Carlos Canudas-de-Wit, carlos.canudas-de-wit@cnrs.fr
septembre 2024
12 + 12
INRIA Rhône-Alpes (Grenoble)
Request Background
Optimization, Applied mathematics, physical modeling, IA
Working Context.
This research will be conducted by the DANCE research team (webpage): DANCE (“Dynamics and Control of Networks”) is a joint team of GIPSA-lab and Inria Grenoble–Rhône-Alpes. Our team has a strong expertise in modeling, estimation and control of large-scale networks with application to Electromobility. The research will be part of the PEPR Digitalisation et Décarbonation des Mobilités (MOBIDEC: https://pepr-mobidec.fr/), and the sub-project PC1: FORBAC “FORecasting impacts of mobility, BACkcasting optimal decisions”.
Scientific Context.
Our group has developed eMob-TwinV1 build upon the findings of the ERC-AdG Scale-FreeBack (emob-twin.inrialpes.fr), resulting in an e-mobility simulation tool driven by digital twin technology. eMob-Twin serves a wide range of purposes including forecasting, analysis, and unlocking EV flexibility, catering to the needs of companies, stakeholders, and electricity markets. Initially designed for the Grenoble metropolitan area, a new version currently under development, eMob-TwinV2, will have the capability to encompass any other metropolitan city in France, incorporating auto-calibration functionalities. Primarily focused on electric vehicle (EV) mobility and their state of charge, it also integrates multi-power charging stations. In the context of the PEPR-FORBAC initiative, we aim to tackle two primary challenges. Firstly, we seek to optimize the placement of charging stations (PC location) to ensure efficient coverage. Secondly, we aim to optimize power delivery density per unit area, utilizing the steady-state solution of the dynamic electromobility model while adhering to energy equilibria constraints (ensuring EV charging demand matches the charging station supply infrastructure). Results will be integrated as a toolbox in the eMob-Twin V2 software.
Work program.
The work program includes several key areas of development, focusing on model extensions, large-scale optimization challenges, and software implementation:
- Expanding the existing model referenced in [1, 2] to incorporate additional nodes representing Charge Stations with actual capacity, driver-user price models, and their integration with the grid,
- Following ideas in [2], devising highly efficient optimization algorithms (including learning and IA strategies) for charge station placement and power delivery density per unit area within the expanded model framework. These solutions should be scalable to cover the entire graph model,
- Integrating the developed algorithms (both the model extension and optimization algorithms) into eMob-Twin V2, and conducting comparative analyses of optimization outcomes with those provided by policymakers and urban planners (SDRIVE).
The integration of software components will be undertaken with the assistance of our local team of Research Engineers.
References
- Rodriguez-Vega, M., Canudas-de Wit, C., Nunzio, G.D., and Othman, B. (2023). A graph-based mobility model for electric vehicles in urban traffic networks: Application to the Grenoble metropolitan area. In 2023 European Control Conference (ECC), Bucarest, Ru.
- Mourgues, R., and Canudas-De-Wit, C. , Rodriguez-Vega, M., (2023). Optimal location of evs public charging stations based on a macroscopic urban electromobility model. In 2023 62nd IEEE Conference on Decision and Control (CDC), 3122–3129.
