[PaulS91]

Hi All,

i guess as most of the new users in this Forum i have got a Problem, which as of right now i cannot solve.

My transient calculation has already run for over 16000 Timesteps (first URANS, then SAS). With a Timestep of 1.25e-05.

My SAS Simulation runs for 24h each run on a University Cluster. Round about my 5th run (so after 5x24 hours) i started getting the following error(example):

+--------------------------------------------------------------------+
| Convergence History
+--------------------------------------------------------------------+

+--------------------------------------------------------------------+
| Writing transient file 15118.trn
| Type : Selected Variables
| Option : Timestep Interval
+--------------------------------------------------------------------+

+--------------------------------------------------------------------+
| ERROR #001100279 has occurred in subroutine ErrAction.
| Message:
| c_fpx_handler: Floating point exception: Invalid Operand
| |
| |
| |
| |
| |
+--------------------------------------------------------------------+

+--------------------------------------------------------------------+
| ERROR #001100279 has occurred in subroutine ErrAction.
| Message:
| Stopped in routine FPX: c_fpx_handler
| |
| |
| |
| |
| |
+--------------------------------------------------------------------+


The strange thing with it is/was, that usually i could fix the error just by deleting and restarting the job. Which went OK until this week. Restarting wont fix the Error anymore and i keep running into it.

The Error occurs right in the beginning without any Timesteps.

I have already tried to further reduce the timestep but this doesnt help.
When i searched for the Error in this Forum i only found explanations like "bad mesh, bad setup" but what bothers me, is that the simulation already ran for quite a long time and did not show any signs of diverging?


Setup:
Advection Scheme: Central Difference , Bounded CBS
Transient Scheme: Second Order Backward Euler
Turbulence Numerics High Resolution

If any further information is needed, iam glad to provide.

Thanks alot in advance!

Paul

[ghorrocks]

You are correct, this error does not seem to be bad mesh or bad setup But most problems on the forum are caused by that.

The error comes straight after writing a results file. Is the written results file OK? Can you post more of your output file so we can see the surrounding actions?

The error usually means something has gone divide by zero or diverged to infinity. So have a close look at the results file and output file to try to find what has gone bezerk.

[PaulS91]

Hi Glen thanks for your reply.

It doesnt actually write the TRN file. It kinda crashes while trying to.

Further investigation has led me to the conclusion, that this is a communication Problem between ansys/linux and the Uni cluster.
As of right now, all of my jobs crash with this error, even the simplest RANS with very simple Meshing... so i guess it might not be a CFX Problem after all.

I will keep the Forum updated about it.

Paul

[chiragsvnit]

Hi Anon,
I experience the same problem on the cluster. Have you found the cause?
That will help me to fix at my university cluster.
Thank you.

[tsibel]

Originally Posted by PaulS91

Hi Glen thanks for your reply.

It doesnt actually write the TRN file. It kinda crashes while trying to.

Further investigation has led me to the conclusion, that this is a communication Problem between ansys/linux and the Uni cluster.
As of right now, all of my jobs crash with this error, even the simplest RANS with very simple Meshing... so i guess it might not be a CFX Problem after all.

I will keep the Forum updated about it.

Paul

Hi Paul


I am running two phase flow (boiling) with the clusters. After some iteration (around 1200) I have the same error (invalid operand). I have tried everything to solve it. But I am suspicious that it is related with clusters. Have you solved this problem?
Many thanks