External Airflow Simulation

OBJECTIVE:

The primary objective of the project is to simulate an external airflow over an Ahmed body in STAR CCM+ and to validate the results by comparing the velocity profiles at different locations with the experimental results done originally. This is done for two geometries with two different slant angles( 25 degrees and 35 degrees) and for two different turbulent models. Also, to calculate the drag coefficient and lift coefficient for all the configurations.

PROCEDURE:

Creating the 3D CAD model:

The first step is to create the CAD model as per the original dimensions done in the experiment. It is as follows,

Fig-1:Ahmed body dimensions

For slant angle = 25 degrees:

So 2D sketch is drawn as per the given dimensions and it is extruded for the given width. The sketch is as follows.

Fig-2:2D sketch of Ahmed body

Fig-3: 3D CAD model of Ahmed body with slant angle of 25 degrees

Fig-4:3D CAD model of Ahmed body with virtual wind tunnel

Wind tunnel dimensions:

The dimensions of the wind tunnel is taken as 1.87 m high and 1.4 m wide and the length is specified based on the requirement. These dimensions are taken from the experimental data of the original paper.

For slant angle = 35 degrees:

Fig-5:2D sketch of Ahmed body with slant angle of 35 degrees

Fig-6:3D CAD model of Ahmed body with slant angle of 35 degrees

Surface repair:

Once the CAD geometry is created, new parts are created from them and some boolean manipulations are done to extract the required volume out of the geometry. Then the part is checked for surface issues and if any, it should be repaired.

For slant angle = 25 degrees:

Fig-7:Surface check of Ahmed body with slant angle of 25 degrees

For slant angle = 35 degrees:

Fig-8:Surface check of Ahmed body with slant angle of 35 degrees

As it can be seen, both the geometries are free of critical issues like non-manifold edges, non-manifold faces or pierced faces. There are only face quality issues which can be easily rectified during surface remeshing. SO, both the models are ready for meshing.

Assigning regions and setting boundary conditions:

As a next step, these parts are assigned to regions with separate boundary for each part surface. After that, appropriate boundary conditions are given for each boundary as below.

1. Inlet - Velocity inlet
2. Outlet - Pressure outlet
3. Ground before the ahmed body - Wall (Slip)
4. Ground starting from the body till the end - Wall (No slip) (This is done to achieve correct boundary layer thickness)
5. Ahmed body - Wall
6. Other sides of wind tunnel - Symmetry

Meshing :

After setting up the boundary conditions, with appropriate parameters, mesh model is generated.

Meshers:

1. Surface remesher
2. Automatic surface repair
3. Trimmed cell mesher
4. Prism layer mesher

Default controls:

1. Base size : 20 mm
2. Maximum core/Prism transition ratio : 2

Surface control - Ahmed body :

1. Target size : 10 mm
2. Minimum size : 5 mm
3. Number of prism layers : 5
4. Prism layer thickness : 6.66 mm
5. Wake distance : 1 m
6. Wake spread angle : 0.1 radian
7. Wake refinement isotropic size : 15 mm

Surface control - Ground :

1. Target size : 40 mm
2. Minimum size : 40 mm

Surface control - Wind tunnel :

1. Target size : 60 mm
2. Minimum size : 60 mm

Mesh validation:

After generating the mesh, it should be diagonosed for validation. So, the above mesh model is diagonosed and found it is valid without any negative volumes. And the total number of cells are 870618.

Y+ value:

For this simulation, it is essential to place the first cell the wall in log-law layer(30 < y+ < 500). So for y+ = 55, the first cell thickness has to be 5 mm for the given velocity, density and viscocity. So the prism layers are generated in a way that the first cell thickness is 5 mm.

For slant angle = 25 degrees:

Fig-9:Mesh model

Fig-10: Mesh model of Ahmed body with slant angle of 25 degrees

Fig-11: Sectional view of the mesh model (Slant angle=25 deg)

Fig-12: Prism layers (Slant angle=25 deg)

For slant angle = 35 degrees:

Fig-13:Mesh model of Ahmed body with slant angle of 25 degrees

Fig-14: Sectional view of the mesh model (Slant angle=35 deg)

Physics continua and models:

The next step is to choose the appropriate solvers and physical models in order to define physics continuum in a right manner. Here, both the mesh models (25 degrees and 35 degrees) are simulated with two different physical configurations with different turbulence models.

First Configuration : K-EPSILON :

1. Space : Three dimensional
2. Time : Steady
3. Material : Gas (Air)
4. Flow : Coupled flow
5. Equation of state : Constant density
6. Viscous regime : Turbulent
7. RANS : Realizable K-Epsilon
8. Wall treatment : All y+ wall treatment ( First cell is placed in log-law layer)

Second Configuration : K-Omega SST :

1. Space : Three dimensional
2. Time : Steady
3. Material : Gas (Air)
4. Flow : Coupled flow
5. Equation of state : Constant density
6. Viscous regime : Turbulent
7. RANS : K-Omega SST ( Highly reliable for boundary layer problem,lift and drag calculation)
8. Wall treatment : All y+ wall treatment ( First cell is placed in log-law layer)

Flow parameters:

1. Inlet velocity : 40 m/s (Velocity used in the original experiment)
2. Outlet pressure : 1 atm
3. Reynolds number : 7.68 × 10⁵
4. Turbulent intensity : 1%

Calculation:

Once all the models and parameters are specified, the simulation is initialized and run for 1300 iterations.

Slant angle =25 degrees, K-Epsilon :

Fig-15:Residuals(25 degrees, K-Epsilon)

Fig-16:Velocity monitor (25 degrees, K-Epsilon)

Slant angle =25 degrees, K-Omega SST:

Fig-17:Residuals (25 degrees, K-Omega)

Fig-18:Velocity monitor (25 degrees, K-Omega)

Slant angle =35 degrees, K-Epsilon :

Fig-19:Residuals (35 degrees, K-Epsilon)

Fig-20:Velocity monitor (35 degrees, K-Epsilon)

Slant angle =35 degrees, K-Omega SST :

Fig-21:Residuals (35 degrees, K-Omega)

Fig-22:Velocity monitor (35 degrees, K-Omega)

Convergence:

From the above plots, especially from the velocity monitor plots for all the four setups, it is clear that the solution is converged.

RESULTS:

1.Velocity contour:

Fig-23:Velocity contour (25 degrees, K-Epsilon)

Fig-24: Velocity contour (25 degrees, K-Epsilon)

Fig-25:Velocity contour (25 degrees, K-Omega)

Fig-26:Velocity contour (35 degrees, K-Epsilon)

Fig-27:Velocity contour (35 degrees, K-Omega)

Inference from the velocity contour:

1. In all the setups, the velocity reaches it's maximum at the front curvatures and at the start of the slant in the rear end. This is due to the fact that the acceleration increases in a curved path. Thus the velocity is directly proportional to the radius of curvature.

2. For slant angle = 25 degrees, the K-Epsilon model failed to capture the flow separation accurately. But the K-Omega SST captured it effectively. It can be seen from the wake region development.

3. For slant angle = 35 degrees, both the turbulence models managed to capture the separation correctly.

4. The separation of flow is high in 35 degrees than in 25 degrees which is an expected outcome.

2. Velocity profiles:

The most important aspect of validating the simulation results with the experimental data is comparing the velocity profiles at various locations. So line probes are created as depicted below and velocity profiles are compared.

Fig-28:Line probes creation

Slant angle =25 degrees, K-Epsilon :

Fig-29:Velocity profile (25 degrees, K-Epsilon)

Fig-30:Velocity profile (25 degrees, K-Epsilon)

Slant angle =25 degrees, K-Omega SST :

Fig-31:Velocity profile (25 degrees, K-Omega)

Fig-32:Velocity profile (25 degrees, K-Omega)

Slant angle =35 degrees, K-Epsilon :

Fig-33:Velocity profile (35 degrees, K-Epsilon)

Fig-34:Velocity profile (35 degrees, K-Epsilon)

Slant angle =35 degrees, K-Omega SST :

Fig-35:Velocity profile (35 degrees, K-Omega)

Fig-36:Velocity profile (35 degrees, K-Omega)

Inference from the velocity profiles :

1. In the first glance itself, it is evident that the K-Omega SST turbulence model is better in matching the velocity profile with that of the experimental data. This is because of the ability of the K-Omega SST model to calculate the wall shear stress correctly, thus predicting the flow separation(wake region) and therefore the velocity.

2. In all the four setups, it is seen that as the location of the line probes goes farther, the velocity profile tranforms from a diffused low range of values to a fully developed profile(Bell shaped curve).

3. For slant angle = 25 degrees, the curves are very narrow than that of the 35 degrees. This is because of the size of the wake region.

4. The profiles of the lines at the negative distance are mere straight lines since they are above the body itself.

5. Even the slighest of the highs in the middle of the curves are captured correctly by the K-Omega SST model.

3. Drag and Lift coefficients:

Fig-37:Comparison of Drag coefficient plots

Drag coeffcients table:

	Slant angle = 25 deg	Slant angle = 35 deg
K-Epsilon	0.327	0.313
K-Omega SST	0.303	0.294

Lift coeffcients table:

	Slant angle = 25 deg	Slant angle = 35 deg
K-Epsilon	0.822	0.745
K-Omega SST	0.809	0.734

Inference from the force efficients:

1. The drag coefficient from the experimental data for slant angle of 25 degrees is 0.299 and for slant angle of 35 degrees is 0.279. So from the above table, it is evident that K-Omega SST model is better in calculating the drag coefficient. This is primarily due it's ability to predict the near wall damping and to calculate wall shear stress correctly with the help of viscocity limiter.

2. Increase in the rear slant angle causes a decrease in the drag coefficient.

4. Streamlines:

Slant angle =25 degrees, K-Epsilon :

Fig-38:Streamlines (25 degrees, K-Epsilon)

Fig-39:Streamlines (25 degrees, K-Epsilon)

Slant angle =25 degrees, K-Omega SST :

Fig-40:Streamlines (25 degrees, K-Omega)

Slant angle =35 degrees, K-Epsilon :

Fig-41:Streamlines (35 degrees, K-Epsilon)

Slant angle =35 degrees, K-Omega SST :

Fig-42:Streamlines (35 degrees, K-Omega)

Streamlines comparison:

Video-1:Animation of vortices

Video-2:Animation of streamlines

Inference from the streamlines:

1. The recirculation region captured more accurately by the K-Omega SST model in both the slant angles. This again due to the fact, this algorithm can predict flow separation better than other 2 equation models.

2. The vorticity is high in the model with the slant angle of 25 degrees. But the length of the vortex along x-direction is lower than that of the slant angle of 25 degrees.

3. The vortices are more attached in the 25 degrees in constrast to the 35 degrees model which is highly detached.

CONCLUSION:

Thus an external airflow simulation around the Ahmed body is done and the results are in agreement with the experimental data. From the simulation, the first thing to be concluded is that a change in the rear slant angle of the Ahmed body has huge effects on the velocity fields,velocity profiles, force efficients and also on the vortex generation. And it is clear that irrespective of the slant angles, the K-Omega SST turbulence model is much better than K-Epsilon turbulence model in calculating the flow properties for the flow around the Ahmed body. Also, it should be noted increasing the rear slant angle reduces drag noticeably.