A Second-Order Stochastic Dominance-Based Risk-Averse Strategy for Self-Scheduling of a Virtual Energy Hub in Multiple Energy Markets

Integrated energy systems are considered a practical solution to fulfill low-carbon energy systems. Accordingly, the concept of a virtual energy hub (VEH) and its capability to participate in different energy markets have attracted significant attention recently. In this regard, the self-scheduling problem of VEH, including a wide variety of uncertainties capable of participating in multiple energy markets, is addressed in this paper. To this end, a two-stage stochastic optimization has been implemented to solve the scheduling problem of a VEH equipped with renewable energy resources as well as conventional units as internal suppliers, different types of energy storage systems, hydrogen vehicles (HVs), and electric vehicles (EVs) in the intelligent parking lot (IPL). The studied VEH can participate in gas and hydrogen markets as well as day-ahead (DA) and real-time (RT) power and heat markets. The impact of flexible units, including energy storage systems and demand response programs, on the expected profit of the VEH is investigated accurately. Based on the obtained results employing a battery energy storage system (BESS), thermal energy storage system (TESS), hydrogen energy storage system (HESS), and cooling energy storage system (CESS) increases the profit of VEH by 0.88%, 0.62%, 1.5%, and 0.64%, respectively. Also, the profit of VEH can be increased by 1.02% and 0.25% by applying the electrical demand response program (EDRP) and thermal demand response program (TDRP), respectively. As risk management is critical for the participation of VEH in multiple energy markets, second-order-stochastic-dominance (SOSD) constraints are imposed on the scheduling problem instead of employing typical risk measures such as conditional value-at-risk (CVaR). Although the proposed risk-management method can shape optimal profit distribution based on the operators’ attitude toward risk, benchmark selection is the main obstacle to the mentioned approach. To this end, the CVaR-based benchmark selection method is applied to overcome the stated obstacle and guarantee the problem’s feasibility.