[D3sT]

Hello all.

I am writing a 2D cell-centered Cartesian laminar solver accelerated by multigrid technique in C++.

I have a problem at cases having low reynolds number. For determining time step, I used local time stepping in the article "Multigrid Solution of the Navier-Stokes Equations in Triangular Meshes" by D.J. Mavriplis. It gave good result when multigrid is not used.

However, when multigrid is used, pressure became minus so quickly. Euler equations are solved in coarser grids even if the flow is viscous. For time step, the local time stepping above is used.

When I change time step calculations to the time stepping for inviscid case or I solve Navier-Stokes equations for coarser grids, pressure became minus again in the first cycle.

How can I determine time steps for coarser grids? What is my mistake, here?
Thanks for all replies: )