Dear all,

i've written a code based on Kim & Choi (2000, J. Comput. Phys. 162, 411) 's fractional step on 2d structured grids. it uses non-staggered grid with cell center and normal face center velocities.

suppose now i want to model objects which move. after reading through a few sources (including this forum), it seems that the ALE formulation with moving/deforming grid is very suitable.

however, i did not manage to find any references which uses Kim's fractional scheme for moving grid.

from what i've read, it seems that the most important thing is to modify the convective face velocity from U to U-U(g), where U(g) is the grid velocity.

Do i need to modify the poisson equation or the face normal velocity computation? i'm not very clear about this. Hope that someone can give advice.

tks

Hi,

maybe I can help. I developed a code on moving meshes using a SIMPLE-like scheme. You are right in formulating the convective fluxes with the "relative" mass flow U-U(g). The formulation of U(g) may be critical if you have deforming cells, s. publications of Peric et al. on space conservation. SIMPLE is an implicit scheme, so here the time discretisation is not critical for moving meshes, any implicit discretisation can be used.

Regarding the Poisson eq.: In my SIMPLE like scheme, I modified the divergence term to be formulated with the "relative" mass flow, this was the only modification on the continuity equation. However, I am not an expert on fractional step methods nor on the exact solution of the Poisson pressure equation.

Nevertheless, hope that helps, Romuald

Look for papers by Wei Shyy.He has done lot of work in ALE and also authored a book titled CFD with moving boundaries.

-H

tks Harish. i've read his book but it seems so difficult to understand.

Btw Romuald, just to confirm, are u using the finite volume method? Becos i've a friend whose using the SIMPLE scheme as well and finite difference. but he told me that there's no change in his poisson equation.

anyway, i'm trying your suggestion. so the RHS of the poisson eqn should be k*sum(U-U(g))dl which k=constant. is that so?

Lastly, is there any reliable verification data for comparison? tks!