Browse by author
Lookup NU author(s): Dr Mazhar AbbasORCiD, Dr Simon LambertORCiD
Full text for this publication is not currently held within this repository. Alternative links are provided below where available.
© 2026 Elsevier Ltd. All rights are reserved, including those for text and data mining, AI training, and similar technologies. Knowing the aging trajectories of lithium-ion batteries (LIBs) in electric vehicles – referred to as state-of-life (SOL) prediction – is essential for timely replacement and effective post-first-life management. In real-world applications, however, operating conditions vary significantly, and this uncertainty must be considered to ensure the generalizability of SOL prediction methods. Existing literature provides limited analysis of how such varying conditions challenge SOL prediction, including the availability of laboratory-like features, the robustness of relationships between features and SOL indicators, and the resulting prediction errors. Similarly, the claimed importance of SOL results for post-first-life decision-making has not been demonstrated in a way that provides a foundation for further progress. To address these gaps, this study proposes a systematic procedure for conducting an in-depth analysis of the impact of uncertainties. A range of input features (IFs) is selected from experimental data; these features capture operational uncertainties and maintain explainable relationships with SOL. These IFs are used to train and validate a deep neural network model for SOL prediction. The analysis confirms that uncertainties influence both the availability and sensitivity of IFs, and that accounting for them is critical to achieving generalizable SOL predictions. The highest mean absolute error observed for SOL prediction was approximately 3%. Finally, the study demonstrates how SOL results can be applied to battery second-life (SL) decision-making and recycling management.
Author(s): Abbas M, Kim G, Lambert S, Kim J
Publication type: Article
Publication status: Published
Journal: Journal of Energy Storage
Year: 2026
Volume: 178
Issue: Part A
Print publication date: 15/11/2026
Online publication date: 10/07/2026
Acceptance date: 01/07/2026
ISSN (print): 2352-152X
ISSN (electronic): 2352-1538
Publisher: Elsevier Ltd
URL: https://doi.org/10.1016/j.est.2026.123404
DOI: 10.1016/j.est.2026.123404
Altmetrics provided by Altmetric
