Asymmetric Current Distribution Within Parallel-Connected Battery Cells and Its Influencing Factors

摘要

The number of applications that have an energy demand, which cannot be fulfilled with a single lithium-ion battery cell, grew fast in the last years. Large storages for electric vehicles or stationary storages often use series as well as parallel connected cells. The effects of different cell parameters in series connections were investigated deeply in the past and battery management systems can handle the influences. The parallel connection has not been in the focus that much. Most time, a symmetric current distribution between parallel connected battery cells is assumed at the system design. This is not the case because inhomogeneities lead to an asymmetric current distribution. An asymmetric ageing can be assumed because ageing is strongly dependent on the charge throughput. With a detailed knowledge of the current distribution and its influencing factors, fast charge strategies can be developed for long life battery systems. The poster illustrates a multitude of reasons for asymmetric current distribution within parallel-connected lithium-ion battery cells. The basis is a temperature difference of two parallel connected cells with 15℃ difference in temperature. The study uses this setup to analyse the influencing factors impedance, entropy, open circuit voltage (OCV) and hysteresis. It compares two cell types (Sony US26650FT and LG INR18650MJ1) with different cathode material (lithium-ion phosphate (LFP) and nickel manganese cobalt oxide (NMC)). For the measurements, the cell state of charge (SoC) were equivalised by charging both cells in parallel connection to 100 % of SoC respectively discharge to 0 % SoC with a constant current (CC) followed by a long constant voltage (CV) phase till the current drops below a c-rate of C/20. At the second step, a CC phase approaches a specific SoC followed by a relaxation of several hours. Then the parallel connection is cycled by +/- 20 % of SoC beginning with charge respectively discharge current. In the end, the cells relax again several hours. This load profile represents the charging, standby and the utilisation of a storage. For further analyses of the measured data, an electrical equivalent circuit model parametrised individually for all cells gives overvoltages and the SoCs. The results show that CC current of 1 C can lead to a SoC difference of two LFP cells more than 20%. In the relaxation phase, the cell current equalises these difference partly. At medium SoC, a SoC difference of more than 17% was measured after more than 3 h of relaxation and a drop of cell currents below 3 mA. The same experiment with an NMC cathode shows up to 9 % of SoC difference before the relaxation phase. In addition, lower current at the CC phase by setting the SoC leads to lower differences. A lower c-rate strengthens the effect of the OCV slope in relation to other factors of influence like impedance. The reason for SoC differences after a long relaxation is supposed to come from the OCV hysteresis, which is more important at flat OCV curves like LFP. The SoC differences after cycling confirm this assumption. In terms of charge throughput, a ratio between two LFP cells and the specified temperature difference was found of 2/3. This ratio is independent of load profile, SoC and c-rate.