[Yannian]

I want to simulate a vortex going towards the propeller by MRF. I want to know whether the interface between rotating and stationary domain can transfer the vortex? Now it seems that it doesn't work, and the vortex can't be transfered continuously through the interface.

[ghost82]

Yes, there are no problems..
I'm attaching a simulation with mrf of an impeller and vortex formation.

[Yannian]

Originally Posted by ghost82

Yes, there are no problems..
I'm attaching a simulation with mrf of an impeller and vortex formation.

Thank you for your reply. It is a really positive news to me.
I saw this case in CFX tutorial before, and is your computation in CFX as well? I am using Fluent now, i don't know whether there is difference for these two codes dealing with the interface. I can't get good result from Fluent untill now.

[ghost82]

I used Fluent for the simulation.

Daniele

[Yannian]

Originally Posted by ghost82

I used Fluent for the simulation.

Daniele

Hey Dani,
I explain to you my case and you can have a better understanding.
I build the propeller in a rotating domain and the rest is stationary domain. My stationary domian is not axisymmetrical because i am simulating a propeller running near the ground, and there is a ground in the stationary domain which is close to the propeller. There is vortex from the ground to the propeller.
Do you have any clue what can casue the vorticity not continuous? i have even tried conformal mesh because i thought the problem is due to the 'hanging' nodes. However it doesn't help.

[ghost82]

Post some pictures (contours) of what you obtained and what you think to obtain

[Yannian]

flow domain.jpg

vorticity_isosurface.jpg

Originally Posted by ghost82

Post some pictures (contours) of what you obtained and what you think to obtain

The flow domain and the vorticity isosurface are in the attachment and could you please have a look.

[Yannian]

Originally Posted by ghost82

Post some pictures (contours) of what you obtained and what you think to obtain

axial velocity in the plane in front of the propeller.jpg
As you can see from the vorticity isosurface and axial velocity, the flow is not continuous at the interface.

[ghost82]

Hi Yolk,
First question is: are you sure your solution is converged?
I know, also from the fluent user guide, that some discontinuities could be present across the interface, however, if you did not try, you can avoid interface for mrf calculation (i.e. connect in pre processing the faces shared by the rotor/stator) and see what's happen.

Daniele

[Yannian]

Originally Posted by ghost82

Hi Yolk,
First question is: are you sure your solution is converged?
I know, also from the fluent user guide, that some discontinuities could be present across the interface, however, if you did not try, you can avoid interface for mrf calculation (i.e. connect in pre processing the faces shared by the rotor/stator) and see what's happen.

Daniele

Hi Dani,
I would say the solution is converged: first, the thrust on the propeller is converged; second, the vortex strength on the ground is converged.
I still think the problem is from the interface. Now i am trying a simple case by adding a vortex in the inlet and let it go from a stationary to a rotating domain, and see what happens.
How can you connect the faces shared by two domains in FLUENT?
Yolk

[ghost82]

Originally Posted by Yannian

How can you connect the faces shared by two domains in FLUENT?

You can't in fluent; you have to go back to preprocessor, connect the faces and remesh the domain.
You still have the two fluid domains (rotor and stator) but they share adjacent faces, instead of having interfaces.

Daniele

[Yannian]

Originally Posted by ghost82

You can't in fluent; you have to go back to preprocessor, connect the faces and remesh the domain.
You still have the two fluid domains (rotor and stator) but they share adjacent faces, instead of having interfaces.

Daniele

Okay, I see your point. When building the geometry, the two domains are combined under one part. There will be two domains in the ANSYS mesh, but there is only one face between the two domains. The face shared by two domains becomes interior in FLUENT boundary set up.

[ghost82]

Yes, I don't use ansys meshing, but your description is ok.

[Yannian]

I tried the case with vortex from the inlet and vortex goes through stationary and rotating domain. The vorticity isosirface seems continuous although there is some discontinuity at the interface. The mesh doesn't have interface and there is interior face between two domains. So i think you are right that the interface can transfer the vortex. One issue is that the interface in this case is a plane not a circular surface, and maybe the circular surface makes the problem more difficult. Anyway, i will try the latter cace later.

[Yannian]

Originally Posted by ghost82

Yes, I don't use ansys meshing, but your description is ok.

Hey Dani,
I upload two images with circular interface perpendicular to the vortex axial velocity. In the case for the first image without frame motion in the inner domain, the vortex can go through it continuously, In the case for the second image with frame motion in the inner domain, the vortex can't go through it continuously, and the vortex isosurface shown here looks like the image I showed before.
I think the circurlar interface, and rotating frame together make the vorticity with discontinuity. Do you have any idea how can solve this problem?

[ghost82]

But in the second picture, I think you are rotating the domain around y axis, don't you? If so, I think it's correct that the vortex is broken.

Daniele

[Yannian]

Originally Posted by ghost82

But in the second picture, I think you are rotating the domain around y axis, don't you? If so, I think it's correct that the vortex is broken.

Daniele

Yes, you're right, the axis of the rotating frame is in the y direction. Why do you say that the vortex is broken is correct? My intention is to let it go through the inner cylinder. Is there a reason for its breaking?

[ghost82]

Well, I don't know...usually mrf, as you already know, is used to take a "snapshot" of a transient simulation, in which inside the rotor there are moving parts.
But in your test case there aren't rotating parts inside the rotor zone (the rotor is empty), so I think that this is not the correct approach to see is the vortex is broken at the interface.
Maybe that cells in the rotor are rotating with the assigned rotational velocity, so I wrote that it is correct that the vortex is broken.
To see if the cells are rotating post a contour plot of velocity in a xz plane.

[Yannian]

Originally Posted by ghost82

Well, I don't know...usually mrf, as you already know, is used to take a "snapshot" of a transient simulation, in which inside the rotor there are moving parts.
But in your test case there aren't rotating parts inside the rotor zone (the rotor is empty), so I think that this is not the correct approach to see is the vortex is broken at the interface.
Maybe that cells in the rotor are rotating with the assigned rotational velocity, so I wrote that it is correct that the vortex is broken.
To see if the cells are rotating post a contour plot of velocity in a xz plane.

Hey Dani,
I don't think the cells in the inner domain is rotating from the velocities.

[Yannian]

Originally Posted by ghost82

Well, I don't know...usually mrf, as you already know, is used to take a "snapshot" of a transient simulation, in which inside the rotor there are moving parts.
But in your test case there aren't rotating parts inside the rotor zone (the rotor is empty), so I think that this is not the correct approach to see is the vortex is broken at the interface.
Maybe that cells in the rotor are rotating with the assigned rotational velocity, so I wrote that it is correct that the vortex is broken.
To see if the cells are rotating post a contour plot of velocity in a xz plane.

A new case without inflow vortex in the outdomain and with frame motion in the inner domain is presented. There is vorticity from the inner cylnder.