CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   Main CFD Forum (https://www.cfd-online.com/Forums/main/)
-   -   Converge and diverge (https://www.cfd-online.com/Forums/main/4457-converge-diverge.html)

 Jack February 23, 2002 07:36

Converge and diverge

Hello, everyone,

I am using low-reynolds k-e model to simulate the flows in ducts with periodic condition. The reynolds stress is model by EASM. However, I have difficult to make the program converge. The iterations always oscilate within a certain range. However, the results with the lowest converge critia are very close to the correct solution. Could anybady give me any idea how to improve it?

Thank you very much

Jack

 wang February 23, 2002 10:14

Re: Converge and diverge

As I understand, you use time-dependent solution. What kind of time discretization do you use? or Have you try to ever decrease the time step?

 Jack February 24, 2002 08:02

Re: Converge and diverge

it is a steady state calculation.

 wang February 24, 2002 09:21

Re: Converge and diverge

Hallo Jack,

my first question is, how can you simulate the periodic condition with steady solution?

If your computation is a steady one, and it doesn't converge, it may mean that 1). The solution is an unsteady one, hence you can not get steady solution. There is numerical diffusion that prevents the solution be not diverged. Try a time-dependent solution. 2). Grid quality. It may be that you have some coarse resoltuion at locations where there is strong gradient. Try to check your grid.

hope this help.

 john February 24, 2002 23:43

Re: Converge and diverge

I second to Wang's suggestion on checking your grid. Personally, I had used low Re model in StarCD, and sometimes it never get to the standard convergence criteria. Because I used very fine mesh, and maybe also bad mesh quality, I could pick up turbulence without using turbulence model, to some extent. So, never had a steady state solution.

 All times are GMT -4. The time now is 22:00.