I am implementing a grid generation code by using elliptic differential equation for 3D SURFACE. However, I have a problem with choosing control function P and Q for generate grid for 3D SURFACE. Are there anyone have experience about this? Help me!!!

I suggest you start with this classic reference:

http://www.hpc.msstate.edu/publicati...ook/index.html

Are you trying to implement grid clustering only or both grid clustering and orthogonality with the control function?

I am trying to implement both grid clustering and orthogonality. I think control function is uesd to transfer the grid distribution on the boudary to the interior grid.

Thanks for your answer

My experience has been that the Ps and Qs can work to some degree but are awkward to use in practise. I had better success modifying the coefficients to control the grid distribution rather than the souce terms although my interest was in non-conformal orthogonal grids and so this experience may not map to what you are trying to do.