I think here the point is exactly to see the divergence, since we are trying to optimize the relaxation factors. If you see the residual plots, they oscillate like crazy. If you see the simulation log, the error gets very high quite early in the simulation. For example for time=82:
time step continuity errors : sum local = 3.76563610418e+82, global = 1.4748516605e+80, cumulative = -1.21876836849e+81
Even if the simulation hadn’t broken at this point, there seems to be no point in continuing. The first way to fix this is of course lowering the relaxation factor. Changing the mesh or the solver could also help, but the focus here is on the relaxation factor.