Dataset - Feedback control bounds variability in large-scale electrochemical systems

Controlling cell-to-cell heterogeneities is critical for the safe and reliable operation of high-energy lithium-ion battery packs, which integrate hundreds to thousands of cells into series--parallel architectures. Although considerable effort is put into optimizing manufacturing processes to minimize cell-to-cell variability, small imperfections still remain. It is widely assumed that cell-to-cell differences inevitably amplify with ageing and use. Here, we challenge this view. We introduce intra-pack heterogeneity metrics evaluated directly from battery management system (BMS) field data. Using data-driven analytical tools to inherently noisy measurements collected from real-world operation of a passenger electric vehicle, we show that beginning-of-life variability does not amplify over the operational lifetime of the pack. The results are then validated in controlled laboratory conditions through pack disassembly and module-level testing.
The boundedness of cell-to-cell variability suggests that effective voltage balancing and thermal control by the BMS maintain intra-pack consistency, highlighting their role in reliable long-term operation. Building on this, we propose deployable onboard methods to monitor intra-pack heterogeneity without additional sensors. The framework scales from EV to grid systems. The dataset—3.5 years of field BMS data with two laboratory validation campaigns—is publicly available and uniquely combines long-term operation with pack-to-cell resolution.