Experimental and computational approach to investigating rapid boron dilution transients in PWRs

At Vattenfall Utveckling AB (VUAB), a 1/5 model of a three loop Westinghouse PWR has been built in order to experimentally investigate rapid boron dilution transients (RBDT). In the model, a plug of salt water solution, representing un-borated water, is injected into fresh tap water circulating through the model. By measuring the conductivity of water at different points on a plane above the bottom plate, the temporal development of the salt concentration field adjacent to the core inlet has been determined. Measurements have been carried out for two different flow rates, corresponding to the steady state inlet Reynolds numbers 1.2 x 106 and 0.4 x 106. The experimental results indicate that there is a similarity between the developments of the salt concentration fields in the two cases. In other words, knowing the aforementioned field for the first case, one can deduce the concentration development of the second case on pure scaling grounds. Two conclusions may be drawn from these results: 1) For sufficiently high flow rates, the salt concentration field development is insensitive to the value of Reynolds number, and 2) The transient part of the flow rate variation, during which it is increased from zero to full flow, does not have a considerable influence on the salt field. The research on RBDT at VUAB has been mainly experimental. Recently, however, computational work with the final aim of simulating the boron field development in full scale reactors under realistic conditions, has started. In the first step of the computational work, the flow field in a 1/5 mock-up, similar to that built for the experiments, has been investigated by using the code PHOENICS, version 1.6.6. The purpose of the computations has been to investigate the influence of computational grid on the flow field at steady state. It is shown that, for Re = 1.14 x 106, at least 80 000 cells are required for a good resolution of only the flow field. The bottom plate and neutron pads were not modelled in these grid convergence studies. Computations on the same grid but with different Reynolds numbers (1.14 x 106 and 0.36 x 106) yield similar velocity distribution patterns