This paper evaluates the performance of multiphysics coupling algorithms on a light water nuclear reactor core simulation. The simulation couples the k-eigenvalue form of the neutron transport equation with heat conduction and subchannel flow equations. We compare Picard iteration (block GaussSeidel) with multiple variants of preconditioned Jacobian-free Newton-Krylov (JFNK). The performance of the methods are evaluated over a range of energy group sizes, boron concentrations and core power levels. A novel physicsbased approximation to a Jacobian-vector product has been developed to mitigate the impact of expensive cross section processing steps. Numerical experiments demonstrating the efficiency of JFNK relative to standard Picard iteration are performed on a 3D model of a nuclear fuel assembly.
Steven P. Hamilton, Mark A. Berrill, Kevin T. Clar