Steady-state CFD simulations of an EPR™ reactor pressure vessel: A validation study based on the JULIETTE experiments

Abstract

Validating computational fluid dynamics (CFD) models against experimental measurements is a fundamental step towards a broader acceptance of CFD as a tool for reactor safety analysis when best-estimate one-dimensional thermal-hydraulic codes present strong modelling limitations.In the present paper numerical results of steady-state RANS analyses are compared to pressure, volumetric flow rate and concentration distribution measurements in different locations of an Areva EPR™ reactor pressure vessel (RPV) mock-up named JULIETTE. Several flow configurations are considered: Three different total volumetric flow rates, cold leg velocity field with or without swirl, three or four reactor coolant pumps functioning. Investigations on the influence of two types of inlet boundary profiles (i.e. flat or 1/7th power-law) and the turbulent Schmidt number have shown that the first affects sensibly the pressure loads at the core barrel whereas the latter parameter strongly affects the transport and the mixing of the tracer (passive scalar) and consequently its distribution at the core inlet. Furthermore, the introduction of an integral parameter as the swirl number has helped to decrease the large epistemic uncertainty associated with the swirling device. The swirl is found to be the cause of asymmetric loads on the walls of the core barrel and also asymmetries are enhanced for the tracer concentration distribution at the core inlet.The k–ϵ CFD model developed with the commercial code STAR-CCM+ proves to be able to predict quantitatively well both, the mixing at the core inlet and the mechanical loading at the core barrel for most of the different flow configurations considered.