I am tracking down some issues in our code, and I am seeking some clarification on the way that Elmer handles the non-penetrating condition at the basal boundary condition (i.e. the ice cannot penetrate through the bedrock) in the Navier-Stokes solver.
As I understand it, Elmer rotates the coordinate system of the nodes that are on the bottom boundary so that the x-coordinate is normal to the base. If you set the parameter "Back Rotate N-T Solution = Logical True", it is supposed to rotate the coordinate system back to the original Cartesian coordinate system after DefaultSolve (and by extension SolveSystem) is finished executing. It is not clear, however, if the coordinate system of the input parameters prior calling DefaultSolve is supposed to be bed-normal coordinate system or in Cartesian coordinates (regardless of the value of Back Rotate N-T Solution). The reason I ask, is because I am setting the results of the Shallow Ice Approximation as an initial guess for the velocity of the ice, and this is stored in Cartesian coordinates in our program.
Another question is why it was set up like this. Wouldn't it be easier to rotate the non-penetrating condition to Cartesian coordinates, rather than rotating the nodes to a local bed-normal coordinate system? Then there would only have to be one rotation calculation at the start of the simulation and the coordinate system of all the nodes would remain consistent.