Hello, I am working on our current Formula Student car and am having trouble to finish the simulation. The first and third tries failed with the error message:
“Velocity field started diverging. Please check the mesh quality near the reported location and try refining the mesh. If the problem occured near a boundary, please check the boundary conditions.” with the coordinates on the side. The first time I inspected the mesh and did find a problem. The second time I have already noticed diverging residuals so stopped it myself. I improved the mesh a bit more and restarted, but again I got the error message from above. This time, the mesh seems to be absolutely fine at the specified location and I dont think it is a boundary condition.
I am unsure what to do about that except for refining this part a bit more and hope for the best. Could you please take a look at the mesh and especially the initial conditions for k and omega? I tried to calculate them, but am unsure if they actually are correct/in the correct ballpark.
I shared the project with the Support. It’s name is “s3-23e”. If you need a Link or more specific info, please tell me how to do that/what you need.
Hope to hear from you soon!
Hello, and thank you for using the Forum.
Bad quality elements can form where two surfaces connect with an acute angle. Meshing algorithms usually experience difficulties forming volume mesh elements between those surfaces. One solution is to create a fillet between those two surfaces, so that volume mesh elements can advance smoothly.
A general recommendation though, have you tried using the standard mesh algorithm? Standard algorithm consists of tetrahedral elements near the boundaries, and hex elements near the core. This algorithm is highly automated, and might save you lots of resources/time by reducing the number of elements. I can recommend you this tutorial for your application as well.
Hope this helps.
Thank you for the reply! Through the last seasons I could never really get it to work and with the other algorithm I knew how to get results. So that’s why I haven’t used the standard one. I spent the day trying to get the standard algorithm to work and finally found the issue which caused it to not even generate the external flow field. I’m now continuing along the lines of the tutorial (thanks for the link by the way!) and hope to finally reach some results. I kindly ask you to keep this question open until I get back to you.
Thank you and kind regards!
Hello again! Great to hear that the issue is solved at the moment.
The topics are always open to reply even if it’s marked as solved at the moment. However, for the sake of clearity, it’s always a better idea to post a new question unless it’s related with the current topic.
Good morning! I have tried multiple iterations with different mesh-sizes and still was unable to reach results. The numerics are the same as the ones I used over the last years and with which I never had issues, so I think those won’t be the issue. Therefore, the mesh will be the problem, yet I cannot find any odd looking cells. The mesh in itself is already finer than most meshes I had before and it still always fails after around an hour or two. I looked at the geometry and made sure there are no sheets, no intersecting bodies and united all parts into only three big bodies to make it simpler, but without effect. By now I have absolutely no clue, what the problem could be. I hope you can give me another hint as to what the issue could be or what I could try to make it work. My last idea is to put only the baseline geometry of our car into the simulation and try to get results from there. Then adding a frontwing, re-run the sim, add a rearwing, re-run, and so on, until I find the problem. But that would be very core-hour expensive.
I hope you can help me once again and wish you all the best!
Have you tried inspecting the mesh quality? Seems like some elements are beyond the acceptable range which may be causing the issue. (Aspect Ratio, Non-orthogonality). Since the number of mesh elements are quite high, inspecting the mesh quality will take a while. However, I can already make a suggestion looking at the geometry by experience. The angle between the tires and ground is mostly like one of the regions that causing bad mesh elements. What you can do is either to create small fillet between tires and ground in a way that won’t have much effect on the solution, or to use the surface wrapper for tires and ground.
Once you’re certain the mesh quality is fine, you can also try advancing the simulation one by one instead of advancing the design. For example, start with non-rotating tires and add rotation later etc.
Another method you can try is to play with under-relaxation factors as the error message suggests once you’re confident about the mesh quality. You can obtain more stable solutions by reducing the relaxation factors, however this will increase the number of iteration required for solution to reach a convergence.
Hope these help.