This poster presents an investigation into the degradation behavior and health prediction of second-life LiFePO₄/graphite batteries, combining experimental analysis with data-driven modeling. Retired electric vehicle batteries typically retain 70–80% state of health (SoH), making them suitable for repurposing in stationary energy storage applications until they reach approximately 60% SoH. The study aims to better understand how operating conditions, particularly C-rate, influence aging mechanisms and lifetime in second-life usage.

The experimental work is based on 30 Ah LFP/graphite pouch cells subjected to five different cycling scenarios with varying charge and discharge C-rates (ranging from 0.5C to 2C). Diagnostic techniques include capacity measurements, electrochemical impedance spectroscopy (EIS), and scanning electron microscopy (SEM). These methods allow for both electrochemical and microstructural characterization of degradation processes.

Results show that higher discharge rates significantly accelerate degradation. Cells cycled at 2C exhibit the fastest capacity fade, reaching 60% SoH within approximately 500–600 cycles, whereas low-rate cycling (0.5C/0.5C) can extend lifetime up to around 2000 cycles. EIS analysis reveals that charge-transfer resistance and diffusion-related impedance increase more rapidly under high C-rate conditions, indicating intensified aging. In contrast, low-rate operation maintains lower impedance growth from first to second life.

SEM analysis provides microstructural evidence supporting these findings. High-rate cycling leads to particle cracking and loss of active material contact due to mechanical stress, while low-rate scenarios preserve particle integrity and maintain a stable conductive network.

In addition, the study incorporates a data-driven approach using LSTM models for SoH prediction. The model demonstrates high accuracy, with errors below 5% in optimal cases, and improved reliability as more training data becomes available. It also shows strong generalization across different cycling scenarios.

Overall, the results highlight that C-rate is a critical factor governing second-life battery degradation. Lower operating rates significantly extend cycle life, while high rates shift aging mechanisms from surface-related processes to structural damage. The combination of experimental insights and machine learning provides a robust framework for predicting battery health and optimizing second-life applications.