Modeling the aging of fuel cells: A coupled physical and machine learning approach

With the aim of reducing CO2 emissions in the transport sector, various technologies and energy carriers are being studied to gradually replace fossil fuels: biofuel or hydrogen thermal engines, electrification of the propulsion chain with batteries or low-temperature fuel cells. The latter technology is strongly recommended for heavy mobility (commercial vehicles, buses, heavy goods vehicles, rail, and sea). In order to design and size such a system as well as possible, it is imperative to take into account not only the performance but also the aging factors in order to achieve a solution with a minimal environmental impact. Many approaches have already been explored to model this aging. Many approaches have already been explored to model this aging, but it remains a challenging task. Therefore, this work takes on this challenge by developing a multi-scale numerical model which will be able to account for the impact of microscopic aging phenomena on the macroscopic scale and under conditions representative of use. 
The subject of the thesis combines an experimental and modeling approach.  This approach consists in determining via statistical and machine learning methods the evolution of the fuel cell parameters in order to model the aging process. The methods developed must be based mainly on the characterization of the fuel cell by electrochemical impedance spectroscopy during its aging and by the exploitation of these spectra by the calculation of the distribution of relaxation time.
This thesis is a collaboration between IFP Energies Nouvelles, FCLab and FEMTO-ST. It adopts a multidisciplinary approach, both experimental and theoretical, and brings together skills in electrochemistry, signal processing, electrical engineering and scientific programming supported by the different laboratories involved.

Keywords: PEM fuel Cell, modelling, aging, machine learning, electrochemistry

• Academic supervisor    Pr Marie-Cécile PERA FEMTO-ST, marie-cecile.pera@univ-fcomte.fr (https://www.femto-st.fr/fr/personnel-femto/mcpera)
• Doctoral School    ED37, SPIM Science Physiques pour l’Ingénieur et Microtechniques  http://ed-spim.univ-fcomte.fr/
• IFPEN supervisor    Dr Quentin CACCIUTTOLO, Département électrochimie et matériaux, quentin.cacciuttolo@ifpen.fr
• PhD location    IFP Energies Nouvelles, Lyon, France and FEMTO-ST, Belfort, France
• Duration and start date    3 years, starting in the fourth quarter of 2023
• Employer    IFP Energies Nouvelles, Lyon, France
• Academic requirements    University Master degree in relevant disciplines (electrochemistry, chemical or electrical engineering…)
• Language requirements    Fluency in English, willingness to learn French
• Other requirements    Knowledge of scientific programming (python or matlab), mathematics and in electrochemistry is a plus