convergence problem
 

Hello,

I'm working on a gas flow throught a thermal exchanger with 3D simulations (400 000 cell). I performed about 10 000 iterations and the residuals are fluctuated around 1.e-03 (velocity and energy) and 5.e-04 (mass conservation). How many iterations more will be necessary for having a converged result ?

Thanks.

Philippe.

Re: convergence problem
 

There isn't an exact number of iteration to obtain a converged solution. However 10000 seems very high. As always, check all boundary conditions and try to underrelax momentum and energy equations.

Anyway, if all conditions of your problem are ok, maybe your initial solution is good, so residuals can't reduce of the quantity you require. You can understand this from residual plotting: if they have been stable for many iterations, maybe this is your case.

Hi

ap

Re: convergence problem
 

10000 iterations is a lot.

Could be a couple of reasons. First try lowering the relaxation rates, especially for the energy equation.

If this doesn't work then you may want to try switching to a transient solver (1st-order if you want a steady-state answer). You may have some small fluctuations in your flow, so that no steady-state solution exists. A transient solver also increases the diagonal dominance, and thus the robustness, of the discrete equations. This has often worked for me, even if arriving at a steady-state answer takes a long time.

Hope this helps,

Bryce