Erroneous switch behaviour

7 posts / 0 new
Last post Nucl...
Erroneous switch behaviour

Dear devs and/or users,

I would like to report a case of a mysterious output of a switch signal. I have recently recalculated a LOCA event in a PWR using some new neutron data in an older input file and while analysing the results I've seen none of the reactor scram signals have activated, even thoug they were expected. Then I discovered that a certain switch did not flip in an expected time. Moreover, I saw it produced a range of values, not just 0 and 1. 

Image icons-log-err.png(Ss-start and TR-start were = -150 and 0)

What could explain this? It was not noticed before and I am unable to track the specific change in input file which caused it. Some other switches in this output file are also affected and also NOT signals do not produce expected reversion of their inputs. 

Thank you for any suggestions 


Dear Colleague,

Thank you very much for this comment! The behavior of the switch is unclear. Unfortunately, from the information provided, any further investigation is very difficult, basically not possible. A similar behavior wasn't observed before. Is there any possibility to provide GRS the respecitve input data set for further analysis? Which ATHLET code version was used here?

Best wishes,

Philipp Schöffel


Dear Colleague,

please also check that the key and pd-files are from the same calculation and fit together. To ensure that the post-processing tool isn't responsible for the unexpected results, you could also consult the main edits in the out-file for cross-checking.

Thank you for your help.

Philipp Schöffel Nucl...

Thank you for the reply - is the picture not visible ? It should be attached to the original text. 

The version was 3.2 as was included in the 2019.1 package. The signal definition is:

S---- Logical one ()
0 1. %TR-START% %SS-START% 0. 0.
0. 0. 0


Yes, the figure is attached. But I am afraid that without any further information from the input file it will be impossible for us to reproduce this behavior. BTW, could you already check if the main edit (I hope your print output includes some) shows the same behavior? Thanks! Nucl...

Yes I have found the signal in a main edit at 89 seconds. 

 Indeed it looks the value is still there. 

I will send you an email with further details so we can try to identify the issue. 

Thank you! 


Dear Colleagues,

The cause for the reported mysterious behavior of a switch could be clarified. A bug in the code results in an erroneous calculation. The bug affects controllers of the types INTE; SWITCH and TSSWITCH, which are assigned to a GCSM block using INTEK=2 (predictor-corrector-integration). The only possible workarround is to move controllers of aforementioned types into GCSM blocks with INTEK=1.

The code versions ATHLET 3.2.x and ATHLET 3.3.0 are affected. A patch for ATHLET 3.3 will be prepared.

Thanks a lot for reporting the error!

Philipp Schöffel