Abstract

Widespread acceptance of all electric vehicles faces two major challenges. First being the higher price tag compared to a similar utility IC engine vehicle, while giving equal or lesser range. Second being the under-developed infrastructure support for refueling. Current trends in Electric Vehicle (EV) industry shows an increase in battery capacity and higher charging speed capabilities owing to an increased adoption of EVs. This thesis focuses on the second challenge of range anxiety of EV users due to lack of enough charging infrastructure compared to their gasoline powered counterparts. Public fast charging infrastructure is proposed as the solution to solve range anxiety and wider acceptance of EVs by public. As setting up the public charging stations at the initial stages of Electric Vehicle (EV) market penetration can be budget demanding, it is therefore crucial that the locations chosen should cover maximum demand at least cost and best convenience.
This thesis discusses the review of research publications focused on optimal placing of Alternative Fuel/Electric Vehicle Charging Stations (EVCS), by considering various approaches and models they have used. Heuristic methods of solving optimization problems was given an additional focus in the review. This thesis addresses the discrete, multi-objective, capacitated location allocation problem of electric vehicle charging station, using agent-based modeling. The developed model of EV trips in an urban environment and their charging events was fed with real data from city of Montreal, allowing the model to replicate real charging demand situations at the charging stations at various locations across the city of Montreal. A pareto optimization method is developed using a simple evolutionary genetic algorithm to find all the best trade-offs between each objective value. The multiple objectives considered are utilization of charger resources, the average reroute distance of EVs to reach a charging station and number of infeasible trips and optimization is done through running the agent-based model iteratively through the genetic algorithm, evolving its solutions in each iteration. Proposed solutions for each optimal objective value and solution with the best trade-off between the objectives are discussed.