Forecasting for day-ahead offshore maintenance scheduling under uncertainty

Offshore wind farm maintenance operations are complex and dangerous, and as such are subject to strict safety constraints. In addition, crew and vessels must be scheduled in advance for both planned and reactive maintenance operations. Meteorological forecasts on many time-scales are used to inform scheduling decisions, but are imperfect. Short-term maintenance scheduling is therefore a problem of decision-making under uncertainty. This paper proposes a probabilistic approach to the short-term scheduling problem based on a cost-loss model for individual maintenance missions, and probabilistic forecasts of appropriate access windows. This approach is found to increase the utilisation of possible access windows compared to using deterministic decision rules. The impact of forecasting on the availability and operational costs of offshore wind is then examined using a Monte Carlo-based cost model. This has quantified the impact on availability and revenue performance under a range of site conditions.