An experimental and numerical investigation into scaling considerations for moderator circulation experiments

The moderator system in a CANDU reactor provides the unique ability to provide emergency cooling to the fuel in postulated severe accidents during the early phases of the transient. A key criteria which dictates the effectiveness of heat removal during these events is the integrity of the fuel channel assembly. This in turn relies on the prevention of dryout on the calandria tube. The main tool used to demonstrate adequate margins to dryout on the calandria tubes is Computational Fluid Dynamics (CFD). While several experimental datasets are available for validation of these tools, the historical facilities lack the specific geometrical features present in some of the older CANDU designs. The historic tests were typically scaled based on the so-called Archimedes number and non-dimensional heat flux. To extend the validation database two unique 1/16th scale (based on vessel diameter) experiments are performed which include features found in the moderator systems of two operating CANDU designs. The primary goal of these new experiments was to generate thermalhydraulic data for validation of CFD codes, but several additional tests were performed to examine the potential impact of scaling on the results. This paper demonstrates a consistent conservative bias in the CFD predictions with respect to maximum temperature in the experiments over the range of conditions tested. The combination of CFD and experimental observations concludes that the likely cause of the bias is due to CFD's over prediction of inlet jet dissipation leading to less cold fluid penetration in the heated core of the test section. In terms of scaling, a combination of experimental evidence and supporting CFD simulations indicate that the historical scaling parameters do provide some level of qualitative similarity in the mixed-convection regimes in the vessel, but some revision to account for geometrical and flow pattern deviations may be necessary.