TL;DR
This paper introduces an "Online Fast Charging Model without Lithium Plating" (OLFEM) for large-format lithium-ion batteries (LIBs) used in electric vehicles. By integrating distributed reference electrodes and employing an Extended Kalman Filter (EKF), the model accurately estimates the negative electrode potential, enabling a fast-charging strategy that prevents damaging lithium plating and preserves battery health.
What Problem Does This Paper Address?
Fast charging is crucial for electric vehicle adoption, but it poses a significant risk of lithium plating on the negative electrode. This plating degrades battery capacity, reduces lifespan, and can lead to safety hazards like short circuits. A key challenge is that in large-format cells, the internal potential distribution is not uniform, making it difficult to accurately predict and control the negative electrode potential across the entire cell. Existing models often struggle with this spatial inconsistency and are sensitive to temperature and charging rates, hindering the development of reliable, safe fast-charging strategies for real-world applications.
Methods
The core of this work is the Online Fast Charging Equivalent Circuit Model (OLFEM). This model builds upon a first-order RC equivalent circuit to represent the battery's electrochemical behavior.
Key methodological innovations include:
- Distributed Reference Electrodes: The researchers ingeniously implanted multiple reference electrodes within a long-dimensional cell. This allowed them to measure the negative electrode potential at different locations, specifically comparing the positive and negative tab sides. This addresses the spatial inconsistency problem.
- Extended Kalman Filter (EKF) for Closed-Loop Estimation: To improve accuracy and robustness, an EKF is used for closed-loop estimation of the negative electrode potential. This algorithm uses the discrepancy between the model's predicted terminal voltage and the actual measured voltage as a feedback signal to continuously correct the estimation. This makes the model less susceptible to initial errors in state-of-charge (SOC) or state-of-health (SOH) estimations.
- Current Control Algorithm: Based on the real-time estimated negative electrode potential, a current control strategy is implemented. This strategy dynamically adjusts the charging current to ensure the negative electrode potential never drops below a critical threshold (50 mV in this study), thereby preventing lithium plating.
- Parameter Identification: The model parameters were identified using the Least Squares (LS) algorithm in MATLAB, trained on data from cells with implanted reference electrodes.
Datasets
The study utilized experimental data from a custom-built, long-dimensional LFP/graphite cell (188 Ah capacity). The data was collected under various conditions:
- Capacity tests: To establish baseline performance.
- Hybrid Pulse Power Characterization (HPPC) tests: Enhanced with multiple rate pulses at different temperatures to obtain model parameters.
- Multi-Rate Constant Current (MRCC) tests: To validate the model under static working conditions.
- New European Driving Test Cycle (NEDC) tests: To assess model performance under dynamic conditions.
The paper does not explicitly mention publicly available benchmark datasets; the experiments were conducted on their specifically prepared long-dimensional cells.
Key Insights and Ideas
- Spatial Potential Gradients are Critical: The study definitively shows that in long-dimensional cells, the negative electrode potential is not uniform. Specifically, the potential on the positive tab side is consistently lower (averaging 30 mV lower) than on the negative tab side during charging. This finding is crucial for understanding and mitigating lithium plating in large cells.
- Reference Electrodes are Feasible (and Necessary) for Large Cells: While not practical for every cell in a battery pack, implanting reference electrodes in representative large-format cells is a viable method for obtaining crucial internal potential data for model calibration and validation.
- EKF Significantly Improves Robustness: The closed-loop estimation using EKF effectively compensates for initial errors in SOC/SOH and sensor inaccuracies, making the model more reliable for real-world vehicle applications.
- Preventing Plating is Achievable with Potential Control: By actively monitoring and controlling the negative electrode potential, a fast-charging strategy can be implemented that avoids lithium plating, even at high charge rates.
Key Results
- High Accuracy: The OLFEM achieved an estimation error for negative electrode potential below 25 mV and for terminal voltage below 5 mV across various temperatures and charging rates.
- Robustness to Errors: Even with simulated initial errors in SOC/SOH (up to 20% capacity reduction or 5% SOC increase), the EKF-corrected model maintained low RMSE for negative electrode potential (within 20 mV) and terminal voltage (within 5 mV).
- Effective Fast Charging: The proposed strategy successfully charged the cell to 96.8% capacity within an hour (specifically, 183 Ah in 3555 s at 25°C, 0% SOC, with a 50 mV negative electrode potential limit).
- Improved Battery Lifespan: After 100 fast-charging cycles, the cell experienced less than 2% capacity degradation, and SEM analysis confirmed the absence of lithium plating.
Limitations and Future Work
- Reference Electrode Implantation: While successful, the implantation of reference electrodes is still a complex process and not feasible for every cell in a large battery pack.
- Model Applicability: While the authors suggest the calibration methods can be applied to other cell types, the primary focus and validation were on long-dimensional LFP cells.
- Future Work:
- Monitor and compare negative electrode potential in both length and width directions of large cells to pinpoint the absolute lowest potential region.
- Further investigate the model and charging strategy for the cell's entire cycle life.
- Explore applying these calibration methods to a wider variety of cell types and chemistries.
Summary updated