Surface preparation and Boundary Flagging of Port Fuel injection Engine

OBJECTIVE

To perform surface preparation and boundary flagging on a Port fuel injection (PFI) engine and set up a No-hydro simulation.

PROCEDURE
The geometry from the CAD software is imported in STL format and diagnosis is performed on the geometry for all the possible errors present in the design. So, after performing the diagnosis it was found out that there were errors associated with the intersection between the triangles, Open Edges, Non-manifold edges, and normal orientation which has to be rectified before proceeding with case setup for CFD simulation.

The majority of the errors were associated with intersections between triangles which creates more problems while running the simulations. Each and every error should be removed before setting up the case for CFD and this process of cleaning the geometry is called Surface Cleanup.

The complete geometry has to be viewed properly to identify the errors and try to eliminate them before moving further.

So, before starting the surface cleanup we need to flag the boundaries. For that, we make use of the ‘boundary flag’ option to select the surface and assign them to an individual component. During this process we also make use of ‘boundary fence’ to clearly define our boundary, because sometimes the angle method does not work efficiently it automatically starts propagating from the location and whenever an edge appears, the selection is going to continue until the angle between the normal is lesser than 20 degrees. So, in those cases, the boundary fence method plays a very efficient role and hence reduces the human effort.

After close analysis, it was found out that the valve from the intake port is shifted a bit upward which is superimposing with the cylinder head, creating intersection errors. It can be clearly seen from the image below that the valves are intersecting with the intake ports.

Before resolving this issue, it would be easier if we first flag the remaining components of the geometry so that we are left with minimum components to handle. The intake and exhaust valves are flagged as top, angle, and bottom boundaries respectively as shown in the figure to resolve the flow and to get correct disconnect triangles.

To shift the valve a small distance we make a three-point arc on the circumference of the circle and get the arc normal. The coordinates of the vector are taken and we translate by 1mm. This pushes the intake port and the problem with superimposing triangles are resolved as shown in the figure.

After again running the diagnosis, we find problems remaining with open edges and normal orientation. By using the Repair-Patch option the open edges are resolved successfully. The normal orientation errors are the simplest type of errors, which can be solved by making sure that the normal is pointing inside the fluid domain, so, we select a particular region by angle method and then click on transform which changes the direction of normal orientation. Hence, all our errors are resolved to make it fit for the case setup for CFD simulations.

One modification is made for simplifying the simulation before we move on to set up initial and boundary conditions we divide our intake port into two sections because fuel is injected though only one half of the intake port and other part has no fuel spray. To initialize the fuel that comes into the combustion chamber we split the intake port into two different regions namely Intake port A (with fuel) and Intake port B (without fuel) respectively as shown in the figure.

After we are done flagging the boundaries to the individual component we assign regions. As we group all the triangles together to form boundaries, we also group all the boundaries to form regions. A region should always be a closed volume. One region can interact with the other but the interface should be clean. We have created 4 regions namely, cylinder, Intake port-1, Intake port-2, and Exhaust port.

While defining the regions, we should make sure that the bottom valves of both the intake and exhaust side should lie inside the cylinder regions and rest lie in the respective intake and exhaust regions.

DISCONNECT TRIANGLES CONCEPT:

Whenever there is a small gap in between two regions, converge automatically creates something called disconnect triangles. When these disconnect triangles are created flow through them is restricted. The disconnect triangles form a perfect cylinder or it is aligned with how the valves are aligned. These disconnect triangles should be as small as possible. So, smaller the gap, larger the simulation time. These disconnect triangles should be made carefully because if our valve is quite open assuming that they are closed, we are under predicting the mass flow entering into the domain.

INITIAL DATA:

The physical parameters of an IC Engine are as shown below:

Here the combustion process is mainly focused on the compression and expansion, so the analysis of the model is made from -480֯ bTCD (i.e. inlet valve closure) to 240֯ aTDC (i.e. Exhaust Valve Closure).

For our simulation, we are provided with ‘mechanism files’ which provides us with chemical reactions and for each gas-phase species are provided with thermodynamic properties. We are only interested in mesh generation and not running the CFD combustion simulation, so we undergo a No-hydro simulation where we can analyze whether of our valve is opening and closing at proper time and meshes are being formed properly. We are using isooctane (iC8H18) whose physical properties are close to gasoline fuels.

Table 1. shows the regions and boundary conditions for every single part present in the port fuel injection engine.

1. The inflow boundary must have a mass fraction equal to 1 which contains the constituent of air i.e. Nitrogen N₂ (0.77292) and Oxygen O₂ (0.20157) and iC₈H₁₈ (0.025508) respectively.
2. The combustion reaction of our fuel is given by the equation:

C₈H₁₈ + 12.5*(O₂ + 3.76N₂) = 8CO₂ + 9H₂O + (3.76*12.5)N₂

3. The outflow flow boundary contains Carbon Dioxide CO2 (0.192304), Water H2O (0.088559), and Nitrogen N2 (0.719137) having stoichiometric concentrations.
4. We assume that combustion is stoichiometric and the concentration of species in our outflow boundary when the combustion products are released is given by the following table:

STOICHIOMETRIC CALCULATOR:

Species	Moles	Molecular Weight
CO2	8	44.01
H20	9	18.01528
N2	47	28.0134

The different regions are provided with boundary conditions shown in the figure:

Based on the valve timing for the intake and exhaust profiles we must connect and disconnect from the combustion chamber using the cyclic option in the Event box.

Since we are having regions and the two intake ports A and B are separated we provide an additional permanent event to it as shown in the figure.

The turbulence modeling is all about how you calculate turbulent viscosity, and for this several models are present out of which we make use of the standard ‘RNG K-epsilon model’ as this turns out to be a more efficient model for IC Engine applications. The base grid of 4mm is used in the entire geometry.

As we are taking into account the combustion modeling, we will need to redefine the mesh at particular locations at the time of combustion to reduce numerical diffusion. As the combustion takes place we understand there is a lot of change in velocity and temperature which needs to be captured for efficient calculations. So, we make our meshes temperature and velocity-dependent with Adaptive mesh refinement (AMR) using a maximum embedding level of 3 for each case. The AMR automatically increases the grid resolution based on the curvatures of velocity and temperature. The grid refines when the change in velocity (Δv) is greater than 1m/s and change in temperature (ΔT) is greater than 2.5K.

Fixed Embedding is provided to refine the grid at specific locations to improve the accuracy of our solutions. We have used a fixed Embedding technique to refine the meshes at the intake port-1 and the exhaust ports with a scale of 3 which scales down to 0.5mm cells at those locations which help to capture the combustion products at exhaust more accurately. An additional embedding is given for our spark plug so when energy is sourced onto spark plug the combustion propagates, so as to capture those energy more efficiently thus reducing numerical diffusion and inch closer to more accurate results. The embedding is visualized in the picture given below as small and large embedding with 2mm and 1mm cells respectively.

CONCLUSION

We have setup the geometry and running a No-hydro simulation to understand the opening and closing of the valve and proper mesh generation at a particular instant. If this No-Hydro simulation turns out to be satisfactory, which means that our valve and meshes are performing in a particular manner at the time, we will then perform a Full-Hydro simulation taking into account the Navier Strokes Equation and combustion equations required for the CFD analysis.