Predictive Framework for Locality-Level Residential EV Load Growth

Anticipated growth in electric loads due to increased EV penetration at the residential level is a concern for distribution utility planners. Usually, EV growth predictions are made for larger geographical areas. However, for the utility to plan distribution infrastructure upgrades effectively, it becomes crucial to predict the EV load profile at a more granular level. Therefore, this paper adopts a top-down and bottom-up approach to present a framework that can predict residential EV load growth at the locality level within a city, primarily focusing on rising metropolises in the developing world. The framework utilizes data-driven regression models to predict the total number of EVs at the city level using high-level data such as environmental and economic indicators. Simultaneously, at the locality level, EV distribution within the city and charging patterns are predicted using low-level data such as local traffic data, rent prices, and income distribution through Monte Carlo simulations. These two approaches are combined to determine a probabilistic prediction of the EV load profile for the 33 localities of the city of New Delhi for the year 2035.