I am trying to implement a subroutine to calculate the SIFs in 3D. It requires to create an independent grid of hexahedral cells around the point at the crack front. This grid has its own coordinate system and all the quantities should be expressed with it. I do not know whether getfem could introduce a subgrid with its own coordinate system such that the derivative and integration could be performed based on its local orientation. If not, I think I am able get the derivative by applying the transformation to the results obtained respect to global coordinate but I do not know how to set up the generic assembly to perform the integration since the operations inside the comp command should be defined respect to the global coordinate system. Could you give me any suggestions? Thanks.