Can anyone suggest a good timestep formula for explicit compressible viscous flow on unstructured grids. On quadrilateral (2D problem), if i stretch cells at wall boundaries solution is diverging after some number of iterations even if i use small cfl number. problem is converging for triangular grids as amount of stretching is less compared to quadrilateral grid.

dt < rho*Pr*L^2/(4*gamma*visc)

where L is a charateristic dimension of the cell

check Turkel et al.(1991), NASA TR.

Thank you for your response.

Earlier for inviscid computation i am using fallowing timestep.

dt = sqrt(area of cell)/(sqrt(u^2+v^2)+ c).

u and v are two velocity components, c is wave speed.

the time step which you have suggested is alone sufficient for viscous computation or i need to combine both inviscid, viscous time steps? how can i get the paper which you have suggested.

Thank you.