Comparison of active energy-balance architectures for second-life battery dynamic equalization

The repurposing of exhausted automotive batteries in second-life stationary applications helps to overcome the ethical concerns related to Lithium-Ion battery disposal. However, the aged cells composing second-life batteries present large capacity variability that limits the battery usable energy. The usable energy can be increased using an active balancing architecture implementing dynamic equalization that moves charge from stronger cells to weaker ones during the normal battery operation. However, the high quantity of transferred charge makes the choice of the balancing architecture challenging. This work compares the Direct-Cell-to-Cell, Adjacent-Cell-to-Cell, Cell-to-Pack, and Pack-to-Cell active balancing architectures in different realistic scenarios. Three case studies are analyzed considering Gaussian, Weibull, and bimodal capacity distributions. For each distribution, a Monte Carlo simulation of 10,000 battery modules composed of 10 series-connected cells is carried out. The usable capacity obtained by applying dynamic equalization to each investigated architecture is calculated. The Direct-Cell-to-Cell architecture shows the best performance. However, the large performance gap among the architectures usually found in the literature appears to reduce when realistic scenarios are considered. Therefore, both Adjacent-Cell-to-Cell and Pack-to-Cells architectures emerge as good trade-offs between performance and complexity for dynamic equalization of second-life batteries.