Utilities' Managed Home Charging Programs for Electric Vehicles

Abstract

Experts estimate 20 million electric vehicles will be on U.S. roads by 2030, and the majority (around 80%) of the electric vehicle drivers will use home charging. Many utilities are designing managed home charging programs to centrally manage charging times to reduce cost, avoid new and aggravated peaks and blackouts, and ensure grid stability. These managed home charging programs are either active, in which the utility continuously controls the charging while the vehicle is plugged in, or passive, in which the participants decide when to charge based on pre-announced low-rate episodes. We study jointly designing and executing these active and passive programs. We present a program-design model, which produces a menu of the charging programs, tailored for each driver type, and a load-management model, which dynamically manages the load supply to each individual participant. The load-management model consists of a large number of nonhomogeneous participants, and it is a large-scale mixed-integer nonlinear stochastic problem. We present an effective approximation method, conduct thorough theoretical and numerical analyses of our approximation, and provide worst-case bounds for its error components. Our methodology provides detailed insights on the amount and timing of the improvements achievable in cost and demand variability by offering managed home charging programs, and by customizing the passive programs. It also offers detailed insights on the significance of the trade-off between cost and demand variability. We find promoting a culture of charging electric vehicles every night may significantly increase utilities' total cost if passive programs have high participation levels.

Keywords: electric vehicles, managed charging programs, home charging, large-scale optimization, approximation, error analysis