I am modeling a simple transient cooling process of a hot solid object (natural convection) using chimney approach. I still get high velocities and divergence even after smoothing the mesh and using different time step size or inner iterations.
Does anyone faced the same problem, or have some hints?
I think the problem is simply the timestep? You need to do the sensetivity study of timestep independence, by gradually reducing the timestep such that you see the convergence at every timestep.