Errors in coupled transport

Questions and discussion about applications, input, output and general user topics
Post Reply
jakin
Posts: 21
Joined: Thu Apr 16, 2020 1:31 am
Security question 1: No
Security question 2: 7

Errors in coupled transport

Post by jakin » Tue May 25, 2021 2:52 am

Hello all,

I have been trying to execute coupled transport runs in serpent. The runs work perfectly in single core and MPI mode when neutrons are whats being transported. However, in coupled transport mode I get the following error with set opti 2 (using this selection due to memory allocation failure with 3 and 4) on a single core and mpi mode:
Screen Shot 2021-05-24 at 9.58.38 AM.png
Screen Shot 2021-05-24 at 9.58.38 AM.png (16.69 KiB) Viewed 205 times



These are the settings used that produced the above error:
Screen Shot 2021-05-24 at 10.02.16 AM.png
Screen Shot 2021-05-24 at 10.02.16 AM.png (37.26 KiB) Viewed 205 times



Thinking I can avoid this by not using set opti 2, I switched to set opti 1 which runs well on a single core but breaks the simulation without much information as to how it failed with MPI enabled:
Screen Shot 2021-05-24 at 3.45.47 PM.png
Screen Shot 2021-05-24 at 3.45.47 PM.png (75.13 KiB) Viewed 205 times



These are the settings used for that run:
Screen Shot 2021-05-24 at 3.47.22 PM.png
Screen Shot 2021-05-24 at 3.47.22 PM.png (37.68 KiB) Viewed 205 times


I would appreciate help in tracking down what could be causing these errors.

Thanks,
Jerry

User avatar
Jaakko Leppänen
Site Admin
Posts: 2442
Joined: Thu Mar 18, 2010 10:43 pm
Security question 2: 0
Location: Espoo, Finland
Contact:

Re: Errors in coupled transport

Post by Jaakko Leppänen » Tue May 25, 2021 9:51 am

The second error (opti 1 in MPI mode) looks like one of the parallel tasks terminated in an error or crashed. In such case you may not get a proper error message. This probably happened at the end of inactive cycles, when the photon production is switched on.

Is this using the latest Serpent update (2.1.32) or some older version?
- Jaakko

jakin
Posts: 21
Joined: Thu Apr 16, 2020 1:31 am
Security question 1: No
Security question 2: 7

Re: Errors in coupled transport

Post by jakin » Tue May 25, 2021 8:13 pm

The version of serpent I have running is 2.1.31. So maybe updating to the latest version will fix this issue. Ill try that and post again if it fixes the issue.

For the error produced with opti 2, is it known what could be causing that?

Thanks,
Jerry

Ana Jambrina
Posts: 551
Joined: Tue May 26, 2020 5:32 pm
Security question 1: No
Security question 2: 7

Re: Errors in coupled transport

Post by Ana Jambrina » Tue May 25, 2021 9:52 pm

Optimization mode 2 calculates majorants in a different way than other modes. The error comes from a check in 'calculatemgxs.c' rotutine (where Serpent calculates the coarse multi-group majorant cross sections for the nuclides) that evaluates if the nuclide uses or not the unionized grid: there is a mismatch between the (ACE) energy grid from for a given (nuclide-wise) reaction and the generated unionized grid.
- Ana

jakin
Posts: 21
Joined: Thu Apr 16, 2020 1:31 am
Security question 1: No
Security question 2: 7

Re: Errors in coupled transport

Post by jakin » Fri May 28, 2021 8:28 pm

Thank you Jaakko and Ana. By updating to the most recent version of serpent, it returned an error of insufficient gbuf. I was able to bump that up and get the simulation running.

Jerry

Post Reply