Computational Methods for Coupled Fast and Thermal Spectrum Reactor Core Analysis

Abstract

Most nuclear reactors are designed to be operated under either thermal or fast neutron spectrum, depending on the energy range of dominant neutrons that sustain the fission chain reaction. Almost all the commercial power reactors are thermal reactors while many Generation-IV reactor designs are fast spectrum based. In addition to fast and thermal spectrum reactors, coupled fast-thermal spectrum reactor (CFTR) concepts, which have both fast and thermal spectrum regions, have been adopted in several research reactors, including the initial design of the Versatile Test Reactor (VTR). For the analysis and assessment of reactor core designs, numerous computational methods have been developed and implemented into legacy codes. These legacy methods and codes adopted different simplifications and approximations in nuclear data representation, multigroup cross section (XS) generation, and whole core transport and fuel cycle analyses, tailored for specific reactor types. Hence, their applicability to other reactor types is limited. Except for Monte Carlo (MC) methods, there is no existing core analysis method directly applicable to CFTRs. This dissertation work is intended to develop an efficient two-step method in replacement of full-core MC simulations for CFTR core design and analysis. Since a CFTR would exhibit combined neutronic characteristics of both fast and thermal reactors, among the legacy two-step methods, the fast reactor core calculation method is preferred in that the whole-core depletion calculation capability with transport flux solution is readily available. To be applied to CFTRs, several improvements were made in the current fast reactor analysis methods based on the MC2-3, VARIANT, and REBUS-3 codes of Argonne National Laboratory. A new computational procedure was developed by combining MC simulation and MC2-3 for multigroup XS generation, incorporating partial current discontinuity factor (PCDF) into VARIANT for nodal transport analysis, and enabling the use of burnup dependent XSs and PCDFs in REBUS-3 depletion calculation. The new XS generation scheme was developed primarily based on MC simulations in assembly supercell models while MC2-3 is used to produce consistently weighted anisotropic scattering cross sections. Supercell models are used because of the significant spectral interference in CFTR environments. The long-range environmental effects are considered by imposing approximate source boundary conditions or including background source zones in the supercell model. Thorough verification tests were performed using two CFTR problems. To reduce assembly homogenization errors in VARIANT nodal transport calculations, PCDF was derived in a consistent formulation for arbitrary angular approximation orders and is fully compatible with the efficient red-black iteration scheme of VARIANT. Practical approaches were developed to generate PCDFs using the reference nodal average flux and surface averaged partial currents obtained from MC simulation. For practical applications, fixed source calculation and imbedded PCDF correction strategies were developed to treat long-range environmental effects on the PCDFs generated from supercell models. Enhancements of REBUS-3 were made in the depletion chain construction and utilization of burnup dependent XSs and energy dependent fission yields in a general manner. The modified REBUS-3 with these improvements and the improved VARIANT nodal transport method was verified using a 2D CFTR whole-core depletion problem. Compared to the original REBUS-3 depletion with constant isotopic XSs and VARIANT transport calculation without PCDFs, the new two-step method reduced the k-effective error by ~1000 pcm. It also predicted fuel assembly powers accurately within a 1% deviation from the reference MC depletion results.