Data-driven multi-stage distributionally robust scheduling for coupled electricity-hydrogen-refinery systems

Green hydrogen production via electrolysis has fueled decarbonization for refineries, revealing a promising roadmap towards an electricity-hydrogen-refinery integration. This paper proposes a novel multi-stage distributionally robust scheduling framework for a coupled electricity-hydrogen-refinery system under renewable energy uncertainty. Power systems, hydrogen devices, and process units are seamlessly integrated to facilitate the comprehensive optimization of the refinery, achieving both economic and sustainable benefits. To effectively hedge against the high-dimensional uncertainty arising from scheduling-stage proliferation and uncertainty-type multiplicity, we develop an innovative structured moment-Wasserstein-based ambiguity set, along with theoretical set-inclusion relationships and probabilistic guarantees. Based on this ambiguity set, the refinery scheduling problem is then formulated to achieve robust optimal dispatch strategies under a given production target. In this formulation, the variables for refinery processes are considered as here-and-now decisions due to inflexibility and production requirements, while electricity and hydrogen dispatches are treated as multi-stage recourse decisions for flexible regulation. To efficiently solve the resulting scheduling problem, we exploit a lifted decision rule, based on which an equivalent mixed-integer linear programming reformulation is derived. Case studies demonstrate the effectiveness and superiority of the proposed approach compared with state-of-the-art decision-making methods. The electricity-hydrogen-refinery coupled scheduling scheme exhibits significant advantages over the conventional scheduling scheme without distributed hydrogen generation, achieving a 3.77 % reduction in scheduling costs and a 20.59 % decrease in carbon emissions.