In order to establish a sustainable society, a large variety of secondary batteries are necessary, each of which is a battery pack composed of many battery-cells connected in series and parallel. Such a battery pack must have an optimal connection suitable to its application. However, the optimality cannot be claimed deterministically in the design stage, since the battery behavior heavily depends on the environment of usage and the variability of production. To cope with this situation, we proposed a statistical method for analyzing the distribution of lifetime of a battery pack composed of plural cells, when aging curve and its distribution of a single cell are given[1]. In order to evaluate the performance of the proposed method, we have conducted several experiments of cycle degradations of single cells, parallel-connected modules, and series-connected modules. In this paper, we report some analytical results of the experiments. We have executed 8 types of experiments using totally 210 Li-Ion 18650 battery cells in thermostat chambers (25°C). In the experiments, we repeated 0.7C cc-cv charge to 100% SOC and 1C discharge to 0% SOC. Among the experiments, 2 experiments are conducted for investigating the cycle degradation of a single cell, one experiment uses 50 new cells and the other uses 10 cells which have been left for 3 months before repeating charge-discharge cycles (we call such cells 3-months cells). From these experiments, we found that new cells deteriorate linearly as the cycle progresses, that is, the average and the standard deviation of SOHs of 50 cells decreases and increases almost linearly, respectively. However, in the second experiment, cells show different degradation behavior, that is, they deteriorate linearly until 300 cycles, but after 300 cycles they deteriorate quadratically, and the distribution of degradation curves spreads more than those of new cells as the cycle progresses. Through these experiments, we found that the distribution of charge/discharge capacity of cells can be represented by initial variability and variation of degradation curves. We conducted 4 types of experiments for 50 modules with 2 cells connected in parallel. As for 20 modules using new cells and for 10 modules using 8-months cells which have been left for 8 months before the experiment, we found cycle degradation curves similar to those of new single cells and 3-months single cells. Namely, modules using new cells deteriorate linearly, but modules using 8-months cells deteriorate quadratically after 300 cycles. The distribution of degradation curves after 300 cycles of 8-months cell-modules was much wider than those of 3-months single cells. We investigated the degradations of 10 modules composed of two parallel-connected cells with different SOHs. The cells with different SOHs were produced by repeating charge-discharge cycles to new cells. Then, as can be readily imagined, the cell with smaller SOH deteriorated faster than the other cell with larger SOH after repeating charge-discharge cycles. Namely, the difference between SOHs of parallel-connected cells becomes smaller after charge-discharge cycles. However, if we disconnected cells in each module and investigated cycle degradation of each cell, then degradation curves were similar to those of new cells obtained before connecting them in parallel. We conducted 2 types of experiments for 15 modules with 4 cells connected in series, in which SOCs of cells in a module were balanced by a battery management system (BMS). An experiment for 10 modules composed of 8-months cells shows degradation curves similar to the experiment for 2 parallel-connected modules composed of 8-months cells, but three series-connected modules deteriorate much faster than parallel-connected modules. Namely, they deteriorate quadratically after 200 cycles and more rapidly after 300 cycles. Therefore, the distribution of degradation curves is wider than that of parallel-connected modules. Another experiment for 5 modules composed of 3-months cells is conducted to investigate the effect of BMS setting. Namely, the charge-discharge cycle is controlled by the voltage of a module equal to the sum of voltages of 4 cells, but the voltage of each cell in the module is controlled by BMS. However, we did not found any distinguished difference in the degradation curves and the distribution of degradation curves was similar to that of 3-months single cells. Since we are now checking the impedance of rapidly degraded cells, we may be able to report the results in the symposium. [1] Daisuke Sasaki, Shuji Tsukiyama, Mariko Matsunaga, Osamu Ishibashi, Shingo Takahashi, “A Statistical Method for Analyzing Lifetime of a Battery Pack,” IEEE Power & Energy Society General Meeting, Paper no. PESGM2015-001311, 2015.
Read full abstract
Aug 7, 2024
Lorenzo Breglio + 7
Open Access