What is the efect of cell size on convergence rate, using unstructured grid.

Ingeneral, cell size itself has not the major effect on the convergence. But, in general, coarse grid has better convergence. Mesh skewness, edge ratio(dx_i/dx_(i-1) ...) and aspect ratio have critical effect on the convergence.

Sincerely, Jinwook

What type of convergence do you mean in this case? Convergence to a mesh independent solution, which is somewhat tricky to guage on an unstructured grid. Or, convergence of residuals of algebraic equations?

Dan.

The convergence in FEM used to unstructured grids is of the order of cube of h(cell size parameter,less than 1.0).

What is the trend in FVM used to unstructured grid?

The convergence always depends on the CFL number when we use explicit scheme. CFL number on unstructured grid generally depend on the volume/area of a control volume.

When the mesh size is very small, the time step must be reduced to get the converged solution.

In my experience, if a mesh is not highly stretched, geometry of a mesh is not an important factor.