Physics-based lifetime modeling and parameter identification of lithium-ion batteries under various degradation conditions

Abstract

This paper presents a systematic methodology to identify parameters of a physics-based model throughout the lifetime of lithium-ion batteries from fresh to calendar and cycling aged cells. The proposed method relies on a single-particle model of the battery coupled with several possible degradation mechanisms, from which solid–electrolyte interphase layer growth and surface cracking propagation were the ones that better described the capacity loss trajectory. Two main parameter identification (ID) steps were considered, namely the ID of fresh cells followed by the ID of aged cells over their lifetime. Each step included two substeps, namely fresh cells ID required the sequential estimation of equilibrium and dynamic parameters, whereas aged cells ID required the sequential estimation of calendar and cycle aging parameters. This strategy allows one to consider a simpler model according to the operating condition, which in turn simplifies the estimation problem and increases the accuracy of the parameter estimates. Considering the maximum root-mean-squared error, results show that the capacity error for calendar-aged and cycle-aged cells is respectively 50% and 54% of the smallest errors reported in the literature. The identified models capture the overall electrochemical behavior of the battery before and during calendar and cycle aging. • Single particle model with degradation used for lifetime battery modeling. • Optimization-based methodology developed for the identification of model parameters. • Battery cells tested across different operating conditions. • Identified model able to capture fresh and (calendar- and cycle-) aged cell behavior. • Data-model errors smaller than other identification frameworks previously reported.

Topics & Concepts