CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   what should I do if the solution didn't converge? (http://www.cfd-online.com/Forums/fluent/33422-what-should-i-do-if-solution-didnt-converge.html)

 Beeny April 7, 2004 09:04

what should I do if the solution didn't converge?

hi all!

If the solution (mass) didn't converge, what in general should I do to check it and amend it?

Benny

 John April 7, 2004 10:04

Re: what should I do if the solution didn't conver

try changing the termination restriction in Solve..>Controls..>Multigrid from it's default value 0.1 I Think to 0.001.

bye

 ap April 7, 2004 17:57

Re: what should I do if the solution didn't conver

First of all, try to reduce under-relaxation factors a bit. If this is not enough, try to obtain a solution using the first order discretization scheme, then switch to higher order.

Also a good quality grid and proper boundary conditions are required to get good convercence rate.

Hi :)

ap

 rob sharma July 27, 2011 09:36

Guys i have a big problem like my solution is not converging in simple case. I have a pipe 800 mm lenth and 15 mm dia. I am using axis-symmetric pipe. I simulate that pipe with inlet velocity 0.486 by using 800 nodes in length and 8 in readiusof pipe for laminar flow. It is simulated at around 246 iterations. But after after increasing nodes in length that is of 1200 and 10 in radius of pipe, it does not converger. If i give initail velocity less like 0.470, then it converges. But the result does not come appropriate. all my values come under the convergence criterion of 0.001. but the continuity does not come. i dont know what to do. If anybody tell me i will bw very thankful...kindly...

 teguhtf August 12, 2011 23:31

Hi Rob,
I think the convergence of simulation based on:
1. Complexity of geometry
2. Mesh size, quality.
3. the precision of the simulation (single or double)
4. Turbulence model choice
5. Discritization method (first order,second order and so on)
6. Our physical input
7. setting under relaxation factor (the setting of under relaxation factor is depend on gauss seidel method)
I think you can analyze your case based on those things.
Hope it helps
teguh

 rob sharma August 13, 2011 06:09

convergence problem

Its a laminar flow with 1200 nodes in length and 10 in radius. Anyone could suggest me the book or reason why does the solution not converge. I need to know the reason.....kindly please help me.......:)

 teguhtf August 27, 2011 00:04

He there,
For internal flow, i think you should choose turbulent mode. because, i think there will be little turbulence generated near the wall. If you're sure that there is no turbulence, you can choose laminar mode.
Teguh

 All times are GMT -4. The time now is 11:26.