|
There is nothing stopping you using porous domains with moving meshes, but there are some issues with the resistance/loss models in CFX-10. The main problem is that the standard loss models use global frame velocity instead of mesh-relative velocity. Secondly, a porous baffle would also have low resistance normal to its plane, but high resistance transversely and so the issue is anisotropic, and the direction may also change as the mesh moves.
One approach is to set up a vector Additional Variable equal to the mesh-relative velocity (Vfluid-Vmesh) and create some general momentum sources/sinks using this variable. For simplicity, you could ignore the frame problem with an isotropic loss model, but still have the streamwise 'correction' proportional to relative velocity via a general source. That may be a reasonable approximation. Alternatively, you could do the maths and work out the proper correction in all three directions so that the relative velocity is correctly implied. This would take advantage of the existing linearisation for isotropic loss. A 'full' approach would involve general sources using the relative mesh velocity, appropriate source coefficients for stability, and anisotropic loss coefficients that are a function of the orientation of the baffle.
|
|
