Developing a Logistic Regression Method for Valuation of Grid-Level Energy Storage Systems

Abstract

Logistic regression models can serve as important tools in developing a framework to establish the value of electrical energy storage systems (ESSs). This study provides models that aggregate use-case scenarios of five battery types, as well as pumped hydro-electric storage systems. The grid applications include: bulk energy at generation, auxiliary services at transmission and distribution, and end-use customer services at distributed generation. The data is derived from 1,261 real world systems. Five different models were developed for short, medium, and long-duration grid services. The models are designed to be technology agnostic and are not sensitive to either performance characteristics or operating conditions of the ESS. The results indicate the probability that an energy storage project will provide an individual service use case given that it may also yield another service, and how technology types and multiple selected applications influence those probabilities.