Introduction

As the need for energy is raising day by day, more efficient methods for implementing the natural sources should be achieved. Obtaining energy by combustion is one of the most classical methods, yet the most important [1],[2]. Many research efforts, such as [3],[4], [5] and [6] have been dedicated in the last decades to reach an optimum burning of the fuel in the Internal Combustion Engines (ICE).

 The using of numerical analysis techniques to predict different aspects of the combustion process has become popular recently in the discipline of (ICE) [7],[8],[9]. However, due to the extreme complexity of the combustion process, the researchers tend to include more factors in simulation to obtain results closer to the empirical ones. Nevertheless, adding large number of factors and constants could lead to the desired results only for one specific engine at one well-determined working regime. Each time the numerical analysis is applied for different engine or different working regime, a large set of constants needs to be calibrated once again by comparing the simulation results with the experimental ones, if the latest are available [10].

Therefore, the authors of this paper propose approaching the problem from innovative point of view. Instead of adding more constants, the problem should be divided into smaller and simpler problems. At the end, when all the involved phenomena are investigated, the connection between the problems can be made intellectually.

Thus, the present paper aims to gain a better understanding of the combustion process in diesel engines by simulating the flow of the gases inside the cylinders. For this purpose, three different simulations are included in this research: intake and exhaust gases flow during the valves overlap opening, gases flow at the end of the compression stroke and fuel flow from the injector nozzle. All these simulations are performed on CFD analysis tool provided in Ansys Workbench.

Simulation description

Conform the main goal of the paper stated above, three simulations are performed on the in-cylinder processes and they are outlined in the following way:

2.1. Simulation I: Fluid flow during the scavenging phase

            The scavenging process has a significant effect on many parameters of the combustion process, from which we mention: temperature and flow shape of the gases inside the cylinder before ignition and temperature of the exhaust gases [2],[11],[12].

2.2. Simulation II: In-cylinder fluid flow at the end of compression stroke

The flow shape of the gases in the pre-ignition phase influences the mixture formation, ignition delay and combustion products [13],[14]. The gases flow can be divided into two distinct motions: rotational around the vertical axis of the piston (swirl flow) and radial  towards the center of the cylinder (quench flow) [2],[15].

2.3. Simulation III: Fuel spray

The injector nozzle should be designed to assure an optimal distribution of the fuel in the combustion chamber, which leads to form an homogeneous mixture. In addition, the fuel flow in the injector and in the combustion chamber affects the droplets size and number. In order to obtain a maximum surface of contact between fuel and oxidizer, the droplets size is required to be as small as possible while the droplets number is needed to be as large as possible [16].

Simulation stages

Simulations include three basic stages: preprocessing, processing and post-processing. The details of every stage differ depending on the type of the simulation and the programs used. CFD analysis in Ansys Workbench consists the following sub-stages: geometry, mesh, setting, solution and results. These steps are described in detail in [17] and [18].

The distinct feature of the simulations presented in this paper over other methods of approaching the combustion process is the ability to apply the same general steps for all engine types and working regimes if the geometric model, boundary conditions and solver options are assumed correctly. Furthermore, although three different fluid flow simulations are performed on three different components of the engine, the simulation stages are almost the same, except some small particularities which will be pointed clearly.

3.1. The geometric model

Three geometric models are built for the three simulations. Each one of these models is built in such a way that permits studying the related simulation accurately. Generally, these models can be constructed in any CAD program, such as: SolidWorks, AutoCAD Mechanical, Ansys and Catia. The authors chose Catia for its easy exporting feature into Ansys.

The shape and dimensions of the geometric models are based on the engine detailed description provided in [19]. Some of the tiny details can be neglected due their small influence on the simulation results. The geometric models for the simulations I, II and III are shown in Figure 1, Figure 2 and Figure 3 respectively.

Fig.1 Geometric model for analyzing the scavenging process

Fig.2. Geometric model for predicting the shape of the gases flow at the end of the compression stroke

Fig.3. Geometric model used to simulate fuel flow in the injector nozzle

3.2. Meshing

As the simulations of this research are based on the Finite Elements Method (FEM), the geometric model should be divided into a large number of tiny elements.

The equations which describe the fluid flow are written for each element separately. After that, the fluid flow is predicted in the whole body by the common solution of the equations obtained in these elements.

It is well known that a better prediction of the fluid flow can be achieved when the elements size is smaller. However, the reduction of the elements size is limited by the time needed to solve the mathematical model which depends mainly on the computer properties. Thus, if we keep in mind that the longer the time needed for simulation, the higher the costs, a compromise between the elements size and simulation time should be considered.

The method followed in this paper to reach this compromise is to start with a mesh having low quality, then the mesh quality is increased as long as the time elapsed to perform the simulation is considerably acceptable.

A chief indicator of reaching an optimum quality for the mesh is when the mesh is refined and the same results are obtained. Another indicator is the homogeneousness of the elements throughout the geometric model [20].

It is worth mentioning that during the meshing stage, it is important to deal with the locations of the body which have great influence on the flow shape with a higher attention. Ansys provides the user with many options to achieve a high-quality mesh such as: sizing, contact sizing, refinement, mapped face meshing, match control, pinch and inflation.

The final mesh adopted for the simulations I, II and III can be seen in Figure 4, Figure 5 and Figure 6 respectively.

Fig. 4. Meshing of the geometric model adopted for the simulation I – sectional view

Fig. 5. Meshing of the geometric model adopted for the simulation II – sectional view

Fig. 6. Meshing of the geometric model adopted for the simulation III – sectional view

The boundary conditions are determined in this stage. Basically these conditions should be set as close as possible to the real exploitation conditions. However, depending on the simulation purpose, the boundary conditions can be chosen to answer the question: “What would happen if these were the boundary conditions?” Therefore, the boundary conditions of the simulations in this paper are firstly chosen close to real conditions. Bearing in mind the papers main aims, small deviations from the real engine working regime are added to make the results more visible.

3.4. Solution

            The time needed to obtain a stable solution depends on the computer properties, the elements size and number and the solution options. The method followed by Ansys to solve the mathematical model is based on giving an approximate solution for a specific number of iterations. The solution is considered satisfactory when an acceptable convergence of results is reached. The average time elapsed to reach the results convergence for these simulations is about 33 minutes within approximately 60 iterations.

3.5. Results

            The outcome of the numerical analysis can be represented in tables, graphical curves or in 3D illustrations. The results illustrations are presented in Figure 7 and Figure 8 for simulation I, Figure 9 and Figure10 for simulation II and in Figure 11 and Figure 12 for simulation III.

In Figure 7 we can see the streamlines of the gases and the pressure on the internal walls of the studied model. Colors from red to blue indicate the magnitude of speed and pressure ranged between the highest and lowest values respectively.

A top-view of the gases streamlines are shown in Figure 8 where the walls of the model are made transparent to visualize the internal space.

Fig.7. The flow of the gases during the scavenging phase – sectional front view

Fig.8. The flow of the gases during the scavenging phase – top view

Fig.9. The flow of the gases at the end of the compression stroke

Fig.10. The flow of the gases at the end of the compression stroke – sectional front view

Fig.11. The flow of the fuel at the injection moment – sectional front view

Fig.12. The flow of the fuel at the injection moment – top view