Maritime energy projects compete for capital and installation time, while their benefits depend on operating conditions and implementation performance. This study develops a digital-twin-oriented decision framework that converts updated vessel and economic states into a risk-aware intervention portfolio.
Methods: A mixed-integer mean–conditional-value-at-risk model is evaluated using a transparent synthetic benchmark. Published evidence informs only selected intervention-effect ranges; capital costs, installation durations, remaining effect ranges, economic distributions, and risk preferences are explicitly treated as experimental design assumptions. The model is trained on 1,000 scenarios and evaluated on 20,000 independent common test scenarios. Bootstrap confidence intervals and 30 independent training seeds are used to assess uncertainty and selection stability.
Results: Expected-value optimization selects HVAC optimization, voyage optimization, and waste-heat recovery, achieving a mean five-year net present value of EUR 908.5 thousand and a fifth-percentile value of EUR 111.2 thousand. With a conditional-value-at-risk weight of 0.80, the model selects the sensor/data layer, HVAC optimization, voyage optimization, and advanced energy management. Mean net present value decreases to EUR 792.4 thousand, while the fifth percentile increases to EUR 144.5 thousand and loss probability declines from 2.02% to 0.86%.
Conclusions: Explicit tail protection changes the selected intervention portfolio and improves downside performance at a measurable expected-value cost. The results demonstrate model behavior within a reproducible synthetic benchmark, not a vessel-specific investment case or an operational digital twin.