The nonlinear effect of combining uncertainties on the energy yield of an offshore wind farm

Offshore wind energy is expected to grow in the coming years: future plans for offshore wind farms total more than 98 GW. Financing is needed to realise these plans. Investment decisions partly depend on the uncertainty in energy yield predictions. It is therefore important that these energy yield predictions and their corresponding uncertainty are determined as accurately as possible. Current methods for determining annual energy production assume that there is a linear relation between input uncertainties and output uncertainty, allowing the use of simple methods for determining annual energy production and its uncertainty. It is however known that this assumption is incorrect: nonlinear relations do exist. This means that it is unclear whether the use of these simple methods can be justified. This thesis has developed a methodology that can be used to determine if, and how, the nonlinear effect of combining two uncertainty sources should be incorporated in the energy yield prediction. This has been done by investigating the case study of the nonlinear effect of combining array efficiency and availability. The investigation was split up in five main steps. First, the physical relations between array efficiency and availability have been explored, revealing that downtime of a turbine affects the array efficiency of a wind farm. Figuring out the state-of-the-art methods pointed out that this interdependency is currently not taken into account. By adjusting the current models, an adapted model was developed that is able to consider this effect. Simulations have been performed on both the current and the adapted model. The results show that for a typical mean availability value of 96.2%, the differences between the current and the adapted model are smaller than 1%. This means that the current methods used by industry can be justified. However, if availability values drop, the difference between both models becomes significant. Due to the high development and computational time of the adapted model, an approximation of the mean annual energy production was developed that uses existing tools. This approximation yields accurate results: the difference between the approximation and the results of the adapted model is lower than 0.5%. Since the approach that was used in this research has proven to be successful, it can be translated to a generic methodology. This methodology can be followed to determine the nonlinear effect of two other uncertainty sources.Wind EnergyAerospace Engineerin