This scientific piece of work aims at modeling the aging of lithium-ion batteries, depending on the vehicle stress (electric or hybrid type). More specifically, this study intends to optimise the design of battery packs for vehicle and power management strategies. A original mechanical fatigue approach has been selected as potentially able to model complex and varied demands. This approach was developed for a specific graphite / NCA lithium-ion battery. It appears that a simple damage accumulation is not entirely relevant, and that two contributions to aging are ongoing: one based on charge throughput and the other based on time. Multiple aging tests were performed and have shown the important influence of temperature, current power and state of charge for each contribution. They led to the establishment of equations linking each of these parameters to battery degradation rate. Thanks to these equations, a computer model for aging prevision has been built, able to take into account both cycling and calendar ageing. The model gives, for slightly to moderately complex solicitations, a very small predicting error. Postmortem analyses were also performed on the batteries to understand the mechanisms involved. Several analytical techniques (physicochemical and electrochemical, impedance spectroscopy) make possible to connect the main mechanisms of ageing to the contributions: an alteration of the crystalline structure of the positive electrode active material for the fatigue contribution, passivation of the negative electrode active material for the time contribution. These analyses provide a more complete view of aging and justify the hypothesis made during the implementation of the model. These electrochemical results allow us to consider a generalisation to other lithium-ion battery technologies. Indeed, an attempt to generalise the model to another commercial battery has made possible to check the reliability and to detect limits of this approach.
