MRF model - Meshing and Interface set-up
 

Hello all,

I'm using ANSYS-14 and currently working on a mixing tank (with baffles) problem. However, I am encountering the issue of bad mesh quality (with min. orthogonal quality around 0.03-0.04 and average around 0.9). The mesh was swept automatically, and all I did was apply inflation for near-wall regions. The mesh looks fine (with hex cells), though and I would like to know if a low value of min. orthogonal factor is something of great concern, or if it is acceptable as long as the average value is high(close to 1). If it is not acceptable, could you please tell me how it can be improved?

Also, once I clearly name the impeller fluid region and the exterior fluid region (the one outside the rotating frame) in meshing, fluent seems to identify these "zones", and hence, three surfaces starting with the word "interior" appear when I click on "mesh display" in fluent. Also, in boundary conditions, I'm unable to change these zone types to "interface". So, does this "interior" naming mean that fluent has recognized the interface regions, and will create the interface automatically? Or am I mistaken and there's a manual way to do it?

Also, I used the RNG k-epsilon model and after setting up the solution and running the calculation, the solution started to diverge, and I had to cancel it. But the same thing for an unbaffled tank worked, where I had a min. orthogonal quality of about 0.2, and I got a proper solution.

So I really need some help regarding the above three points, please do provide your inputs ! Thanks a lot !

Anybody? Please help ! :confused:

Hi dear friend
I have some experience in stirred tank simulation(gambit+fluent) but I dont have experience in workbench. if you know this useful for you, tell me to help you. I use structured mesh by GAMBIT for stirred tank.

Hello jamalf64,

I've really never used GAMBIT before and so I don't know if I'll be able to link whatever you suggest to the procedure in FLUENT and Workbench Meshing. Thanks anyways. :)

please send to me a picture of your geometry and its dimensions. I create that in gambit and will send to you.

I will post them here so that others might as well provide their inputs.

Dimensions (Standard Rushton Turbine) : http://i62.tinypic.com/2a5ey2w.jpg

Geometry (sliced into four parts for sweep meshing, and two fluid zones: one around the impeller, and one exterior to the first zone) :

http://i60.tinypic.com/fc76lu.jpg

My Mesh (tetrahedral around impeller, hexahedral in the rest of the domain) :

http://i60.tinypic.com/1z3yv00.jpg

I used the realizable k-epsilon model and got the solution to converge, and the convergence was quite stable, not with many oscillations, with the criteria set to 1e-04 and I hope that is sufficient. My mesh min. orthogonal quality remains at 0.14 and I still don't know how I can improve it. Any ideas anyone?
I'm also not sure if the interface got created properly by fluent but I think it did because I get to see the interface (named "interior_region1_region2", where region1 and region2 are the two zones on either side of the interface) in the "display" option in Fluent.
And also I don't get why the RNG model failed to give convergence while the realizable one could do it. Please help ! Thanks !!

For improve mesh quality you can split MRF zone to 6 volumes like attached pictures. now you must mesh these volumes using structured elements.

Ok jamalf64, I'll try that and get back to you. Is it like some thumbrule, having more regions improves mesh quality? Also, could the mesh quality be affected because I'm having tet elements near the impeller region? Thanks.

1- In fact, by split a volume can create volumes that have ability to structured grid.
2- Improve mesh quality cause to convergence problem be easy.

forgive for bad English writing.:D

lol its ok, and thanks, I'll try that suggestion and get back to you.