Energy Storage in Electricity Markets: Equilibrium Models and Learning-Enabled Participation

Energy storage resources - particularly batteries - are rapidly becoming essential sources of flexibility in modern power systems and active participants in electricity markets. By arbitraging price differences, they generate profit while reducing peak demand and mitigating renewable variability. Yet, their effective market participation requires accounting for future price opportunities and inherent uncertainties. As a result, storage operators may rationally withhold capacity for economic reasons, rather than offering strictly based on physical cost. These dynamics create challenges both for storage operators seeking to optimize returns and for regulators aiming to prevent market power abuse and ensure socially efficient outcomes.

This talk presents game-theoretic equilibrium frameworks alongside optimization and machine learning approaches for enabling efficient energy storage participation in electricity markets. I will demonstrate that strategic storage participation converges to efficient market outcomes under the existing two-settlement U.S. market design. Building on this foundation, I will examine the energy storage arbitrage problem in detail, framing it within a dynamic programming perspective, and present solution strategies ranging from classical stochastic optimization to decision-focused learning pipelines.

Speaker: Bolun Xu, Stanford University

Room 102