Hello:

I am simulationg such problem, fluid around a wall boundary, this wall maybe cylinder,So when the wall moving (rotation or translation ), the grids also deform with the wall boundarya grid deformation routine, that means a cell holds the flow field values changes its position from time to time and it's position is then not the same position as in the flow.

So i have to consider that the grids have a velocity, from the peric's(1997) compute fluid method book, the grid velocity can be calculated by a face swepting volume in dt. And the grid movement affects only the mass fluxes, so for all the faces, we must substract the moving grid's mass fluxes from prime mass fluxes, these all can be done.

But, for the wall boundary, I do not know how to do? for the wall boundary faces, the muss fluxes is zero usually. Now the wall faces are moved, these faces also have the grid moving mass fluxes, then the wall faces' mass fluxes are minus grid moving mass fluxes, so now how to deal with the wall boundary?

Who can tell his experience for dealing with such problem? Thanks!

the grid should move but the wall should be stationary! This means that the fluid is moving round the wall.

YOU NEED TO SET THE WALL AS ROTATING/MOVING BUT LEAVE IT AT 0RAD/S.

what is your mean, phil first, the wall is moved, then all the grid also deform some with the wall boundary.

normally with a moving boundary you use multiple moving reference frames or sliding meshes. You can also use deforming meshes, I would need to see the problem.

The usual approach to a moving boundary is to set the wall moving a 0m/s,rad/s,etc, then set the fluid's reference frame moving at the velocity with which the boundary should be moving. This simulates the flow over/through a moving wall, which could be anything from a compressor to a car.

IMAGINE THE FLOW OVER A CYLINDER IN A WIND TUNNEL. WOULD IT NOT BE THE SAME AS A CYLINDER MOVING THROUGH STILL AIR AT THE SPEED THE WIND TUNNEL FLOWED AT?

For ınviscid flows, moving wall BC does not require any special treatment. In case of viscous flows, extrapolation of velocity components into ghost cells are changed !

CFD Student

what?

>For ınviscid flows, moving wall BC does not require any special treatment.

I had the same reaction as Phil (what???) to this statement. Please, discard this as easily misleading. Even in inviscid flow the effect of boundary motion on the wall-normal velocity component will enter the boundary condition. You can, however, easily formulate this in a general way by making sure that the difference between flow velocity and wall velocity has a zero wall-normal component (and a zero tangential component in viscous flow). This is generally the case, including the stationary situation (zero wall velocity). So whether you call it "special" treatment or not is up to you, just make sure you understand that all fluxes have to be formulated using the flow velocity formulated in the local moving reference frame (i.e. cell). Instead of using flow velocity to calculate the fluxes, you will now use flow velocity minus cell face velocity (where appropriate). If you do this, the flow equations have the same form as in the case of a stationary grid. Is suggest you look at one of the many papers explaining this in detail.