Severe accident investigations for VVER-reactors including radiological impact and quantification of uncertainties

Safety concern on mitigating potential impact of a possible release of Source Term (ST) to the environment largely increased after the major severe accidents of Chernobyl and Fukushima. Severe accident studies are conducted by regulators using deterministic integral tools to predict the accident progression and the resulting ST including the radiological impact. Based on such analysis, emergency response teams develop emergency management plans considering the best-possible, reliable and available information just after the accident happens. In this dissertation, a simulation platform is established consisting of the integral severe accident code ASTEC, the atmospheric dispersion and decision support tool JRODOS to predict the transport and release of the ST during a severe accident, and the radiological consequences of the released ST-inventory to the environment. It also includes the codes for the Fission Product inventory of realistic VVER-1000 core loadings such as KORIGEN, and ORIGEN. In addition, uncertainty quantification and sensitivity analysis (U&S) of the ASTEC and JRODOS codes is performed by using UQ-tool named KATUSA. With this aim, physical models of ASTEC code to simulate in-vessel phase of a severe accident are validated by using VVER-specific test data such as the QUENCH-12 experiment. The prediction capability of the ASTEC code simulating VVER-specific in-vessel severe accident phenomena is evaluated by comparing the calculation results with the experimental data related to physical models describing main key phenomena of accident progression such as oxidation, heat transfer and oxide scale growth. Based on the validation work, it can be stated that the ASTEC-code is able to predict most relevant in-vessel severe accident phenomena e.g., oxidation, hydrogen generation, oxide layers of the fuel rods for LOCA-conditions including the quenching phase. The underestimation of hydrogen production during quenching phase is caused by missing models for oxidation of VVER related material. The VVER-1000 plant is modelled by ASTEC including all the primary and secondary circuit components, containment rooms, and active and passive safety systems to be able to simulate progression of severe accident scenarios and release of FPs to the environment. Risk-relevant scenarios that lead to core damage and ST release were identified. The Large Break Loss of Coolant Accident (LBLOCA) on cold and hot leg coincident with the Station Blackout (SBO) were selected and this scenario was simulated until the rupture of the reactor cavity. Following ASTEC simulation, JRODOS analysis is performed for selected sites with selected season to predict the dispersion of the radiological inventory to the environment. Finally, KATUSA tool is used to quantify uncertainties and determine the most sensitive variables within the selectedvi parameters impacting the source term prediction. Based on the severe accident results obtained with ASTEC, following findings are highlighted: The generated hydrogen mass is about 790 kg when the break is located on cold leg. The highest hydrogen mass of about 1270 kg is predicted in case of a LBLOCA on the hot leg due to slower progression of severe accident. The activity released to the environment is about 1.2x10${}^{18}$ Bq at the end of simulation in case of LBLOCA on cold leg. The retention levels are significantly high due to retention on the walls of steam generators. On contrary, 7.0x10${}^{18}$ Bq activity is released when break is located on hot leg since the pathway to the containment is significantly shorter than the one break on the cold leg. Based on the subsequent prediction of the radiological impact using the JRODOS code, the following statements can be made: Analysis shows that the contamination reaches about 22.9 MBq/km${}^2$, resulting with 1.38x10${}^3$ mSv annual effective dose for an adult from all pathways with the predicted ST for the case of LBLOCA on cold leg. Higher ST prediction in case of LBLOCA on hot leg results with 11.5 GBq/km${}^2$ aerosol deposition and 1.75x10${}^5$ mSv annual effective dose for an adult. Based on the performed quantification of the Uncertainties of JRODOS using KATUSA it can be stated that the maximum dose prediction is about 2.9x10${}^{18}$ Bq for cold leg break and 1.30x10${}^{19}$ Bq for hot leg break. In both cases, the shape factor relative to Stokes velocity is the dominating factor on the results which the change of this parameter impacts the gravitational settling of aerosols. Finally, the difference on the contamination of Cs-137 isotope is about 3.2 MBq/km${}^2$ for the cold leg break scenario whereas this difference is about 0.5 TBq/km${}^2$ in case of hot leg break based on the radiological impact analysis using best-estimate and worst-case ST inventories. The established simulation platform consisting of the tools such as ORIGEN, ASTEC, JRODOS and KATUSA was essential to predict accurate, reliable and fast information about the progress of a severe accident, the potential radiological consequences and the range of radiological dispersion for any nuclear site, any accident scenario and any meteorological condition. That kind of information are of paramount importance for regulators and early response teams in order to develop plans to minimize the radiological consequences of severe accidents and for the implementation of effective and efficient measures to protect the peoplevii and environment. In addition, this dissertation demonstrates the unique and extended capability of the simulation platform capable to evaluate the consequences of a severe accident from the early in-vessel core degradation phase to the dispersion of ST around the nuclear power plant considering a realistic nuclide inventory of any burnt core loading.