Flgvli.f Fortran runtime error

Running and Runtime Errors
1

Dear experts,

When I am running the below simulation I get an Error code 2 in Flair:
“Started 2021.07.28 21:58:33
- Running: unimelb_bunker
Dir: /Users/matteovolpi/Dropbox/Fluka_flair/fwdflukaXBOXMELBOURNE
Cmd: /usr/bin/nohup /Users/matteovolpi/fluka4-1.0/bin/rfluka -M 5 unimelb_bunker
F L U K A
Dir: /Users/matteovolpi/fluka4-1.0
Data: /Users/matteovolpi/fluka4-1.0/data
Exec: /Users/matteovolpi/fluka4-1.0/bin/fluka
Input: /Users/matteovolpi/Dropbox/Fluka_flair/fwdflukaXBOXMELBOURNE/unimelb_bunker.inp
Initial seed copied from /Users/matteovolpi/fluka4-1.0/data
Running fluka in /Users/matteovolpi/Dropbox/Fluka_flair/fwdflukaXBOXMELBOURNE/fluka_6472
======================== Running FLUKA for cycle # 1 ========================
Error code: 2”

Looking at the log file it seems there is a problem with the fortran flgvli.f file :

“At line 69 of file comlat/flgvli.f
Fortran runtime error: Bad value during floating point read
Error termination. Backtrace:
#0 0x10c7c79d0
#1 0x10c7c8248
#2 0x10c7c8c52
#3 0x10c8f9951
#4 0x10c8fc01d
#5 0x10c8fcd9e
#6 0x10c8fa970
#7 0x1016d8fb7
#8 0x1016f0e88
#9 0x10196dd95
#10 0x1019ef870
#11 0x101cb7ac0”

You can find the inp file (unimelb_bunker.inp) and all the outputs at this link:

Any ideas ?

Thank you very much in advance.

Cheers

Matteo

3

Hi Matteo,

As indicated by the error there is an issue of precision which I believe has to do with your geometry definition. You defined the different rooms as RPP bodies which you then subtract from the “Walls” region. However, in quite a few cases you have RPP bodies which share a surface boundary. Given that you do not provide many digits to define the coordinates, FLUKA runs into trouble with precision of these values (see slide 30 in this FLUKA course lecture on geometries).

To fix this issue I believe the best to do is to use infinite bodies, in your case I assume you can define all regions using infinite planes (XYP, YZP, XZP). The rooms then need to be redefined to be bounded by these infinite planes but there will be no precision issues anymore since the boundaries are then unambiguously defined. This is in any case a standard practice to avoid tracking issues and is more efficient computationally. Please have a look at the FLUKA geometry lecture link above for examples.

Hope this helps!

Andreas

1 Like