Accuracy analysis of the State-of-Charge and remaining run-time determination for lithium-ion batteries

This paper describes the various error sources in a real-time State-of-Charge (SoC) evaluation system and their effects on the overall accuracy in the calculation of the remaining run-time of a battery-operated system. The SoC algorithm for Li-ion batteries studied in this paper combines direct measurement of the Electro-Motive Force (EMF) and book-keeping systems. The major error source appears to be the value of the EMF, which changes with temperature, with previous (dis)charging conditions and due to aging over time. It is shown that with a new method for modelling the SoC–EMF relationship and by predicting the battery’s State-of-Charge-left, an error of less than 1 min or 1% in the remaining run-time can be obtained.