State-of-charge and capacity estimation for MWh-scale LiFePO4 peak-shaving battery energy storage stations based on real-world operating data

Battery energy storage stations (BESSs) are typically composed of thousands of individual lithium iron phosphate (LFP) battery cells connected in series and parallel to enhance voltage and capacity levels. The complex series-parallel configurations of battery systems, coupled with inherent cell-level inconsistencies, hinder precise state estimation in BESSs. Furthermore, the unique open-circuit voltage (OCV) characteristics of LFP batteries, such as hysteresis and plateau regions, present significant challenges in extracting meaningful information for precise state estimation. This study develops accurate estimation algorithms for the capacity and state of charge (SOC) of MWh-scale LFP energy storage battery stations based on real-world operating data. Using the two-point method and voltage-capacity rate curve transformation method, we accurately estimate the LFP battery capacity during operation within the OCV plateau, with an estimation error of less than 3.36 % for 1596 battery units. Considering the series-parallel configuration of the BESS and the inherent capacity inconsistencies among individual cells, we propose a hierarchical SOC estimation strategy for the complete battery system. This method calculates the SOC of the battery with the minimum discharge capacity within each cluster and employs an SOC envelope method between clusters. It improves SOC accuracy by 22.69 % compared to the system's built-in ampere-hour integral SOC estimation methods. The proposed SOC estimation method for a 2.45 MWh storage station yields an annual electricity savings of 1110 USD under the assumed conditions, highlighting its economic value for large-scale battery systems.