COSI Test Case Simulation With CFX Code

Abstract

During a Small Break Loss-of-Coolant Accident (SB-LOCA) in a Pressurized Water Reactor (PWR), cold water (9°C) is injected into the cold leg and mixes with hot primary fluid (285°C) which flow down to the Reactor Pressure Vessel (RPV) downcomer. Such a scenario may result in a Pressurized Thermal Shock (PTS) on the structural components, first of all the walls. In order to access to the structural behaviour of the RPV during a PTS, reliable predictions of thermal loads on the RPV walls and fluid temperature distribution in the downcomer are needed. This second parameter depends on temperature mixing in the cold leg between cold water coming from Emergency Core Cooling (ECC) and hot primary fluid. The two-phase flow issues are Direct Contact Condensation (DCC), stratified flow, jet impacting a free surface or a cursed wall. A Computational Fluid Dynamic (CFD) simulation is needed in order to have an accurate prediction of these phenomena and then to calculate the distribution of fluid temperature in the cold leg and the downcomer. The CFD-code CFX seems to be appropriate to deals with such issues but it is not validate to simulate physical phenomena which occur in the cold leg during a PTS. The paper presents CFX assessment on COSI experiment. COSI facility is a scaled model of a 900 MWe PWR under LOCA thermal hydraulic conditions and has been developed to study condensation which can occur in the cold leg during a Pressurized Thermal Shock. The paper focuses on one test corresponding to a situation with a high water level in the cold leg. The COSI test has been simulated with CFX code where a specific model coming from literature has been used for heat transfer between liquid and interface description. The use of this model is necessary in order to obtain a good agreement between COSI experimental and simulated results. In fact, with this model, the code is able to reproduce correctly all the main features, mean error between experimental and simulated results being about 10%.