Authors: A. Novak, J. W. Peterson, L. Zou, D. Andrš, R. Slaybaugh, R. Martineau
Year: 2019
Industry: Nuclear & SMR
Keywords: Use FSE, Steady state, Dynamic, Pebble Bed, Loss of Coolant, Validation, Comparison
ABSTRACT
A significant challenge in the core modelling of pebble bed reactors (PBRs) is the complex fuel-coolant structure. At the expense of approximating local flow and heat transfer effects, porous media models can provide medium-fidelity predictions of complicated thermal-fluid systems with significantly less computational cost than high-fidelity Computational Fluid Dynamics (CFD) models. This paper presents a new porous media code, Pronghorn – a fast-running core simulator intended to accelerate the design and analysis cycle for PBRs and provide boundary conditions for systems-level analysis. This paper describes the physical models in Pronghorn and demonstrates the capability of a friction-dominated model for predicting gas-cooled PBR decay heat removal by presenting simulation results for all 52 of the steady-state axisymmetric German SANA experiments, which include two different fluids and three different types of pebbles. The pebble temperature in all 52 cases is predicted with a mean error (predicted minus experimental) of +22.6 °C with standard deviation of 54.6 °C. To demonstrate Pronghorn’s capability for modelling bed-to-plenum heat and mass transfer, one open-plenum SANA experimental case is also simulated. A code-to-code comparison with Flownex and GAMMA shows that Pronghorn is comparable in accuracy to other porous media simulation tools, with the additional advantages of 1) an arbitrary equation of state; 2) 3-D unstructured mesh capabilities; and 3) multiphysics coupling to other Multiphysics Object-Oriented Simulation Environment (MOOSE) applications. Finally, the effect of several porous media closure selections, in particular the porosity, the near-wall treatment for effective solid thermal conductivity, the interphase drag and heat transfer, and the fluid thermal dispersion, on temperature predictions are quantified.