ED Sciences Physiques et de l'Ingénieur
Analysis of the use of electric vehicle charging flexibility in a real collective self-consumption case: design of energy management systems
by Garazi ETXEGARAI AZKARATEGI (ESTIA-Recherche)
The defense will take place at 10h30 - Guillermo Martinez Room University of the Basque Country Europa Plaza, 1 20018 Donostia-San Sebastián Spain
in front of the jury composed of
- Mahamadou ABDOU TANKARI - Maître de conférences - University of Paris-Est Créteil - Rapporteur
- Najiba MRABET BELLAAJ - Professeure - University of Tunis El Manar - Rapporteur
- Haizea GAZTANAGA ARANTZAMENDI - Directrice de recherche - Ikerlan (technology centre) - Examinateur
- Ekaitz ZULUETA GUERRERO - Professeur assistant - University of the Basque Country (EHU) - Examinateur
- Ionel VECHIU - Directeur de recherche - Université de Bordeaux - Examinateur
In the current context of accelerating electrification of mobility, the increasing penetration of electric vehicles has raised concerns regarding their impact on the electrical demand. Simultaneously, the power system is undergoing a transition from a centralized structure towards a more decentralized model, characterized by the integration of distributed energy resources. These two trends converge into a promising opportunity: using locally generated photovoltaic energy in buildings to supply electric vehicle charging demand. Furthermore, when electric vehicles are treated as flexible loads, their charging process can be actively managed to better align with photovoltaic generation. This enables an increase in the self-consumption rate within collective self-consumption schemes, leading to both economic benefits and a more sustainable balance between energy generation and demand. This thesis investigates how electric vehicles, considered as flexible loads, can contribute to increasing the self-consumption rate in a real collective self-consumption case study combining photovoltaic generation, building demand, and public charging infrastructure. The research is motivated by the simultaneous growth of electric mobility and distributed renewable generation, which creates both challenges and opportunities for local energy management. To address this problem, two energy management systems are developed and evaluated. The first is a rule-based system that implements an indirect control strategy through three tariff schemes designed to encourage users to shift charging towards periods of photovoltaic surplus. The second is an optimization-based system that applies direct control by scheduling charging time and power using predictive models and a genetic algorithm. The optimization-based system is assessed under four different scenarios, representing different levels of flexibility in charging time and power. The thesis also develops day-ahead forecasting models. Building electricity consumption forecast employs feed-forward neural networks and long short-term memory networks, with input features and hyperparameters selection optimized via genetic algorithm. Photovoltaic generation forecast uses analytical expressions based on meteorological data, acknowledging potential improvements through machine learning approaches already developed in other case study. Additionally, an experimental analysis of electric vehicle charging pattern prediction is conducted. All models and control strategies are validated through simulations using real data from a collective self-consumption project involving seven buildings, photovoltaic generation, and a public electric vehicle charging station. Results demonstrate that both approaches can improve the self-consumption rate, although their performance depends on the flexibility available, the local load profile, the photovoltaic system size, and user behaviour. The rule-based system offers a simple and low-cost solution, while the optimization-based system can handle multiple constraints and more complex operating conditions. For the specific case studied, the optimization-based approach achieves superior improvements when sufficient flexibility is available. The proposed frameworks are scalable and suitable for applications in energy communities, microgrids, and charging stations.
Design and characterization of advanced electronic packaging solutions for electromagnetic shielding and moisture protection.
by Victor MAHAUT (Laboratoire de l'Intégration du Matériau au Système)
The defense will take place at 10h45 - Auditorium Thales Research & Technology, 1 avenue Augustin Fresnel, Campus Polytechnique, 91767 Palaiseau
in front of the jury composed of
- Tristan DUBOIS - Associate Professor - Université de Bordeaux - Directeur de these
- Alexandrine GRACIA - Associate Professor - Université de Bordeaux - CoDirecteur de these
- Philippe BESNIER - Directeur de recherche - CNRS, IETR - Rapporteur
- Christophe GUIFFAUT - Directeur de recherche - CNRS, XLim - Rapporteur
- Isabelle DUFOUR - Professeure - Université de Bordeaux - Examinateur
- Benoit GORAL - Docteur - Thales SIX - Examinateur
This thesis, conducted in partnership with the IMS laboratory and Thales Research & Technology, focuses on electronic packaging. It is structured around two main research axes and deals with the implementation and evaluation of novel protection solutions based on additive manufacturing. The first axis aims to evaluate the electromagnetic shielding performance of two of these processes. The objective is to understand the underlying physical phenomena, first at the isolated material level, and then within a complete system. This approach, combining experimental characterization and finite element simulation, integrates the influence of the printed circuit board design on the overall performance. It also offers a comparison of several characterization methods to identify the most relevant ones depending on the usage scenarios. The second axis investigates the moisture uptake of plastic packages. Water absorption within these encapsulating resins is initially characterized through gravimetric analysis. A novel in-situ measurement method, employing embedded sensors, was subsequently developed and correlated with the gravimetric results. This original methodology made it possible to evaluate the influence of atomic layer deposition on the absorption capabilities of these packages.