Stochastic Wind Park Modelling and Maintenance Scheduling under Uncertainty

Abstract

In this thesis, my main objective was to enable improved decision making for the scheduling of maintenance for offshore wind farms. The decisions depend on many factors which are subject to uncertainty. The factors were studied in this thesis and it was identified where variations and uncertainty should be accounted for. Subsequently, an in depth analysis of two factors was conducted. A set of indicators were defined, which can be used to monitor a wind farm’s performance in terms of production, reliability, maintenance, finances and safety. The presented indicators enable the comparison of different maintenance strategies. Weather was identified as an influential factor. Different methods for weather modeling have been presented and discussed as well as the influence of weather forecasts studied. The duration of maintenance actions - the repair time - was also identified as influential to the decision making. Multiple ways to incorporate the repair time into decision support models have been presented and discussed. The included case studies confirm the importance of including the variation of the repair times as an input to maintenance scheduling models. Following the identification and investigation of influential factors, a novel method for decision support has been presented. The included case study indicates a possible alternative maintenance strategy for future energy prices. The findings from the previous chapters were integrated into a serious game that can be both used as a teaching tool for newcomers to the wind industry and as a general outreach tool. The thesis concludes with recommendations for future work.