[lostking18]

Hi, there

Recently I ran a cooling channel simulation in CFX as following:
1) I built two meshes for the same cooling channel: one with fine mesh; one with coarse mesh.;
2) I ran a fine mesh simulation and get converged; the flow is steady simulation with no-slip wall boundary condition at the bottom X_Y plane;
3) I set up the coarse meh simulation with the same boundary conditions as the fine mesh simulation, then I use the fine solution as the initial guess to the coarse mesh simulation and ran the simulation for just one time step.

Ideally since the difference between the fine and the coarse simulation is solely the mesh resolution and at the wall for both cases the velocity is 0(no-slip boundary contions), after one time step the momentum residual should be zero because there is no difference in momentum equation between these two simulations. However, I noted that the residual at the wall is not zero. Could you give some explanation on this observation? Is it because the residual is interpolated from the center of the mesh cell? Where does CFX store residual? On the vertex or at the cell center?

Thank you very much!

[ghorrocks]

You should read the CFX documentation on discretisation approach.

The finite volume which includes the wall has the wall on one side and internal faces on the other side. The centroid of this finite volume is therefore off the wall, which means it has a non-zero velocity and therefore a residual.