| Abstract [eng] |
The need for experimental and numerical analysis of the physical phenomena that lead to severe nuclear accidents has significantly increased after the Fukushima Daiichi accident. The main goal of this dissertation is to develop a methodology for numerical analysis of severe accidents leading to partial fuel bundle melting by integrating experimental results with advanced modelling approaches for symmetric and asymmetric geometries and applying the best estimate approach for the calculation results. In this work, experimental data from the PHEBUS and QUENCH programs were used. Numerical investigations were performed using the severe accident code RELAP/SCDAPSIM together with the SUSA tool for the best estimate approach application. Also, the CORSOR-M model, for the evaluation of fission product release, was implemented in the RELAP/SCDAPSIM code. The CORSOR-M model provided reliable estimates of Cs/I release in early bundle degradation phases, while later stages require more advanced models. The uncertainty analyses confirmed the reliability of simulations for symmetric geometries. For asymmetric geometries, the investigation results showed that the pseudo-symmetrical approach allows accurate predictions under quenching phase, where this model underestimates hydrogen generation. The use of component-level approach for the simulation of asymmetric geometries improved predictive accuracy of the calculation results. As a result, this first-of-its-kind methodology extends the applicability of severe accident codes, enhances predictive accuracy, and provides new insights into simulation and investigation of hydrogen generation, cladding oxidation, and fission product release under severe accident conditions. |
