[mohammad]

Dear all,

Recently I am doing a heart pump design project. Meanwhile, I am simulating different configurations for creating the data sheets and curves, so sometime I need to solve different models ranging from single discs to the 3D models.
As a common method, I usually use MRF for 3D rotating impllers. HOwever, I was thinking if I could activate and apply the wall velocity to the impller part option in stationary FRAME without using MRF.

What would be the differnce between the following cases?
1) using a MRF technique with 2 MRF/STATIONARY domains;
2) using the STATIONARY domain and applying the rotation to the blades.

Thanks

[Opaque]

Option 2 is ill-defined. You cannot move the wall normal to the without deforming the mesh, or moving it as a rigid body.

Option 2 should be defined as: solving the problem in the STATIONARY frame, but moving the mesh as a rigid body and applying the rotation to the blades.

Now the two problems are equivalent on paper, but the differences will come out of discretization errors. In the moving mesh case, you are approximating the velocity of the mesh while in the rotating frame those are exact (though there other approximations there as well).

Hope the above helps..

[mohammad]

Originally Posted by Opaque

Option 2 is ill-defined. You cannot move the wall normal to the without deforming the mesh, or moving it as a rigid body.

Option 2 should be defined as: solving the problem in the STATIONARY frame, but moving the mesh as a rigid body and applying the rotation to the blades.
....

Hello and thanks Opaque,

So, what you mean is that that moving mesh is actually for dynamic mesh...Am I right?

Then,please consider two parallel circular discs ( like two CDs):
CASE 1) Both are rotating with the same RPM in the same direction...I considered the stationary domain and I just applied the rotational velocity to both discs ...the convergence is good, No error, and results ( Vel. profiles, etc...) are logical...

Question1) according to your comment, I need to use MRF. right? Then my question is that in the MRF, the two discs must be the MRF boundary? I have always worked with MRF in two domains ( inner and outer and a MRF interface).... Is it possible that I apply MRF to a single domain?

CASE 2) Only one of the discs is rotating and the other is stationary. How should I define the MRF domain and what should be the BCs?


Thanks for your favor, in advance.

[ghorrocks]

Moving mesh, dynamic mesh... same thing.

You need to use MRF if the mesh which defines the domain rotates. So if your impeller has blades then you need MRF. If it is just a rotating disc and has axial symmetry then it can be modelled as a stationary frame of reference with a tangential wall velocity. But note the wall velocity can only be tangential for the stationary frame of reference case. If the wall velocity has any normal component this will not work.

[mohammad]

Originally Posted by ghorrocks

So if your impeller has blades then you need MRF.

Dear Glenn Horrocks,

My model has impeller. We have first simulated a model in which the impeller is surrounded by the blood, and we got severe recirculation in the gap between the bottom of impeller and the bottom of the casing. Then we decided to remove the gap between these two part. So, the new model may look like the image I have attached. ( At this point we do not consider if it is possible to make such a product or not).
as you see, the bottom of the rotating impeller is in fact the bottom part of the MRF domain. My question is how should I define the location of MRF interface in radial direction ( left side of the photo) and at the bottom?

Thanks

[ghorrocks]

This can be modelled easily using MRF. Have a look at the CFX tutorials for how to do this. You have a lot of flexibility in where you put the interfaces. Anywhere beyond the impeller will be fine.