Reconfiguration-Assisted Charging in Large-Scale Lithium-ion Battery Systems

Large-scale Lithium-ion batteries are widely adopted in many systems such as electric vehicles and energy backup in power grids. Due to factors such as manufacturing difference and heterogeneous discharging conditions, cells in the battery system may have different statuses such as diverse voltage levels. This cell diversity is commonly known as the cell unbalance issue, which becomes more critical as the system scale increases. The cell unbalance issue not only significantly degrades the system performance in many aspects, but may also cause system safety issues such as the burning of battery cells and thus increase system vulnerability. In this paper, based on the advancement in reconfigurable battery systems, we demonstrate how to utilize system reconfiguration flexibility for achieving an efficient charging process for the battery system. With the proposed reconfiguration-assisted charging, the cells in the system are categorized according to their voltages, and the charging process is evolutionarily carried out in a category-by-category manner. For the charging of cells in a given category, a graph-based algorithm is presented to obtain the desired system configuration. We extensively evaluate the reconfiguration-assisted charging through small-scale implementation and large-scale trace-driven simulations. The results demonstrate that our proposed techniques can achieve a 25% increase on average on charged capacities of individual cells while yielding a dramatically reduced variance.