[Mohammad74]

Hello,

I've been trying to simulate the air flow around and more importantly above an array of 5*5 cubic buildings. At first I generated a structured hex mesh using Gambit (since im much more familiar with it) but at some locations it did not give me acceptable results. I asked for a solution on the official Ansys forum and they told me that the error is due to the jump in cell dimensions in my original mesh and suggested to use polyhedral mesh.

H9ZZE0XCA108.jpgROSIOLUT6ZTY.jpg

So I went ahead and used ICEM to mesh my geometry using Delaunay tetrahedral scheme. Then smoothed it to be within the acceptable quality range and finally converted it to polyhedral mesh inside the Ansys Fluent. But now after the simulation, I'm getting a rough and weird velocity contour on the x-z plane:

ADLAAI4BB7D0.jpgS6TBMK3EMTXL.jpg

I suspect that this is most likely a mesh problem, but since I'm pretty new to ICEM and poly mesh, I don't know how to fix it. Any suggestions???

[Indra]

Originally Posted by Mohammad74

Hello,

I've been trying to simulate the air flow around and more importantly above an array of 5*5 cubic buildings. At first I generated a structured hex mesh using Gambit (since im much more familiar with it) but at some locations it did not give me acceptable results. I asked for a solution on the official Ansys forum and they told me that the error is due to the jump in cell dimensions in my original mesh and suggested to use polyhedral mesh.

Attachment 88295Attachment 88296

So I went ahead and used ICEM to mesh my geometry using Delaunay tetrahedral scheme. Then smoothed it to be within the acceptable quality range and finally converted it to polyhedral mesh inside the Ansys Fluent. But now after the simulation, I'm getting a rough and weird velocity contour on the x-z plane:

Attachment 88297Attachment 88298

I suspect that this is most likely a mesh problem, but since I'm pretty new to ICEM and poly mesh, I don't know how to fix it. Any suggestions???

has your solution converged, doesn't look very converged to me?

Secondly, for your original hex mesh, you could have always reduced the growth rate of the mesh to get a better result, for your 2D case there is no reason why poly would be better than hex. I don't see a boundary layer in the poly mesh, is it there?

[Mohammad74]

Originally Posted by Indra

has your solution converged, doesn't look very converged to me?

Secondly, for your original hex mesh, you could have always reduced the growth rate of the mesh to get a better result, for your 2D case there is no reason why poly would be better than hex. I don't see a boundary layer in the poly mesh, is it there?

hi,

yes it is converged and both of them are 3d problems. I did generate a prism layer near the walls in the poly mesh.

I also figured the problem with the poly mesh. I'm using a custom velocity profile at the inlet and my mesh wasnt able to capture the velocity gradient properly normal to the shown contour plane, so I had to refine it.

thanks for your answer.

[aero_head]

I think you should look into the concept of Y+ in order to help guide your mesh refinement.