CFD analysis of water vapor condensation in large containments: Numerical model, verification and validation

In the event of a loss of coolant accident (LOCA) in a Pressurized Water Reactor (PWR), large quantities of water vapor and hydrogen might be released into the containment building. Increasing pressure and the risk of hydrogen deflagration can threaten the containment integrity. Water vapor condensation on cold containment structures modifies the pressure load on the containment structures. On the one hand, the total pressure reduces due to the decrease of the water vapor content of the containment atmosphere. On the other hand, the concentration of hydrogen increases, which can reach locally the threshold of deflagration or detonation. A Computational Fluid Dynamics (CFD) model is developed and tested in order to predict transients of the pressure and the concentrations of gaseous species in reactor containments after a LOCA. Since the free volume of the containment of typical French PWRs is between 50,000 m3 and 80,000 m3, the CFD model requires significant simplifications.The physical model and its simplifications as well the implementation in the CEA OpenSource CFD code TrioCFD are described. Model tests are presented in two steps. In a first step, verification tests are discussed to show that the condensation model is implemented correctly in the CFD code. In a second step, the model is validated against experimental data of the International Standard Problem ISP47. In this benchmark, steady state conditions between vapor injection and condensation were reached experimentally in the MISTRA test containment of the CEA. In phase A of the benchmark, an equilibrium was achieved between water vapor injection and water vapor condensation on temperature controlled cold walls. One incondensable gas, namely air, was present in the test vessel. In phase B, a second incondensable gas was added to the containment atmosphere, namely helium. Both steady state situations were analyzed with the CFD model. The calculations represent well the experiment when the predominant condensation paths are modelled. It is shown that the pressure in the containment vessel as well as the mean mass fractions of water vapor and air, as well as of helium in phase B, are calculated in accordance to the experiment. The temperature in the containment is overestimated. Measured vertical profiles of the species concentrations are reproduced correctly.