Hello,

I just read about the Sliding-mesh-method which can be used for simulating stirrers and so on. I also read that Chimera-mesh-methods can be used for such kind of problems. After reading a bit about these methods I wondered what is the difference. As I understood both methods seperate the rotating and the static part into grids that do not match at the interface and so there is interpolation needed in that area of the grid. But what is the difference beside that chimera grids are always structured, whereas sliding meshes can be unstructured (at least as far as I know)?

Help is much appreciated! It would also be cool if somebody knew literature on that topic!

The major difference is in Chimera-mesh same volume is occupied by 2 mesh. For example, if you consider a point in flow domain, it may lie inside 2 cells.

But in sliding mesh that is not the case.

Yes, to clarify that, Chimera meshes overlap, sliding meshes obviously don't. Sliding mesh cabailities are found in many codes, but Chimera is a lot rarer. For simple rotation about one axis, sliding (with a cylidrical interface) is by far the easiest option. For more complex motion with translation or eccentric rotation, Chimera makes more sense.

Thanks, your answers helped me a lot. I have also read about deforming meshes. Which of those methods (sliding and deforming) can be considered best for those problems? I think the deformations make really bad elements after some degrees of rotation, so that sliding mesh is more accurate. Right?

If you have situation where you can have a nice smoothly sliding cylindrical mesh, that more or less has to be your best option. If you have translation or an angular geometry rotating, I would prefer Chimera, because you more or less know what will happen to the mesh. A deforming mesh * may * end up giving your poor quality or even negative volume cells, which will kill your solution. You have to watch a Chimera solution quite carefully though, they tend to be very robust, to the point of not failing even where the situatuation has gone totally non-physical. I had one where the moving mesh popped right out of the side of the background domain. No warnings, no failures, the code just ran happily on! It only showed up when I plotted the trajectory afterwards.

The bullet to be used depends on the target to be shot.

The increasing complexity of the problem needs an increasing complexity of the algorithm.

If you are dealing with turbomachinery, for example, the best is sliding meshes because you can create two separate meshes one moving (the rotor) and one fixed (the stator) that communicate only via the fluxes at the boundaries.

If you are dealing with a helicopter you must have a grid attached to each blade translating, rotating, flapping,pitching; all these grids move into a background grid attached to the fuselage (translating). At each time you have intesection of the blade-grids with the background-----> Chimera is the only way!!