One of the major customers‘ demands on battery electric vehicles is the lifetime of the battery as a potential cost driver in the future and influencing factor on sales value. Battery aging is influenced by degradation mechanisms, driven by storage and operating conditions, known as calendric and cyclic aging. These mechanisms lead to a loss of capacity and an increase in impedance. Investigating aging behavior during development is one key to success for market penetration and customer acceptance.

Given the time and resources required to perform aging tests, computational modeling and simulation can provide insights into the lifetime of batteries much faster and more economical. Furthermore, it can drive decisions on relevant tests to perform. Different aging modeling approaches can be used to predict the degradation rate of batteries, including physics-based, semi-physical and data-driven models. Physics-based and semi-physical approaches describe diverse physical and chemical reactions, which occur in the battery, through mathematical formulations, providing reliable predictions of aging behavior even under extreme conditions. In contrast, data-driven models, which include machine learning and deep learning algorithms, are entirely developed based on experimental data with no physical interpretation. These models use datasets to identify patterns and predict aging in lithium-ion batteries. They exhibit excellent adaptability and accuracy, particularly in capturing complex, nonlinear relationships within the data.

FEV Europe GmbH aging tool chain covers the whole range of the models described. An overview of the pros and cons is given, together with details on physical and data-driven methodologies to address the lifetime concerns in batteries and enable investigations to extend the lifetime described.