# Convergence issue due to reverse flow

 Register Blogs Members List Search Today's Posts Mark Forums Read

 December 7, 2023, 17:11 Convergence issue due to reverse flow #1 New Member   Join Date: Jul 2021 Posts: 8 Rep Power: 5 Hello, I know people offer to modify (for example extend) geometry to avoid reverse flow but it is not avoidable in my simulation. I am trying to simulate a piezo fan. During oscillations of fan it creates vortices and these vortices reach to outlets and rotation cause reverse flow as you can guess. Moreover, low pressure regions caused by the fan lead air to rush in flow domain from outlets. Therefore reverse flow must be the part of the simulation. The problem is (I guess it is the problem) these reverse flows cause inconsistent reduction in residuals from time step to time step. Contiunity residual sometimes reduces to e-05, sometimes e-02 but usually it reduces to e-03. Do you have any recomandation for making better (maby consistent) reduction in contiunity residual?

December 7, 2023, 17:38
#2
Senior Member

Filippo Maria Denaro
Join Date: Jul 2010
Posts: 6,815
Rep Power: 73
Quote:
 Originally Posted by coffee_break Hello, I know people offer to modify (for example extend) geometry to avoid reverse flow but it is not avoidable in my simulation. I am trying to simulate a piezo fan. During oscillations of fan it creates vortices and these vortices reach to outlets and rotation cause reverse flow as you can guess. Moreover, low pressure regions caused by the fan lead air to rush in flow domain from outlets. Therefore reverse flow must be the part of the simulation. The problem is (I guess it is the problem) these reverse flows cause inconsistent reduction in residuals from time step to time step. Contiunity residual sometimes reduces to e-05, sometimes e-02 but usually it reduces to e-03. Do you have any recomandation for making better (maby consistent) reduction in contiunity residual?

Reverse flow in outlet is not a cause of numerical instability, provided that the correct BCs are prescribed.

Incompressible or compressible flows model require different BCs, subsonic or supersonic are different.

 December 8, 2023, 02:39 #3 New Member   Join Date: Jul 2021 Posts: 8 Rep Power: 5 The case is cooling of unconfined vertical hot wall by piezo fan. Flow velocities are quite low, so compressible flow is out of scope. The fan itself is the only reason for the excitment of air. No velocity defined at the inlet. The picture below is taken from a paper. Authors claim they validated their simulation by experiments under these boundary conditions. Last edited by coffee_break; December 8, 2023 at 05:57.

 Tags convergence issue, reverse flow