Session: Validation Methods

Paper Number: 152559

152559 - Validation of Turbulent Mixing in the Chimney of a Pool-Type Reactor 

Abstract: 

Nuclear reactors are often located within pools of water. Allowing access to the reactor core from the pool surface is achieved with a chimney. The chimney passively guides flow from the primary loop out of the core and into the outlets which then loop back into the core. Within this chimney exists a region of turbulently mixing water as the stagnant pool is resting on top of the upward-flowing water from the core. When the stationary water and moving water collide there exists turbulent mixing until the upward flow is entirely stopped by the pool. The details of this flow are not well documented in the present literature, and so a validation procedure using the existing experimental data from the literature is necessary. With this validation, confidence in present models is established for future work to be complete for the NIST Neutron Source (NNS). Turbulent mixing of water in a junction of four paths is simulated and experimentally investigated by Sengupta et al. who have extensively characterized the hydraulics of this flow. The experimental data collected for two different cases are tested against present models of the same cases. A case without any bypass flow into the pool and a case with maximal bypass flow into the pool are studied. Sengupta et al.’s geometry is copied and commercial CFD software ANSYS Fluent is used to generate a mesh and solution to the problem. Specific methods are documented within this work, and mesh convergence is studied by method of a Grid Convergence Index (GCI). The researchers which provide experimental data also validated their own data with ANSYS Fluent, so present results are also compared to these older Fluent results. Present models were non-convergent under a steady state solution regime. To compare with the literature, simulations are carried out until the RMS value of the residuals for each solution parameter approach a singular value. At this time a selection of iterations spaced 5 apart are taken and averaged together to provide the average steady state values. These averaged values match experimental measurements better than the older models. Present errors from experimental data average around 3%, and the Grid Convergence Index predicts 0.3% numerical uncertainty in model results. This validation procedure found high agreement with experimental data and concludes that the newer fluent model with specified options is better suited to model the velocities in this problem, however it is important to note that the jet height is not predicted as accurately as Sengupta et al.’s computational solutions.

Presenting Author: Breken Wallar Texas A&M University

Presenting Author Biography: Senior Nuclear Engineering student at Texas A&M University interested in thermal hydraulic simulations of nuclear systems.

Authors: