CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CFX (
-   -   Convergence issue (

Omer March 4, 2007 15:07

Convergence issue
I have a model of a straight pipe, I am using to get a fully developed profile to use in my pipe junction.

I am using an SST model. Funnily, I seem to have no convergence even after about plenty of iterations. The behaviour is bouncy but it does not even try to go down even a tiny bit.

I tried playing with the timestep. But that didn't work.

But funnily enough, when I refine the mesh, I get convergence as low as E-12 in just 5 iterations. So what can my problem be?

I use CFX mesh to create the mesh. While trying to keep the Y+ less than 1.

Thanking in anticipation

Astrid March 4, 2007 18:41

Re: Convergence issue
Something is really very wrong when you get convergence to 1e-12 in 5 iterations. This cannot be a converged solution!

When you use SST and find a lot of flow separation at delicate locations you won't ever get convergence. Have you exported residual values to your res-file? What do these tell you?

Have you tried k-e with scalable wall functions? This should converge quite fast as it clips to y+=11.

What do you define on the in- and outlet in terms of turbulence?


Omer March 7, 2007 11:37

Re: Convergence issue
Well, my inlet velocity is 210 m/s. I did a little calculation for my branch of diameter- 16 inches. And found out that the trubulence intensity would be around 2.3%.

My outlet is a pressure out, at 0 atm. I will try to see how K-epsilon does.

Have I exported residual values to your res-file? I am not really sure what that is, and how it may help? Would you please help me understand?

Thanx for replying Astrid!

All times are GMT -4. The time now is 16:24.