Charging time and loss optimization for LiNMC and LiFePO4 batteries based on equivalent circuit models

Xiao Hu, Shengbo Eben Li, Huei Peng, Fengchun SunJournal of Power Sources 2013 · DOI

TL;DR

This paper tackles the critical challenge of optimizing electric vehicle battery charging by developing a dual-objective strategy that balances charging speed and energy loss. By employing sophisticated battery models and advanced control techniques, the authors demonstrate how to achieve a more efficient and user-friendly charging experience, especially considering factors like battery age and temperature.

What Problem Does This Paper Address?

The widespread adoption of electric vehicles (EVs) hinges on addressing two key charging concerns: charging time and energy loss. Drivers want to recharge their vehicles quickly to minimize inconvenience, similar to refueling a gasoline car. However, faster charging often leads to increased energy dissipation as heat (energy loss), which not only wastes electricity but can also accelerate battery degradation. Traditional Constant-Current Constant-Voltage (CCCV) charging methods are simple but don't explicitly optimize for these conflicting objectives. While various advanced charging strategies exist, few have explicitly formulated and solved the problem of simultaneously minimizing charging time and energy loss, especially considering the complex dynamics of modern Li-ion batteries like LiNMC and LiFePO4. This paper aims to fill that gap by developing and evaluating a multi-objective optimal charging control strategy.

Methods

The core of this paper lies in formulating a dual-objective optimal charging control problem. The authors leverage detailed equivalent circuit battery models for LiNMC and LiFePO4 chemistries, which were developed and validated in their prior work. These models capture the internal states of the battery, allowing for more accurate predictions of charging behavior.

The optimization problem is defined with two conflicting objectives: minimizing charging time ($t_f$) and minimizing total energy loss ($\int P_{loss} dt$). A weighting factor ($\alpha$) is introduced to balance these objectives, allowing for a trade-off between speed and efficiency.

To solve this complex, nonlinear optimal control problem, the authors employ the pseudo-spectral technique, specifically using the Radau Pseudo-Spectral Method (RPM). This method discretizes the continuous-time dynamics into a nonlinear programming (NLP) problem, which is then solved using a powerful NLP solver (SNOPT). The use of an hp-adaptive mesh refinement algorithm further enhances the accuracy and efficiency of the solution by iteratively adjusting the discretization.

Datasets

The study utilizes battery data collected from testing LiNMC (UR14650P) and LiFePO4 (APR18650M1A) cells. The testing schedule, detailed in Figure 1, includes characterization tests at various temperatures (10°C, 22°C, 35°C), static capacity tests, hybrid pulse tests, DC resistance tests, DST tests, FUDS tests, and impedance tests. This comprehensive data collection allows for the development and validation of the accurate battery models used in the optimization.

Key Insights and Ideas

  • The inherent conflict: Charging time and energy loss are fundamentally at odds. Faster charging generally means higher currents, leading to increased ohmic losses ($I^2R$).
  • The power of advanced models: Using detailed equivalent circuit models (like RC and RC-plus-hysteresis) is crucial for accurately capturing battery dynamics and achieving a truly optimal tradeoff, as simpler models can lead to suboptimal results.
  • Temperature matters: Higher (but safe) operating temperatures can be beneficial for charging efficiency, as they reduce battery resistance, leading to lower energy loss and potentially faster charging.
  • Aging is a significant factor: As batteries age, both their capacity and internal resistance change, leading to longer charging times and increased energy loss. This necessitates recalibrating charging strategies for aged batteries.
  • Voltage threshold is a key lever: Using a lower upper cut-off voltage (closer to the Open Circuit Voltage at the target State of Charge) can reduce energy loss, though it might slightly increase charging time.

Key Results

  • The developed dual-objective optimization strategy successfully generates Pareto fronts (Figure 2, 4, 6) illustrating the trade-off between charging time and energy loss for both LiNMC and LiFePO4 batteries across different temperatures and aging levels.
  • At 22°C, using the upper cut-off voltage (4.2V for LiNMC, 3.6V for LiFePO4) resulted in lower energy losses compared to using a higher voltage limit.
  • Lower temperatures (e.g., 10°C) generally led to longer charging times and higher energy losses compared to higher temperatures (e.g., 35°C). For instance, at 35°C, the minimum energy loss for a 10-hour charge was lower than at 10°C.
  • Battery aging increased both charging time and energy loss. For example, an aged LiNMC battery showed a decrease in optimal charging performance compared to a new battery.
  • Comparison with simpler models (IR model for LiNMC, first-order RC for LiFePO4) showed that while they might offer seemingly better individual metrics (e.g., faster time), they resulted in a non-optimal tradeoff between the two objectives when applied to more complex battery models.

Limitations and Future Work

  • Thermal management assumption: The study assumes an effective thermal management system is available to maintain a constant battery temperature during charging. In reality, charging generates heat, and this dynamic thermal behavior could influence optimal strategies.
  • Model complexity vs. controller capability: While more complex models (RC-plus-hysteresis) provide better accuracy, their computational demands might be prohibitive for some embedded charge controllers. The paper acknowledges that simpler models might be acceptable when computational resources are limited.
  • Future work could explore:
    • Integrating dynamic thermal modeling into the optimization.
  • Investigating adaptive charging strategies that can continuously recalibrate based on real-time battery state and aging.
  • Further exploring the impact of different charging protocols and their interaction with battery degradation mechanisms.
  • Real-world experimental validation of the proposed strategy.

Summary updated

BibTeX
@article{hu2013charging,
  author = {Xiao Hu and Shengbo Eben Li and Huei Peng and Fengchun Sun},
  title = {{Charging time and loss optimization for LiNMC and LiFePO4 batteries based on equivalent circuit models}},
  journal = {Journal of Power Sources},
  year = {2013},
  doi = {10.1016/j.jpowsour.2013.03.157},
}
Figure