Posters

Abstract

As wind turbines (WTs) grow in complexity and scale, the need for accurate and predictive reliability models becomes increasingly important to optimise operations and maintenance (O&M) strategies. This research presents an advanced subsystem reliability modelling framework which describes the failure behaviour over a turbine’s operational lifetime for major WT components. The model is trained on real world maintenance data of over 1000 WTs covering over 4000 WT operating years. The reliability modelling approach is based on a Bayesian mixture model. The model captures the evolving risk profile of critical components over the turbine’s operational lifetime while accounting for the effect that technological development and design choices have on the failure behaviour. It does so by first evaluating the intensity of early, intrinsic and deterioration failures separately via Bayesian updating; then superimposing these separate models via a final Bayesian update. The output of the approach is expected subsystem failure rates per turbine per year.  To demonstrate the utility of the reliability modelling approach, we predict the failure behaviour of two use-case turbines and use the outputs as inputs to the OffshoreTIMES O&M simulation tool. We compare the resultant OPEX estimation to figures derived using average failure rates – which is by far a more common practice in the field. Based on the disparity in failure rate estimates over time compared to mean failure rates, we anticipate significant differences in the respective OPEX estimates.  This work therefore contributes to improved planning and operations at offshore wind farms. More accurate estimation of WT failure rates provides a better basis upon which to make a successful maintenance strategy.