Are collar grids unavoidable for chimera method applied to penetrating bodies,

Thanks,

Bill.

Not if you are clever in how you grid around the protruding body. For example, consider a wing grid where j goes around the body and k moves outboard along the wing (l then is the direction out from the wing). If you make the k = 1 surface conform to the fuselage surface you will not need a separate collar grid. The k = 1 surface becomes part of the solid boundary definition for the fuselage, and represents the part of the fuselage surface removed by the presence of the wing. Of course, sometimes it's easier just to build a collar grid in order to not have orphans.

I agree, but my point is: collar grids are also quite complex to generate. The strength (on paper) of chimera method is the use of totally independant meshes. But it seems to me that we just displace the problem to the collar grids.

The strength of the Chimera method is that instead of trying to generate one grid for a complex geometry (wing-fuselage-inlet-etc) I can generate simpler grids for each part and have them mate in a fairly general overlapping manner. A collar grid is significantly less complex than what you need if you don't use some type of domain decomposition.