Validation and Uncertainty Quantification for Two-Phase Natural Circulation Flows Using TRACE Code

The work presents validation of the TRAC/RELAP Advanced Computational Engine (TRACE) code for natural circulation two-phase flow in a vertical annulus. Natural circulation experiments were recently conducted for a vertical internally heated annulus at the Multiphase Thermo-Fluid Dynamics Laboratory at the University of Illinois. The experimental matrix consists of 107 experiments with system pressure in the range of 145 to 950 kPa and heat flux up to 275 kW/m(2). Void fraction, gas velocity, and interfacial area concentration were measured in five axial locations along the test section with six measurements of bulk liquid temperature and pressure. To validate the capability of the TRACE code under natural circulation flow conditions, a complete model of the experimental facility was created and validated using forced convection and single-phase natural circulation data. Sensitivity and uncertainty quantification were performed. The sensitivity to important simulation parameters was studied using Sobol's variance decomposition and the Morris screening method. The sensitivity of boundary conditions on void fraction measurement was investigated. The sensitivity study has shown significant differences in model sensitivity between different experimental conditions. With heat flux being the most influential parameter for high-pressure cases without flashing and pressure, temperature and heat flux have a combined strong effect in the case of low-pressure experiments when flashing occurs. Additionally, higher uncertainty in void fraction prediction was observed for experimental conditions at low pressure with flashing.