BUG: Grid coarsening in theta direction and spherical geometry loses angular momentum #347
Description
Describe the issue:
When Grid Coarsening module is used in spherical geometry and in theta direction (X2), and with 3 components, there is a loss of angular momentum in z-direction. Indeed, the average procedure in the coarsen_flow function is done on the velocity and not on the angular momentum. It implies that for two neighboring cells at radius r, angular velocity om and angles th1 and th2, the phi velocities are v_phi1 = romsin(th1) and v_phi2 = romsin(th2) and thus the angular momentum is L_z = r^2om(sin^2(th1)+sin^2(th2)) before averaging. However, after averaging, the common phi velocity in both cells is rom(sin(th1)+sin(th2))/2 and the new angular momentum is L_z' = r^2om(sin(th1)+sin(th2))^2/2. which smaller than L_z for all values of th1 and th2.
Error message:
runtime information:
most recent idefix version, independant of platform