Due to the broad use of parallel-connected cells across a wide variety of applications, it is essential to understand the current distribution between them. Variations in resistance, temperature and capacity can lead to an inhomogeneous current distribution and have a deleterious influence on ageing and safety. It is therefore crucial to investigate the current distribution within such systems. Many researchers have designed distinct test-benches to measure the current distribution. In any test-bench, known and unknown resistances caused by wiring, interconnections and measurement equipment can adversely affect the current distribution. The aim of this work is to design a low-complexity test apparatus that does not itself affect the current measurement.This work describes a novel measurement technique [1] to analyse the current distribution within parallel-connected cells without the use of complex custom-made test benches. While the cells are physically connected in conventional test benches, the proposed technique utilizes the software of a commercially available battery cycler to connect the cells in parallel. Because the software replaces the parallel junction within equations, this approach is called “virtual parallel connection“. Each cell is individually connected to the battery cycler and each cell voltage or current can be controlled separately. While a current pulse is used to determine the voltage across the reference cell, the control unit imposes the same voltage across each cell. Therefore, each cell benefits from the 4-wire measurement of the battery cycler, which generates no additional interconnection resistances. For this reason, only the Open Circuit Voltage (OCV), the capacity and the resistance of the cell itself, as well as its relationship to other cells can influence the current distribution. Additionally, any desired connection resistance can be chosen by adding a voltage drop in the equation of the battery cycler. Connecting cells virtually has many advantages, such as the ability to decouple the cells for check-ups and recouple them for cycling without ever touching them physically, independent location of each cell, scaling to n-parallel constellations, low assembly effort on the test-bench and defined contact resistances. In addition, different cell formats can be investigated with no additional effort.In this work, the virtual parallel connection will first be demonstrated, validated and discussed for constant current-, constant voltage- and rest phases using of a conventional test bench [2] within two NMC cells. Next, to understand the influence of additional resistances in two parallel-connected cells, two measurement studies were carried out in which interconnection resistances were varied between 0 and 5 mΩ. The first study investigates the influence of an inhomogeneous resistance increase within one of the parallel paths. Subsequently, the second study examines a homogeneous resistance increase within both parallel-connected paths.The conclusion of both studies is that the height of the local minima and maxima of the current distribution are mainly dependent on the resistance ratio of the parallel legs. The minima and maxima divide relative to the ratio of the combined cell and interconnection resistance in each pathway. In contrast, neither the local minima and maxima measured at various cumulative charge throughputs, nor the intersection points of the current distribution are affected by varied resistances. Meanwhile, OCV interactions between both cells determine the shape of the current distribution.Additionally, the shape of the Differential Voltage Analysis (DVA) showed correlations with current distribution. Some local minima and maxima within the current appear in almost the same region as the local minima and maxima of the DVA. By assigning different peaks of both half-cell profiles of the DVA to the anode and the cathode, the current distribution of both currents is expected to be affected by these characteristic peaks.
Read full abstract