An Investigation Into The Flow In Candu Fan Nozzles

Abstract

The moderator within the CANDU calandria vessel is an important element of the reactor serving various operational and safety functions. A validated computational fluid dynamics model to simulate the fluid within the calandria vessel is required to predict how it would behave in different situations. Given is high level of complexity (flow from complicated inlets inlet, outlet, volumetric heating, flows around fuel channels, etc…) validation typically proceeds from separate effect studies for each component as well as integral validation where all complex phenomena are included. This work is focused on the separate effect studies aimed at phenomena relevant to the inlets of Pickering A.

A scaled down model of the Pickering CANDU moderator inlet nozzle was constructed and used to verify simulations done on the same nozzle. Physical measurements were taken using a PIV laser system on the scaled down nozzle. Seven measurements planes were taken on two of the outlets of the nozzle at two different flowrates (0.2 kg/s and 0.1 kg/s). Given a single PIV measurement plane contains approximately 1000 velocity points this represents approximately 28000 new velocity measurements for CFD validation. A CAD model of the nozzle was imported into STAR-CCM+ CFD software to perform the simulations. The simulations used the k-ω SST turbulence model and were performed using the same flowrates as in the experiments. The k- model was also used as a comparison for the k- SST model. The same locations were used in calculations for the seven measurement planes so that the same areas could be compared for the models employed.

The CFD results were found to agree qualitatively with the measured ones but no quantitative comparisons should be made with care. Additional investigation is required for a more comprehensive comparison. While both, measured and calculated results, showed similar velocity distributions, the velocities values differed by varying degrees. In particular, the simulations did not show the same jet expansion as the physical measurements indicated. There is not enough data from the measurement planes to determine if the flow in the physical experiments are more diffuse and the velocity field gradients are lower than obtained in the current study. With large portions of the nozzle unobserved by the measurement planes, it cannot be concluded that there are no other peaks or large velocity gradient regions in between the planes. For future work it would be beneficial to have a larger model of the nozzle for better gradient resolution, as well as using a measuring system that allows for more precise data collection. Different turbulence models and measurements further away from the nozzle should also be used to determine which would be best in a larger simulation.