Mixed-integer linear programming for optimizing fleet electric vehicle charging

Abstract

Background: With electrification of the worldwide vehicle fleet increasing rapidly in recent years, a problem is beginning to arise on a large scale. This is the problem of charging an ever growing number of electric vehicles without requiring significant costly and time consuming infrastructure upgrades both to the grid and to power generation facilities. Typical unoptimized electric vehicle charging loads are large and peaky, and so it is desirable to find a way to handle these loads with the existing infrastructure. This idea has been explored in the existing literature, but practical models for use in the real world are scant. This was the motivation for this work, completed in collaboration with New Zealand based electric vehicle company EVisi. Process: This work included a literature review of the problem domain and context, the development of a mixed-integer linear programming formulation to optimize electric vehicle charging on a single site scale, and the enhancement of this formulation through the use of two heuristics. Reformulations were made throughout the process to increase performance and real-world accuracy, with the goal of producing a prototype solving pipeline that could be integrated with EVisi’s existing cloud-based services to interface with electric vehicle chargers and optimize vehicle charging through varying charge rates. The optimized charging aimed to balance improved cost outcomes for consumers with reduced infrastructure impacts for grid operators. Outcomes: The output of this work was the model described above, which is capable of solving problems containing up to 100 vehicles and 100 connectors within five minutes. The solutions it produces are sensible and logical, and exhibit the desirable trends and patterns related to improving cost and infrastructure outcomes. With its performance and level of detail the produced solving pipeline meets the requirements defined by EVisi for it to be integrated into their systems.