[Jeremie84]

Hi,
I would like to submit a question to the Fluent users about the Multiple Rotating Reference Frame (MRF).
After reading carefully the Fluent User’s Guide, I still have this question. For a steady-state analysis, what is the difference between the MRF approach and simply setting wall velocities (moving wall) without any rotating reference frame? Does MRF only help for convergence in that case?
Finally, with the MRF approach, should the boundaries of the rotating reference frame be placed far enough from the rotating elements?
Thank you for your help.
Jérémie

[vicarious]

Hi,
As You know There are 3 approaches for simulating the flow in such problems : MRF, Mixing plane(MP) and Sliding mesh (SM). Both MRF and SM methods assume the flow steady and are used when the stationary and rotating frame interaction is weak or when it is needed just an proximate solution. The SM model assumes the flow unsteady and is used when the interaction is important (such as compressor and turbine simulation). Briefly, if your problem deals with such devices, The SM model is the best and gives you an accurate solution. I've been using the SM model and you need to place the inlet and outlet boundaries far enough to avoid the potential effects. The upper and lower boundaries could be assumed as periodic linked meshes.

Best regards.

[Jeremie84]

Thanks Pedram
I am working on a vertical axis wind turbine. As a first approximation, I’d like to perform steady-state simulations. And I really wonder what would be the difference between a solution obtained in a rotating reference frame (MRF approach), and a solution obtained in a stationary reference frame (with a moving wall condition for the wind turbine blades). Is it easier to get a converged solution using the MRF approach?
Regarding the boundary, I meant the boundary between the rotating reference frame and the stationary reference frame (in the MRF approach). Should the boundary be placed far enough from the wind turbine blades?
Thanks again
Regards,
Jérémie

[vicarious]

Yes it's easier and better to consider the MRF method since it's been recommended. If you mean the interface between your rotary and stationary frames,(for example boundary between rotor and stator cascades); No it is not needed to place the boundary far from the blades. It could be at any distances regarding to the test stages or other setups.

Regards.

[ghost82]

Originally Posted by Jeremie84

Hi,
I would like to submit a question to the Fluent users about the Multiple Rotating Reference Frame (MRF).
After reading carefully the Fluent User’s Guide, I still have this question. For a steady-state analysis, what is the difference between the MRF approach and simply setting wall velocities (moving wall) without any rotating reference frame? Does MRF only help for convergence in that case?
Finally, with the MRF approach, should the boundaries of the rotating reference frame be placed far enough from the rotating elements?
Thank you for your help.
Jérémie

Hi, you can use moving wall (rotation) only when your rotating object is a surface of revolution; also this surface must be centered around the rotation axis; for example you can use moving wall when you have a cylinder (surface of revolution), centered in 0,0,0, positive in Z direction, rotating around z axis; you cannot use moving wall for example for a rushton turbine, as it is not a surface of revolution, even if it is centered in 0,0,0.

Daniele

[Jeremie84]

Thank you very much for your answers.

@Daniele: after reading your comment, I read this on the Fluent user's guide:
"Note that the modeling of tangential rotational motion will be correct only if the wall bounds a surface of revolution about the prescribed axis of rotation (e.g., a circle or cylinder)."

Thanks again for your help.

Regards,

Jérémie

[Far]

Originally Posted by ghost82

Hi, you can use moving wall (rotation) only when your rotating object is a surface of revolution; also this surface must be centered around the rotation axis; for example you can use moving wall when you have a cylinder (surface of revolution), centered in 0,0,0, positive in Z direction, rotating around z axis; you cannot use moving wall for example for a rushton turbine, as it is not a surface of revolution, even if it is centered in 0,0,0.

Daniele

You can use the offset option in fluent if axis are not passing through centre of the domain.