Project 1 : Flow over Ahmed Body

AIM: To simulate airflow around an Ahmed Body using STARCCM+ and compare the results with the given experimental data.

PROCEDURE:

The Ahmed Body was created using the in-built 3D CAD modeller in STARCCM+. The dimensions of the body were provided with two different slant angles of 25 and 35 degrees. The simulation was run for both the slant angles using the k-epsilon and k-omega SST models for each case.

The initial steps from geometry modelling to simulation set-up were same for all the four cases and hence will be described in the beginning. The post-processing results will be discussed separately.

STEP - 1: Geometry Modelling:

The side view of the geometry was sketched and then extruded to get the 3D block shape. Make sure that the rear end of the Ahmed Body lies on the origin since the experimental data provided is with respect to the rear end at the origin.

Dimensions of Ahmed Body

On the lower face of the block, 4 circles were sketched at the appropriate given distances to create the stubs of the Ahmed Body.

These circles were extruded below by 50mm. Fillet option was used to provide the fillets to the front surfaces.

Ahmed Body Geometry

The next step is to create a Wind Tunnel around the Ahmed Body. The wind tunnel was created in two blocks to take into account the slip and no slip conditions at the road to get the desired boundary layer thickness.

The dimesnsions of the wind tunnel are 15x1.16x1.5m.

Wind Tunnel around the geometry

Wind Tunnel Boundaries

The common wall between the two blocks was deleted to create a single block. Only the part of the road on which the Ahmed Body rests was given the Wall boundary condition. The inlet and outlet were given the Velocity Inlet and Pressure Outlet BCs while all other faces of the wind tunnel were given Symmetry Plane BC.

STEP - 2: Surface repair and surface wrapping:

Initially, the Ahmed Body had 66 pierced faces along with face quality errors. Surface wrapping was done to remove these pierced faces errors. A base size of 40mm was used along with custom controls for the Ahmed Body to create a finer mesh. Contact prevention was used between the Ahmed Body and the road surface.

Surface wrapped geometry

After wrapping the surface, it was again checked for any surface defects. Only two face quality errors were found.

STEP - 3: Mesh Generation:

To get the required boundary layer thickness at the Ahmed Body, we need to do some calculations. To calculate the Reynold's Number, an inlet velocity of 40 m/s was used. We need to calculate Rex, that is the Reynold's number for flow length 'x'. Here, the flow length (x) is 2m, which is the distance from the start of the 'Wall' BC of the road to the front of the Ahmed Body. The air flow parameters of density and dynamic viscosity at 25 C are 1.184 kg/m^3 and 1.849*10^-5 Pa-s.

To find the Boundary Layer Thickness, the formula for turbulent flow over a flat plate was used. The formula is -

The value of boundary layer thickness was found to be 0.0308m or 30mm. This will be the total thickness of the Prism Layers along the wind tunnel.

After repairing the surface, the part is to be assigned to Regions. For meshing, the following 4 meshers were selected -

Surface Remesher

Automatic Surface Repair

Trimmed Cell Mesher

Prism Layer Mesher

A base size of 0.3m was selected. The Total Prism Layer Thickness was provided as 30 mm (calculated from the boundary layer thickness formula). Custom surface controls were used for the Ahmed Body and the legs(stubs). A cubic volume was created around the Ahmed Body to create a finer mesh for better resolution. A custom size of 15% relative to the base size (or 4.5cm) was used for this volume.

The Base Size of the volumetric control could have been even smaller but was kept moderate to get a balance between solution accuracy and the available computational power.

A plane section was created at the z=0 axis to get a better representation of the mesh.

STEP - 4: Setting up the simulation:

After mesh generation, we need to provide Physics conditions. The following physics models were used for the k-epsilon case:

The following physics models were used for the k-omega case:

Inlet velocity was considered as 40 m/s.

STEP - 5: Setting up reports:

The following reports and plots were set up -

Drag Force Coefficient

Lift Force Coefficient

Surface Average Velocity

The Frontal Area reports were also used as input values for the force coefficient reports.

Line Probes were created to get the values along them. The probes were created at the given values of the X & Z co-ordinates at y=0 plane.

Line Probes created at the given co-ordinates

The stopping criteria was set to 3000 iterations.

RESULTS & POST-PROCESSING:

After the simulation results were obtained, an X-Y plot was plotted using the obtained results. On the X-axis was the length in x direction and on the Y-axis were the velocity magnitude values along the Line Probes.

To plot the data points of the experimental values from the given excel sheet, we need to first copy the required columns in another excel sheet and save it as a .csv file. This file is then imported into STARCCM+ in the Tables section under Tools and later plotted on the X-Y plot using the Data Series functionality.

CASE - 1: Slant Angle 25 & K-Epsilon turbulence model

1. Residuals and Surface Averaged velocity plots:

The residuals plot indicated the convergence of the solution. In the above plot, the residual values have converged upto a very small magnitude. The plot of a couple of values, namely the Continuity and X-momentum do show some oscillations of small magnitudes. However, the solution can be treated as a converged solution as the surface averaged values (in this case the surface averaged velocity) have stabilised. Hence, the above solution can be treated as a converged solution.

2. Scalar Contours:

Velocity and Pressure contours are very helpful to study the flow distribution around the Ahmed Body.

3. Drag and Lift Coefficients:

The drag and lift coefficient plots gave the following values -

Drag Coefficient - 0.364683

Lift Coefficient - 0.112101

4. Velocity Magnitude vs X co-ordinate plot:

This plot was plotted to compare the experimental and simulation results. First, an X-Y plot was created which gives the velocity magnitude values along the probes (which were created with reference to the available experimental values) and the experimental values were added to the plot for better comparison.

The red lines indicate the values given by the probes in the simulation whereas the blue dots represent the experimental values provided. As can be observed, most of the values given by our simulation match perfectly with the experimental results. A few experimental values do vary from those obtained through simulation. The main reason for this difference is the variation in inlet conditions of velocity and pressure as well as the different dimensions of the Wind Tunnel. The wind tunnel dimensions and inlet parameters were assumed for the simulation since exact values used in the experiment were not available. However, the two results do match to an acceptable level.

CASE - 2: Slant Angle 25 & K-Omega turbulence model

1. Residuals and Surface Averaged velocity plots:

In this case as well, oscillations of small magnitude persist, but the surface averaged values have stabilised. Hence, the obtained solution is treated as converged.

2. Scalar Contours:

3. Drag and Lift Coefficient:

The drag and force coefficient values are:

Drag Coefficient - 0.358914

Lift Coefficient - 0.1157027

4. Velocity Magnitude vs X co-ordinate plot:

As in Case 1, most simulation values in the above plot do match with the experimental values upto an acceptable level.

CASE - 3: Slant Angle 35 & K-Epsilon turbulence model

1. Residuals and Surface Averaged velocity plots:

In this case as well, oscillations of small magnitude persist, but the surface averaged values have stabilised. Hence, the obtained solution is treated as converged.

2. Scalar Contours:

3. Drag and Lift Coefficient:

The drag and force coefficient values are:

Drag Coefficient - 0.40020624

Lift Coefficient - 0.07125547

4. Velocity Magnitude vs X co-ordinate plot:

As is with the above cases, most of the simulation values are comparable with the experimental values.

CASE - 4: Slant Angle 35 & K-Omega turbulence model

1. Residuals and Surface Averaged velocity plots:

In this case as well, oscillations of small magnitude persist, but the surface averaged values have stabilised. Hence, the obtained solution is treated as converged.

2. Scalar Contours:

3. Drag and Lift Coefficient:

The drag and force coefficient values are:

Drag Coefficient - 0.39108433

Lift Coefficient - 0.03854110

4. Velocity Magnitude vs X co-ordinate plot:

CONCLUSION:

• As can be observed from the results, the inconsistency in the simulation and experimental results is mostly in the recirculation region behind the Ahmed Body. This is due to the fact that the recirculation zone is not correctly captured in the simulation and the simulation values are over-predicted (higher than the actual experimental values).
• Both the k-epsilon and k-omega turbulence models work fine in this case and gave similar results for drag and lift coefficients. However, the k-epsilon model gives better accuracy in freestream regions (away from the boundary) and the k-omega gives better accuracy for near-wall regions. Hence, depending on the area of interest, a suitable turbulence model should be used.
• In conclusion, a simulation with a finer mesh size (finer mesh wasn't used due to the limited computation power available), the actual input parameters and wind tunnel dimensions will give us a solution close to the observed results.