When changing the solver from TASCflow to CFX-5 there is a problem in convergence. The attached file shows the two residual histories. We have tried to solve the problem by checking all settings etc. The grid is exactly the same in both cases. Why is this happening?

The problem is related to using stage interfaces perpendicular to the flow direction where recirculating flow exists. This leads to recirculating flow across the interface. The problem is smeared out in TASCflow but not in CFX-5, since the CFX-5 solution is more accurate. If you change the stage interface to frozen rotor in CFX-5, this makes the solution much more stable. In the attached example, it was possible to converge the solution. In particular there was an improvement in convergence for heat transfer.

Show Form
No comments yet. Be the first to add a comment!