Hi everyone,
I'm simulating a simple solid-liquid fluidized bed case using cfdemSolverPiso. The coupling interval is specified as 1 which results in a CFD time step 1e-5 and DEM time step 1e-5. Everything went well at the begin until the simulation aborted at somewhere between 35s and 36s and gave fatal error as:
"cfdemCloud::dataExchangeM().timeStepFraction()>1:Do not do this, since dangerous. This might be due to the fact that you used an adjustable CFD time step. Please use a fixed CFD time step."
Actually, the runTimeModifiable was set as no. So the CFD time step couldn't be the issue. I don't know whether it's a bug or something wrong with my settings. What I guess is the precision of the time value in the simulation. The attached are my main case files and error. Could you kindly have a look at this problem?
Kind regards,
Linhan
| Attachment | Size |
|---|---|
 couplingproperties.txt | 4.69 KB |
| controldict.txt | 2.48 KB |
| in.liggghts_run.txt | 2.18 KB |
 error.jpg | 108.13 KB |
alice | Mon, 07/17/2017 - 11:01
Hello Linhan,
Hello Linhan,
I'd suggest to update to the latest version of LIGGGHTS and CFDEMcoupling (3.7.0) since the precision of this check was changed in the last release.
Cheers,
Alice
gelinhan | Thu, 07/27/2017 - 07:48
sill have the problem
Hi Alice,
I found the check of the time step is stricter in the latest version. I still got the same error message.
Regards,
Linhan
j-kerbl | Fri, 07/28/2017 - 10:52
Hi Linhan,
Hi Linhan,
do you still get the same error message with the latest version? I understand it is very odd if this happens after 30s of simulation time.
Just above the error message should be some output similar to "timeStepFraction() =". What does this say in your case? I am trying to reproduce this behaviour on my side.
What you can do for now is to change the "FatalError" to "Warning" in https://github.com/CFDEMproject/CFDEMcoupling-PUBLIC/blob/master/src/lagrangian/cfdemParticle/cfdemCloud/cfdemCloud.C#L781 on your side and recompile the code. Then it should run through.
Cheers,
Josef
gelinhan | Fri, 07/28/2017 - 12:21
above the error message
Hi Josef,
Thanks so much for your reply. The above message is like:
timeStepFraction() = 1.0000001
And every time it stopped exactly at 34.1461s. FYI, I'm running this case on a cluster and using OpenMPI version 1.10.7. If I restart the case before the breakpoint, it works fine.
Looking forward to your solution.
Cheers,
Linhan
gelinhan | Thu, 08/03/2017 - 11:06
reproduce the bug
Hi Josef,
I found you might be able to reproduce the bug by setting CFD timestep = 0.0001, DEM timestep = 0.00001,coupling interval = 10. Besides, after I changed the "FatalError" to "Warning", the simulation could run without problems, but "TotalForceImp: (0 0 0)" appeared.
Cheers,
Linhan
Rachel | Mon, 01/15/2018 - 18:58
About the same problem
Hi Linhan,
I met the same problem in my simulation, and after running about 16s, my CFDEM coupling code stopped with the error message,'timeStepFraction()>1: Do not do this, since dangerous. This might be due to the fact that you used a adjustable CFD time step. Please use a fixed CFD time step'. The same as yours, I didn't use any adjustable CFDEM time step either. Have your problem solved? Could you please give my any clues about how to solve this problem? Many thanks.
Rachel
tala | Wed, 01/17/2018 - 16:58
the same error
Dear all
I have the same error, I have used timeStep for writeControl, but the problem didn't fixed.
Any help is appreciated,
Talat
paul | Wed, 01/17/2018 - 22:49
Re-Read the thread, j-kerbl
Re-Read the thread, j-kerbl provided a temporary fix by turning the error into a mere warning.
Alternatively, I recommend reverting this:
https://github.com/CFDEMproject/CFDEMcoupling-PUBLIC/blob/b23726463721d5...
back to this:
https://github.com/CFDEMproject/CFDEMcoupling-PUBLIC/blob/641f6c694fd16b...
I believe this is exactly the kind of issue that warrants a dot release ;)
j-kerbl | Mon, 01/29/2018 - 11:38
Hi everyone,
Hi everyone,
as I see you still do experience the problems with the latest 3.8 release of CFDEM, correct?
Just to confirm, you can still reproduce it with the above posted case, or was it a new one?
Thanks,
Josef
gelinhan | Thu, 03/22/2018 - 11:49
no problem
Hi all,
In my case, the problem seems to be fixed in the latest 3.8.0 release.
Cheers,
Linhan
mardussa | Tue, 08/06/2019 - 02:35
Adding my experience
Hi all,
I would just like to add my experience with this same error.
I received the same fatal error due to the timeStepFraction > 1.0000001 (in cfdemCloud.C). What I found strange was that in terminal output, the line before the error was thrown gave the timeStepFraction() = 1. Now I don't know if it was just truncating the true value, perhaps it was something like 1.0000002 but only shows 7 significant figures in the terminal output.
Anyway, I did as suggested in one of the previous comments and removed the fatal error and replaced it with a warning only. My CFDEM simulations no longer crash at random times before my set endTime.
What I also just noticed now in my currently running simulation is that my timeStepFraction() = 0.999995, leading me to believe that my above suspicion concerning not enough significant figures present in the output to be the case. What I am not sure about is why my timeStepFraction is actually getting smaller than 1, where as previously I had an error because the TSF was larger than 1.
Just thought I would share my experience with this.
Regards,
Robert