A statistical approach for modelling the effect of hot press conditions on the mechanical strength properties of HTGR fuel elements

摘要

The high temperature gas-cooled reactors (HTGRs) are the next-generation reactors with high safety, because the cores of HTGRs would not melt. But it is important to improve the oxidation resistance of the fuel in case of a huge oxygen ingress into the core to further improve the safety of HTGRs, because most of the volume of the core of the HTGRs consists of graphite. Coated fuel particles (CFPs), of which the diameter is around 1 mm, are used in HTGRs. A small sphere containing fissile materials is sealed up with ceramics coating layers to form a CFP. A mixture of CFPs and starting materials of binding material (matrix) is sintered to form fuel compact. Currently the matrix is graphite, which would easily oxidized in case of a huge oxygen ingress into the core. In this study, the development of oxidation resistant fuel compact, of which the matrix is a mixture of SiC and graphite, has been carried out. Simulated CFPs (alumina particles) and starting materials of matrix (Si powder, graphite powder and resin) were molded and hot-pressed into simulated fuel compacts. In order to maintain the structural integrity of fuel elements for HTGR under accident conditions, not only oxidation-resistant but also high-strength fuel compacts should be further developed. Hot press conditions such as pressure would be one of the factors affecting the strength of the HTGR fuel elements. In order to identify the optimal hot press conditions for preparing the high-strength fuel elements, the effect of the hot press conditions on the mechanical strength properties of the HTGR fuel elements should be evaluated quantitatively. In the present study, the response surface model, which represents the relationship between the hot press conditions and the mechanical strength properties, has been constructed by introducing statistical design of experiments (DOE) approaches, and the optimal hot press conditions were estimated by the model.