Time-Series Modeling of Aggregated Electric Vehicle Charging Station Load

Abstract

The widespread proliferation of Electric Vehicles (EVs) can have a transformative effect on the electric power system. The power and energy consumed by EVs when charging is substantial, which has consequences on power system operation and planning. This paper identifies, evaluates, and proposes time-series seasonal autoregressive integrated moving average (ARIMA) models of aggregated EV charging station load. The modeling is based on 2 years of time-stamped aggregate power consumption from over 2400 charging stations in Washington State and San Diego, California. The different data sets allow the influence of time-of-use pricing on the time-series models to be explored. Weekday, weekend, and near-term and long-term models are developed and analyzed. The best performing near-term weekday models are (2, 0, 0) × (0, 1, 1)24 × (1, 0, 0)120 for Washington State and (2, 0, 0) × (1, 1, 0)24 × (0, 0, 1)48 for San Diego. Applications of the seasonal ARIMA models to aggregate EV charging station load forecasting and creation of synthetic time-series at different penetration levels are discussed.