CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   FLUENT (https://www.cfd-online.com/Forums/fluent/)
-   -   Clicking/Sliding Grid (https://www.cfd-online.com/Forums/fluent/27449-clicking-sliding-grid.html)

L.carvalho January 18, 2000 08:36

Clicking/Sliding Grid
 
Hello

Has anyone any idea how to simulate a sliding grid in Fluent? Image you have a room and one of its wall move with v=x m/s. It is obvious an unsteady problem, where your domain/grid changes with the time. What could I do here? Another question: What is clicking grid? I am sorry, but I have never heard about it!!! Thanks for responding!

Bye L.C.

Jonas Larsson January 18, 2000 09:15

Re: Clicking/Sliding Grid
 
If you just have one moving wall which is flat and moves parallel to the domain (ie the domain volume is always the same), then you don't need any slidig grid - you can just set a translational velocity for that wall. This case is not by definition unsteady, it might well give a steady flow-field. If the wall is moving so that the shape/volume of the domain changes then Fluent can not handle it.

You can do sliding interfaces in Fluent. Typical applications are for example two cars meeting, stator-rotor interaction in turbomachines, ... The implementaion of these sliding interfaces in Fluent is not very stable though. You need to learn a lot of tricks to get it to work - For example, when setting up a stator-rotor case I have to do a "check grid" each time before I define a new "rotationally periodic sliding interface", another way is to first define them translational, delete them, and redefine them as rotational... not that obvious ;-) To get it to run in parallel demands even more witchcraft... but it does work eventually in most cases.

With "clicking grid" I assume that people mean a sliding interface which is implemented so that the meshes on each sides have to match and that the grid is moved one cell in each time-step. Fluent does not have this demand - they have a general interface and you can have different grids on different sides and move them with whatever time-step you chose.

L.carvalho January 19, 2000 08:35

Re: Clicking/Sliding Grid
 
Thank you very much for responding! Well, the problem I have is to model the flow in a rotary displacement machine, which has two rotors that rotates with a defined velocity. First I want to calculate the flow field in 2 dimensions. We have always a different grid by a different position of the rotors. Since the grid can not "move" I am trying to find a solution for calculating it in Fluent. I thought I could do that:

(1) read grid 1 "grid1.msh" (2) Model: rotational velocity for the walls, calculates steady until convergence (or maybe unsteady until the next position is reached - the time here is known and fixed) (3) Save data-file "position1.dat" (4) Read grid 2 (with the same number of cells) (5) See (2) (6) Save data-file "position2.dat" (7) Read grid 3 and so on...

I think that is a way to calculate the flow field for that model, but I am not sure if is a good way! I don't know either if Fluent can read a new grid keeping the data results of the past simulation. I would like to hear your opinion and from others too! Thank you, LC.

Jonas Larsson January 19, 2000 17:41

Re: Clicking/Sliding Grid
 
I don't fully understand what you are trying to do, but it sounds very impractical. Does your grid really change with time or are just some parts rotating? If some parts are just rotating then you can use a sliding interface in order to have those parts of the grid rotate with the geometry. The rotating parts of the grid can "slide" against the stationary parts of the grid. I don't know if this is applicable to your applications though.

Fluent might be able to interpolate one solution from your old grid onto a new grid, but using that to simulate a deforming grid sounds very complicated and probably impossible.

L.carvalho January 20, 2000 09:10

Re: Clicking/Sliding Grid
 
Thank you again for your opinion. The grid is continuously in motion and changing in shape indeed. It is not like rotor-stators! The geometry of the grid may be defined at all the times and its motion described by rotation in case of 2D. If a general moving grid could be built into the numerical calculations, it would be nice here. The sliding mesh from Fluent will not probably help me in this case! Do you think is it possible to create a macro or an user-define-function which gives the next mesh at a determined time? I mean the programm will read/know exactly the new nodes values each time step. I have to say that I work at the university, in other words - I would have the time to develop things like that, but I would like first to hear opinions like yours!

Best regards, LC

Jonas Larsson January 20, 2000 09:57

Re: Clicking/Sliding Grid
 
I doubt if it would be possible to implement moving/deforming grid functionality in Fluent via a UDF, someone please correct me if I'm wrong. This is something that you should be talking to Fluent about though, they can surely give you a better answer. If the moving/deforming grid feature is essential to you I think that you would be better of switching to another code. The next version (6) of Fluent will include moving grid features, but it is not due until late this year or next year I think.

Kai January 20, 2000 12:42

Re: Clicking/Sliding Grid
 
Hi Larsson,

Have you ever tried sliding mesh with translational interface, such as in the case of two cars met, any ideas of setting up the boundary conditions for this type of problem. I tried several times, but it seems the flow fields do not look right, it looking like the car is moving inside a wind-tunnel...

Best KK

Jonas Larsson January 20, 2000 13:26

Re: Clicking/Sliding Grid
 
Nope, I've never used a translational sliding interface in Fluent. I'm working in the rotating world of turbomachinery. The rotational sliding interfaces work quite well if you have a fine enough grid and use a proper time-step - wakes and pressure waves most often pass without too much distortion. I can't see why the translational interfaces should work any worse, they probably have the same basic schemes.


All times are GMT -4. The time now is 14:48.