Presentations

Abstract

Offshore wind farms face complex maintenance scheduling challenges that have a significant impact on their operational profitability. Operators must continuously balance maximising power production through high turbine availability while performing necessary maintenance activities, taking into account multiple dynamic constraints such as weather windows, vessel availability, spare parts logistics and maintenance costs. Traditional planning methods are often using an intuitively evaluation of the lost production based on the turbine locations, this study presents a new maintenance planning tool that quantifies effective lost production by including the wake effects in the optimisation. The optimisation tool can help wind farm operators to optimise their maintenance activities, focusing on two critical aspects: maintenance slot selection and team route optimisation.  The maintenance slot selection process integrates weather forecasts that include relevant meteorological properties such as wind speed, direction, significant wave height and air density. The inclusion of a wake model allows to take into account the mutual influence of turbines. While conventional wisdom might suggest prioritising maintenance on downwind turbines to minimise production losses, a simple analysis reveals a more nuanced reality. Shutting down upwind turbines can sometimes be beneficial due to the compensating increase in production from downstream turbines that would otherwise operate in their wake. The optimisation algorithm combines high-resolution wind forecasting with wake modelling to quantify these complex trade-offs. In addition, the system incorporates wave height constraints as a critical safety parameter that can prevent maintenance activities, thus ensuring operational safety while optimising maintenance scheduling.  The route optimisation addresses the challenge of minimising travel time and associated costs in three key phases: initial deployment from the port or Service Operation Vessel (SOV) to the first maintenance destination, inter-turbine movements and return logistics. As maintenance activities often extend over several days, turbines are typically put back in operations between work periods to minimise production loss. This operational requirement creates significant discontinuities in the cost function, as the available wind energy can vary significantly between consecutive days. Our analysis showed that traditional local optimisation methods are inadequate for this complex problem space.  After evaluation of several optimisation approaches, a specialised evolutionary algorithm that efficiently navigates the discontinuous solution space to identify globally optimal solutions has been implemented. This choice was driven by the algorithm's ability to maintain solution diversity while dealing with complex, non-linear constraints. Moreover, the algorithm is known to be robust in dealing with the multi-objective nature of the optimisation problem, balancing production losses against maintenance efficiency.  The solution was implemented as an interactive Python-based tool with a simple user-friendly interface that allows maintenance planners to visualise, optimise and export their scheduling decisions into standardized formats. The proof-of-concept tool is currently used by wind farm operators, providing data-driven support for daily maintenance scheduling decisions. Initial deployment results show significant improvements in maintenance efficiency and reduced production losses compared to previous scheduling methods.