Assessing the effect of using supercells instead of lattice blocks on multigroup cross sections of the MHTGR-350 reactor
Abstract
Uncertainty and sensitivity analysis methods are essential in nuclear reactor design and safety. These methods can be used to predict the probability distributions in output parameters, treat discreet events and handle large amounts of input data. The International Atomic Energy Agency (IAEA) initiated a co-ordinated research project (CRP) a few years ago to quantify the effect of modelling uncertainties on High Temperature Gas-cooled Reactors. This was done in parallel to a similar uncertainty analysis project on light water reactors overseen by the Nuclear Energy Agency of the Organisation for Economic Corporation and Development. The High Temperature Gas-cooled Reactors have unique features, for example the fuel comprises tri-structural-isotropic (TRISO) coated fuel particles, graphite is used as the moderator and structure material for the reactor core, and helium is used as the coolant. Thus, another important goal in addition to quantifying the uncertainty is to establish whether these differences influence the uncertainty and sensitivity propagation methodology significantly compared to those used for the Light Water Reactors cases. In the International Atomic Energy Agency co-ordinated research project, two systems are defined based on the pebble bed design and the prismatic design. The prismatic reactor design is called the MHGTR-350. In order to study the uncertainty quantification and associated sensitivities of the HTR, the IAEA CRP has been divided into a number of phases, with each phase in turn having a number of exercises. This dissertation is based only on exercise I-2c, which is concerned with the generation of groups constants using supercells. Other aspects of the IAEA CRP, for example, comparing the uncertainties propagated for the HTR with those of the LWRs are beyond the scope of this work. In preparing collapsed group constants for nodal calculations, the traditional method is to use lattice calculations for the fuel blocks, in which single fuel blocks are modelled with reflective boundaries. However, the real environment of a fuel block is not an infinite array of fuel blocks, but rather has neighbouring blocks with different compositions, for example spent fuel or reflector blocks. The International Atomic Energy Agency Coordinated Research Project has an exercise (I-2c) based on modelling the fuel block within a mini-core (a supercell) to capture the uncertainties that can arise when modelling a fuel block in an infinite environment or otherwise. In this study, the effect on the multiplication factor and the collapsed constants was studied using the definition in the benchmark document provided for exercise I-2c of the International Atomic Energy Agency Coordinated Research Project. The main sources of uncertainties considered were the type of model (reflected fuel block or supercell), the library version (ENDF/B-VII.0 or ENDF/B-VII.1), the nuclear data uncertainties, modelling in continuous or multi-group mode and the code used. The 4-group collapsed macroscopic cross sections of the MHTGR-350 supercells were obtained and were compared with equivalent MHTGR-350 fresh fuel block, and the difference was found to be significant. Furthermore, when comparing these differences with the uncertainties in the nuclear data, it was found that these differences were quite significant. It is therefore concluded that the model selected (fresh fuel block or supercell) generates a significantly larger uncertainty for the collapsed constants than that due to the uncertainties propagated by nuclear data uncertainties.
Collections
- Engineering [1403]