Critical comparison of equivalent circuit and physics-based models for lithium-ion batteries: A graphite/lithium-iron-phosphate case study

Accurate modelling of Li-ion batteries is essential for optimising performance and safety across a range of applications, from electric vehicles (EVs) to grid storage. This paper critically evaluates two prevalent battery modelling methodologies: Equivalent Circuit Model (ECM) and Physics-Based Model (PBM), using a 60 Ah prismatic graphite/lithium‑iron-phosphate battery as a case study. The focus of this work is on developing, parameterising, and cross-validating these approaches through a comprehensive set of electrical tests at different ambient temperatures under constant and variable current densities, including the Worldwide Harmonised Light Vehicles Test Cycle (WLTC) protocol. This evaluation not only assesses the accuracy and reliability of the ECM and PBM but also underscores their strengths and limitations. The ECM shows advantages in computational speed, ease of calibration, and accuracy within its calibration range and for variable current profiles. However, its accuracy diminishes at higher currents, especially for prolonged current pulses, and beyond the calibration range, as evidenced in charging scenarios beyond 1C. Conversely, the PBM maintains accuracy beyond the calibration dataset but necessitates estimation of many physical parameters, a laborious calibration process, and extended computational times for variable current scenarios. Within the range of conditions investigated (from C/3 to 2C between 10 °C and 40 °C), the average errors in voltage prediction are 51.5 mV for ECM and 19.3 mV for PBM, while 0.9 °C for ECM and 0.4 °C for PBM in temperature prediction. In summary, while the ECM is suited for reproducing constant discharges or WLTC-like profiles with brief and low-intensity charge pulses, the PBM strength lies in its predictiveness for high-rate operations, making them complementary tools for simulating realistic EV load operations and for optimising fast-charging protocols, respectively. These insights contribute to the ongoing advancement of battery technology, focusing on realistic and applicable model development and parameterisation.