A degradation-based stochastic optimization framework for inspection and maintenance in marine energy systems

Marine energy (ME) is a renewable energy with abundant resources, potential flagship of the blue economy, and key integral to achieving sustainable development goals (SDG) of the United Nations (UN). Despite the promise of ME to ensure these goals, it has not been realized yet due to significant industrial adoption barriers caused by inefficiencies in operation and maintenance (O&M) policies. O&M of ME systems has unique dynamics separating itself from other energy systems, which are well-studied in the literature. While investments in ME assets are constantly increasing, there is limited research on making the best of these assets. This study presents an integrated framework for inspection and maintenance (I&M) designed to enhance the utilization and reliability of the assets in ME systems. It combines (i) predictive analytics methodology that uses real-time sensor data to predict future degradation levels, with (ii) a novel I&M optimization model that takes into account the uncertainties associated with the degradation levels of ME assets. The objective is to minimize costs arising from maintenance, inspection, and crew routing while maximizing asset availability and utilization of asset lifetime. The proposed framework employs observation errors to capture uncertainty in asset degradation and a unique chance constraint approach to limit failure risks associated with those errors. Chance constraints are embedded in the model by the use of sample average approximation. The model makes use of constraints to control maintenance and failure statuses and coordinates them with crew routing decisions so that opportunistic I&M schedules are situated. Key contributions of the proposed framework are the integration of maintenance and inspection decisions, crew routing (affected by the accessibility conditions), and chance constraints. The tradeoff between failure risks, availability, and lifetime utilization is modeled as a function of uncertainties driven by observation errors and degradation trajectories. A comprehensive set of experiments is conducted using a representative ME system with 20 assets to demonstrate the advantages of the proposed framework over traditional policies. This study aligns with UN’s SDG 7 (Affordable and Clean Energy) by advancing ME systems through optimal I&M, directly contributing to cleaner, more reliable energy sources, SDG 13 (Climate Action) by promoting renewable energy adoption, and SDG 9 (Industry, Innovation, and Infrastructure) through innovative predictive analytics and optimization frameworks.