I have a very simple problem. Flow over a 2D rectangular object. I am using .ini file of the flow over sphere example with some minimal modifications.
I have generated three mesh configurations rect_simple.msh, rect_simple_fine.msh and rect_simple_finest.msh using gmsh. The only difference between these three files is size of meshcells.
Meshcells are refined using gmsh “Refining by splitting” button. For the two coarser mesh files, the simulations run beautifully, but with the finest one, check_nan fails in the very first test.
I use openmp backend. What could be the source of this problem? Is this some limitations of openmp?
I have attached mesh files.
rect_simple.msh (344 KB)
rect_simple_2d.ini (1.95 KB)
rect_simple_fine.msh (1.4 MB)
rect_simple_finest.msh (5.85 MB)