Title:On the Benefits of Stochastic Economic Dispatch in Real-Time Electricity Markets

Abstract:Independent system operators (ISOs) in the US clear their real-time electricity market every five minutes, optimizing energy and reserve dispatches to minimize operating costs while meeting technical and regulatory constraints. Currently, many real-time markets are cleared by solving a single-period, deterministic, security-constrained economic dispatch (SCED). Nevertheless, given the growing penetration of renewable generation and the forthcoming retirement of conventional generation units, it becomes increasingly challenging to manage operational uncertainty at the real-time stage via the current SCED formulations. This limitation is best illustrated by the recent introduction into the real-time market of multiple short-term ramping products, which aim at bridging the gap between deterministic and stochastic formulations. In contrast, this paper explores the scalability and potential benefits of explicitly considering uncertainty in real-time market formulations by combining multi-period look-ahead dispatch (LAD) and stochastic look-ahead (SLAD) formulations. An accelerated Benders' decomposition is presented to solve the resulting problems efficiently. The paper conducts extensive numerical experiments on a real, industry-size transmission grid that benchmarks the proposed approaches and evaluates their benefits. The results demonstrate that stochastic optimization is now tractable enough to be used in real-time markets. Furthermore, the combination of multi-period and stochastic look-ahead offers significant benefits in both reliability and cost, as SLAD can better position online generators to accommodate future ramping needs, thereby reducing future operational costs and violations. Overall, SLAD reduces import costs and the risk of transmission violation and saves an average of more than 2% of costs compared to SCED.