Learning Forum

[murali666]

I am doing both RANS (k-w SST) and LES simulations to simulate a 3D vector field around a bridge pier.
I am using Hexa mesh in the whole flow domain developed in ICEM and providing the inflation layers near to the walls (bottom surface and cylinder) to accurately simulate the near wall effects. These inflation layers significantly effecting the mesh aspect ratio.
The mesh metrics of the developed mesh are
Aspect ratio = 1 (min) - 300 (max)
Erikson skewness = 0.8 (min) - 1 (max) (Recommended range of values as per ICEM manual is 0.5 to 1)
Mesh Quality = 0.8 (min) - 1 (max) (close to 1, better the mesh)
Determinant 3*3*3 = 0.92-1 (close to 1, better the mesh)
All the mesh metrics other than aspect ratio are in the acceptable range. The mesh elements with the worst aspect ratio are present in inflation layer region. The improvement of aspect ratio in that region leads to a significant increase in the number of mesh elements, such high number of mesh elements are not necessary in the regions away from the wall.
So please explain me the importance and aspect ratio in both RANS (k-w SST) and LES simulations and what will the suggested action in the above explained situation.

[Nikhil Narale]

Hello Murali,

High ARs in the boundary layer cells are normal and that too without affecting the convergence and accuracy*. It might go up to 1000 or even more in some cases. The benefit of high AR in the boundary layer mesh is reduced cell count. This logic stands strong in the boundary layer (near wall) region, where we expect substantial flow variation only in one direction. It make sense to have high mesh density in the normal direction (to capture large gradient) while having low mesh density in the stream-wise direction, where the gradients are small, which is achieved by stretched cells or high AR cells.
*This is applicable unless you have to deal with the separated flows. When the flow separates, the gradients in the stream-wise direction is also large near the wall, which needs to be captured using high density mesh in the stream-wise directions. This will result into low AR mesh with high cell count.
As per my understanding, it makes sense if we have high AR in the attached flow regions (bottom surface), while maintaining low AR on the cylinder. Also, I would suggest to check y+ value and keep it within the recommended valve, depending on the turbulence model selected.
I hope this helps!

Nikhil