[liz]

Hi,,
I'm trying to model heat transfer of a rotating periodic domain, but settings for opening boundary condition confuse me. When I set opening condition's frame type stationary, terrible imbalance of energy equation happens(larger than 1 sometimes). But if I change frame type to rotation, imbalance of energy equation reduces to zero but convergence of momentum becomes difficult. Do you know the causes?
Model details: k-epsilon, scable wall function, thermal energy, rotating periodic interfaces;walls with fixed temperature, stationary relative to frame type; opening boundary with 0 relative pa, fixed open temperature
Any discussion is welcomed.

[Opaque]

Not sure if it matters for your case, but rotating domains should always be run using Total Energy unless you already know that the angular speed will not make much difference in the solution, i.e. it could have been run as stationary anyways.

You are using opening instead of inlet ? Are you expecting recirculating flow at the boundary ? If you do not expect such flow, you are better off running as an inlet directly.

[liz]

Originally Posted by Opaque

Not sure if it matters for your case, but rotating domains should always be run using Total Energy unless you already know that the angular speed will not make much difference in the solution, i.e. it could have been run as stationary anyways.



You are using opening instead of inlet ? Are you expecting recirculating flow at the boundary ? If you do not expect such flow, you are better off running as an inlet directly.

Thanks for your reply. It's a pity that total energy doesn't give a better result.
ImageUploadedByCFD Online Forum1417138323.321329.jpg
This problem is about a vented disc, rotating with pumping effect, so inlet velocity is not very clear. The model is similar to attachment.And the vented part may produce mesh motion in stationary frame. Rotating frame is needed.

Model looks like this. ATTACH]35586[/ATTACH]

Do you think opening boundary needed to be changed?

[ghorrocks]

Have you done the standard stuff when you have convergence difficulties? This FAQ discusses it: http://www.cfd-online.com/Wiki/Ansys...gence_criteria

[liz]

Originally Posted by ghorrocks

Have you done the standard stuff when you have convergence difficulties? This FAQ discusses it: http://www.cfd-online.com/Wiki/Ansys...gence_criteria

Hi,
Almost every method proposed in the link page has been tried，but it doesn't work. Something important may be ignored, can you give me some advices? or could you read the .cfx file?
Thank you so much!

[ghorrocks]

What do your residual histories look like?

Have you increased the time step size and run it for a bit longer? It might not be converged yet.

[liz]

Originally Posted by ghorrocks

What do your residual histories look like?

Have you increased the time step size and run it for a bit longer? It might not be converged yet.

Yeah, you're right. Set MAX residual criteria as 1e-4, and with a better initialization, I got energy balance. To accelerate convergence, a larger local timescale factor(=50) was set.
This problem has been worked out.
ImageUploadedByCFD Online Forum1417489337.261514.jpgImageUploadedByCFD Online Forum1417489659.185954.jpgImageUploadedByCFD Online Forum1417489673.290496.jpgImageUploadedByCFD Online Forum1417489683.594484.jpg

Thanks so much for your kind help!