Learning Forum

[TingWangUNO]

Hello,

While using Fluent 2020R2, I've had a simulation crash consistently at 5320 timesteps. The error files generated are for an abnormal exit, and contain no other information. Has anyone heard of anything similar happening and/or had a similar issue? If so, fixes? 

The simulation is a multiphase transient setup, modeling the interface between air and water and the energy equation is turned on. The case is set to autosave every 2500 timesteps, and every 5 timesteps a tecplot file containing some data required for post processing is saved. 

Thank you for your help and your time.

[Rob]

Does the transcript show anything else? Ie, is it showing converging iteration output data and then Abnormal Exit, or is there something else?  If you run to 5300 timesteps, save, restart the machine, resume do you see a crash? 

[TingWangUNO]

Thank you for your response,

The transcript shows nothing else; the case does not diverge, and the only error presented is the abnormal exit. The case can be reintialized from the autosave at timestep 5000, where it will run from timestep 5000 to timestep 10320 before crashing again.  

If the case is reinitialized from the start after a system reboot, the problem will still persist.

[Rob]

OK, that could be a system issue - ie a queue timing out. Or it's linked to repetitive activities in Fluent not fully releasing RAM. How many images/reports/monitors are you creating? 

[TingWangUNO]

Only two monitors are set: both area-averaged values, for heat flux and wall shear. No images, no reports. Other than that, its just the solution data export and autosave.

[Rob]

That shouldn't cause any problems. Given the issue is reliably at 5320 I'd check with IT for queue/run durations. 

[sanchit2208]

Hello, I'm also suffering from the same issue. The crash happens exactly after 5310 timesteps for me. I'm also using 3 monitors, saving some data in plt and cgns format for postprocessing after 100 timesteps. Autosaving of dat file is after 400 timesteps. 

Did you find out the solution? It would be really grateful for your reply.

[Rob]

Can you run on from the last autosave? You may need to edit the monitors to remove overlapping points. 

[sanchit2208]

Yes, I'm able to run from last autosave 'dat' file.

Can you please elaborate more about editing the monitors? I'm using three monitors, two for monitoring the mass flow rate and pressure through exit (since it's an internal flow problem) and a third monitor to check the total mass unbalance in my domain.

I'm also creating 3 files: 2 plt files for saving some flow variables in 2 different planes (since it's a 3D problem); and 1 cgns file to save some flow variables in the whole 3D domain.

[Rob]

When you restart and assuming you don't overwrite the existing monitor files you'll have a section that's repeated, ie from step 5300 to 5310 as the last autosave is likely at 5300. The output is text so it's just a case of deleting a few lines. That's only needed if you're going to do something else with the data. 

[sanchit2208]

I copy the data from the existing files into some other files at different location. Then I delete these monitor files. Every time when I restart, new monitoring files are created.

[Rob]

And at how many iterations/steps does the restart fail at?

[sanchit2208]

Again after 5310 time steps..

[Rob]

Have you checked with IT for queue/job limits? 

[sanchit2208]

I have not checked with IT as I was'nt having this problem uptil now. For the same geometry, mesh but with different boundary condition I've never run into this issue. 

I'm using a transient pressure profile as a boundary condition. For relatively lower back pressures, I didn't run into this issue. But lately, with a higher back pressure this issue persists. The simulation runs for roughly 26hrs and then crash.

[Rob]

Odd, is there anything different with the profile? 

[sanchit2208]

No

[Rob]

Can you monitor RAM usage over the run? 

[sanchit2208]

Yes..RAM usage is in range of 25-30%

[Rob]

Even as it crashes? Just wondering if we're leaking memory. What operating system are you using?