Agent-based Simulation of Electric Vehicles

Abstract

Major global concerns for today’s world are the implications of climate change and future energy security. The transportation sector plays an important role within this context, as it currently heavily relies on fossil fuels. In order to break this dependence, electric vehicles could play a key role, especially due to their greater energy conversion efficiency compared with conventional vehicles. Furthermore, by using electricity these vehicles can play an important role in the energy system of the future, where energy generation is envisioned to be more sustainable, incorporating a higher share of renewable energy resources. However, as many of these energy sources are intermittent and require energy storage capacities, the batteries of electric vehicles could take up this role; by exchanging information between electricity demand and supply stakeholders in real-time (“smart grid”), an electric vehicle would charge at times of electricity oversupply and stop charging or even supply energy back to the grid for short periods in times of electricity generation shortage, in order to stabilize the electricity network (“vehicle to grid”). But there are also concerns that the electricity grid, which has not been designed with dynamic demands in time and space in mind, could suffer from the large scale integration of electric vehicles. This could manifest itself in powerline and transformer overloads on lower levels of the electricity network distribution infrastructure. This security and stability of the gird is further at risk due to increased distributed energy generation (including alternative energy) and the liberalization of electricity markets. In this case electricity is traded beyond national borders, leading to possible congestion at powerlines. In order to support the analysis and future design of such complex systems including electric vehicles, integrated modeling of energy demand and supply is needed. This dissertation proposes a framework for such modeling, with particular focus on electricity demand modeling for electric vehicles. As many problems within this context require disaggregated models in time and space, e.g. to uncover bottlenecks in the electricity grid, an existing agent-based travel demand simulation called MATSim is used, which allows the modeling of individual preferences. In order to prepare MATSim for simulation of large scale disaggregated electric vehicle scenarios, a new traffic micro-simulation model is implemented together with other performance enhancements to the framework, making use of parallel computation. Additionally, the current parking model in MATSim is rexi placed by a new parking model, which takes parking supply constraints into account and also supports special parking for electric vehicles with integrated electricity charging facilities. The parking choice model has been developed further towards an initial parking search model in the course of this dissertation. Based on this work, a framework has been developed that integrates various models, including a vehicle fleet definition, vehicle energy consumption models and electricity charging models. In addition, various types of charging infrastructure are modeled including stationary infrastructure with plugs and inductive charging along roads. Furthermore, several types of charging schemes are available including smart charging, where an intelligent central entity in the smart grid is assumed which controls the charging of vehicles. During the course of this dissertation it became evident that there is a lack of integrated and detailed electricity demand and supply models, which hampers interdisciplinary work in the field. Therefore, the framework is being generalized and published as open source under the name “Transportation Energy Simulation Framework”. For many models only basic implementations and interfaces are provided. The idea is that other researchers who are experts within their fields can build on top of it, for example models for “vehicle to grid” applications. A case study for the city of Zurich is presented in this dissertation, which highlights the capabilities of the framework to uncover possible bottlenecks in the electricity network. Furthermore, the case study also highlights the ability of the models to support policy design. To the best of the author’s knowledge such integrated modeling is the first of its kind, in terms of methodology, spatial and temporal resolution and scenario size.