Improvement of Geant4 Neutron-HP package: Unresolved Resonance Region description with Probability Tables

TL;DR

This paper enhances Geant4's neutron transport accuracy in the Unresolved Resonance Region by implementing and validating a probability table method, enabling better modeling of resonant structures and their impact on neutron flux calculations.

Contribution

The paper introduces a probability table approach for Geant4's URR treatment, validated against MCNP6 and Tripoli-4, and compares pre-processing tools NJOY and CALENDF.

Findings

Probability tables improve URR modeling accuracy.

Self-shielding effects are significant for inelastic cross sections.

Validation shows consistency with established Monte Carlo codes.

Abstract

Whether for shielding applications or for criticality safety studies, solving the neutron transport equation with good accuracy requires to take into account the resonant structure of cross sections in part of the Unresolved Resonance Region (URR). In this energy range even if the resonances can no longer be resolved experimentally, neglecting them can lead to significant numerical biases, namely in flux-based quantities. In Geant4, low energy neutrons are transported using evaluated nuclear data libraries handled by the Neutron High-Precision (Neutron-HP) package. In the version 11.01.p02 of the code, the URR can only be described by average smooth cross sections that do not take into account the statistical resonant structure of the cross sections. To overcome this shortcoming, the treatment of the URR with the use of the probability table method has been implemented in Geant4 and successfully validated with the reference Monte Carlo neutron transport codes MCNP6 (version 6.2) and Tripoli-4 (version 12). These developments will be taken into account in the next release of Geant4. All the validations of Geant4 have been performed with probability tables generated from both the NJOY and CALENDF pre-processing tools. Therefore Geant4 now has this unique feature to study the relative impact of the strategies involved during the production of probability table by the two pre-processing codes. This has been used to show that self-shielding is important also for inelastic cross sections in the example of 238U. The tool to generate probability tables usable by Geant4 either from NJOY or from CALENDF is made available on a dedicated GitLab repository and will be included in Geant4.