[Defence of thesis] 25/02/2025 - Alix Dupont: "Operator strategies for User-Driven Electric Vehicle Charging in Public Spaces" (UPR LIA)

Mr Alix DUPONT will publicly defend his thesis entitled: "Operator Strategies for Charging Electric Vehicles in Public Spaces with User-Driven Behaviour", on 25 February in Palaiseau.

Laboratory

UPR 4128 - LIA - Avignon Computer Science Laboratory

Composition of the jury

MR YEZEKAEL HAYEL	Avignon University	Thesis supervisor
Ms Luce BROTCORNE	INRIA Lille Nord-Europe	Rapporteur
Mr Dieter FIEMS	University of Ghent	Rapporteur
Mr Panagiotis ANDRIANESIS	Mines Paris - PSL	Examiner
Mr Olivier BEAUDE	EDF	Examiner
Mr Alban JEANDIN	Izivia	Examiner
Ms Tania JIMENEZ	University of Avignon	Thesis co-supervisor
Mr Jean-Baptiste BREAL	EDF	Thesis co-supervisor
Mr Raphaël PAYEN	EDF	Guest

Summary

Electric Vehicles (EVs) are seen as a key solution for decarbonising the transport sector. However, the charging capacity of existing Electric Vehicle Charging Infrastructures (EV-CFIs) remains limited. Increasing the power available on a site entails significant costs for the electricity grid, while installing a large number of charging stations is also expensive for operators. As a result, EV charging often has to be managed in a constrained and congested environment, which can adversely affect the quality of service for users. In this context, this thesis aims to shed light on the decision-making of public charging stakeholders. A decentralised point of view is adopted, where users' decisions are taken individually. In the first part, these decisions are modelled by a queuing game, making it possible to understand and anticipate congestion levels at the various IRVEs. For standard (typically 7 kW to 22 kW) and opportunistic charging, IRVE occupancy is modelled by a lossy queuing system. An approximation of the average energy received by an EV is proposed and the quality of this approximation is measured theoretically and numerically. Congestion impacts the blocking probability and the individual charging power. The resulting interaction between users is modelled by a set of potentials. The uniqueness of a Wardrop equilibrium is demonstrated, characterised and associated with bounds on the price of anarchy. In the context of an urgent need for fast charging, occupancy at IRVEs is represented by lossless queues. Congestion at an IRVE affects the waiting time for recharging. The results show that users tend to make greater use of attractive IRVEs (in terms of charging power or location) than they would in a centralised system, where charging choices are decided by a central entity rather than by the users themselves. However, the inefficiency of the decentralised system remains moderate. In the second part, real-time charging schemes are explored, using a two-tier approach. These schemes aim to mitigate the effects of congestion, and take account of coupling over time. In the context of an IRVE limited by the number of charge points and offering standard charging, the proposed pricing policy imposes a cost on users exceeding a predefined parking time. Occupancy is modelled by a continuous-time Markov chain, partially controlled by the pricing. The pricing dynamics depend on the instantaneous occupancy level, the time of day, or a combination of both. Numerical results indicate that incorporating the instantaneous occupancy level does not significantly improve performance. Finally, the possibility of punctually reducing aggregate power to one or more IRVEs in exchange for a fee is examined. A menu of charging power levels is offered to users. The prices for each option are adjusted dynamically, taking account of variations in electricity prices and aggregate power (where this has an impact). In this context, the two-level problem is transformed into a PLMNE (Mixed Linear Integer Programming). The numerical results provide insights into how a load operator would respond to incentives for a one-off reduction in aggregate power.

Keywords Electric vehicle, Queuing game, Markov chain, Pricing