Hello! I'm simulating a swirl pulverized coal flame, using discrete particle method.

Considering a first mesh (with 180k cells) the calculation time of discrete phase is reasonable (about 50% of the overall CPU time).

Considering a finest mesh (about 250k cells), refined near the inlet walls and in the flame core zone, the calculation time of the discrete phase increases enormously, and the overall simulation becomes too long.

Is it a normal thing that DP increases in this way? What are the DPM parameters that influence the DPM calculation time and the accuracy of solution?

Thank you for your attention!

Michele

Hi Michele

The few points I can give, as I do not fully understand the DPM, are: I think (high) step length factor and (low) step length are the main methods to control (increase) the number of iterations performed for the DPM (but the DPM never steps past a cell face in a step so small cells set a limit before these). If you have particles being trapped in the domain each DPM cycle will have to run to the full Max Number of Steps (for those particles) which would increase it. Otherwise unsteady tracking, turbulent dispersion models, continuous phase coupling would increase the computation. You may need these for accuracy however.

Not performed real combustion simulations myself so maybe a-whole-nother level.

Kris

Thanks for your answer, KJ!

I will try your advices, in particular the step length factor.

Michele