Validation of Computational Fluid Dynamic Tools for Single-Phase Flow Distribution in CANDU Headers

Abstract

<p>Canada Deuterium Uranium, CANDU, nuclear reactors use forced convection cooling to remove heat from the nuclear fuel and transport it to the power production systems. Flow is supplied by large capacity heat transport pumps and is distributed to each separate fuel assembly through headers. The determination of thermalhydraulic parameters of the CANDU headers is important because hydraulic behavior in the headers governs the void fractions of fu el channels connected to them and influe nces the fuel bundles cooling efficiency during postulated accidents.</p> <p>This work presents the validation of FLUENT 6.3.26, a three dimensional Computational Fluid Dynamics (CFD) code, for header flow distribution simulations by comparing predictions to experimental data. The experimental data were obtained for three different header geo metries: horizontal header with four vertical outlets (case study data obtained from literature), horizontal header with two vertical inlets and header with two horizontal inlets (experiments done in this study). The experiments were carried out using 1.0 m long, 3.67 cm 10 horizo ntal cylindrical header with two symmetrical distributed vertical inlets or two horizo ntal inlets at the two header ends. The flow is distributed to five horizontal and five vertical outlets along the header with 0.92 cm 10.</p> <p>In the flrst validation, FLUENT provided good predictions of flow distribution and pressure gradients along the header for different inlet flow rates (Re number between 800 and 4,800). In the second and third validations, simulations for both vertical and horizontal inlet configurations were examined and with varying levels of inlet flow imbalance. The experimental data consists of a set of outlet flow rates as a function of inlet flow rates. The effects of flow inlet velocities, flow modeling and grid density on the computational accuracy are also presented. The CFD technique was found to be an efficient tool to predict the flow distribution in the headers studied.</p>