Master Thesis Defense by Xaver Felix Kramer

Title: Assessing the Monetary Impact of Extreme Wind Gusts on Renewable Energy Assets in a Changing Climate

Abstract: 
This thesis evaluates the risks that extreme wind events pose to renewable energy infrastructure, focusing on solar power plants and wind turbines. A machine learning framework is developed to estimate wind gusts from climate model wind speed data. The model is trained on ERA5 reanalysis data by linking instantaneous wind speeds to the corresponding gusts. After training, the model is applied to two CMIP6 HighResMIP simulations. The climate change impacts are evaluated by comparing a historical baseline (1980–2010) with a high-emission future scenario (2015–2050). Extreme wind gusts are quantified using the Peak Over Threshold (POT) method, with a Generalized Pareto Distribution (GPD) fitted to the exceedances. This distribution is used to estimate 30-year return levels of wind gust speed at each grid cell. Model bias is evaluated by comparing historical return levels against ERA5 data. Risk at the asset level is assessed by combining the GPD derived probability functions with technology-specific vulnerability curves.

This approach enables the estimation of expected annual damage (EAD) as well as the potential damage associated with 30-year wind gust return levels. Projected changes in wind-related losses at this spatial resolution are primarily influenced by shifts in extratropical cyclone (ETCs) activity. Shifts in extreme wind gust hazards could be explained by a combination of changes in storm track density and wind speed. Local assessments highlight that relying solely on changes in 30-year return levels is inadequate for understanding evolving wind hazards. A full probabilistic analysis of the entire extreme wind distribution is necessary to account for variations in both the frequency and intensity of hazardous events. However, this analysis does not isolate individual events contributing to overall risk changes. Further research is needed to better understand the sources of bias, the driving mechanisms, and the magnitude of projected changes in extratropical cyclone activity within HighResMIP simulations.

Supervisors: Larissa Nora Van der Laan, Jens Hesselbjerg Christensen