This master thesis examines the possibility of performing asimplified uncertainty analysis on a probabilistic safety assessment(PSA) of the Oskarshamn 3 nuclear power plant. The aim of the thesiswas divided in two parts, first to examine the uncertainty parametersof a PSA-model for Oskarshamn 3 and the second part was to developand examine a simplified method of uncertainty analysis as comparedto a more regular method of Monte Carlo-simulation. The thesis ismostly concerned with examining the core damage frequency. Theexamination of uncertainty parameters showed that many parameterswere missing from the model and thus further investigation areneeded, if a full Monte Carlo is to be performed. The simplifiedmethod for uncertainty analysis that was developed consisted ofassuming a lognormal distribution for the frequency of basic eventsand then using the minimal cutset-list to calculate an approximationto the end distribution. The simplified method was then compared tothe Monte Carlo-analysis for Oskarshamn 2 for different MCS-lists anda preliminary uncertainty analysis was performed for Oskarshamn 3.

The scope of this project is to investigate if the software Ecolego is suitable for creating models in accordance with level 3 PSA, with the goal of creating a model calculating the possible outcomes for the radiological impact at 20 km distance resulting from a fictional severe nuclear accident at the Forsmark Nuclear Power Plant. This report aims to answer the questions “What is the concentration on the ground (kBq/m2), at a distance of 20 km, 30 days and 10 years after the fictional severe nuclear accident, with and without filtered venting” and “Is Ecolego a suitable software for level 3 PSA models?” The source term for the fictional severe nuclear accident is made to resemble the actual source term from the Fukushima Daiichi accident including the radionuclides Cs-134, Cs-136, Cs-137, I-131, I-132, I-133 and Te-132. In the model, three source terms are created and tested. Two source terms in which the total emissions are released in 24 hours, for which one contains the total emissions from the Fukushima Daiichi accident and one containing 1 percent of the emissions from the Fukushima Daiichi accident due to reduction of emissions when passing the filtered venting. The thirdsource term is made to resemble the time-dependent emissions for the Fukushima Daiichi accident, with emissions varying in intensity over 50 days. The transport of the radionuclide concentration is only due to atmospheric dispersion in the model, for which the Gaussian Plume Model (GPM) is applied under undisturbed condition where only the concentrations in the centerline of the plume are calculated. Probabilistic variation is performed with Monte Carlo simulations where probability density functions (PDFs) are assigned to wind speed andprecipitation, with 5000 iterations. Simulation of the model with one of the two source terms which has the release period of 24 hours gives reasonable results. However, to run the simulation with the time varying source term the model needs to be developed to generate reasonable results, for which necessary development actions are presented. The calculations of the resulting concentration on the ground 30 days and 10 years after the fictional accident shows that there are multiple possible outcomes, which makes it impossible to give a single answer to the expected concentration on the ground. Due to their short half-lives, there will be no concentrations of I-132 and I-133 on the ground at the distance 20 km after 30 days or 10 years. For the remaining radionuclides, the ranges of the possible outcomes for the concentration on ground are presented. It is concluded that Ecolego is suitable for PSA level 3 with the risk metric of environmental impact based on the results of the investigation. However, due to the time limit of the investigation, the applicability of creating an Ecolego model with the risk metrics health effects and economic impact are not investigated. But the report discusses suitable development of the model to contain the risk metric health effects in accordance with level 3 PSA. With such development the conclusion is that Ecolego is suitable for level 3 PSA.

Nuclear power is an important part of Sweden's energy system and contributes with about 30 % of the supplied electric energy. Interest in new construction is currently high and one type of reactor that may be built is small modular reactors, SMRs. A method that is used to evaluate the safety of a traditional nuclear power plant is the so called probabilistic safety assessment (PSA). An important question is if PSA is applicable to SMRs. This thesis examines differences between a BWRX-300 type SMR and a generic boiling water reactor to investigate if the traditional PSA method is applicable to SMRs and what characteristics that may affect the results.  To investigate the topic, a simplified PSA analysis is carried out at level 1 for five different initiating events: extreme snowfall, loss of cooling accident, loss of main feedwater, loss of offsite power supply and transient. These are then compared to a model of a generic BWR reactor.  The conclusions that emerge from the analysis are that the traditional PSA method can be applied for SMRs of the BWRX-300 type. The features that distinguish the BWRX-300 compared to a generic BWR are mainly that the SMR is smaller in size and power and the influence of the passive safety systems. This results in a lower core damage frequency. Some the areas where the PSA method could be developed for SMRs are the analysis of passive components and passive functions.