Fusion neutronics @FNG: Serpent2 vs MCNP6.2

Share your results and discuss how Serpent compares to other neutron transport codes or experimental data
Post Reply
User avatar
Posts: 17
Joined: Mon Feb 24, 2020 11:00 am
Security question 1: No
Security question 2: 7

Fusion neutronics @FNG: Serpent2 vs MCNP6.2

Post by sjjamsa » Tue Mar 03, 2020 12:25 pm

Dear fellows,

We have this modelling of a Frascati Neutron Generator (FNG) irradiation and benchmark case. (This is in preparation for actual measurements.)

There is a neutron source, some LiPb, steel, plastic etc. next to the neutron source, as well as 7 pairs (up+down) of locations for activation measurements at various depths of LiPb b/w the source and the tally location. In each location, there are 10 reactions tallied as well as the neutron flux as a function of energy.

Both codes are using mostly Jeff3.3 cross sections. MCNP uses IRDFF 1.05 for most activation tallies, while Serpent uses Jeff3.3 except for the Nb and In reactions IRDFF-II is used.

The bottom of the post shows the results. One can notice that:
  • The neutron flux matches within a couple of % at energies below ~1MeV
  • The activation calculations show that Serpent overestimates the activation by 10%-15%, increasingly so while the moving deeper into the material.
  • Above ~1MeV Serpent seems to consistently overestimate the flux, more the deeper one is in the material.
Is this 10%-15% "good enough"? Where do the difference originate from?

- Simppa -

flux.png (224.25 KiB) Viewed 1132 times
rates.png (74.68 KiB) Viewed 1132 times
ratesRatio.png (76.11 KiB) Viewed 1132 times

User avatar
Jaakko Leppänen
Site Admin
Posts: 2446
Joined: Thu Mar 18, 2010 10:43 pm
Security question 2: 0
Location: Espoo, Finland

Re: Fusion neutronics @FNG: Serpent2 vs MCNP6.2

Post by Jaakko Leppänen » Thu Mar 05, 2020 11:16 am

10-15% between Serpent and MCNP is a lot. When we generate new cross section libraries, we run systematic tests for each nuclide to confirm that the reaction physics is consistently handled. One set of test results can be found at:


With a few exceptions, the results are a near perfect match. Basically, if your geometries are consistently defined and you use the same ACE libraries , the differences can be expected to be within the range of statistical precision.

At least this applies to results from the transport calculation (flux). As for the activation, there may be discrepancies in the way the IRDFF activation data is handled (this data has not been as thoroughly tested). But before looking into activation, try to get a better mach for the flux.
- Jaakko

Post Reply