State-of-Charge estimation algorithms and their implications on cells in parallel

State-of-Charge (SOC) of a battery is one of the most important parameters used by a Battery Management System (BMS). It is subsequently very important for the efficient functioning of an electric/hybrid vehicle. Since, the SOC is a derived quantity, it is important that the algorithm, on which the estimator is based, is robust and accurate. The algorithms developed thus far, for different battery chemistries are dependent mainly on the current and voltage data received from individual cells or the entire battery. It is possible that an accurate SOC-estimation algorithm could be the key to increasing the efficiency of a typical hybrid-electric vehicle and hence the real-world applicability. This accuracy should be relevant to both normal and failure cases associated with any drive cycle. This in turn depends on the current and voltage signals received, the positing of such sensors, etc. Also, as battery is comprised of a combination of series and parallel strings, it is important that the difference between the SOCs for an individual cell and that of the entire battery is negligible. The objective of this paper is to compare three SOC estimation algorithms: the current based Coulomb-Counter approach; the voltage dependent model-based approach; and the mixed algorithm that adapts the complementary behavior of the other two methods. In this, the positioning of current and voltage sensors; and the reliability of data, i.e. cell SOC and pack SOC, have been considered. Models created using MATLAB/Simulink, based on literature, have been used. It is seen that the SOC estimation algorithm that is based on both current and voltage data is the most accurate under difference cases of normal and failure cases including short-circuit, open-circuit, etc.