Data-informed Healthy Cell Clustering Technique for Second-life Applications of Retired Electric Vehicle Batteries

• Main Authors: Samanta, Akash, Karnehm, Dominic, Neve, Antje, Williamson, Sheldon
• Format: Conference Proceeding
• Language: English
• Subjects: Accuracy; battery management system; battery recycling; Clustering algorithms; Clustering methods; data driven techniques; Electric vehicles; machine learning; Renewable energy sources; state of health; Transportation; transportation electrification; Voltage
• ISSN: 2473-7631
• DOI: 10.1109/ITEC60657.2024.10598850

The electric vehicles (EVs) adoption rate has been significant in the last decade and keeps on increasing. Recently, the retired EV battery packs which no longer provide satisfactory performance to power an EV started appearing in the market. Typically, EV batteries are declared unfit for EV applications when the battery capacity is reduced to 70-80% of its nominal capacity. Various end-of-life (EOL) options are proposed by researchers before sending the retired batteries to recycle such as low-demanding e-mobility, grid-tired energy storage systems, storage of renewable energy, and home emergency power supplies. It is predicted that this may create new value pools in the energy and transportation sectors shortly. It is noticed in lithium-ion battery packs that all cells do not age similarly, thus clustering of healthy cells among the retired batteries is very crucial for safe and reliable operation in second-life applications. Thus, an efficient cell clustering method utilizing voltage relaxation curves powered by a data-driven BIRCH (Balanced Iterative Reducing and Clustering using Hierarchies) algorithm is proposed in this paper. The efficiency and efficacy of the proposed algorithm compared to the state-of-the-art cell sorting technique are also demonstrated in the paper.