seeking full potential expert

Alberto

Dear All,

I would like a little advice from you. I programmed a Finite Element code to solve the full potential equation (steady) on unstructured grid. I am having lots of troubles when the shocks get strong, ie the solution diverges. I am using density upwinding but still handling the nonlinearity updating the density with the solution at the previous step. I understood that this cannot work since I violate the flow physics when the equation gets hyperbolic. What is the best way out of this problem, in your opinion? I was thinking about solving the unsteady equation to steady steate implicitly in time, but how can I linearize it? Any of your suggestions would be of great help!

Thanks a lot

Alberto