A comprehensive framework for harnessing energy arbitrage benefits through an electric vehicle aggregator

Abstract

This paper proposes a comprehensive framework for an EV aggregator to operate as an energy arbitrage entity, aiming to provide a steady income stream to EV owners while reducing grid stress by maximising the system's Load Factor (LF). The framework integrates spatial EV behaviour analysis, machine learning-based forecasting of electricity demand and prices, and a two-stage optimisation process. The first stage utilises Mixed-Integer Linear Programming (MILP) to optimise charging and discharging schedules, thereby maximising EV owner revenue. The second stage employs Gradient-Based Sequential Quadratic Programming (GD-SQP) to optimise the number of participating EVs for grid benefits. The model is built on real-world data from Queensland, Australia, including EV mobility derived from travel surveys and historical 5-minute resolution data from the National Electricity Market. Forecasting performance indicates that the Fine Tree Algorithm (FTA) is most effective for electricity demand prediction, while the Ensemble Bagged Tree Algorithm (EBTA) performs best for price forecasting. Simulation results on the IEEE 9-bus and 39-bus systems validate the framework's performance. In the 9-bus system, LF improves by up to 6.08 % in the H-EV Over-optimum scenario. In contrast, HW-EV scenarios yield reduced LF due to misalignment between revenue-optimised dispatch and system objectives. In contrast, the 39-bus results show that HW-EV Optimum can achieve up to 6.2 % LF improvement, highlighting the influence of grid topology and spatial flexibility. The framework also achieves up to 6.08 % reduction in transmission loss, with additional benefits including peak load shifting, deferred infrastructure investment, and increased revenue for aggregators. An important finding is that longer EV connection durations, as in HW-EV cases, do not guarantee better grid performance due to a trade-off inherent in the two-stage optimisation. These results confirm that EV participation must be carefully managed to strike a balance between economic and technical objectives. Overall, this study demonstrates that data-driven EV coordination, supported by forecasting and dual optimisation, can unlock substantial benefits for both EV owners and the power grid in future smart energy systems.