Prognostics and Health Management (PHM) technologies have been introduced into wind turbines to forecast the Remaining Useful Life (RUL), and enable predictive maintenance opportunities prior to failure thus avoiding corrective maintenance that may be expensive and cause long downtimes. For a wind turbine, when an RUL is predicted, a predictive maintenance option is triggered that the maintenance decision-maker has the managerial flexibility to decide if and when to exercise before the turbine fails. By implementing the predictive maintenance, the high cost of corrective maintenance can be avoided; however a portion of the RUL will be thrown away that can be translated into cumulative revenue loss.

In this dissertation, a simulation-based European-style Real Options Analysis (ROA) approach is used to schedule the predictive maintenance for a single wind turbine with an RUL prediction managed using an as-delivered payment model. When an RUL is predicted for the wind turbine, the predictive maintenance value paths are

simulated by considering the uncertainties in the RUL prediction and wind speeds. By valuating the European-style predictive maintenance option at all possible predictive maintenance opportunities, a series of predictive maintenance option values can be obtained, and the predictive maintenance opportunity with the highest expected predictive maintenance option value can be selected.

By extending the approach for a single wind turbine, a wind farm managed using an outcome-based contract, specifically a Power Purchase Agreement (PPA), with multiple turbines indicating RULs concurrently can be analyzed. The predictive maintenance value for each wind turbine with an RUL indication depends on the operational state of all the other turbines, the amount of energy delivered, and the energy delivery target, prices and penalization mechanism for under-delivery defined in the PPA. A case study is provided demonstrating that the selected predictive maintenance opportunity for a PPA-managed wind farm is different from the same wind farm managed using an as-delivered payment model, and also differs from the selected predictive maintenance opportunities for the individual turbines with RULs managed in isolation.

Finally, the magnitude of the life-cycle benefit that the developed approach can bring to the wind farm owner is estimated through a simple case study. Using the European-style ROA approach to determine the wind farm maintenance policy, the improvement to the wind farm expected life-cycle net revenue is significant compared with the state-of-art wind farm maintenance policies, i.e., up to 25% higher than the corrective maintenance policy, and up to 83% higher than the predictive maintenance at the earliest opportunity policy.