4 Replies Latest reply on Dec 5, 2016 6:03 AM by Alex Clarke

    how many cells is reasonable for solidworks flow simulation?

    Justin Brim

      I have a very complex model I am working on and having a hard time meshing due to the size of the internal flow channel sizes etc.  I am wondering if there is some sort of ball park number for solidworks in general for computation time?  Ie I'm not wanting the computer to sit and crunch for days but overnight would be alright once the meshing is done.

       

      I do realize there area a ton of different variables depending on model geometry but I'm wondering if anyone has a ballpark idea of what will bogg a computer down for a week vs maybe 6 hours when converging on a simple flow solution for pressure etc.

        • Re: how many cells is reasonable for solidworks flow simulation?
          Siavash Khajehhasani

          As many cells as your simulation results can be considered to be grid independent.

          • Re: how many cells is reasonable for solidworks flow simulation?
            Bill McEachern

            Well, it depends on a lot of things. What sort of physics do you have turned on? If you have a transient heat transfer with flow and radiation and long real times it can take a long time. Gernally though you can get pretty good results with modest cell counts but some problems are just inherently large due to geometric complexity. I try to keep them below 10 million and I really don't like them that big - a few million is very workable for most things and 500k is a few hours for most things. I run a 3.3 GHz 12 core dual socket machine with 192 GB of RAM just for reference. Everyone resource limited. Just change the problem to be something you can solve and solve that. You figure out what's important to the answer and simplify the problem to match your resources.

              • Re: how many cells is reasonable for solidworks flow simulation?
                Justin Brim

                Thanks much!  Wow that is some computer.  Thats a perfect ballpark idea for me.  I had a model with something like 16 million half cells etc.  I think I've got it narrowed down now.  The minimum gap size adjustment was the key for me.  I was asking for too small of a gap size to start with and it was making the mesh way too refined for what I need.  Once I tripled the gap size, it seemed to perform pretty well. 

                  • Re: how many cells is reasonable for solidworks flow simulation?
                    Alex Clarke

                    You should also try using manual mesh settings and mesh controls- This allows you to use a coarser global mesh, and refine the mesh for phenomena that you are interested in - For example, imagine you're analysing flow through a car radiator - You've got the whole car modelled, but you're only really interested in what goes through the grille. You would use a coarse global mesh, then use a manual mesh control on the front face of the car. Within that mesh control, you can change the refinement for different parameters - Cells across a narrow channel is a good one for our car radiator - This should usually be set to 5 cells across a narrow channel (The refinement being how many times the mesher is allowed to split the global mesh to achieve that).

                     

                    Every problem is different, but using the manual mesh settings certainly gives you much more control and, if used correctly, should greatly reduce solve times.

                     

                    For reference, I was able to reduce the cell count of my fan simulations from 2 million to 500,000 with no detrimental effect.

                     

                    Hope this helps.