CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   Grid Interfaces (http://www.cfd-online.com/Forums/fluent/71644-grid-interfaces.html)

soggy316 January 11, 2010 01:15

Grid Interfaces
 
I am solving a case where a flow is past a vibrating cylinder. After much heartburn, I have got the dynamic mesh working for a small rectangular neighborhood around the cylinder, the rest of the domain being stationary. As a result, I have to define interfaces between the two zones.

Now when I create a grid interface FLUENT creates a wall at that location. and I do not require that as the the two zones are the same fluid.

Is there a way to prevent the creation of those walls?

-mAx- January 11, 2010 02:20

you should create your interfaces in Gambit (2 superposed interfaces set as 2 separated interfaces --> interface_1 and interface_2)
Then in Fluent, go and define interface as one set, say interface_A, and choose interface_1 and interface_2.

herntan January 11, 2010 03:51

cannot understand what u trying to do. Please post picture and describe more detail.

soggy316 January 11, 2010 05:33

Here is a screenshot of my problem domain

http://i46.tinypic.com/2igzity.jpg

I create 2 mesh files

The yellow box around the circle is the boundary between the two mesh files

The common boundaries of both meshes are set to "interface" boundary condition.

the grid is assembled using tmerge

Now I bring the entire mesh into fluent, and assign grid interfaces as left, right, top, bottom, using the overlapping interface boundaries of both meshes

Now, when I do this , it creates a number of walls at the interface location, which totally screws up the flow solution.

My question is - how do I create the grid interface such that either no walls are created, or how do I delete/deactivate these walls that have been created automatically

herntan January 11, 2010 05:57

soggy,

Grid interface will only useful on 2 touching surface (not common one) and after create grid interface, this 2 surface will allow fluid flow through each other.

For your case , if its a common surface, change it to "interior" then the flow can pass through each other already.

If u want to have movement of the yellow box, u need to create 2 surface touching like i said in order for the touching surface sliding on each other.

hope this helps.

soggy316 January 11, 2010 06:05

^ so as I understand it, you are saying that I cannot have an interface between two fluid zones in fluent.

The issue that I am facing is that the circle moves up and down inside the yellow box, the mesh in the yellow box is thus deforming, but the surrounding mesh is stationary, so at the border , the mesh will be non-conformal, so I went for grid interfaces.

If fluid-fluid interfaces cannot be created, then does anyone have any idea on how to get the remeshing option to work in dynamic meshes?

-mAx- January 11, 2010 13:32

Quote:

Originally Posted by soggy316 (Post 242121)
^ so as I understand it, you are saying that I cannot have an interface between two fluid zones in fluent.

The issue that I am facing is that the circle moves up and down inside the yellow box, the mesh in the yellow box is thus deforming, but the surrounding mesh is stationary, so at the border , the mesh will be non-conformal, so I went for grid interfaces.

If fluid-fluid interfaces cannot be created, then does anyone have any idea on how to get the remeshing option to work in dynamic meshes?

what you are trying to do seem to be ok from my side.
If your motion is a translation, I would let move your box (and the fluid domain inside, as rigid body), and use remeshing option (--> no need interfaces).
But your model should work (you set up 4 interfaces pair).
Can you provide the .dbs file (from Gambit)
Why are the outter edges of the big domain yellow?

herntan January 14, 2010 21:36

soggy,

i guess you have to post more picture on what u trying to do with pointer pointing out how u set it.

i agree with -Max- comment on having ur box as rigid body.

another method is setup interface pair. u need to change ur gambit model and break up the common surface then create interface out of it.

soggy316 January 14, 2010 23:24

sorry for the delayed response...

The flow simulation is working with the setup that I have created (with four interface pairs) I was getting confused because Fluent-Post was screwing up the streamline plot, the vector plot confirms that the fluid is able to flow through the interface without problems.

Thanks for your help - herntan and -max- :D

alex-fan February 2, 2010 21:53

your job is great can you email a example of dynamic mesh thank you verymuch

enry March 10, 2010 12:44

Hi Soggy,
I also use interface between 2 portion of fluid: moving and stationary fluid.
I also noted that FLUENT creates 2 new wall, that is mentioned in User manual, but I don't understand if I have to change the boundary conditions that FLUENT assign to the new walls. Flow go through the interface, so I realized that everything is correct, but I would like to know why FLUENT create that walls.
Thanks.

zwsok May 4, 2012 01:40

Moving mesh inside another mesh
 
Quote:

Originally Posted by -mAx- (Post 242191)
what you are trying to do seem to be ok from my side.
If your motion is a translation, I would let move your box (and the fluid domain inside, as rigid body), and use remeshing option (--> no need interfaces).
But your model should work (you set up 4 interfaces pair).
Can you provide the .dbs file (from Gambit)
Why are the outter edges of the big domain yellow?

Hi all,

I just saw this suggestion on the forum and would like to consult people on the following problem:

I am currently working on a moving airfoil (translational and rotational). Since I need resolution near the airfoil as I want to observe vortex shedding, I would like the mesh around it to be structured and preferably fixed. The motions will be accommodated by the unstructured mesh outside the structured one, with smoothing and remeshing applied to it.

I am working on Fluent and the Design Modeler and Mesher from ANSYS workbench. Should I build 2 meshes or one? And how do I make one move and have the other one respond accordingly?

Your help will be so very much appreciated. Thanks!

Regards,
Sok

samiam1000 May 17, 2012 05:17

Dear Friends,

I read very carefully this thread and I have found it very useful.

I have some questions for you: first of all, I would like to know which solvers can be used to threat a moving mesh. I would need a multi-region domain solver, is it possible?

Also, one more point: is it possible - instead using moveDynamicMesh - to create 10 different meshes (with the fluent mesher) and then to perform a simulation on those meshes? If so, how can I pass the solution at time n as initial guess at time (n+1)? This could be a challenging point, since the meshes are differet at each time step.

I think that's all for now. Thanks for the help you will provide.

Samuele


All times are GMT -4. The time now is 15:52.