Hello,
I’ve ran into trouble in some of my CFD runs as the flow begins to diverge. I have recognised the face that is causing the divergence, and it is because of the aspect ratio of the mesh elements there, however I am unsure on how to make the mesh better. I have tried creating more fine meshes, less fine meshes and adding region refinements to the point in question but it doesn’t seem to better the aspect ratio.
Any help on this would be greatly appreciated!
Hi there, thank you for using the forum!
I took a look and have some comments regarding your project:
Let me know if this works well 
Best regards,
Fillia
Ahh okay thank you, I’ll look through that and give that a go!
Hi, I made the refinements as suggested, but it still did not change the mesh values. It actually increased the edge, volume and aspect ratios?
I suggest you follow the instructions of the tutorial, that include:
Of course you don’t have to use the exact same values for the length inputs. For example, the local element size you are using has a really small value that results in a very fine mesh only on these surfaces.
I followed the tutorial and it seems like the sizing of the mesh elements are ok, but the mesh values in the log are still too high?
Hi,
When it comes to divergence due to poor mesh quality, the two main workflows are trying to improve the mesh settings and/or cleaning up the CAD model around the problematic regions.
This article that I wrote shows the usual troubleshooting steps (which are essentially the same for every simulation). By following the steps in the article, you will notice that the simulation is diverging around this region:
Having a look up-close, you will notice a very small gap and a lot of thin faces:
Checking around the opening of the gap:
Judging by the model, and these small faces cleaning up the model is definitely a good idea. Thin faces and gaps are not really desired, in terms of mesh quality and simulation stability.
So if you are willing to clean up the model (which is usually my recommendation), that would definitely help.
Alternatively if you are not willing to clean the model, here are some things that are worth a try:
In any case, using a clean CAD model always helps!
Good luck.
Amazing thank you so much, found a few bits in the CAD that were messing it up and re-ran the simulation and it run without a problem!!
Small features suppression worked very well in the past. I’m in truble now, as ‘Small feature suppression’ is ignored by meshing algo, even with manual sizing of standard mesh elements.
My model is 0.8 x 0.25 m and I wanted to suppress elements < 0.001 m (1 mm) and next try < 0.0001 m (0.1 mm) without success. Meshing process decides, that I want excessive suppression and re- sets it to 0.0.
That way my mesh is ~ 6 times bigger then expected and takes at least 20 times longer to simulate.
In my opinion it is a regression and I suggest at least a warning that my parameter setting may cause a problem down the line but not removing my setting. Another solution would be a switch to accept automatic evaluation (on / off).
Current message in log file is:
Small feature tolerance may be too big. 36.30 % of the model faces would be suppressed. Rolling back to the original model
Model import took 1.601533965s.
Maximum precision of model and its entities: 6.45587989493963e-07 m.
I’ve seen that Riccardo (@Ricardopg) was handling it at that time and I do not know who can take care of it now.
