Experiments and CFD calculations on coolant mixing in PWR - Application to boron dilution transient analysis

The coolant mixing is significant for safety assessment of boron dilution and cold water transients. For the investigation of the relevant mixing phenomena, the Rossendorf test facility ROCOM was constructed. ROCOM is a 1:5 scaled Plexiglas model of the German PWR Konvoi disposing of four loops with fully controllable coolant pumps. Mixing of both overcooled and deborated water is investigated by adding a tracer salt solution, performing conductivity measurements by wire mesh sensors and velocity measurements by the LDA technique. The tracer concentration fields established by coolant mixing under transient flow conditions were investigated. The experiments aimed at the validation of CFD codes. CFD calculations were carried out with the code CFX-4. During start-up of the first pump being relevant for boron dilution transients, the maximum of the tracer concentration appears at the core inlet at positions opposite of the loop with injection, while the maximum boron dilution strongly depends on slug size, slug initial position and mass flow ramp. A good agreement was achieved between measuring results and CFD calculations. A semi-analytical perturbation reconstruction model (SAPR) has been developed allowing the description of the coolant mixing inside the reactor pressure vessel of PWRs by superposition of response functions on nearly Dirac-shaped perturbations, which can be determined experimentally or from CFD calculations. SAPR provides realistic time-dependent boron concentration fields at the core inlet as boundary conditions for the best-estimate analysis of a hypothetical boron dilution event after start-up of the first main coolant pump in a generic four-loop PWR. Core calculations were performed with the 3D reactor dynamics code DYN3D. By varying the initial slug volume it was found, that for the given core loading pattern slugs of less than 20 m3 do not lead to re-criticality of the shut-off reactor. Calculations with the bounding slug volume of 36 m3 show, that the corresponding reactivity insertion does not result in core damage. (author)