the problem about the relaxation time

Hi, everyone,
Sometimes, we will encounter that the relaxation is very small and near to 0.5, (e.g. tau= 0.501), then the computational result is diverging.
How does this can be solve?
Looking forwad your help.

You can decrease the grid spacing (ie, add more grid points), but this can be computationally expensive. Using MRT can greatly improve numerical stability.