Why the time step could not be reduced?

Min Zhang's picture
Submitted by Min Zhang on Mon, 03/23/2020 - 18:26

When I use deltaT_CFD=2.5e-5s and deltaT_DEM=2.5e-7s, the case could run.
Part of the output info:

#cfd coupling
fix cfd all couple/cfd couple_every 100 mpi
nevery as specified in LIGGGHTS is overriden by calling external program (../cfd_datacoupling_mpi.cpp:77)
fix cfd2 all couple/cfd/force
restart ${restartstep} ../DEM/restart/liggghts_coupled.*
restart 800000 ../DEM/restart/liggghts_coupled.*

run 1
Setting up run ...
Import and parallelization of mesh ins_face containing 632 triangle(s) successful
Import and parallelization of mesh walls_0 containing 87256 triangle(s) successful
Import and parallelization of mesh inlet_ containing 632 triangle(s) successful
Import and parallelization of mesh outlet_ containing 320 triangle(s) successful
INFO: Particle insertion ins: 85.763582 particles every 4000000 steps - particle rate 85.763582 (mass rate 0.000402)
21440 particles (mass 0.100402) within 996000000 steps
Memory usage per processor = 9.42321 Mbytes
Step Atoms KinEng RotE ts[1] ts[2]
0 0 -0 0 0 0
INFO: Particle insertion ins: inserted 86 particle templates (mass 0.000403) at step 1
- a total of 86 particle templates (mass 0.000403) inserted so far.
1 86 2.0136627e-06 0 0 0
Loop time of 0.0263526 on 48 procs for 1 steps with 86 atoms
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

When I use deltaT_CFD=1e-5s and deltaT_DEM=1e-7s, the case could not run.
Part of the output info:
run 1
Setting up run ...
Import and parallelization of mesh ins_face containing 632 triangle(s) successful
Import and parallelization of mesh walls_0 containing 87256 triangle(s) successful
Import and parallelization of mesh inlet_ containing 632 triangle(s) successful
srun: error: nid00012: tasks 0-16,18-23: Floating point exception
srun: Terminating job step 2766343.0
slurmstepd: error: *** STEP 2766343.0 ON nid00012 CANCELLED AT 2020-03-23T11:53:44 ***
srun: error: nid00013: tasks 24-47: Floating point exception
srun: error: nid00012: task 17: Floating point exception
TACC: MPI job exited with code: 136

TACC: Shutdown complete. Exiting.

Min Zhang's picture

Min Zhang | Tue, 04/14/2020 - 00:45

For the same case, when I run LIGGGHTS only and deltaT_DEM=1e-7s, I got the (specific) error message shown below:
ERROR: Fix insert/stream (id ins): Particle insertion: Overflow - need too long for particle insertion. Please decrease # particles to insert or increase insertion rate (../fix_insert.cpp:380)

Hello Mohan,
Thanks for your reply. I am using timestep check:
fix ts all check/timestep/gran ${ckstep} 0.1 0.1