Free Surface Problem - Convergance
I am running some 2D cases of planing flate plates with an unsteady free surface (air-water) model, using an explicit scheme.
This is fine, except that I seem to have to use very small time steps (of the order of 1x10^(-6)) for the solution to converge to residuals of 1x10^(-4) at each time step. The flow velocity is quite high ~ 20 m/s so this is probably to be expected.
Since I am not really interested in the transient behaviour of the solution, I beleive it is okay if the residuals do not converge at each time step, as long as they are within reasonable limits. Of the simulations I have run so far, I certainly seem to get unreasonable results when the residuals do not converge sufficiently.
What I would like to know is whether there is any other ways of speeding the convergance? I have used implicit body force formulation which seemed to increase the convergance rate somewhat, but am not sure if I could make any other simplifications which might make the solution a bit faster?
|All times are GMT -4. The time now is 17:18.|