Question

2. Based on initial guesses, a mesh for a CFD simulation is set up for an external ow around an object. The
ow does not experience signicant separation on the body surface. The results yield y+ values between 4
and 40 on the the body surface. Explain why this result is not adequate. Explain how you would proceed to
improve the results, and how you would know when they are acceptable.

Answer 1

I am mentioning some key points, Kindly go through them thoroughly

1. The results are not adequate because for the simulation of external flow with boundary layer separation, the y⁺ value should always be less than 1 i.e., y⁺ < 1. This is because for accurate prediction of point of separation of the boundary layer it is required to resolve the laminar sub-layer and thus, y⁺ <1.

Ways to improve

2. I think you are using RANS for modeling turbulence, specifically go with SST k-w turbulence model with low-Re correction, as it is suitable; for resolving low-Re regions in the flow i.e. boundary layer of the flow.

3. Go finer mesh near the wall so that the y⁺ < 1, by using the inflation layer in the mesh.

4. Keep the aspect ratio of the mesh in the main volume to be less than 30 and, in the inflation layer to be less than 100 if you are using Fluent like Solver else with CFX solver the aspect ratio in the inflation layer can be up to 300.

5. Then, check for y⁺ again. It should be less than 1. Else refine the mesh further.

6. Finally check the predicted value of the drag coefficient against the experimental value. A match obtained with 12 - 16 % error is acceptable.

The results based on this methodology will be surely acceptable.

Please RATE and VOTE.