INTRODUCTION The use of large-size lithium-ion battery has been expanding to not only automotive but also other industrial fields such as AGV, railway, load-leveling, etc. This expansion promotes its various operation conditions and, therefore, requires the way to be able to estimate its life accurately. Many efforts have been devoted to estimate the life even in the several applications. According to the report [1], the battery capacity loss for the estimation is divided into calendar and true cycle capacity loss. The calendar capacity loss is caused by active Li trapping mainly due to SEI growth at negative electrode, of which rate is dependent on SOC level, temperature and time [1-3], whereas the true cycle capacity loss is dependent only on the amount of charge-discharge cycle number in specified DOD range [1]. However, the true cycle capacity loss and its mechanism of large-size battery has not been clarified in the diversified cycle containing high and low frequency of fluctuational SOC. In this study, the factors that affect the true cycle capacity loss with clarification of its mechanism have been investigated in detail, and then updated life estimation algorism based on the result of the investigation has been proposed to enable high-accuracy life estimation of the battery operated even under complicated condition. EXPERIMENTAL The endurance tests for large-size lithium-ion cells were conducted under various conditions of different charge-discharge route (SOC route) in order to clarify the factors that impact the capacity loss. Each amount of the Li trapping at negative electrode, and degradations of the positive and negative electrodes were investigated by discharge profile fitting analysis [4] and post-mortem analysis. RESULTS AND DISCUSSION According to results of the cells tested under various conditions, the capacity loss depends on the SOC route despite of the same accumulative amount of SOC fluctuation and test time. The relation between relative coefficient of true capacity loss and the SOC fluctuational range as controlling the designated value of center SOC at 60% is shown in Fig. 1. The coefficient of the loss drastically increases with expansion of the range, which means that the true cycle capacity loss should be estimated depending on the individual range in the diversified SOC change. Furthermore, post-mortem analysis and discharge profile fitting analysis for the tested cells prove that the total capacity loss of the cell is accelerated mainly by the factors of the range and its center value because the rates of both SEI growth at the negative electrode and degradation of positive electrode are affected by these factors. Therefore, updated model describing true cycling capacity loss has been proposed. Finally, the verification test including high and low frequency of fluctuational SOC under various operating temperature for 550 days was carried out together with estimation of capacity retention by the conventional and the updated model. As a result, the accuracy of battery life estimation even in long-term test with complex patterns is dramatically improved from 31% to within 2% of the actual capacity retention by updated model considering SOC route the cell experienced. REFERENCE H. Yoshida, N. Imamura, T. Inoue, and K. Komada, Electrochemistry, 71, 1018-1024 (2003).H. Ploehn, P. Ramadass, and R. White, J. Electrochem. Soc., 151, A456-A462 (2004).N. Shinha, T. Marks, H. Dahn, A. Smith, J. Burns, D. Coyle, J. Dahn, and J. Dahn, J. Electrochem. Soc., 159, A1672-A1681 (2012).K. Honkura, K. Takahashi, and T. Horiba, J. Power Sources, 196, 10141-10147 (2011). Fig. 1. (a) Relative coefficient of true capacity loss at the same accumulative amount of SOC fluctuation in various SOC fluctuational range. The values are normalized by true capacity loss between 30 and 90% of SOC cycle test (SOC fluctuational range = 60%). (True capacity loss [%]) = (Total capacity loss [%]) - (Calendar capacity loss [%]) (b) Cycling protocols of SOC endurance test shown in Fig. 1 (a). Charge: CC 1.0 CA 4.10 V at 25 ℃. Discharge: CC 1.0 CA 2.75 V at 25 ℃. Figure 1