FEM/moving boundarys
 

I am beginning studies in the subject of moving boundarys, and I have noticed that grid movement is quite important. I am using an ALE method to adapt the N-S equations and a Laplace eq. for the grid velocity. But for the free-surface the FEM-Galerkin formulation for this Laplace eq. won't work because the oscilations of neibour nodes is quite diferent an then the interior nodes get a little "crazy"... Any sugestions for a method that can be quite general for free-surface and moving boundarys? Thank you Astrid

Re: FEM/moving boundarys
 

Laplace eq.s are time dependent and are used for moving control volume. It may be that some of your interior nodes are touching saddle point and this may happen due to singularity effects. I would suggest if you would change the grid spacing distance and see the results every time you change the grid spacing by manual trial and error method.

Re: FEM/moving boundarys
 

You are whright, I tried using a convection diffusion eq. independent of time and it seems the results have improved. For a rigid body it worked out well using the convection velocity the same as the body velocity. For the free-surface I am going to try out using the free-surface vertical veloc., but I haven't had time for that experience.

I am doing this so I can use a moving rigid body toghether with the free surface, so I must use the same method for both.

Thank you for your suggestion, any other please tell me, Sincerelly yours, Astrid Barros