I am modelling cyclonic separators with everything second order and a pressure-velocity coupling of PISO. However when the swirl angle reduces (i.e. the tangential velocities reduce) it becomes difficult to reach a converged solution. The turbulence model is RSM. Does anyone know if there is a lower swirl number limit for the RSM model and then maybe the rke or rng with swirl should be used?

As an extra note I find that the use of PISO gives the solutions at higher swirl angles more stability probably due to the increased degree of closure.

Regards,

Dave

(1). You can change the swirl condition gradually from the converged high swirl solution. This will provide a better initial flow field to avoid un-necessary problem. (2). There is also a slight possibility that at certain swirl condition, the flow field in your configuration is not compatible with the physical design. Such as flow separation at the exit, or ??? (3). Otherwise, pay attention to the mesh distribution and the boundary conditions. It is also possible that the flow field is transient at that condition.

Thank you for the reply.

It was not really possible to use a solution from the high swirl case as they are different geometries since I also modelled the swirl element. However what I did find that solves the problem is to set the initial under-relaxation values of the pressure at 0.3 and momentum at 0.7. Then steadily increase pressure and decrease momentum at all times ensuring they sum to 1.

Dave

Dave

Just out of curiosity what is your swirl angle distrubtion. Maybe you are approaching the critical swirl ratio.

Regards Oliver