I am simulating particle transport in a rough fracture, and the rough fracture geometry and simulation screenshots are attached.
I simulated for both Newtonian fluid (fluid viscosity = 1cP, 10cP, 100cP) and non-Newtonian fluid (shear-thinning fluid).
In a lot of cases, I did not have the particle disappearing issue. But I am having the particle disappearing issue in some cases: (1) Newtonian fluid viscosity = 100cP and (2) Non-Newtonian fluid.
The total particle number in the simulation domain changing with time plot is attached.
So in this example case, the particle disappearing issue happened after ***Time=73s***, and please see the attached screenshot for more info.
Thanks!
| Attachment | Size |
|---|---|
 fig_fracturedsandstone.jpg | 12.09 KB |
| particle_transport.png | 199.86 KB |
| particle_transport_2.png | 2.01 MB |
| total_particle_number_changing_with_time_plot.png | 117.11 KB |
| particle_disappearing_issue_happened_after_time73s.png | 1.25 MB |
Min Zhang | Fri, 08/28/2020 - 16:47
The job crashed when I am using m m m boundary
Since I saw this post (https://www.cfdem.com/forums/missing-particles-variable-inlet-velocity), so I tried m m m boundary.
When I was using f f f boundary, the particle disappearing issue won't happen until physical time = 73.8s.
But when I used m m m boundary, my case crashed at Time = 0.196825s without any error message.
After I looked into the log file carefully, I found that one big difference between these two boundary types is memory usage.
m m m boundary: Memory usage per processor = 1214.49 Mbytes (but actually it has very few particles: Loop time of 0.442151 on 48 procs for 100 steps with 86 atoms)
f f f boundary: Memory usage per processor = 9.54139 Mbytes (Loop time of 0.0276629 on 48 procs for 100 steps with 1008 atoms)
This is the detailed log file of case m m m boundary: I am very curious about why it did not give a specific error message since I am having the same kind of crash for a totally different case (please see this post: https://www.cfdem.com/forums/what-options-i-have-when-i-have-memory-issu...). Your valuable comments would be greatly appreciated!!
Time = 0.196825
Courant Number mean: 0.000188309 max: 0.00195368
Coupling...
Starting up LIGGGHTS
Executing command: 'run 100 '
run 100
Setting up run ...
Memory usage per processor = 1214.49 Mbytes
Step Atoms KinEng RotE ts[1] ts[2]
787201 86 20.080622 0.51858829 0.10050491 0.04215273
CFD Coupling established at step 787300
787301 86 20.080586 0.51858829 0.10050491 0.04215273
Loop time of 0.442151 on 48 procs for 100 steps with 86 atoms
Pair time (%) = 0.00136133 (0.307887)
Neigh time (%) = 0 (0)
Comm time (%) = 0.0143952 (3.25572)
Outpt time (%) = 0.0167195 (3.7814)
Other time (%) = 0.409675 (92.655)
Nlocal: 1.79167 ave 57 max 0 min
Histogram: 46 0 0 0 1 0 0 0 0 1
Nghost: 0.104167 ave 4 max 0 min
Histogram: 46 0 1 0 0 0 0 0 0 1
Neighs: 1.14583 ave 39 max 0 min
Histogram: 46 0 0 0 1 0 0 0 0 1
Total # of neighbors = 55
Ave neighs/atom = 0.639535
Neighbor list builds = 0
Dangerous builds = 0
LIGGGHTS finished
timeStepFraction() = 1
srun: error: nid00872: task 2: Killed
srun: Terminating job step 3022098.0
slurmstepd: error: *** STEP 3022098.0 ON nid00872 CANCELLED AT 2020-08-28T07:35:03 ***
srun: error: nid00872: tasks 1,3-23: Killed
srun: error: nid01310: tasks 24-36,38-44,46-47: Terminated
srun: error: nid00872: task 0: Killed
srun: error: nid01310: tasks 37,45: Terminated
srun: Force Terminated job step 3022098.0
TACC: MPI job exited with code: 143
TACC: Shutdown complete. Exiting.
Min Zhang | Sat, 09/05/2020 - 01:25
New updates on this issue
1. I used f f m BC, and I still have the particle disappearing issue. Please find the comparison of f f f BC (the initial run) and f f m BC from the attachment.
2. I restarted the initial run from Time=73s, and then I haven't had the particle disappearing issue until now, and the current Time is 84.6s.
You valuable comments would be very much appreciated!