Event-Driven Real-Time Multi-Objective Charging Schedule Optimization For Electric Vehicle Fleets

TL;DR

This paper presents an event-driven, real-time multi-objective optimization approach for EV fleet charging that reduces costs, minimizes battery degradation, and enhances ride availability, outperforming baseline policies.

Contribution

It introduces a novel multi-objective optimization framework for EV fleet charging that considers real-time electricity prices, battery health, and operational demands.

Findings

33.3% reduction in peak electricity load

53.2% savings in charging costs

16% lower battery capacity fade

Abstract

The utilization of Electric Vehicles (EVs) in car rental services is gaining momentum around the world and most commercial fleets are expected to fully adopt EVs by 2030. At the moment, the baseline solution that most fleet operators use is a Business as Usual (BAU) policy of charging at the maximum power at all times when charging EVs. Unlike petrol prices that are fairly constant, electricity prices are more volatile and can vary vastly within several minutes depending on electricity supply which is influenced by intermittent energy supplies like renewable energy and increased demand due to electrification in many industrial sectors including transportation. The battery in EVs is the most critical component as it is the most expensive component to replace and the most dangerous component with fire risks. For safe operation and battery longevity it is imperative to prevent battery capacity fade whenever the EVs are under the control of the fleet operator such as during charging.Fundamentally, the fleet operator would like to service as much demand as possible to maximize the revenue generated at a particular time instance.This is achieved by minimizing the EVs time spent on charging and thereby increasing their availability for rides.The three goals of reducing charging cost, battery capacity fade and maximizing ride availability are formulated as a multi-objective optimization problem. The formulation is tested using the Gurobi solver on two cases from the real-world ACN dataset involving low and high EV charging densities over a week long period. The results of the proposed solution show 33.3% reduction in peak electricity loading period, 53.2% savings in charging cost and 16% lower battery capacity fade for the fleet operator.