Learning Forum

[chuanx]

When using HEAT, I often encounter a situation where the structure meshes successfully but the simulation does not run. The error message is:

[0] process exited without calling finalize

and

...thermal-engine-msmpi.exe ended prematurely and may have crashed. exit code 0xc0000005

This happens, for example, when I change the dimensions of one of the (rectangular prism) objects in my structure. In this case, even a very small change can trigger the problem. With one set of dimensions, the programs happily runs, but with the other it crashes with the above messages. What is going on here?

BTW, sometimes when this happens, I tweak the HEAT mesh settings, and the problem disappears, but not always. I suppose without any understanding what is going on, I cannot come up with a reliable solution. It is very frustrating.

Chuan

[kghaffari]

Hi Chuan Like you pointed out, in finite element simulations especially in HEAT finding an appropriate mesh that works can be the more difficult and time-consuming portion of setting up the project.
In some cases, this means setting the meshing values (most importantly min/max edge length in the solver or the value of a local mesh constraint) such that thin layers or any other small features can be resolved by the mesh.
In other cases, feature size might not be the issue; it could be that the structure is in general difficult to mesh (usually this will be a large simulation including fairly complicated structures). In this scenario a small change in mesh settings will be the difference in attaining the mesh. IÔÇÖd recommend once you have a mesh that works to visualize it (visualize HEAT > grid) and confirm that the mesh refinement in different regions satisfies your requirements.
Best regards Khashayar