CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   Need how many iteration times (http://www.cfd-online.com/Forums/fluent/28355-need-how-many-iteration-times.html)

Harry Qiu March 26, 2001 07:15

Need how many iteration times
 
Hi,Mr or Mrs:

I'm modeling laminar flame(methane/air).I considered laminar flow, radiation(fluid and solid), chemical reaction.Probably, because the problem is comlex(too many sub-model),no convergence is obtained even if the computer iterates for more than 10,000 times(I note that several handred of iteration time is needed in many FLUENT tuterial guides).Can you tell me that so many iteration times is correct or not?

Thank you very much! Sincerely yours, Harry.

John Patrick March 26, 2001 18:45

Re: Need how many iteration times
 
I'm modeling flow around a automobile model and on certain boxy designs, convergence establishes at about 1000 iterations, yet on my newest model, a curved design, the convergence state hasn't been achieved yet even at ~200,000 iterations, and residuals keep lowering. So I too am interested in this.

How are iterations measured? Can one apply a time or length to them?

~JP

John C. Chien March 26, 2001 18:59

Re: Need how many iteration times
 
(1). It is one of the most frequently asked question. (2). I don't know whether my experience is applicable to you or not. But I used to read the cfd results into a third party post-processor, where I can put one contour plot on the top of another contour plot at different time (or iteration). (with identical contour levels set for both cases) (3). In this way, you can easily identify whether the steady state has reached or not. At the steady state, you can see only one contour plot, because they are then identical. If not, you can see the areas where they are different. This method should be very accurate. (naturally, it is useless for oscillating flows) I have used the technique for ten years in my own code. (4). Another cheap method is to monitor the flow variable at a fixed point. But then it is hard to see the whole picture.

manaf hasan April 3, 2001 15:28

Re: Need how many iteration times
 
i am working in fluent about a year and i had this problem of convergense before if the residuals smothly going towards the converganse thats good you can go(but first save your work)incrae under relaxation for all the equations all the reidual will jumb wait for 50-60 iterations to see how this change work this change should slop the residual dowan very fast if that doesnt help moniter your mass inlet and outlet if the risdual is large you have truble in the your mesh. good luck



All times are GMT -4. The time now is 07:08.