NEWT failure to get the flux data

[Antonio Dambrosio]

Hello, 
while trying to calculate the cross sections for the radial reflector of a VVER-1000 rector with TRITON, I've encountered these errors:
"
   Outer iteration sweep begins.

 Outer  Eigen-  Eigenvalue Max Flux  Max Flux     Max Fuel  Max Fuel     Inners
 It. #   value     Delta    Delta   Location(r,g)  Delta   Location(r,g) Cnvrged
 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
    1 0.81000   0.000E+00 6.849E+04 (    6,251)   1.000E+00 (  921,  0)    F
    2 0.84825   4.509E-02 3.096E+01 (   22,  2)   2.281E+00 (66753,  0)    F
...

[omitting the iterations in the middle]

...
   28 0.73830   1.023E-04 4.410E-03 (    3,238)   1.626E-03 (72305,  0)    F
   29 0.73837   8.772E-05 3.895E-03 (    3,238)   1.408E-03 (72305,  0)    F
   30 0.73843   7.986E-05 3.438E-03 (    3,238)   1.221E-03 (72305,  0)    F
 k-eff =       0.73842747     Time=      0.00d  Branch=1
 Property changes for this branch:
  Density factor 0.999089 for mix=4
  Density factor 0.999089 for mix=44
  Density factor 0.999089 for mix=54
  Density factor 0.999089 for mix=64
  Boron ratio 1.85446 mix=4
  Boron ratio 1.85446 mix=44
  Boron ratio 1.85446 mix=54
  Boron ratio 1.85446 mix=64


 Critical Spectra Calculated with the B1 Method
 Input buckling of  0.00000E+00 gives a corresponding k-eff of  7.32829E-01
 Equivalent core height is infinite.
***Error:  Wrong header  size for  newt newt flux file fluxrat record

***Error:  Wrong header  size for  newt newt flux file fluxrat record

***Error: NEWT failed to get the flux data.
***Error: Transport execute failed.
***Error: Application failed - see messages file for details.
"
After many trial and error attempts, I have found out that the errors seem to be due to the too large number of meshes used globally, i.e. for both the reflector parts and the surrounding fuel assemblies. I'm saying this because, after moderately reducing the number of meshes of the elements, the inputs started working.
Since I could not find any documentation of similar errors in the manual, I have two main questions:
- did I hit a threshold or a non-documented "hard" limit for the meshes number? If not, what's the main cause of these errors?
- is there a different way to overcome them, other than reducing the total number of meshes?

P.s. the errors are generated in both SCALE6.2.4 and SCALE6.3.1


Thanks,
Antonio Dambrosio

[Yevgen Bilodid]

Hi  Antonio, 
I 'd like to advice you to converge inner iterations.

Best regards,
Yevgen

[Antonio Dambrosio]

Hello Yevgen,
mine was only a demo run to show my problem. I've run newer problems, up to convergence, but getting the same error. This time I have noticed that even the number of burnup points I define in the burndata card can trigger this error. Namely, I have used the same model, only with different burndata cards. The inputs that failed literally had only one burnup point more than the input that completed successfully, so I believe that NEWT has a hard limit for the number of flux points it can calculate. I assume that when the combination of mesh grids + burnup points can actually overflow the header, array or whatever NEWT uses to store them.

"
   26 0.84689   5.555E-04 5.218E+00 (   29,  1)   1.611E-03 (10694,  0)    F

   27 0.84730   4.823E-04 3.826E-01 (   29,  1)   1.386E-03 (10694,  0)    T

   28 0.84765   4.173E-04 2.087E-02 (   29,  1)   1.192E-03 (10694,  0)    T

   29 0.84796   3.603E-04 1.210E-03 (   30,235)   1.025E-03 (10694,  0)    T

   30 0.84822   3.106E-04 1.039E-03 (   30,235)   8.818E-04 (10694,  0)    T

   31 0.84845   2.676E-04 8.926E-04 (   30,235)   7.584E-04 (10694,  0)    T

 k-eff =       0.84844897     Time=      0.00d Nominal conditions

 Critical Spectra Calculated with the B1 Method

 Input buckling of  0.00000E+00 gives a corresponding k-eff of  8.49755E-01

 Equivalent core height is infinite.

***Error:  Wrong header  size for  newt newt flux file fluxrat record


***Error:  Wrong header  size for  newt newt flux file fluxrat record


***Error: NEWT failed to get the flux data.

***Error: Transport execute failed.
"

[Yevgen Bilodid]

I'm sorry I couldn't help :)

субота, 2 листопада 2024 р. о 16:49:27 UTC+2 Antonio Dambrosio пише:

[Ugur]

Dear Antonio,

SCALe in general doesn't have hard coded limits unless mentioned in the manual. However, if the dark areas in your core model are just dark because of meshing, I am not surprise you are running into problem. The error message is coming from not being able to read the flux file. It looks like the file exists; however, corrupted or not written completely. You might be running into disk space or time out issues. I would just check the temp directory and file sizes. If you send us an input file we may be able to help you more. 

thanks

Ugur