Sannolikhetsbaserade dammhaveriberakningar i fyllningsdammar

Dam breach modelling is associated with significant uncertainties due to the complex and often unpredictable nature of breach development. This thesis explores the application of probabilistic methods to simulate breach scenarios for embankment dams, aiming to assess how these methods can improve risk assessments and decision-making within dam safety evaluations. Using Monte Carlo simulations in the two software Breacher and Embrea, two dam failures were analyzed. For each case, many stochastic simulations were conducted, varying dam breach parameters such as breach bottom width and breach time within defined intervals. The analysis focused on the resulting probabilistic distribution of outflow hydrographs and peak flows, as well as identifying key sensitivities and uncertainties. The results show that probabilistic methods can help to visualize uncertainty through percentile-based analyses and distribution plots. This supports a more transparent assessment of possible outcomes compared to traditional deterministic methods. It was further found that the choice of probability distribution has limited influence on results, while the assumed parameter intervals are critical. Limitations were mainly linked to the practical usability of the software and the inability to integrate stochastic output into hydrodynamic flood models. Nevertheless, the findings highlight a strong potential for probabilistic methods to complement and enhance traditional dam breach assessments by enabling better risk communication, parameter sensitivity evaluation and uncertainty quantification.