Estimation of State of Charge (SOC) and Water Loss for Absorbed Glass Mat Lead Acid Battery

Abstract

Accurate and fast estimation of state of charge (SOC) and water loss during battery operations plays an important role in the prevention of over-charge and over-discharge and optimal control strategy of charging, which requires a model that has great performance in accuracy, algorithm robustness, computational efficiency, numerical stability, and cost. For SOC estimation, most researchers focus on electric equivalent circuit model (ECM) and electrochemical model. The latter is based on electrochemical and thermal principles which are capable of representing the details of cell behavior, it is more accurate. However, it cannot be applied to real time applications, due to high computational time. The ECM is relatively simple, but limited to represent a narrow range of operating behaviors not considering the effects of temperature and aging. Therefore, there is a need for the development of a method that considers the effects of temperature and aging and also has real time capability. In water loss estimation, the qualitative analysis is proposed. A Second order ECM with an extended Kalman filter (EKF) is used to estimate the SOC of an AGM lead acid battery. Considering the model parameter dependence on temperature and aging, the EKF is designed to identify model parameters online. Because the ECM shows poor performance when the battery is under constant voltage (C.V.) charging, the new EKF and Coulomb counting are combined. Then, given the relationship between capacity and temperature, the capacity-temperature model is added to the new method. With this proposed method, SOC estimation error can be reduced to 3% at various temperatures and aging processes. In water loss estimation, this thesis is the first to propose and test a method to measure the mass of the decomposed water of an AGM lead acid battery. Through calibration of the reaction rate of decomposed water, a water loss estimation algorithm is presented. The comparison between simulation and experimental results proves the accuracy of this water loss estimation method. The algorithm shows that water loss can be minimized by limiting the maximum voltage and temperature during charging.