Closed Prasanna2989 closed 5 years ago
With a 1:2 ratio for the element resolution the elements geometry will be quite stretched. This can effect numerical robustness, especially running 1k+ timesteps. Consider increasing the number of "angular" elements to make the element geometry closer to a 1:1 ratio.
Thanks Jule. r:2 pi r (1:5) ratio models are woking fine.
Cheers
Hi All,
I was running a mantle convection model in an annulus with UW2 regional Mesh and encountered with some instabilities in the Temperature solution in the Eulerian domain (Swarm is not involved). For first 2000 steps model works fine and after that it creates the instabilities as per attached images. I have tested this model with 2 different resolutions of (64x128) and (128x256). Issue exists for both cases. I herewith include the code for your reference. This code was build based on the Van Keken 2001 example in publications in UW2 regionalMesh. The model is iso-viscous. What could be the reason for this behaviour and How should I fix this?
cheers Prasanna