35 milliseconds (DF mode)

.

The energy share (ES) of the diesel pilot fuel in the total energy supplied to the engine in the DF mode is calculated by Equation (2).

$${\mathrm{ES}}^{\mathrm{diesel}}=\frac{\mathrm{mDiesel} \times \mathrm{LCV} \mathrm{of} \mathrm{Diesel}}{\left(\mathrm{mDiesel} \times \mathrm{LCV} \mathrm{of} \mathrm{Diesel}\right)+\left({\mathrm{mCH}}_4\times { \mathrm{LCV} \mathrm{of} \mathrm{CH}}_4\right)}$$

(2)

where, ${\mathrm{ES}}^{\mathrm{diesel}}$ is the ES of the diesel pilot fuel, $\mathrm{mDiesel}$ and ${\mathrm{mCH}}_4$ are the diesel and CH₄ masses that were supplied to one cycle, respectively. The $\mathrm{LCV} \mathrm{of} \mathrm{Diesel}$, and ${\mathrm{LCV} \mathrm{of} \mathrm{CH}}_4$ are the lower calorific value (LCV) of diesel and CH₄, respectively.

2.3. Simulation Cases

The engine in this study is a generator engine. It supplies mechanical energy for a 2740 kW/60 Hz generator. Due to it being an alternative current (AC) generator engine, it always operates with a fixed speed of 720 rpm and produces a power of around 2880 kW. It does not run at other speeds. Therefore, this study investigated the effect of the injector SA on the combustion and emission characteristics of the engine only at this full load. Eight simulation cases were carried out for both the diesel and DF modes. In each mode, the SA of the pilot fuel injector was changed from 145 to 160° with an angle step of 5°. The SA of the fuel injectors affects the combustion and emission formations of the ICEs because it determines the targeting point of the fuel injection. The degree of consumption of O₂ in the combustion process also strongly depends on the targeting injection point [27,28,29]. If the SA of the injectors is too narrow, the targeting injection point is in danger of approaching the opposite wall of the combustion chamber of the engine, i.e., the piston surface [30,31]. This causes the formation of liquid fuel films on the piston surface, due to the wall-wetting phenomena. In contrast, if the fuel injector SA becomes too wide, the outer edge of the injection region is in danger of colliding with the cylinder liner, causing the formation of the wall-wetting liquid fuel films on the cylinder liner [20,21]. Liquid fuel films, due to the wall-wetting of the injected fuels on the combustion chamber surfaces reduce the quality of the fuel combustion. Especially, this significantly increases the amount of soot and carbon monoxide (CO) emissions owing to the incomplete combustion. For this reason, the SA of the injector of the engine in this study has been limited to the range of 145 to 160°. A schematic of the injector spray angle and interaction between injected fuel droplets and the piston surface is presented in Figure 3.

To clarify the impact of the fuel injector SA on the engine combustion and emission characteristics, only the injector SA was adjusted, other simulation parameters and the engine operating conditions, including the supplied fuels remained unchanged. The simulation cases in this study are listed in

Table 3. Simulation cases.

SA	145°	150°	155°	160°
Diesel Mode	Di-145	Di-150	Di-155	Di-160
DF Mode	DF-145	DF-150	DF-155	DF-160

.

2.4. CFD Models

The AVL FIRE code with its advanced CFD models has been demonstrated to be suitable for modeling the combustion and emission formations in the cylinder of ICEs, such as diesel, gasoline, and DF engines with a high accuracy [21]. In the present study, the AVL FIRE ESE diesel platform was employed to simulate the working process of the engine from the IVC to EVO. The simulated results were then compared to the measured results to verify the CFD models.

The k-𝜁-f is a well-known four-equation turbulence model. It was developed from the k-ε two-equation model [32]. It has a better stability and a higher accuracy than the original k–ε model. Therefore, it was used to model the fluid flow turbulence inside the engine cylinder in this study. In the combustion problems, the transport and mixing processes of the chemical species are governed by solving the conservation equations that describe the diffusion, convection phenomenon, component species concentrations, and reaction sources in the system. In this research, the 3-zone species transport model named the 3-Z extended coherent flame model (ECFM-3Z) [26,33] was employed to model the combustion of the fuels inside the cylinder of the engine. Based on the assumption that in ICEs, the chemical time scales are much smaller than the turbulent time scale, the ECFM combustion concept was applied. The ECFM is applicable to both non-premixed and premixed combustions, based on a laminar flamelet concept. In this concept, the velocity and thickness of the flames are mean values, integrated along the flame front. They only depend on the pressure, temperature, and richness of the fresh gases. The laminar flamelet concept assumes that the chemical reactions take place within relatively thin layers, which separate the unburned gases from the fully burnt gases. With this assumption, the mean turbulent chemical reaction rate is calculated as the product of the laminar burning velocity and the flame surface density [26,33]. The direct injection of the pilot fuel was modeled by using the diesel nozzle flow model [26,34]. The diesel nozzle flow offers a simple approach to correct the velocity and the initial diameters of the fuel droplets caused by the cavitation phenomena. This model is presented in detail in Appendix A. The evaporation and breakup of the fuel droplets were respectively modeled by the multi-component and WAVE models [26,34]. The ignition of the diesel oil inside the cylinder in the diesel and DF modes was simulated by the auto-ignition and diesel ignition gas engine models, respectively [26]. Regarding EGEs, the well-known extended Zeldovich model [26,35] was utilized to simulate the NO emission formation inside the engine cylinder. This model consists of seven species and three chemical reactions. It has also been demonstrated to be able to accurately predict the thermal NO emission in ICEs over a large range of fuel-air equivalence ratios. For modeling the formation of soot during the combustion of the engine, a kinetic soot mechanism was used [26,35]. The interaction between the engine combustion walls and the injected fuel droplets was modeled using the Walljet1 model [26,34]. The CFD models used in this study are summarized in

Table 4. CFD models.

Model	Description
Turbulence	k-𝜁-f
Combustion	Extended coherent flame models (ECFM)
Emissions	Thermal NO	Extended Zeldovich mechanism
	Soot	Kinetic soot mechanism
Pilot ignition	Diesel mode	Auto-ignition
	DF mode	Diesel-ignition-gas-engine
Pilot fuel atomizations	Breakup	WAVE model
	Evaporation	Dukowicz (Diesel mode)
		Multi-component (DF mode)
Droplets–Walls interactions	Walljet1

. For the convergence criteria, in this study, the absolute convergence criteria of ${10}^{-3}$ for the continuity, momentum, velocity, k, and 𝜁 were set. Meanwhile, a tighter convergence criterion of ${10}^{-6}$ was set for the energy equations. These criteria are technically suitable for high-turbulence combustion problems in ICEs [26].

2.5. CFD Model Validation

The CFD models in this study were validated by comparing the simulated results to the experimental results reported in the technical measurement data of the researched engine. The experiments (tests) were conducted in an engine test workshop. During the experiment process, the power of the engine was recorded by a dynamometer. The mass flow rate of the fuels was controlled by mass flow meters. The in-cylinder pressure and temperature were recorded by pressure and temperature sensors, respectively. The mass fraction of the emissions in the exhaust gas was analyzed by an exhaust gas analyzer. The measured values were then compared with the simulated results for the CFD model validation purposes.

In the diesel mode, the engine cylinder was fueled with a diesel mass of 0.0020667 kg/cycle. In the DF mode, the diesel pilot fuel and CH₄ masses were 0.0001 and 0.001682 kg/cycle, respectively. The total fuel mass supplied to the engine cylinder in the DF mode was thus 0.001782 kg/cycle. The total fuel mass utilized in the DF mode was less than that in the diesel mode caused by the higher LCV of CH_4, compared to the diesel oil. The air excess ratios ($\lambda$) were 1.52 and 1.56 for the diesel and DF modes, respectively. With these fuel masses and air excess ratios, the engine cylinder produced the powers of 486.9 and 485.7 kW in the diesel and DF modes, respectively.

Figure 4 presents the comparison results between the simulated and measured data for both the diesel and DF modes with the original injector SA (155°). It is clear that the simulation and the experimental results are in good agreement with each other. The deviations between the simulated and measured cylinder power in the diesel and DF modes were only 1.44% and 1.19%, respectively. The deviations between the simulated and measured peak pressure in the diesel and DF modes were 1.30% and 1.96%, respectively. Regarding the emissions, the deviations between the simulated and measured NO and CO₂ emissions were 7.73% and 2.66%, respectively, in the diesel mode. Meanwhile, in the DF mode, the deviations between the measured and simulated NO and CO₂ emissions were only 3.53% and 3.8%, respectively. Once the CFD models were verified, they were applied to modeling the combustion and emission formations occurring inside the engine cylinder for all simulation cases in this research.

2.6. Mesh Independence Analysis

The accuracy of the final CFD numerical results is greatly affected by the mesh quality (or mesh resolutions). On another aspect, the mesh resolutions significantly affect the calculation time. In general, a finer mesh may produce a better mesh quality resulting in a higher accuracy for the CFD results. However, it also always prolongs the computation time. Therefore, to ensure the accuracy of the final CFD simulation results and the reasonableness of the computation time, a mesh independence analysis was carried out. Three simulations with various mesh resolutions, including a coarse, medium, and fine mesh were performed and the simulation results were analyzed and compared. The simulations using the coarse, medium, and fine-resolution meshes took place in 24, 36, and 92 h, respectively, by using a 32 GB RAM workstation with a 12-core Intel Xeon processor. The mesh metrics and the corresponding computation time of these three mesh resolutions are listed in

Table 5. Mesh metrics and the calculation times.

Mesh Resolution	Coarse	Medium	Fine
No. of faces of the 2D mesh at the TDC	12,949	17,715	39,307
No. of cells of the 3D mesh at the BDC	586,796	882,620	1,593,732
Calculation time	24 h	36 h	92 h

.

Figure 5 presents the final CFD numerical results for the three various mesh resolutions. As can be observed, the final results were no longer dependent on the mesh resolution. Therefore, all three of these mesh resolutions can technically be utilized for simulations, to achieve both highly accurate and mesh-independent CFD numerical results. The medium mesh, however, was selected for the simulations in this study since it not only gave accurate results but also in a more reasonable time. Additionally, it also has an appropriate resolution for good contour presentations in the post-processing step. In this study, as only the injector SA had been adjusted, all of the simulations using the medium-resolution mesh took place in approximately 36 h.

3. Results

3.1. In-Cylinder Pressure

The in-cylinder mean pressure and rate of heat release (RoHR) in all simulations are shown in Figure 6. The simulation results showed a slightly lower in-cylinder peak pressure in the DF mode, compared to the diesel mode. The lower peak pressure in the DF mode is owing to the lower amount of pilot diesel fuel (approximately 5% of the total supplied energy) injected directly into the engine cylinders to provide the ignition source for the NG-air premixed mixture. Typically, the combustion of diesel DIEs (working with diesel only) is divided into four stages: (1) ignition delay (ID) stage; (2) premixed combustion stage; (3) diffusion combustion stage; and (4) late stage of combustion. In these four stages of combustion, stages 1 and 2 play a critical role in the pressure rise rate (PRR) and thus the in-cylinder peak pressure of the engine. The longer the ID stage and premixed combustion, the higher the PRR and peak pressure. The combustion process of the engine in the DF mode, however, typically occurs in three stages [36]. The first stage begins at the time the diesel pilot fuel is injected directly into the engine cylinder, which already contained a gaseous fuel-air premixed mixture. Since the injected diesel droplets need time to receive heat from the surrounding premixed gas-air mixture for the heating and vaporization processes, the diesel pilot fuel will not ignite immediately but after a few seconds. This period is the ID time of the DF mode. During the ID time, the injected diesel oil droplets will mix with air inside the engine cylinder to form a diesel-air pre-mixed mixture. The ID stage finishes when the premixed diesel-air mixture auto-ignites under a high-enough temperature condition inside the engine cylinder. This stage of combustion, therefore, includes the premixed combustion of the premixed direct-injection-diesel air mixture and a small portion of the premixed port-injection-gas air mixture. However, because the gaseous fuel (CH₄) has a higher activation energy, leading to a higher auto-ignition temperature, the premixed combustion of the premixed diesel air mixture is the main contributor. During this stage, the in-cylinder pressure and temperature rapidly rise due to the initiation of the diesel premixed combustion. The second stage begins immediately after this initial combustion of diesel pilot fuel owing to an ignition source available inside the engine cylinder. During this stage, the continuing injected diesel pilot fuel is immediately ignited. Thus, this stage involves the diffusion combustion of the direct-injection-diesel pilot fuel along with the initiation and flame propagation of the gaseous fuel due to the presence of the ignition source. In this stage, the diffusion combustion of the diesel fuel is mixing-controlled, while the premixed gas-air flame propagates throughout the engine cylinder from the ignition kernels. Lastly, the diffusion combustion of the remaining diesel fuel and the late combustion phase of the port-injection-gas fuel is completed in the 3rd stage of combustion [37,38,39]. In the DF mode, only approximately 5% of diesel oil energy share was employed for the ignition, the ID stage and premixed combustion stages were thus very short. These reduced the peak pressure inside the engine cylinder. The combustion stages of both the diesel and DF combustions are presented in Figure 6.

As shown in Figure 6, the ID time (stage A-B) in both the diesel and DF modes were almost the same. This implies that introducing CH₄ gas to the engine cylinder during the intake stroke of the engine has no significant influence on the in-cylinder temperature at the end of the compression stroke, the moment when liquid fuel (diesel oil) is injected. This was also confirmed by the in-cylinder temperature diagram shown in Section 3.2. At the start of the combustion (S.O.C), the RoHRs in the diesel and DF modes were almost the same (stage B-C). This is because the combustions inside the engine cylinder in both diesel and DF modes were controlled by the premixed combustion phase of diesel fuel. However, after this phase of combustion, the RoHRs in the diesel and DF modes were totally different. Following the premixed combustion phase (stage B-C), the combustion inside the engine cylinder in the diesel mode was transferred to the diffusion combustion phase (stage C-D), followed by the late combustion stage (stage D-E). Due to the fuel injection process ending at 380 CAD, the combustion was almost finished at around 400 CAD. Most of the energy was released in a shorter duration, increasing the peak pressure in the diesel mode. In another manner, in the DF mode, after the premixed combustion process of the diesel pilot fuel (stage B-C), the combustion inside the engine cylinder was transferred to the premixed combustion of the NG-air premixed mixture within the engine cylinder, due to the spread of the pilot flame (stage C-D’). Since the fuel and air were premixed, the local concentration of the reactant species in a premixed combustion system was lower than that in a diffusion combustion system. This led to a slower reaction rate or a lower RoHR, as can be observed in Figure 6. The slower reaction rate led to a longer combustion duration. The longer combustion duration creates a higher in-cylinder pressure in the combustion late stage in the DF mode, in comparison to the diesel mode, as shown in Figure 6.

Figure 7 presents the peak pressure in the engine cylinder in all simulation cases. The simulation result showed a slight reduction in the peak pressure as the SA of the injector increased from 145 to 160° in both the diesel and DF modes. This may be due to the increase in the fuel-air mixing quality when the injector SA increases. As the injector SA increases the interaction region between the injected fuel and the charge air increases, accordingly. This increased the fuel-air mixing quality. The increase in the fuel-air mixing quality reduced the ID stage and thus the peak pressure in the engine cylinder when the fuel was burnt. Kim et al. [20] and Yoon et al. [23] also revealed in their study that the smaller SAs (narrow SAs) produced a higher maximum combustion pressure and heat release rate than the wider SAs. The reduction of the in-cylinder peak pressure in the DF mode of the NG-diesel DF engines, compared to the diesel mode has also been reported in previous studies [8,9].

The powers of the engine cylinder in all simulation cases are presented in Figure 8. As a result of the slight increase in the peak pressure as the SA of the injector decreased from 160 to 145°, the engine cylinder power also had the same increase trend in both the diesel and DF modes. Compared to the originally designed injector SA of 155°, the SA of 150 and 145° increased the cylinder power by approximately 0.062 and 0.144%, respectively, while the SA of 160° decreased the cylinder power by approximately 0.123% in the diesel mode. In the DF mode, the SA of 150 and 145° increased the cylinder power by approximately 0.165 and 0.288%, respectively, while the SA of 160° decreased the cylinder power by approximately 0.226%, in comparison to the originally designed injector SA of 155°.

3.2. In-Cylinder Temperature

The mean temperatures inside the cylinder of the engine in all simulations are shown in Figure 9. The results showed a lower in-cylinder peak temperature in the DF mode, compared to the diesel mode. However, the in-cylinder mean temperature during the late stage of combustion in the DF mode was higher than that, in comparison to the diesel mode. The lower peak temperature and higher mean temperature during the late-stage combustion are interesting characteristics of the port-injection premixed combustion, in comparison to the direct-injection combustion. In the port-injection approach, gaseous fuels are injected into the intake port of the engine during the intake stroke and mixed with the charge, to form a premixed mixture inside the engine cylinder. As a result, this fuel-air premixed mixture is compressed during the compression stroke of the engine. Because of this feature, there is always enough time for the gaseous fuel and charge air to mix with each other to form a gas-air premixed mixture, prior to an ignition source being supplied for the ignition. Owing to the fuel-air premixed mixture being perfectly prepared, the combustion in the cylinder of the port-injection engines occurs more uniformly than in the direct-injection engines. The more uniform temperature distribution within the combustion chamber of the engine reduced the peak mean temperature, as shown in Figure 9. As the same as the in-cylinder temperature, the reduction of the in-cylinder mean temperature in the DF mode of the NG-diesel DF engines, compared to the diesel mode has also been reported in previous studies [8,9].

Figure 10 presents the temperature distributions inside the engine cylinder at the TDC in all simulation cases. The temperature contours point out the targeting point of the injection, the diffusion flame shape, and reaction zones inside the engine combustion chamber. The figure showed that a high-temperature combustion region appeared in the flame kernel caused by the fuel injector of the engine. This is because, in ICEs, the temperature of the fuel-rich zone is the highest in the entire engine cylinder, as reported in [40]. In addition, the figure pointed out that the injector SA of 155° and 150° produced the highest local peak temperatures in the diesel and DF modes, respectively.

It is very interesting to observe the change in the diesel flame kernel size when changing the injector SA in the diesel mode. Similarly, the change in the flame curvature when changing the injector SA in the DF mode is also notable. In the diesel mode, when the engine ran with diesel oil solely, the diesel flame kernel tended to become bigger as the injector SA increased from 145° to 160°. Meanwhile, in the DF mode, the curvature of the flame decreased when the injector SA increased from 145° to 150°, and then reversed the curvature as the SA continued to increase to 155° and 160°. These changes in the flame shape might relate to the spatial relationship between the injection axis and the piston and the cylinder head surfaces as the injector SA changed. It is clear that there will always be an interaction between the fuel injection beam and the combustion chamber wall when the fuel is injected into the engine combustion chamber. This interaction will obviously be affected by the relative position between the fuel injection beam inside the combustion chamber and the combustion chamber wall surface. Therefore, when the axis of the fuel injection beam changes, the shape of the flame is also changed, accordingly.

3.3. NO Emissions

The specific NO emissions of the engine in all simulation cases are presented in Figure 11. The simulation results showed a significant reduction in NO emissions in the DF mode, in comparison to the diesel mode. The DF mode reduced the NO emissions by up to 71%, compared to the diesel mode.

NO is responsible for more than 90% of NOx emissions emitted by ICEs. Chemically, there are two chemical mechanisms involved in the NO emission formation: (1) thermal NO mechanism (described by the extended Zeldovich mechanism); and (2) the prompt NO mechanism (described by the Fenimore mechanism) [41]. However, in ICEs, the thermal NO mechanism is prominent, so only this mechanism must be considered when analyzing the NO emissions. This is the reason why the extended Zeldovich mechanism was employed to model the NO formations in this study. In the thermal NO mechanism, NO formations are significantly influenced by the local peak temperature, the O₂ concentration inside the cylinder, and the reaction resident time. In the cylinder of ICEs, NO emissions are mainly formed in the regions where the temperature is equal to or above 1800 K. The rate of the formation dramatically increases with the increase of the local peak temperature [18,33,41,42].

In the DF mode, as shown in Figure 10, the local peak temperatures were significantly lower than those in the diesel mode. This resulted in a significant reduction in NO emissions, as shown in Figure 11. The injector SA of 155° and 150°, respectively, produced the highest local maximum temperature in the diesel and DF modes, leading to the highest NO emissions in these corresponding modes. Moreover, the injector SA of 145° produced the lowest NO emissions in both the diesel and DF modes. The reason is that because this SA of the injector generated the lowest local peak temperature in both the diesel and DF modes. Additionally, in the DF mode, the gaseous primary fuel was injected into the intake port to mix with the charge air, forming a homogeneous fuel-air premixed mixture, prior to being supplied to the engine cylinder. This leads to a significant reduction in the local O₂ concentration in the engine cylinder. This contributed to the NO emission reduction when burning gaseous fuel, compared to pure diesel. This tendency has been reported in the previous studies [8,9,19]. In conclusion, the injector SA of 145° is the optimal value to reduce NO emissions in both the diesel and DF modes.

3.4. Soot

Soot is the main contribution of particulate matter (PM) emissions [18,41,43,44]. Under high temperatures and high fuel-air equivalence ratios (fuel-rich zones), which are typically found in the cylinder of ICEs, the combustion of hydrocarbon fuels strongly tends to form carbonaceous particles, that is soot [45]. Under the normal working condition of ICEs, most soot emissions formed in the early stages of combustion is completely burnt, owing to oxidation with the residual O₂ in oxygen-rich regions within the engine cylinder in the later stages of combustion. Therefore, the completeness of the soot emission oxidation process and the consequently final soot amount actually determine the engine PM characteristics [18,35]. Specifically, the local fuel-air equivalence ratio, in-cylinder temperature, pressure, and the reaction residence time are the key factors influencing the soot characteristics of ICEs [18,35]. Soot particles are typically formed early in the diffusion stage of combustions, due to the dissociation of fuels under a high fuel-air equivalence ratio and high-temperature conditions.

The soot emissions in all simulations are presented in Figure 12. The results showed a significant reduction of soot in the DF mode, compared to the diesel mode. The soot in the DF mode were almost zero. This is because of the significantly higher fuel-air mixing quality in the DF mode (using the port injection method), compared to that in the diesel mode (using the direct injection method). A higher fuel-air mixing quality resulted in significant reductions in the local fuel-air equivalence ratio in the DF mode. This reduced the soot formation in the DF mode. In addition, as well known, CH₄ is the simplest and cleanest hydrocarbon fuel. It does not contain C-C bonds in its chemical structure and contains no aromatics or sulfur. Therefore, it tends to minimally produce soot, compared to other hydrocarbon fuels [2,46]. The reduction tendencies in soot emissions in the DF mode which uses NG as the primary fuel, compared to the diesel mode which uses pure diesel had also been reported in previous studies [8,9].

In both diesel and DF modes, the injector SA of 150° produced the lowest soot emissions. This is thus the recommended SA for the injector of the engine to reduce soot emissions. It reduced 56% and 17% soot in the diesel and DF modes, respectively. Wei et al. [17] also revealed in their research that the injector SA of 150° produces the best emission performance.

3.5. CO₂ Emissions

The CO₂ emissions in all simulations are presented in Figure 13. The results showed a considerable reduction in CO₂ emissions in the DF mode, in comparison to the diesel mode. The DF mode reduced CO₂ emissions by approximately 20%, compared to the diesel mode.

As is globally known, carbon dioxide (CO₂) is a carbon-based emission. Its formation is directly dependent on the number of C (carbon) atoms contained in hydrocarbon fuels. Moreover, its formation is strongly influenced by the combustion quality in the engine cylinder. CO₂ is the final product of the complete combustion of hydrocarbon fuels. In the combustion process of ICEs, hydrocarbon fuel is firstly oxidized by O₂ contained in the charge air to form CO. CO is then oxidized to form CO₂ sequentially by the residual O₂ under high-temperature conditions in the engine cylinder. The higher the O₂ concentration and temperature, the greater the CO is oxidized.

In this study, the reduction in CO₂ emissions in the DF mode was related to two reasons: (1) compared to the diesel mode, the DF mode produced a better fuel-air mixing quality inside the engine cylinder that led to a better combustion quality; and (2) the cleaner characteristics and fewer C atoms of NG, compared to the diesel oil.

Chemically, assuming that in the ideal conditions, diesel oil (C₁₃H₂₃) and CH₄ are completely burnt. The chemical reactions for the combustion of the fuels are expressed as:

$${\mathrm{C}}_{13}{\mathrm{H}}_{23}+\frac{75}{4}{\mathrm{O}}_2\to 13{\mathrm{CO}}_2+\frac{23}{2}{\mathrm{H}}_2\mathrm{O}$$

(3)

$${\mathrm{CH}}_4+2{\mathrm{O}}_2\to {\mathrm{CO}}_2+2{\mathrm{H}}_2\mathrm{O}$$

(4)

Based on Equations (3) and (4), it is calculated that 16 g CH₄ produces 44 g CO_2, while 16 g C₁₃H₂₃ produces 51 g CO₂ if the fuels are completely burnt. In this study, the fuel dosage supplied to the engine in the diesel and DF modes were 0.0020667 g C₁₃H₂₃ and 0.001782 g (including 0.001682 g C₁₃H₂₃ and 0.0001 g CH₄) per cycle, respectively. It is very easy to calculate that, in the case of complete combustions, the diesel mode produces approximately 0.00658 g CO_2, while the DF mode produces only approximately 0.0049 g CO₂. The reduction in the CO₂ emissions when running in the DF mode, compared to the diesel mode, will thus be around 25%. However, due to the fact that the combustion of fuel inside the cylinder of actual ICEs will not be complete. As a result, the actual reduction in CO₂ when using CH_4, instead of diesel in real life will be a little different depending on the combustion efficiency of the engine. Compared to a reduced level of 20% in CO₂ emissions when running the engine in the DF mode instead of the diesel mode in this study, it can be concluded that the simulation results are reasonable. The reduction tendencies of the CO₂ emissions in the DF mode when using NG as the primary fuel, compared to the diesel mode, had also been reported in many previous studies [8,9].

In both diesel and DF modes, the injector SA of 150° produced the lowest CO₂ emissions. This is thus the recommended SA for the injector of the engine to reduce CO₂ emissions. It helped to reduce 0.77% and 0.31% CO₂ emissions in the diesel and DF modes, respectively.

Through the results presented in Section 3.4 and Section 3.5, it is very interesting to note that, in both diesel and DF modes, the injector with a SA of 150° produced the lowest soot and CO₂ emissions. This is related to the wall-wetting phenomena mentioned above. The injector SA of 150° makes the targeting injection point reach the center of the combustion chamber, far from the combustion chamber walls. This helped to avoid the wall-wetting of the liquid fuel on the combustion chamber walls, and thus reduced the soot and CO₂ formations. However, it should be noted that even though the simulation results pointed out the effects of injector SA on the CO₂ emissions of the engine, it is clear that the SA has only a marginal role in the CO₂ emission reduction, in comparison with the role of the alternative fuel (CH₄).

4. Conclusions and Discussion

This study numerically researched the effect of the injector spray angle on the combustion and emissions of a 4-stroke NG-diesel DF marine engine. The ultimate target of the study is to find out the optimal spray angle of the injector for the engine, in order to reduce exhaust gas emissions without using any post-treatment devices.

The major results of the study are listed as follows:

(1)

The in-cylinder peak temperature in the DF mode was lower than that in the diesel mode, due to the more uniform temperature distribution within the engine combustion chamber.

(2)

The DF mode reduced the NO emissions by up to 71%, compared to the diesel mode. The injector SA of 145° is the optimal value for reducing the NO emissions in both diesel and DF modes.

(3)

The DF mode significantly reduced the soot emissions, in comparison to the diesel mode. The soot emissions in the DF mode was almost zero. In both diesel and DF modes, the injector SA of 150° produced the lowest soot emissions. This is thus the recommended SA for the injector of this engine to reduce soot emissions. It helped to reduce 56% and 17% of soot in the diesel and DF modes, respectively.

(4)

A considerable reduction in the CO₂ emissions in the DF mode, compared to the diesel mode was observed. The DF mode reduced the CO₂ emissions by approximately 20%, compared to the diesel mode. In both diesel and DF modes, the injector SA of 150° produced the lowest CO₂ emissions. This is thus the recommended SA for the injector of this engine to reduce CO₂ emissions.

(5)

It is important to note that this study investigated the effects of the injector SA on the combustion and emissions of the engine, without considering the effects of the piston bowl geometry. Actually, the mixing quality of the fuel-air mixture which strongly influences the combustion quality of the engine depends not only on the injector SA but also on the piston surface shape. Therefore, the optimal SA for the injector in this study is applicable to the engines with a ω-type piston surface shape only. For engines with other types of piston surface shapes, such as a U-type or re-entrant types, the optimal SA for the fuel injectors might be different.

Based on the above conclusions, it is highly recommended to use a SA of 145° or 150° for the fuel injector of ω-type-piston engines, to reduce the NO or soot and CO₂ emissions, respectively, depending on which emission regulations need to be met while retaining the engine power.