[frossi]

Hi,

I am running a steady state simulation of a wing. My problem is that my residuals, lift, drag, and mass flow rate converge perfectly until 1000 iterations, and stay like that until 2000 iterations. At this point they start diverging more and more.

Also note that I am doing a parametric study. I keep the wing geometry and mesh the same, while only changing the velocity. My results for lower velocities (from Mach 0.2 to Mach 0.3) are right and stay converged. While when I run Mach 0.35 (the same case as the pictures) you can see how the results diverge. How come this problem arises only at higher velocities?

I would really appreciate help, I have been stuck with this problem for long.



Best regards to all.

[ghorrocks]

This can happen from many sources:
* Some difficult flow condition (a vortex, or a rapid change in fluid conditions like a shock wave) has finally reached the exit boundary and the exit boundary is not stable enough to handle it
* round off errors have built up and trigger a divergence

CFX rarely needs 1000 or 2000 iterations. 100 or 200 is more like it for most cases. This suggests your simulation is either not very numerically stable, or you should increase the physical time step and converge quicker.

All of these issues come back to numerical stability. To improve numerical stability:
* Better quality mesh (this cannot be emphasized enough)
* double precision solver
* better initial condition