CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   Stirred tank Using MRF (http://www.cfd-online.com/Forums/fluent/126213-stirred-tank-using-mrf.html)

gurpreet89 November 11, 2013 05:55

Stirred tank Using MRF
 
[LEFT]hi friends,
i am working on stirred tanks (baffled) with rushton turbine using multiple reference frame for my project work. I am very new to the fluent. i have made the geometry of stirred tank in designmodeler ,i have created an interface between the rotor(moving region) and fluid(stationary region). Now when i am rotating the fluid in contact with the impeller at 180 rpm, keeping the impeller and the outer fluid stationary i am getting the vector plots as given in pictures
http://s7.postimg.org/xlpr80zfv/FLU1.png
http://s22.postimg.org/in686v3cx/FLU2.png
http://s23.postimg.org/cn2e9dwob/FLU3.png

as the fluid crosses the interface it changes its direction, why is it happening?? plz help

ghost82 November 11, 2013 08:09

Hi,
check your cell zones; assign to rotor zone absolute rotational velocity of xxx rpm; make stator zone stationary; in boundary conditions set shaft+impeller walls rotating at a relative rotational velocity of 0 rpm and to baffles/tank absolute velocity of 0 rpm.

PS: positive rotational velocity means conterclockwise rotation

Daniele

gurpreet89 November 14, 2013 03:26

thanks for the reply
 
Actually there are many walls in my case in the 'boundary conditions' i dont understand which one is for rotor and which one is for baffle. i used display option to show those walls but it is not showing any. dont know whether there is any mistake in my geometry.

i iterated it for 1500 iterations but after around 1000 graph becomes constant.do i need to iterate more to get better results?

can you send me one of your example so that i can compare?
gurpreet89@gmail.com

ghost82 November 14, 2013 04:14

Then, turn back on your geometry and fix it.
An example of mrf is from Mr. Bakker,

Stirred tank flow field model

http://www.bakker.org/dartmouth06/engs199/

Daniele

gurpreet89 November 16, 2013 09:46

thanks for replying, i visited bakker's website. He is a very well known personality in the field of stirred tanks and CFD. i downloaded his one of the homework file 'pbt.msh.gc' . i imported that file then putted the same boundary and cell conditions as you told
in cell zone conditions - gave fluid-impeller the 'absolute' volcity of 30 rpm and kept fluid tank as stationary.
in boundary conditions - gave relative velocity 0 to the impeller and shaft, and 'absolute' velocity 0 to baffles and tank wall.

i used 'coupled' discretisation method and did hybrid initialization, result i am getting as below

http://s21.postimg.org/x1yxduot3/FFF.jpg
above image is streamline view

http://s22.postimg.org/4tuu1pfm9/top.jpg
contours (top view)

http://s23.postimg.org/pj7igrynv/topv.jpg
vectors
http://s18.postimg.org/4kgq46ord/topv2.jpg
as in my case, the vectors inside the rotating zone are moving in opposite direction to the stationary zone. plz help :confused:

ghost82 November 16, 2013 09:52

Hi,
can you plot velocity in stn frame in cfd post?

Daniele

gurpreet89 November 16, 2013 10:41

yes ,now i am getting in the same direction :)
thank you very much

ghost82 November 16, 2013 10:54

Just for your info, when you plot velocity in stn frame you have velocity with regard to a stationary frame.
When you plot velocity you have velocity with regard to the moving frame.

Daniele

solmaz aryafar May 14, 2014 11:36

Hello Daniele and gurpreet89,

First orf all thanks for the useful topic!

I have a question!!
I tried the same example as you were talking about above (pbt.msh.gz) , and I don't get the convergence at all. I used the same cell zone conditions and boundary conditions, and just one thing I'm not really sure about is the "direction of rotation" which I selected it to be (1 0 0), is that correct?
So why my solution doesn't converge???

Actually I tried this example, because I have the same problem with my own example, which is a stirred tank as well.
Please help me, I'm running out of time!

Thanks in advance,
Solmaz.

gurpreet89 May 17, 2014 23:20

hi,

(1 0 0) means rotation about x-axis, similarly (010) fr rotation about y-axis and (001) about z-axis. If the shaft of the impeller is in X direction then take (100) but i think in that example the it was (001).

Gurpreet

solmaz aryafar May 20, 2014 08:32

Hello Gurpreet,

Thanks a lot for your answer! I guess now it's getting better, but no in that example the shaft was in the direction of x-axis ;)

Look, I have a question here from you, because as I saw in the forum you work (or worked) on stirred tank simulations, am I right? So here I have a problem!

I created a mesh in Ansys Meshing for my stirred tank which have pitched blade impellers, and I used MRF approach, but I had issues in convergence, I mean the solution did not converge! As I showed the mesh to an expert (which I couldn't reach anymore unfortunately) he said that the quality of mesh is not good, and that I need to try Sliding Mesh approach. Then I researched a little bit in forumes, and also teh website of Prof. Bakker, and now it seems to me that I need to mesh the part arround the impellers with structured cells as the rest of the tank, am I right? so does it mean that I have to use sweep mesh for this part? I imagined like that, so I tried to mesh this part with sweep methos but the Ansys Meshing complains and says: "this part is not sweepable. So my question is that, I need to use ICEM instead of Ansys Meshing? I mean it's not possible to make sweep meshes arround a complex geometry (like the impellers) with Ansys Meshing?

I will appreciate your (or any one else's) help!

Solmaz.

gurpreet89 May 26, 2014 10:33

hi,
what convergence criteria you are following, 10^-3 or 10^-4 ?
In my case, i was able to converge it to 10^-3 using ansys meshing and went closer to 10^-4. i made my rotor and baffles as structured but for fluid zone i used tetrahedral (unstructured).
may be your results are not converging due to some geometry issues, if you want your results to be very accurate with error less than 10- 5 % or so,then you can go for ICEM.
Results don't vary much either you use MRF or SM, lot of journals are there on the same.


All times are GMT -4. The time now is 23:01.