Computation for nuclear reactor systems

[Astronuc]

A lot has changed since the following threads were active.

Computation in nuclear engineering
https://www.physicsforums.com/threads/computation-in-nuclear-engineering.282715/

Using HELIOS Code for Preparing Macroscopic XS in PARCS
https://www.physicsforums.com/threa...for-preparing-macroscopic-xs-in-parcs.347578/

https://www.studsvik.com/key-offerings/nuclear-simulation-software/
CASMO5 is Studsvik Scandpower’s state-of-the-art 2D lattice physics code for modeling square and hexagonal LWR nuclear fuel. By including the latest nuclear data and substantially expanded modeling capability, CASMO5 reaches far beyond previously available lattice physics codes.

Nuclear codes and methods have changed considerably, and the companies back then have changed/restructured dramatically - new owners - new identities.

AREVA split into Orano and Framatome; New Areva became Orano, which refocused on nuclear materials development and waste management. Orano’s activities encompass mining, conversion-enrichment, used fuel recycling, nuclear logistics, dismantling and engineering, i.e., front and back ends of the nuclear fuel cycle. Framatome does the reactor design/development and nuclear fuel.

I attended a presentation on Framatome's new methods related to core design, safety analysis and nuclear fuel performance. The ARCADIA package consists of Apollo2-A, Hermes, and Artemis. Apollo-2A is the lattice physics code, while Hermes, an interface code, collapses the cross sections to fewer groups; the output of Hermes feeds the core-simulator, Artemis.

An earlier (2017) paper describes the early development.
https://www.kns.org/files/int_paper/paper/MC2017_2017_6/P283S06-05MartinN.pdf

AREVA developed the ARCADIA(R) reactor code system including the lattice physics transport code APOLLO2-A. Based on the APOLLO2 kernel developed by CEA, APOLLO2-A features a state-of-the-art methodology designed by AREVA for Light Water Reactor industrial applications. The validation of the code is achieved through comparisons with a comprehensive experimental database and with Monte-Carlo reference codes. In this paper, the main features of APOLLO2-A, the methodology and results from the validation base are presented.

Ref: https://www.researchgate.net/publication/275969363_APOLLO2-A_-_AREVA's_new_generation_lattice_physics_code_Methodology_and_validation

https://www.epj-conferences.org/articles/epjconf/pdf/2016/06/epjconf_wonder2016_01001.pdf

I'd like to touch on Monte Carlo (MC) methods, which basically track a population of neutrons in a system, as opposed to a full blown core simulator (neutronics code), which is based on transport or diffusion theory, or some hybrid method. Later.


Edit/update: An example of available data for benchmarking a neutronics code package.
https://www.ipen.br/biblioteca/cd/physor/2000/physor/183.pdf

In BWRs, which boil water (moderator and coolant) in the core, the local void fraction is a key factor in determining cross-sections and spectral effects. Without moderation, the nuetorn energy spectrum hardens, i.e., the fast neutron flux plays a greater role in causing fissions and transmutation of ²³⁸U into [/sup]239[/sup]Pu and other TU isotopes. Spectral shift can be used to reduce enrinchment, batch size and/or extend cycle length by converting some of the fertile ²³⁸U into fissile [/sup]239[/sup]Pu. Furthermore, MOX fuel produces a slightly harder spectrum than U-based fuel.

 

[rpp]

A lot has certainly changed. One area that has seen a lot of development is in Monte Carlo codes.
MCNP used to be the only code used, but now people seem to be moving towards Serpent and OpenMC.

MCNP is still the gold standard and the developers continue to make big improvements in the code. However, MCNP is becoming harder to obtain through RSICC due to 810 restrictions. MCNP was not traditionally developed for reactor applications, and they are behind in usability and capabilities such as depletion and multigroup cross section generation.
https://mcnp.lanl.gov/

Serpent was designed from the ground up to be used for modeling reactors, and the input is much easier to use for reactor geometries. Serpent depletions continue to be the state-of-the-art and the code is used to generate multigroup cross sections many users, including INL. Serpent can be obtained from RSICC, but it is not classified as 810. Serpent is also easier to obtain internationally since it was developed in Finland. Serpent has an active on-line community and good on-line documentation.
https://serpent.vtt.fi/serpent/

OpenMC is the newest Monte Carlo code and is open source, making it very easy to obtain. The code has a Python front-end, which makes setting up problems easy if you are familiar with Python. OpenMC has lattice features, which make it very easy to set up standard reactor geometries. OpenMC can perform depletions and generate multigroup cross sections, but these features are still in development. The documentation is very good for developers, but can be lacking for users.
https://openmc.org/
https://github.com/openmc-dev/openmc

Standard disclaimer - all opinions are my own.