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/83955-stirred-tank-using-mrf.html)

dishengbin January 16, 2011 21:09

Stirred tank using MRF
 
Hi, I am modeling a stirred tank using MRF. Usually, the fluid zone is divided into two parts: a stationary zone and a rotating one. My question is can I simulate the tank without dividing the zone in the case of only one impeller( no baffle also). If yes, how to set the boundary condition in FLUENT? Is there anyone who can explain how MRF works in FLUENT? Thanks very much!

Sun January 17, 2011 08:31

Hi,
I think there is no other way to simulate the rotation without dividing the zone. To see how the MRF works in the FLUENT check out this tutorial:
http://my.fit.edu/itresources/manual...tg/node135.htm
Good luck :)

dishengbin January 17, 2011 21:36

Thank you, Sun. Can I explain Single Rotating Reference Frame (SRF) modeling like this: the fame is fixed on the impeller, and then the impeller’s relative velocity is zero in the reference frame, the fluid zone is controlled by the N-S equation for a Rotating Reference Frame. And then set the tank wall with an opposite angular velocity?

Sun January 18, 2011 05:57

Hi,
For a rotating frame of reference, Fluent (if you are using it) assumes all the walls rotate at the speed of moving reference. So, I think you do not need to assign opposite angular velocity for tank wall.

red_lemon January 18, 2011 19:02

You can model as SRF as long as it has outer surfaces of complete uninterrupted surface of revolution (no baffles).
Fix fluid as rotating in fluid panel at N rpm and specify walls as rotating at zero absolute velocity.

dv_cfd February 16, 2011 00:50

Regarding MRF multiphase steady state simulations
 
Hi,

I have a query regarding the multiphase steady state simulation. With the iterations, the mass of each phase in eulerian-eulerian simulation is varying. The system is bound by wall from all the sides, so reaching at the converged solution, the mass of each phase with in the system is supposed to be equal to the initial mass of each phase. But, on the other hand, I am getting a deviation in value of the mass of each phase initially present in the tank. So is the case with the total mass present in the system.

Can you please suggest me an explanation?

rr123 February 17, 2011 19:04

stirred tank - MRF or SRF
 
dear sirs, in a stirred tank, the impellers with the blades are the ones that rotate in reality. while, the cylindrical wall of the reactor and the baffles are fixed. the impeller rotates and this rotates the fluid around it.

depending on the points mentioned above, could you suggest which would be the correct approach to take - SRF or MRF and why ? Thank you very much.

Sun February 18, 2011 03:54

hey,

Based on whatever i read and consultant with some ANSYS experts, there would not be a significant difference between these two methods. But if you are modeling a stirred tank with baffles it is recommended to use SRF.

rr123 February 18, 2011 04:51

dear sun,

thank you for your reply.
as you mention the fact that as i am using baffles, then i must use SRF. could u kindly elaborate a little but more in this subject ?
why do the presence of baffles influence the choice between SRF and MRF ?

thank you very much.
hoping for your reply
roy.

Sun February 19, 2011 02:56

Hi,

I think you can capture the transient events with SRF better than MRF. I should check my notes to find the details or finding some references that will be useful for your work. (Maybe tomorrwo or Monday)
Good Luck :)

dj_croog February 20, 2011 20:33

I'm fairly sure you have to use MRF (not SRF) if the tank is baffled, as red_lemon mentioned....

SRF is only applicable as long as the tank walls are a surface of revolution about the axis of rotation.

http://202.118.250.111:8080/fluent/F...ug/node360.htm

You'll probably find academic papers on stirred tanks use the MRF method, usually to create initial values for an SG simulation

Sun February 21, 2011 01:28

rr123,
I made a funny mistake. I have mistaken SRF with SMM (these acronyms !). As dj_croog said. I also think you should use MRF and you can use SRF for initialization.

rr123 February 22, 2011 03:28

dear sirs,

many many thanks for your replies.
i have also seen Tutorial 9 from Fluent 6.3 Tutorial guide in this regard.
coming to our system, so we have stirred tank where i have the impeller stick and its rushton blades in the middle of the tank; and the cylidrical wall has four baffles fixed on it (a typical stirred tank reactor).
now i have been able to mesh the inside completely, through GAMBIT.

what i would like to know is that is it needed to make one cylindrical wall inside the reactor which sort of wraps around the impeller (and its blades system) with a diameter in such a way so that it is almost closer to the baffles. this wall would be designated as interior in Gambit.
this would mean that the fluid near the blades rotate, while the fluid near the baffles are almost stagnant.
am i correct in my approach ?

if not, can you send me a sample case file of your stirred tank reactor so that i can look into it for the settings ?

thank u Sun, red_lemon, dj_croog and others for helping me.

aremu March 7, 2011 10:30

Hi,
I am also trying to get a 3D stirred tank model in Fluent. Did u used the model in the tutorial or a full 3D model ? I have a full CAD model of my impeller as an IGES file draw in CATIA but the problems seem to be the periodic boundary as was done in the tutorial. Can u please share your model of getting the parts meshed in Gambit.Thanks

dv_cfd March 13, 2011 23:11

MRF- steady or unsteady and mass-imbalance problem
 
Hi,

I am still struggling to get a solution for the problem I mentioned. Please have a look at it and try to provide me with suitable explanation.

One more thing, how good are steady state simulations are for stirred tank reactors?

Thanks in advance,

My previous question:

Hi,

I have a query regarding the multiphase steady state simulation. With the iterations, the mass of each phase in eulerian-eulerian simulation is varying. The system is bound by wall from all the sides, so reaching at the converged solution, the mass of each phase with in the system is supposed to be equal to the initial mass of each phase. But, on the other hand, I am getting a deviation in value of the mass of each phase initially present in the tank. So is the case with the total mass present in the system.


Can you please suggest me an explanation?


All times are GMT -4. The time now is 03:15.