Applications of Complexity Science in Modeling Electric Vehicles and Planning Charging Infrastructure

There is a growing need to develop cheap and efficient means of energy storage for electrification of transportation and harnessing solar and wind energy associated with intermittency in order to meet the goal of decarbonization. Among the various energy storage options, lithium (Li) batteries have revolutionized portable electronics and emerged as the frontrunner for transportation and grid storage. The high energy density, long life and stringent safety requirements of transportation and grid storage have made it challenging to use the current state of the art Li-ion batteries. Furthermore, the moderate energy density of current Li batteries means that electric vehicles have limited range and are more expensive than the gas-powered equivalents. The limited range of batteries coupled with a lack of public charging infrastructure results in severe range anxiety among potential customers, hampering the adoption of electric vehicles. Some studies have shown that a majority of range anxiety is a result of the lack of charging infrastructure rather than the limited range of electric vehicles as the current range is about where customers would need it to be. Solving this issue in infrastructure is not as simple as installing more chargers wherever it is convenient to do so. Improperly placed chargers will likely see low utilization rates, negatively impacting the economics of that charging station, and will inadequately service the demand. Determining where to place chargers requires an understanding of the complex interactions between the city dynamics, vehicle dynamics and battery dynamics and how each of those are impacted by the local climate. In this dissertation, I build a complex system analysis tool that can be used to analyze the large scale dynamics of an electric vehicle fleet and use that to optimally place charging stations. I start by developing the different battery models that can be used within the algorithm to capture the dynamics of the battery. Battery models can range in complexity from a simple equivalent circuit model all the way to multiple partial differential equations describing the intricate electrochemical processes that go on inside the battery. The main trade off of these models is made between accuracy and computational cost. Simple battery models can be quickly computed but will often miss the more intricate details of the battery dynamics; while a more complex model will accurately capture those effects in exchange for additional time to compute. Thus, a balanced model must be chosen so that the large complex system model that is run can capture some of those intricate details while not taking too long to resolve the model. The analysis then transitions into studying the performance metrics of both terrestrial electric vehicles and electric air taxis. In these models, a physics model is developed to calculate the power required of the electric vehicles in order to successfully complete their tasks under average conditions. The power profile is then sent to a battery model to compute the battery performance metrics necessary to successfully electrify that vehicle. While the average conditions can be used to evaluate the general performance conditions and corresponding trade-offs, it is not enough to evaluate the electric vehicle performance in a real scenario. Further, other models make large assumptions about the operating conditions of the vehicle which can inaccurately represent the average operating condition of a vehicle within the various environments it might encounter. To augment this investigation, a complex system analysis model is created which will allow for a more dynamic prediction of the performance metrics for the electric vehicles. The model will do so by creating a realistic distribution of simulated vehicles, using the population dynamics inherent to the location, and then calculates the vehicle power requirements based on the environment in which it is operating. The power dynamics are then sent to the battery model which uses the chemistry of the battery within the vehicle and the temperature to determine how the power required by the vehicle dynamically depletes the battery. By tracking the vehicle and battery performance over all of the simulated vehicles, a series of statistics around the performance of the electric vehicle fleet can be determined. This statistical information can be used in a variety of applications such as charging infrastructure as will be described later in this work. To demonstrate the results that can be produced by the software, terrestrial fleet case studies were performed in Massachusetts, Minnesota, Arizona and Colorado where, in each study, the application of each input was weighed against the model with no environmental considerations while a variety of battery chemistries and spawn conditions were tested. An analysis of the model’s applicability to charging infrastructure was explored using Massachusetts and Colorado as the target locations. Throughout these studies, battery chemistry and opportunistic charging is applied to determine how they impact the public charging infrastructure that is needed. A scaling approach to charging infrastructure is then developed and used to compare these results to demonstrate some of the key differences that can emerge. This study is then concluded with a look at the model’s applicability to electric aircraft which is an emerging technology that will be deployed in a variety of cities over the coming years.