Learning Forum

[EL_G]

Hello;

I am running a multiphase model with Eulerian method. After running the model for around 20 seconds, I need to add Mass Species Transfer feature without initializing. I'm wondering if it is possible. If yes, please let me know how.

Thanks in advance for your advice.

[Rob]

Stop the run (at end of time step). Save case & data. Make the changes. Save case & data (add -a or something to the name to avoid overwriting the original). Iterate. Watch carefully as you may need a smaller time step or the like to handle the first "shock" of the change.

[EL_G]

Thanks a lot, Rob for your hints.
Regarding the smaller time step, may I ask like what? My current Time Step Size, before adding the Species Mass Transfer, was 0.0019 [s]. With the Species ... on, I decreased it down to 0.00001 [s] but the model diverged. The simulation passes the shock but it diverges in a couple of Number of Time Steps.
Any advice is highly appreciated.

[Rob]

If you reduce the cell size the courant number will increase so you need to drop the time step to revert to the original courant number.

[EL_G]

Thanks, Rob for your hints about decreasing cell size along with Time Step Size [s] dropped. However, my computer does not support it. My model is a 2D geometry (40 [m] x 8 [m]). With 0.11 [m] global mesh size, the simulation takes around 17 [hours] for a 14 [s] model. If I don't change the mesh size and only decrease the Time Step Size to 1e-5 [s] (1,400,000 Time Steps), the computer needs around 37.5 days (9 [hours] for every single 1% progress).
May I have your hints on how to address this issue?
Thanks in advance.

[Rob]

Have a look at the domain to see if you can tweak anything. If the solution starts to settle down you may be able to increase the time step again. Otherwise, you may want to look at using more computers.