Preliminary Verification of Fixed-Source Sensitivity Analysis in OpenMC

Abstract

Sensitivity analysis capabilities have yet to find extensive use in fusion reactor design applications where they can help understand the impact of nuclear data uncertainties on the tritium breeding ratio (TBR), shutdown dose rates, and nuclear heating. Significant uncertainty exists in nuclear data for fusion applications, and the goal of this work is to explore whether adjoint- and perturbation theory–based eigenvalue and generalized response sensitivity methods recently developed within the OpenMC Monte Carlo code can be extended to fixed-source Monte Carlo radiation transport simulations.

This paper presents a derivation for the extended sensitivity analysis method. The adjoint-based sensitivity coefficients are compared with Monte Carlo SERPENT sensitivity coefficients for a TBR calculation in a simplified ARC-class tokamak. Further verification with the Monte Carlo MCSEN and deterministic ASUSD sensitivity coefficients for the tritium production rate in the Frascati neutron generator helium-cooled pebble bed test blanket module mock-up experiment was performed. The OpenMC sensitivity coefficients were found to agree with the other code systems. The use of the sensitivity coefficients with nuclear data covariance within the sandwich rule for cross-section uncertainty propagation also showed good agreement with the reference total Monte Carlo nuclear data uncertainty.