Containment atmosphere mixing is an important nuclear power safety issue related to hydrogen explosion risk. Extensive experimental and numerical research effort has been applied to this issue internationally, including several international standard problems. One of the findings was that containment atmosphere mixing using lumped-parameter codes common in the safety analyses is complicated due to limitations inherent in the lumped-parameter method. However, it was also found, that in the cases when the user had selected appropriate nodalisation schemes, results obtained with the lumped-parameter codes were satisfactory.
The paper presents simulations of one of the containment atmosphere mixing experiments - MISTRA M5 (formation of stratified conditions by a jet/plume steam release and vertical thermal gradient) - performed using a lumped-parameter code COCOSYS. Special nodalisation features required to compensate for the method limitations are described and used. Additionally, an uncertainty and sensitivity analysis of the simulation is presented. The parameters with large uncertainties having high influence on the results are identified. The importance of an adequate nodalisation to account for the occurring processes, e.g., in this case, plume/jet releases, is showed. (C) 2013 Elsevier B.V. All rights reserved.