Learning Forum

[ZX]

I am trying to simulate a stirred tank with impellers. The tank contains water and air, with air above the water and the impellers submerged in the water.

Regarding the rotation modeling: I understand that the Multiple Reference Frame (MRF) approach is typically used for steady-state simulations with a fixed mesh, while the Sliding Mesh (SM) approach is used for transient simulations with a rotating mesh.

However, I noticed that for MRF, transient simulation is also available. When I use MRF Transient and Sliding Mesh Transient, I can see that, due to the impellers, the surface splashes in the water in both simulations (as shown in the image below). Comparing the turbulent kinetic energy between the two methods, I found the results are similar.

Could someone please let me know the difference between MRF Transient and Sliding Mesh Transient?

Also, in which scenarios should I use Sliding Mesh, and in which scenarios should I use MRF Transient?

Thank you very much.

 

[Rob]

Very simply sliding mesh moves the impellor whereas mrf just mimics that effect. Sliding is likely more accurate as there's no averaging (read the mrf maths) but in cases with no baffles I'd expect the results to be near enough identical. With baffles the difference will be due to the blade-baffle pass interaction. 

[ZX]

Hi Rob, 

    Thank you very much, now I understand that Sliding is more accurate as there is no average. 

    But could you please elaborate on the statement “With baffles the difference will be due to the blade–baffle pass interaction”? Specifically, could you explain why the blade–baffle passing interaction leads to differences between the MRF transient and sliding mesh approaches?

    Thank you.

[Rob]

Think about what happens when the impellor tip passes the baffle. What do you expect to see for a close pass system? Would that flow feature differ from a system with no baffles (but no vortexing) or with a larger gap between blade tip and baffle. 

[ZX]

 

 

Thank you very much.

I also read the math of MRF, and I read that: for the Relative Velocity Formulation, The momentum equation contains four additional acceleration terms. The first two terms are the Coriolis acceleration  and the centripetal acceleration , respectively. These terms appear for steadily moving reference frames (that is, and are constant). 

Therefore, if my moving reference frame is steady, does it mean the relative velocity Vr (the velocity viewed from the moving frame) is constant? As Vr is constant, therefore it can not treat the pulse (like the balde-baffle pass interaction).

Not sure if my understand is correct, thank you.

 

 

[Rob]

It can't see the pulse yes, and depending on the relative positions in mrf you may get a differenet result. 

[ZX]

Thank you very much!