Resilience Aware Digital Twin Co-Optimisation of Off-shore Wind Spare Parts, Vessel Scheduling and Hybrid Energy Storage

Abstract

Offshore wind farms are capital-intensive assets and remote, weather-constrained logistics systems whose availability depends jointly on component degradation, spare-part availability, vessel access, and export capacity. Existing studies typically address these dimensions separately, with vessel-routing models assuming fixed spare availability, inventory models using exogenous lead times, and storage-sizing models assuming fixed asset availability. This separation can underestimate costs and overestimate resilience under increasingly severe weather. This study develops an integrated, resilience-aware framework that co-optimizes a two-echelon spare-parts system, condition-based maintenance triggered by a digital-twin remaining-useful-life estimator, heterogeneous vessel scheduling under stochastic metocean access windows, and hybrid battery–electrolyser buffer sizing and dispatch coupled with a voltage-source converter high-voltage direct current export link. Component degradation follows a gamma process with Bayesian drift updating, while significant wave height follows a Markov-modulated Weibull process with storm severity-adjusted regime intensities. The model is formulated as a two-stage distributionally robust mixed-integer program using a Wasserstein ambiguity set and a Conditional Value at Risk objective for energy not served. Analytical results establish convexity of expected cost in base-stock level, monotonicity of the optimal maintenance trigger with storm severity, and sub modularity of the resilience index in storage design. An accelerated integer L-shaped decomposition achieves a 0.8% optimality gap. For a 1.2-GW, 100-turbine farm, joint optimization reduces annual O&M cost from €4.90 million to €1.79 million, increases resilience from 0.41 to 0.87, and reduces 95% CVaR of energy not served by 44.6%.