NEA PWR Depletion Benchmark Phase I

Share your results and discuss how Serpent compares to other neutron transport codes or experimental data
Post Reply
User avatar
Jaakko Leppänen
Site Admin
Posts: 2447
Joined: Thu Mar 18, 2010 10:43 pm
Security question 2: 0
Location: Espoo, Finland

NEA PWR Depletion Benchmark Phase I

Post by Jaakko Leppänen » Sat Apr 03, 2010 3:52 am

I have run some burnup calculations using specifications from the NEA Depletion Calculation Benchmark, Phase I [NEA/NSC/DOC(2004)11].The geometry is a 17 by 17 PWR fuel assembly with burnable absorbers. The requested results include nuclide compositions in one of the fuel pins after various cooling times. Serpent results calculated using ENDF/B-VI.8, ENDF/B-VII, JEF-2.2, JEFF-3.1 and JEFF-3.1.1 based cross section libraries are found in tabular format at: ... rk/PhaseI/

Zip-compressed input and output files are found at the same directory.

All calculations were run using similar options:
  • Each fuel pin was handled as a separate depletion zone, Gd-pins were divided into 10 annular sub-regions (this results in 66 independent depletion zones when the 45 degree symmetry of the configuration is taken into account).
  • 3 million neutron histories were run for each transport cycle (10 inactive and 500 active cycles of 6000 source neutrons).
  • CRAM method was used for solving the depletion equations.
  • All available fission product cross sections (fpcut set to zero) and actinide cross sections between mass numbers 232 and 252 (new option) were included in the calculation (350 nuclides with cross section data, 4188 nuclides in total).
  • Burnup history was divided into 3 intervals with intermediate cooling periods. A total of 53 burnup steps were run with predictor-corrector calculation (including 4 decay steps at the end without transport calculation).
  • Probability table sampling for unresolved resonance cross sections was not used.
The total running time ranged from 17 to 19 hours on a 3 GHz single-processor Inte Xeon workstation and the calculation required 9 to 13 Gb of memory, depending on the cross section library. Grid thinning (egrid parameter) was used with 1E-5 tolerance.

I have used the same geometry configuration for comparing Serpent to CASMO-4E (see our M&C 2009 paper), but with a simplified irradiation history. I couldn't find any final report or reference results for this particular benchmark, so if you know where this data could be found, please let me know.

A few words about the results...

The origin of the cross section data seems to be a major source of discrepancy for the final compositions. This is especially the case when older libraries (JEF-2.2 and ENDF/B-VI.8) are compared to more recent ones (ENDF/B-VII, JEFF-3.1 and JEFF-3.1.1). These differences are clearly reflected in k-eff as well.

I tried re-running the calculations with probability table data, but I couldn't see any differences in the final compositions. This is certainly a problem, because in some cases the use of probability tables may have a noticeable impact in the final results, and without seeing any differences it is impossible to validate the methodology. I started another topic focused on unresolved resonance probability table data.
- Jaakko

Posts: 6
Joined: Fri Mar 26, 2010 3:17 pm
Security question 2: 0
Location: SECNRS, Moscow, Russian Federation

Re: NEA PWR Depletion Benchmark Phase I

Post by Anikin » Fri Mar 16, 2012 8:26 am

Dear Jaakko, I have found several reports (links are listed below) containing results of Takahama-3 PIE. Some of them contain C/E-1 (%) values, there C was calculated by 2D sequence TRITON of SCALE computer code system, so it's interesting to compare these values with yours.

So here is the official results on the oecd web page: ... &axis=1968

And here is more interesting report with TRITON C/E-1 (%) values: ... cr6968.pdf

Post Reply