CFD Online Discussion Forums

CFD Online Discussion Forums (
-   FLUENT (
-   -   number of iterations for convergence (

CHUBBY April 14, 2001 11:08

number of iterations for convergence
hi everyone, I am basically not from mechanical background .So my queries may look to be much sinple for everyone. I am assigned with the study of air distribution(velocity and pressures) on a porous media. The air penetrates into the porous media with the help of a suction fan. I find many difficulties in post processing of this model. Firstly , I had created the model (3 D). I meshed the model with hexa/wedge and hexa grid structures.I had a serious problem in creation of porous_jump. Whenever I used to assign the porous media as porous_jump, all goes well till the mesh is exported. But when I read it in FLUENT/UNS I get the warning that the porous jump is converted to wall, and I am unable to assign the permeability, or the resustance to the porous media.

treating with the porous media as wall. I tried to converge the solution, I had given about 1500 iteration but the solution doesnt converge. Do the iterations depend on the number of nodes?

I had given the inlet velocity as 30m/s. I tried to converge the solution. But after 1500 iteration as well. I checked in the plot of velocity vs distance that the velocity is varying from 0 to 600 at the inlet.This might be due to the fact that the solution has not yet been converged. I have about 1,00000 nodes in my mesh.

Could any one please try to help me in suggesting solution for these.


Scott W April 17, 2001 09:51

Re: number of iterations for convergence
I had that problem two years ago when the porous jump converted into walls, all I had to do was to change it back to a porous jump within Fluent. I complained to Fluent, and they sent me a new Gambit version that didn't have that flaw (upgraded from Gambit 1.1 to 1.3).

If you treat the porous media as a wall, I don't think you will get the flow to pass through the wall properly...

In general, the number of iterations depends on the number of cells, ESPECIALLY when you model a porous jump. This is since data can only flow from one cell to an adjacient cell each iteration. So if your domain is 1000 cells long, it will take 1000 iterations for the information to reach the end. Fluent overcomes this partially with its multiple grids (higher level grids with fewer cells), but this technique seems to work poorly in Fluent with the porous jump. I have no idea why, maybe someone else here is more experienced than me in this area.

CHUBBY April 19, 2001 14:24

Re: number of iterations for convergence
Hi, thanx for the response. I had tried very hard in order to create the porousjump. I could even suceed in it. But when I was checking the grid . It had given me as grid check failure. There were no errors on the screen while reading the case file , but appeared on grid check.

Could u please through some light on it. Thank u Chubby

Scott W April 20, 2001 11:06

Re: number of iterations for convergence
If you have grid check errors, I think your best bet is to ask your Fluent support person. Common problems I've seen: negative volumes or negative areas (due to manually adjusting a node position or using the boundary layer mesh tool), grid points on both sides of the axis for axisymmetric problems, using the wrong axis for axisymmetric problems (must be the x-axis), and not having the parts of your geometry connected (most common error I've seen, use the face/edge/vertex connect commands). Without any knowledge of your grid or geometry I would hazzard a guess that your model may not be connected (ie the zones which are split by the porous jump are considered two separate regions instead of one complex region). If you don't have one of these simple problems, I cannot help. Good luck.

chubby April 22, 2001 05:27

Re: number of iterations for convergence

Thanx for answering all of my queries. I was successfull in creating my mesh with porous jump . I read it in fluent, there were no errors, and the porous jump was not converted to wall... so thats very good news indeed... It took 650 iterations for me to converge , of course I had reduced the underrelaxation by half. the results were also matching.. thanx for ur help. I am sure u woulod help me in future as well.. thanx for ur help again bye bye CHUBBY

All times are GMT -4. The time now is 04:55.