The Reduced-Moderation Water Reactor (RMWR) is an innovative water-cooled reactor with some of the characteristics of a fast neutron reactor, which can achieve a high conversion ratio of more than 1.0 and a negative void reactivity coefficient. The conception of RMWR is built on established Boiling Water Reactor (BWR) technology and is being researched by the Japan Atomic Energy Agency (JAEA) and Hitachi. The objective of this research is the 3D reactor physics modeling of RMWR.

The current methods of modeling: 2D deterministic codes are not sufficient and 3D Monte Carlo codes are slow due to the need to generate data for a full core axially heterogeneous RMWR model. These drawbacks severely limit the modeling capability of the RMWR. The objective is to observe the differences within the 3D deterministic route to see whether it can be used more efficiently as an alternative to the 3D Monte Carlo in order to achieve reasonable accuracy. The 2D few-group cross-section generation methodology will be insufficient for the full core modeling of such reactors. Although 2D modeling is quick, it is not suitable for the axially heterogeneous RMWR core configuration. Cross-sections generated with 2D lattice calculations will also be insufficient for the full core modeling of such reactors. That is because neutron net current at the boundary of each zone is assumed to be zero in the 2D modeling while in fact, neutrons of different energies will leak from adjacent zones. That is why 3D modeling of an RMWR is necessary. Therefore, the aim of the project is to study the 3D reactor physics modeling of an RMWR fuel assembly and compare the results using three computational codes: WIMS, MONK and EVENT. It is worth addressing that Deterministic code WIMS and Monte Carlo code MONK are the reactor physics software package developed by the ANSWERS Software Service. In addition, EVENT (Even-parity Neutral particle Transport) is developed by the Applied Mechanics and Computation Group in Imperial College London and it can solve the multi-group, steady-state and time-dependent self-adjoint second-order transport equation using the finite element-spherical harmonics approximation.

Two sets of fuels: Pu-UO2 and Th-UO2 have been studied for this project. The comparison among the three codes have been carried out with 1D axial slab, 3D lattice and 3D pincell models of an RMWR fuel assembly model. Three sets of cross-section are used in comparison: (i) P0 scatter with total cross section, (ii) P0 scatter with transport cross section and a corrected self-scatter cross section and (iii) P1 scatter with total cross section. The solution of the multiplication factor, neutron flux and yield in energy are compared using 172-group cross-sections using 2D equivalence theory in WIMS and with cross-sections that have been condensed to 12-group by the use of 2D transport calculations. Therefore, two sets of cross-sections with P0 scattering for WIMS and both P1 and P0 scattering for MONK were used for the comparison of results for the slab, pincell and lattice models with both 172-group and 12-group.

This research shows that an excellent agreement is observed between the WIMS and MONK results with a maximum difference of ∼100-300 pcm. Slightly higher maximum differences of ∼500 pcm are observed when comparing the results of the ANSWERS codes (WIMS and MONK) and EVENT. Furthermore, P0 transport correction is probably tolerable but does introduce errors. It is also observed that the 3D calculation route for WIMS (CACTUS 3D) requires the 172-group cross-section generations in a 2D slice of the model. In addition, solving the 172-group problem directly in CACTUS 3D is computationally expensive. It is preferable to somehow reduce the number of groups before performing the 3D calculation. Convergence tests with higher mesh refinement and with higher angular approximation have been accomplished by using WIMS.