Stochastic real-time dispatch considering AGC and electric-gas dynamic interaction: Fine-grained modeling and noniterative decentralized solutions

In an integrated electric and gas system (IEGS) with high penetration of renewable energy sources (RESs), the minute-level RES fluctuation in the electric power system (EPS), and the uncertainty propagation between the EPS and the natural gas system (NGS) caused by the gas-fired units (GFUs) participating in the automatic generation control (AGC) to balance the RES fluctuations, have posed a challenge for the real-time dispatch security of IEGSs. To address these problems, this paper first presents a multistage stochastic real-time economic dispatch (RTED) model with a fine-grained resolution that considers the EPS-side AGC, the NGS-side dynamic physics of pipelines, and the uncertainty propagation between the EPS and NGS. To solve this large-scale RTED model for IEGSs, a novel notion of the coupling boundary dynamic adjustment region is proposed, based on which a noniterative decentralized solution is developed. To solve the EPS-side subproblem with multistage stochastic optimization (MSSO) structure, a robust sample path-based uncertainty model is established to describe uncertain RES series, based on which the complicated MSSO can be reformulated into a linear robust optimization with optimality guarantee. Case studies demonstrate that the fine-grained RTED model can effectively reduce the risk of EPS-side reserve shortages and NGS-side pressure violations. Additionally, with only a 0.25% optimal cost increase compared to that of the centralized solution, the computational time of the proposed decentralized solution is only approximately 4% of that of the centralized solution, and approximately 7% of that of the conventional alternating direction method of multipliers.