Abstract
The high energy demand driven by industrial development has transformed the power system from a single energy source to multiple energy systems (MESs). These systems, which involve thermal generators, combined heat-and-power (CHP) units, electrolyzers, fuel cells, etc., with realistic forecast uncertainty, are very operationally challenged. This paper proposes a Distributionally Robust Optimization (DRO) based on a Wasserstein-metric ambiguity set, which simultaneously optimizes the annual maintenance schedules and short-term operational dispatch across MESs. The ambiguity set is constructed using joint samples of forecast errors for the three carriers' demand, allowing for a data-driven worst-case distribution approach that mitigates the excessive conservatism typically associated with conventional robust optimization (CRO). The penalties are explicitly enforced for load and renewable energy curtailments across each of the MESs with source-specific value-of-lost-load coefficients. The Wasserstein radius is improved by sensitivity analysis, obtaining a theta value of 0.20 as the cost reduction radius for a 40% RES penetration. Five RES penetration levels are implemented here on the IEEE 39-bus New England network, with CHP, electrolyzer, fuel cell, thermal storage, and hydrogen storage. The DRO reduces the total annual system cost by 56% compared to CRO, while reducing the unbalanced energy.