Title: How much energy storage do modern power systems need?

Abstract: The central question we seek to address in this paper is: How rapidly do the operating cost benefits of grid-scale energy storage decline as installed storage capacity increases? We use a 240-bus model based on the US Western Interconnection, first optimally locating storage in the network and then dispatching it in a unit commitment model with DC load flow. The model uses storage to provide frequency regulation, load following, and arbitrage for each hour of a study year, and we investigate a range of scenarios for fuel price and renewables penetration. We find that value from long-term energy shifting is negligible at all penetrations we investigate, but also that displacing fossil-fueled generators from providing reserves is initially very valuable. However, in most scenarios the value is negligible beyond 10 GWh of storage, or the equivalent of roughly 6 minutes of average demand in the system. Above penetrations of 4-8 GWh, storage operating cost benefits are less than estimated capacity values for storage. We also show that storage has the potential to increase overall carbon emissions in the electricity sector, even when it is not providing significant amounts of arbitrage and is preferentially providing regulation and load following services.