[martindg]

Hi,

I'm having a go at using CFX to simulate the flow past a cyclist with moving legs. This is difficult, as there is surface interaction between stationary and moving parts and I can't see anyway to use a GGI.

My plan was to start with a given mesh, deform it a small distance (~2 degrees of the full cycle), and then interpolate to a new mesh and repeat. The way I've set it up, each run starts off with an initial deformation and then deforms back to its "unstressed" position; I've found this can delay any mesh folding, by minimizing surfaces crossing over each other. This seems to work, however, there is a problem: CFX interpolates onto the un-deformed mesh, even though the specified displacements for that point in time might not be zero. The mesh then jumps from zero deformation to whatever the specified deformation is for that step. This leads to a substantial error.

My question is: is there any way to tell CFX to interpolate onto a new mesh only after applying any specified initial deformation? Or applying the deformation after meshing, but before running the solver? If not, I'll have to have a re-think...

Thank you

[ghorrocks]

Why are you interpolating onto a new mesh at all? Why can't you do this using the normal moving mesh capabilities?

[martindg]

Thanks for the reply.

I have tried doing this with the normal moving mesh capabilities and I run into the same problem. The solution still needs to be interpolated to a new mesh, which CFX doesn't deform before interpolating.

For example, in the example shown in this tutorial:

http://www.edr.no/blogg/ansys_blogge...nch_cfx_remesh

each time a remeshing occurs, the new mesh is undistorted and then deforms as the rotor rotates. The number of remeshes required could be reduced if one started with a deformed meshed. The rotation would then take the mesh from distorted to undistorted and again to distorted, increasing the number of degrees one mesh could cover and reducing the number of meshes needing to be generated. This is what I would like to do and I am wondering if anyone knows if there is some provision for this in CFX.

Thanks again.

[ghorrocks]

You might need to talk to CFX support for examples of remeshing to deformed meshes.

But why can't this be done in a single deforming mesh simulation with no remeshes? Then you do not need to do interpolations.

[martindg]

The way the geometry is set up, the thigh rotates around the hip joint, which is inside the torso. So as the thigh rotates relative to the torso, the intersection between the surfaces of the thigh and the torso moves, which very quickly folds the mesh. There is the possibility of moving surface elements around on the surface of the thigh and torso, so to avoid them getting lost in the intersection, but (I think) using frequent remeshing and keeping the motion simple is the more straightforward option. If you have an idea of how else to deal with this, or could point me at an example which attempts something similar, I'd be interested to hear.

[ghorrocks]

I would certainly try the built-in mesh smoothing options before going to remeshing. Parameters like distance from wall or element size may help, give them a try.

CFX support has several examples of remeshing. If you cannot get remeshing to work then I would contact them and get some demo files from them.

BTW: you can quickly check mesh motion by using an expert parameter to turn all equation solving off (eg solve fluids=f, ditto for turbulence, AV and any thing else) and setting it to 1 coeff loop per time step. Then the solution will proceed and only do the mesh motion. You can use this to quickly check the mesh motion is OK before committing to the bigger simulation with fluid solution turned on.