[drdet]

There are a few rules of thumb that I have been taught about convergence criteria for CFD simulations. One is that the residuals should be less than 1% or 0.1%. In Phoenics, we used to enter 0.001 as the convergence criteria, and I assume that the solver was determining what level that was for each solution variable (i.e. the 0.001 entered was a relative value or a percentage of the actual solution variables). In Flow Simulation, it seems that we are entering an absolute value, though. For example, if the Current Value for GG Av Velocity (X) shown in the solver monitor window is 14.2 ft/s, and I want the solution to converge to a level of 1%, I would need to do the math and enter 0.142 ft/s as the convergence criterion in Flow Simulation, but in Phoenics I would enter 0.01. Am I correct? Also, does anyone know how Flow Simulation chooses the automatic criteria?

[Boris_M]

Hi Dan,

There is some automation behind the convergence criteria that is chose by FloEFD. That depends on the model and the project with the mesh settings. Of course it unlikely to get a 1% convergence if the mesh is too coarse so it choses a higher value.
In the latest version which I mentioned in your other post about the time step it is also possible to enter the convergence criteria manually as a percentage value of the absolute value. So no need to do the math on your side anymore :-)

If you don't have that in your version then it seems you have not yet update to the latest version.

Boris

[drdet]

Hi Boris,

Thanks for all the information! I have version 2015 sitting on my desk - I'll install it and give it a try. It sounds like the enhancements are worthwhile.

Thanks,

Dan