Real-time replication of battery cells using digital twins: A novel experimental methodology to measure current inhomogeneity in n-parallel configurations

Battery cells are commonly connected in parallel to increase battery pack capacity. However, intrinsic and extrinsic non-uniformities across parallel-connected battery cells may result in uneven current distribution, compromising system performance. Experimental studies on how number of cells affects current inhomogeneity in parallel configurations remain limited as they are prone to inaccuracies and resource-intensive. Here, we present a precise and efficient measurement methodology that allows flexible adjustment of the number of cells in parallel through real-time replication of physical cells, while introducing a controlled deviation in path resistance of another physical cell. By systematically varying both path resistance inhomogeneity and number of cells, we analyze their effects on current inhomogeneity across three cell technologies. Our findings show that current inhomogeneity scales linearly with path resistance inhomogeneity and asymptotically with number of cells, regardless of cell technology and whether the deviating cell exhibits increased or decreased path resistance compared to the others. Notably, cells with reduced path resistance also pose challenges. The identified sensitivities allow scaling of current inhomogeneity measured within a specific parallel configuration to any other parallel configuration, thereby significantly reducing measurement effort. Our methodology is adaptable to the study of various non-uniformities in parallel configurations, facilitating battery system optimization.     «

Battery cells are commonly connected in parallel to increase battery pack capacity. However, intrinsic and extrinsic non-uniformities across parallel-connected battery cells may result in uneven current distribution, compromising system performance. Experimental studies on how number of cells affects current inhomogeneity in parallel configurations remain limited as they are prone to inaccuracies and resource-intensive. Here, we present a precise and efficient measurement methodology that allows...     »