AnsweredAssumed Answered

Is having more cells across a tight restriction "better"?

Question asked by Jordan Truitt on Jul 31, 2020
Latest reply on Aug 10, 2020 by Joe Galliera

Is there an ideal amount of cells across a restriction? I think general rule of thumb is 4-5. I'm simulating a tight restriction, 0.05 mm in height, using the adaptive refinements. What I'm noticing is the pressure drop shows good convergence through the first several refinements, the % change flattens out for 2-3 refinements and looks great and I'd call it good. But if I keep refining I eventually I get a jump in pressure that never goes back to those previous "converged" results.

 

-If the goal is the same after a 2nd refinement is that considered converged? I expected the result to eventually flat line, where even a 3rd and 4th refinement would be identical but this doesn't seem to be the case. Is that common?

 

-I can't figure out how to directly measure the boundary layer thickness (is there a way?) but am I hitting the transition point from the different wall functions and that is causing the sudden change in pressure after several "converged" refinements? I attached a quick screen shot showing a mesh size where the results appear to be converged. I have the boundary layer display on and am guessing this mesh is within the thin boundary layer scheme. One more thing that's leading me to think this is the issue is a note from the help regarding two-scale wall functions:

 

"Note, moreover, that when the appropriate boundary-layer approach is selected automatically and the computational mesh is rather fine, the solution accuracy may fall off. The reasons for the accuracy falloff are that the mesh is excessively fine to apply the thin-boundary-layer, but it is insufficiently fine to resolve boundary layers and apply the thick-boundary-layer."

 

Should I have a best practice where I try and stay within the thin boundary layer scheme? The thick boundary layer is going to be computationally expensive and my computer isn't the most powerful so I don't think I could even properly get ~4 cells within the boundary layer. From my understanding the two wall functions should yield very similar results.

 

Thank you

Attachments

Outcomes