Electric vehicles (EVs), including hybrid-electric vehicles (HEV), plug-in hybrid-electric vehicle (PHEV), battery electric vehicles (BEV) and fuel-cell electric vehicles (FCEV), have been developed for decades to reduce the energy consumption during on-road operations through adopting electricity and significant efficiency improvements. The energy consumption estimation is essential in assessing the benefits and costs of using EVs and evaluating the effectiveness of their implementation strategies. However, the traditional methods of estimating energy consumption of EVs were either dependent on numerical analysis of vehicle testing data or using high-performance simulation models, which suffers from the scaling problems for meso- and macro-scale transportation networks. A new approach with the capability of large transportation network projection is needed for generating the energy consumption of EV fleets.
In this study, a modal-based approach for estimating energy consumption and emissions of EVs is proposed. The modal-based approach has been previously implemented in the U.S. Environmental Protection Agency’s Motor Vehicle Emission Simulator (MOVES) and widely applied for conventional engine vehicles, which associates the emission and energy consumption rates with specific operating condition bins. The modal-based approach allows the energy estimation for meso- and macro-scale transportation network and considers variation of vehicle operating conditions. In this study, the advanced modal-based approach will be proposed to estimate energy consumption and emissions of EV fleets as a function of key operation parameters.
First, real-world operation data will be collected and post-processed to obtain the energy consumption and emissions from typical light-to medium-duty electric-drive fleet under all possible current and future conditions. Second, the relationship between energy and emissions as a function of key vehicle configuration and operating factors will be depicted using factorial analysis and regression analysis. Third, the hierarchical tree based regression (HTBR) with stochastic gradient boosting will be applied to classify the energy consumption rates under distinct operating conditions, using significant factors from the previous step. Finally, the modeled energy consumption rates by new bin systems will be tested using a real-world transportation network to verify the effectiveness of the proposed methods. With this approach, the network-level EV energy consumption can be generated and used for various transportation studies, such as assessing grid impact, comparing charging strategies and electrification of shared-autonomous vehicles.
Dissertation is embargoed until Fall 2020.