Posters

Abstract

The continued expansion of offshore wind capacity has increased operational complexity and intensified pressure to control maintenance-related costs. Offshore wind turbines operate in demanding marine conditions that increases the rate at which component wear out while limiting physical access due to weather conditions. Under these conditions, traditional maintenance approaches—whether reactive or time-based are often inefficient and costly, contributing significantly to overall operational expenditure. As offshore projects scale in size and distance from shore, alternative maintenance strategies are increasingly required.  This study examines the role of Artificial Intelligence (AI) predictive maintenance (PdM) in addressing these challenges. Drawing on a systematic review of 85 peer-reviewed publications published between 2020 and 2025, the research evaluates how machine learning techniques applied to SCADA, condition monitoring, and environmental data are being used to detect early-stage faults and support maintenance decision-making in offshore wind operations.   The findings indicate that AI-driven PdM can reduce unplanned outages, improve asset availability, and extend component life through earlier and more accurate fault detection. Given that operations and maintenance costs are estimated to account for approximately 25–40% of the levelised cost of energy for offshore wind projects, these improvements have the potential to deliver meaningful reductions in LCOE and enhance long-term project returns.   Several barriers to effective implementation are identified. From a technical perspective, data quality issues, and limited transparency in complex algorithms remain persistent challenges. Organisational constraints include skills shortages, fragmented responsibilities between digital and engineering teams, and limited executive understanding of AI capabilities and limitations. In addition, cyber security risks and the energy intensity of model development raise emerging concerns around system sustainability.  Based on these findings, the paper outlines a set of practical enablers for AI adoption in offshore wind maintenance. These include the use of targeted pilot deployments, closer integration between operations and data teams, and the establishment of governance mechanisms to support explainability, trust, and regulatory alignment.  The study concludes that AI-based predictive maintenance has the potential to play a significant role in improving the reliability and cost-efficiency of offshore wind assets. Realising this potential, however, requires alignment between technology, organisational capability, and operational strategy. The paper contributes an industry-focused perspective to current discussions on digitalisation in offshore wind and offers guidance for operators seeking to implement these solutions.