A scalable energy modeling framework for electric vehicles in regional transportation networks. (1st July 2020)

• Record Type: Journal Article
• Authors: Xu, Xiaodan; Aziz, H.M. Abdul; Liu, Haobing; Rodgers, Michael O.; Guensler, Randall
• Is Part Of: Applied energy. Volume 269(2020)
• Publication Date: 2020-07-01
• Subjects: Electric vehicle -- Transportation network -- Vehicle drivetrain simulation -- Bayesian Network -- Regional-scale energy prediction; Power (Mechanics) -- Periodicals; Energy conservation -- Periodicals; Energy conversion -- Periodicals; 621.042
• DOI: 10.1016/j.apenergy.2020.115095
• Languages: English
• ISSNs: 0306-2619

Abstract

Highlights: Activity-based model predicts electric vehicle energy usage in large-scale networks. A Bayesian Network method is used to estimate energy rates by operating conditions. The inference model closely reproduces full vehicle simulation but is 100× faster. The modeling approach was demonstrated in an Atlanta case study. The modeling approach is scalable and transferable to other regions. Abstract: Vehicle electrification plays a central role in reducing global energy use and greenhouse gas emissions. Predicting electric vehicle (EV) energy use for future transportation networks is critical for the planning, design, and operations of sustainable transportation systems. However, there is currently a lack of EV energy modeling approaches that are fully-scalable to large transportation network applications and consider actual on-road vehicle operating conditions. Such an approach is required for the accurate assessment of EV energy impact under various transportation scenarios. Here we present a simulation-based quasi-statistical approach to estimate EV energy consumption under various on-road vehicle operating conditions. In this approach, a Bayesian Network method is used to integrate outputs from full-system vehicle simulation tools for specific makes and models of EVs under a wide-variety of on-road operating conditions. These outputs are used to develop inference models that greatly improve computational efficiency, while maintaining most of the prediction accuracy of the complete system models. This approach is both highly scalable and transferable for analyzing the energy impact of EV fleet deployment in different regions, can facilitate the estimation of network-level EV energy consumption, and can be incorporated into a wide-variety of transportation planning models. In our case study of Atlanta, GA, the results indicate that if 6.2% of urban travel distances and 4.9% of rural travel distances were to be driven by EVs, regional fuel savings would be around 4.0% for a typical travel day in 2024.