New electric energy and charging infrastructure demands are emerging in future electric transportation systems. These demands affect many diverse aspects of the transportation infrastructure, ranging from charging station locations to routing and optimal speed profiles. The sensitivity of electric vehicle (EV) energy consumption to environmental factors drastically affects optimal strategies, especially when facing uncertainties in the available data. The proposed methods in this dissertation will aid in achieving the goal of building a future energy efficient and sustainable transportation system infrastructure.
The fundamental concept of this research is an environmentally aware energy consumption prediction model for given routes. Environmental effects on EV energy consumption have been investigated using a sensitivity analysis. Adaptive multi-resolution energy consumption predictions are implemented for various real-time accuracy requirements. Based on accurate energy cost prediction, sustainable driving and infrastructure strategies have been investigated.
In the first part of the dissertation, energy-efficient driving strategies are designed for optimal transportation energy utilization. Sustainable optimization models are constructed for optimal speed profiles. They exploit various environmental conditions to minimize the overall energy consumption along a route. Two different aspects are investigated: minimum energy cost with travel time constraint and minimum time cost with energy constraint. Infinite dimensional optimization models are proposed for exact problem descriptions and approximate discretized convex models are derived for highly efficient solutions. In addition, an eco-routing decision-making framework based on stochastic programming is introduced for optimal routing. Route planning with minimum energy cost expectation is performed when facing random effects of environmental factors and electric drive limits. This problem is tackled using convex relaxation along with a primal-dual interior point method and a path reconstruction algorithm.
In the second part of the dissertation, sustainable strategies are analyzed and introduced for satisfying the EV energy demand. Spatio-temporal models have been constructed to essentially describe three-dimensional (location and time) features of energy demand. Charging urgency models take into consideration many practical factors in order to estimate the perceived demand. All of these factors provide concrete demand information to help building a sustainable charging infrastructure. Optimal charging station placement under an energy demand framework has also been studied. Two different optimization requirements are addressed in this dissertation: maximum number of reachable households and minimum overall energy consumption. A multi-objective optimization model will synthesize a solution using both requirements so as to obtain better decisions for charging infrastructure placement.