Optimization of maintenance and operational policies of an offshore wind farm subject to stochastic wind conditions

Offshore wind power has recently become a highly sought after source of sustainable energy. It consists of the generation of electricity through wind turbine farms located in bodies of water off the coast. While offshore wind farms have the advantage of greater wind availability and not taking up land space, they are also challenging to maintain do to their harsh environment and accessibility issues. We propose a Markov decision process to determine the optimal maintenance and operational policy of an offshore wind farm. The maintenance decision determines which subset of the wind turbines to maintain while the operation decision yields the optimal braking speed of the wind turbine in order to reduce deterioration on the system. Our model takes into account the stochastic wind and weather conditions through both the current conditions and a longer-term forecast in order to reduce the risk of taking wind turbines offline during high production periods. Furthermore, we model the accessibility and maintenance resource issues through a fixed maintenance duration period. The deterioration of the wind turbines is modeled via a stationary gamma process which eliminates the need for a complex physical model or a large amount of historical data. A numerical example is presented to highlight the effectiveness of our model.