Application of an Adaptive COMET Method to a PWR Benchmark Problem with Gadolinium

Abstract

Coarse Mesh Radiation Transport (COMET) is a reactor physics method and code that has been used to solve reactor core problems. The method has been shown to be accurate in both its calculation of core eigenvalue as well as its calculation of fission density for every pin in the core. COMET solutions also enjoy excellent computational efficiency, as calculations are performed in a runtime that is several orders of magnitude less than Monte Carlo techniques. The method is a response-based, decomposing the problem domain into subvolumes, called coarse meshes, which allows for the global eigenvalue problem to be converted into a system of fixed-source problems. This produces no approximation if fluxes on coarse mesh interfaces are known. However, these are not known a priori, so an approximation is made in the form of a flux expansion. Responses to an assumed incoming boundary flux using an assumed expansion function are computed for each unique coarse mesh in a precalculation and stored in a response library. A deterministic iterative algorithm uses these precomputed responses to solve the coupled system of fixed-source problems, resulting in a full-core solution. In an effort to improve the computational efficiency of the code, a method has been developed that allows for automated adaptive selection of flux expansion orders in a COMET solution. An ideal flux expansion should be low in order to ensure computational efficiency but still ensure satisfactory accuracy in a calculation. In previous implementations of the COMET method, this flux expansion has been held constant throughout a problem and has been dependent upon user input and numerical experimentation. The method described in this paper allows the flux expansion to vary based upon both mesh-dependent (e.g., mean free path) and problem-dependent (e.g., imposition of boundary conditions and loading patterns in a reactor core) factors. Previous work with the method has shown that it has selected flux expansions that maintain accuracy in COMET calculations while increasing computational efficiency as compared to a COMET calculation employing maximum expansions in the response library for both small problems as well as a full-core PWR benchmark problem. This work extends the method to another benchmark problem that is larger in size with different material composition. The benchmark problem selected in this paper is a 4-loop PWR core with gadolinium. The adaptive method will select expansion orders to obtain an accurate solution for this problem relative to the benchmark COMET solution, and its performance in terms of both accuracy and computational efficiency will be assessed. Results will be provided in the full paper.