This paper studies how short‑term uncertainty about electricity demand changes the way grid batteries are used and how those operating choices feed back into long‑term capacity decisions. The authors show that when future demand is uncertain, storage systems follow a “precautionary” policy: they deliberately keep some energy in reserve to hedge against possible high‑demand events. That behavior changes the distribution of demand that the rest of the system sees after storage acts, and it interacts with market rules in ways that can distort investment incentives.

To reach these conclusions the researchers built a simple but explicit model of both operation and planning. They represent uncertain net demand with a mean‑reverting stochastic process called an Ornstein–Uhlenbeck (OU) model, then discretize it into demand states. Storage operation is cast as an average‑cost Markov decision process (MDP), which is a standard way to find a steady operating rule that minimizes long‑run average costs. The operational model includes realistic features such as stored‑energy capacity E, power limits Ks, round‑trip efficiency η, constraints that prevent charging and discharging at the same time, and an explicit cost for any unserved load (the value of lost load).

The key technical move is to take the stationary operating policy that comes out of the MDP and embed it inside a stylized capacity expansion framework. That lets the authors trace how short‑run uncertainty and precautionary operation change the long‑run equilibrium mix of storage and conventional generation. They find that uncertainty makes storage reserve energy for possible future scarcity, so the pattern of net demand seen by generators after storage acts is materially different than it would be under perfect foresight. They also show that the “reliability externality” in electricity markets — the fact that system reliability is a public good and short‑run prices are often capped — interacts with this precautionary behavior and leads to qualitatively different distortions in both operation and investment for storage than for conventional generators.