CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   DIVERGENCE with MRF calculation in FLUENT (http://www.cfd-online.com/Forums/main/957-divergence-mrf-calculation-fluent.html)

FLUENT USER June 28, 1999 12:23

DIVERGENCE with MRF calculation in FLUENT
 
Hi everydody,

I'm trying to simulate a laminar viscous flow in a three staged stirred tank my mesh size is : 341640 cells. I was surprized when my calculation diverged , i tried different methods , but no results .

Any idea around ?

Thanks a lot


Sung-Eun Kim June 28, 1999 13:09

Re: DIVERGENCE with MRF calculation in FLUENT
 
Hi,

Can you please contact your support engineer ?

John C. Chien June 28, 1999 13:27

Re: DIVERGENCE with MRF calculation in FLUENT
 
(1). I am not surprised at all. The code I am using can not give me converged solution 50% of the time. If I submit two cases, then one will diverge or something like that. This is good, because it is telling me something. I would be unhappy if it gives me wrong converged solutions. (2). The only suggestion I have for you is " take one step and do one thing at a time". (3). Were you able to get any problem solved using this code? If the answer is "NO". Then, you need to learn how to use the code to solve a problem, any problem. (4). If the answer is "Yes". Then you are in good shape. The next step is to create a simple model of your problem, that is simplify the geometry first. Then put in a small number of mesh points. In this way, you can quickly run the code and observe the results. (5). It is a good idea to get the solution started by using small over-relaxation parameters. (6). For the numerical scheme, try to use the first-order method. (7). Also pay attention to the initial flow field guess. Try to set it to zero initial velocity distribution. (8). If you have something rotating, try to set the rpm to a small number. (9). Also set the condition such that the Reynolds number is small, Re=500, or 1000. (10). The idea behind is to provide the code a good working environment such that you can check out all of the input parameters. (11). Once it starts running, you can relax the constraint one-by-one. (12). Always remember that CFD is a high level guessing game. (13). Don't try to give the code a hard time. It is you who is guiding the code to the solution. Try to use simple geometry, relatively few mesh points, low Reynolds number conditions, with first-order algorithm, initially stationary flow field, a slow rotating speed, and a small (0.1) over-relaxation parameter to get your solution started. (14). Try to use vector plots and contour plots to check the flow field from time to time. This will help you understand the flow field problem. (15). You will learn by making mistakes and succeed by correcting errors. If you expect help, you need to tell us more about your problem set-up.

AndiMiller June 28, 1999 15:44

Re: DIVERGENCE with MRF calculation in FLUENT
 
hy,

is the flow in reality laminar ? if it is not, you calculate with a laminar solver something like a unscaled turbulent flow. try to calculate with turbulence and a very low k at the entrance.


All times are GMT -4. The time now is 18:44.