Condition-Based Maintenance in Offshore Wind Turbines: Modeling Fault Progression, Detection Latency, and Time-to-Repair Under Environmental and Sensor Uncertainty

Abstract

This article presents a reliability-centered condition-based maintenance framework for offshore wind turbines that explicitly models how uncertainty propagates through monitoring, diagnostics, and maintenance scheduling to determine time-to-decision, probability of missed detection before functional failure, probability of opportunistic repair within access windows, and expected energy production loss. A quantitative scenario-based study is developed for drivetrain and power conversion subsystems, comparing four maintenance architectures that span threshold-based monitoring, model-based diagnostics with redundancy, risk-based maintenance scheduling with spares governance, and a two-tier verification architecture that constrains nuisance alarms while enabling staged intervention. Results show that (i) fleet availability is dominated by the upper tail of time-to-repair rather than by mean time between failures, because weather and logistics amplify delays once a fault progresses beyond a controllable stage, (ii) moderate sensor bias and baseline drift can substantially increase false stability events and shift detection later into the fault progression curve, producing outsized downtime penalties even when average alarm rates appear acceptable, and (iii) governance that couples quantile-based alarm thresholds with verification and repair staging provides a superior cost–risk balance by reducing the probability of late detection without driving unsustainable nuisance maintenance. The study provides copy-ready tables and full prompts for data-driven figures suitable for Techne submission and adaptation to site-specific fleet data.