I am trying to finish TFM simulation.
However, time step (dt) diminishes adaptively time by time due to (outer iteration) convergence for unsteady simulation.
I understand that the criteria of reduction of dt is the below, but I check the NIT and flow velocity, but I guessed that they are not the reason why dt diminishes.
(1) NIT becomes over max_iteration
(2) CFL condition is unsatisfied
If you know there are criteria other than the above two,
I would appreciate it if you could tell me.
Hi Mitusuru. If you attach your simulation, we can try to help. Under the main menu, “Submit bug report” will create a ZIP file with all needed input files, you can post that here and someone will take a look.
Unfortunately, I cannot share my input file or log data because of confidential information.
Now, I found that the cause of my problem is due to diverge/stalled check, and I could improve it by increasing linear solver’s iteration number and so on.
I would like to do my best for convergence.
If you have some documents or technical papers about general numerical stability (particularly Numerics in GUI), I deeply appreciate it.
The “Theory guide” and “Numerics guide” listed under “Legacy manuals” at MFiX Documentation - NETL Multiphase Flow Science are a bit old, but still contain relevant information. Other than that, I find the best way to figure out what is going on with MFiX is often to look at the code.