American Nuclear Society
Home

Home / Publications / Journals / Nuclear Science and Engineering / Volume 165 / Number 1

Scaled and Full-Size Three-Loop Reactor Vessel Simulation for Boron Dilution Studies Using Computational Fluid Dynamics

T. V. Dury, M. T. Dhotre

Nuclear Science and Engineering / Volume 165 / Number 1 / May 2010 / Pages 101-116

Technical Paper / dx.doi.org/10.13182/NSE08-90

Current designs of pressurized water reactors (PWRs) employ a boric acid solution in the primary cooling water to control core reactivity during operation and shutdown. However, situations could theoretically occur in which diluted borated water is present in the primary circuit. Scale experiments have been performed for a single-pump start-up, with subsequent computational fluid dynamics (CFD) simulation, to examine the accuracy with which the concentration distribution of diluted borated water entering a reactor core can be predicted. It was concluded that higher-order advection schemes must be used to obtain sufficient resolution of the velocity field and capture the larger-scale effects of the flow but that each turbulence model produces a different core-inlet boron concentration development and distribution. Though it was not the most sophisticated available, the two-equation RNG k- turbulence model produced the closest agreement with experiment. However, mesh independence of the computational results was not achieved. As a sequel to this scaled CFD study, a simulation was carried out of a full-size three-loop Siemens-type PWR featuring a perforated cylindrical flow baffle in the lower plenum. Results again showed different characteristics in time and space, depending on the turbulence model used. Comparative assessment of the results obtained with the code CFX-5 showed that correct geometrical modeling of a perforated flow baffle in the lower plenum is essential, as a porous medium representation of the baffle can lead to serious underprediction of mixing. This occurred particularly with the RNG model but also using more sophisticated turbulence models. Further refinement of the mesh is now necessary to achieve mesh independence of the results. This requires access to a massively parallel computer system.