Application of FEMLAB on Supercritical Hydrogen Components of the High Flux Isotope Reactor Cold Source

Abstract FEMLAB has played a key role in the design and safety analysis of several key components of the new High Flux Isotope Reactor (HFIR) hydrogen (H2) cold neutron source (CS) at Oak Ridge National Laboratory (ORNL). The main components of interest for the detailed analysis capability of FEMLAB are those where the H2 temperature spans a large range causing the fluid proper-ties to change dramatically in both magnitude and slope. This dramatic fluid property change tends to aggravate the problem solution as if it were a two-phase fluid, yet be-cause the pressure is supercritical, is actually only a sin-gle phase. Two HFIR CS components that are designed to experience this dramatic fluid property change are the moderator vessel (MV) and the pressurizer. The MV is a turbulent-flow, forced-convection device with internal heat generation and experiences the H2-specific dramatic fluid property changes in the boundary layer near the walls. The pressurizer is a laminar flow, natural convection de-vice that experiences even greater fluid-property changes over the more global range of the bulk fluid. A brief his-torical perspective discusses how the analysis process has impacted the design of the pressurizer including the use of codes other than FEMLAB early on in the process. The discussions includes subtle details and experiences in using FEMLAB to model and analyze these components with particular emphasis on the pressurizer component. Included in this discussion are choices of equations solved by the model, solution methods, proper utilization of arti-ficial dissipation, mesh generation and associated memory limitations, boundary conditions and implementation, es-tablishment of initial conditions and a stepwise procedure of obtaining a final, fully-coupled solution via restarting from intermediate solutions, and exporting output to ex-ternal graphics applications. Some comparisons with other CFD analyses are discussed for a forced-convection val-idation problem. Finally, some overall conclusions of th