MFIX 25.3 simulation with Cyclic BC fails on HPC but runs on local computer

MFiX
1

Hi,

The simulation attached runs well on my laptop but on HPC (both HPC and my laptop using MFIX-25.3), it fails right at the start after some angry messages:

Beginning DEM settling period: 1 steps.
DEM settling period complete.
t= 0.000000 Wrote SPx: 1,2,3,4,A,B; .RES; DES.RES;

Velocity exceeds limit: 50000.
in cell: I = 2 J = 2 K = 1
Epg = 0.48603 Ug = 0.27423E+06 Vg = 22289. Wg = 0.0000
To change the limit, adjust the scale factor MAX_INLET_VEL_FAC.
Recovered: Dt= 0.90000E-03 :slight_smile:
Velocity exceeds limit: 50000.
in cell: I = 2 J = 2 K = 1
Epg = 0.48603 Ug = 0.14551E+06 Vg = 18478. Wg = 0.0000
To change the limit, adjust the scale factor MAX_INLET_VEL_FAC.
Recovered: Dt= 0.81000E-03 :slight_smile:
Velocity exceeds limit: 50000.

I tried to run and compile in serial and it still fails. I am out of options there.. Thanks for the help

bedload_water_only.mfx (17.5 KB)

particle_input.csv (272.4 KB)

2

Not sure, I cannot reproduce on our HPC. It seems to run fine (currently at time=0.20s). Can you try the batch solver and/or a different fortran compiler?

3

#Yodamode. I am already compiling with build_mfixsolver --batch -j --dmp. I ll try another compiler then.

4

I have used another cluster with very different compilers and can get the case to run in serial or with a small decomposition of the domain in DMP. Once I add all the subroutines I needed for the “actual” case, it does not run though (yet it does run perfectly on my laptop :frowning: )

5

Make sure you build in debug mode when you have custom code.