Learning Forum

[ztjia]

when I simulated a stirred tank with MRF, I found significant difference on the border of station region and rotate region in attached picture. details about model and simulation were as follow:

TFM-RANS (mixture realizable k-e).

phase coupled simple scheme was adopted for the pressure-velocity coupling.

Quick and second-order upwind scheme for spatial discretization of volume fraction and momentum.

the simulated domain was meshed by poly-hex-core with fluent meshing shown in the attached picture. waiting for your reply, thank you sincerely.

[Rob]

That does look odd. Given the likely flow field I'd tend to favour a pure poly mesh, but poly-hexcore should also be OK. Double check for stray wall between the rotating and stationary zones. How good is the convergence, and how long has the model been running? How does the air get out of the domain? 

[ztjia]

thanks for your reply, is there another possible reason for this phenomenon?

[ztjia]

thanks for your reply,Bob. all the residual were below 10e-3, steady simulation was used about 100000 iterations, the air was defined as a velocity inlet, at this condition, the air velocity was 22 m/s, the stirred top was degassing condition. I have checked the mesh, the rotate region and station region have been shared topology in SCDM, the mesh were commom nodes.

[Rob]

If you look in 3d is there an explanation? Please post the velocity plot too. 

[ztjia]

thanks for your reply. the air velocity plot is as follows.

[Rob]

That looks fairly sensible, and the water?