Aging models for second-life batteries play a critical role in determining their viability for repurposed applications such as stationary energy storage. These models focus on estimating residual capacity and forecasting performance to ensure safe and economically feasible deployment. Unlike first-life models, which predict degradation in virgin batteries, second-life models must account for prior usage history, varying operational conditions, and the potential for accelerated aging mechanisms.
The foundation of second-life aging models lies in leveraging first-life degradation data. Batteries retired from electric vehicles (EVs) typically retain 70-80% of their original capacity, making them suitable for less demanding applications. However, their degradation trajectories in second-life scenarios differ due to changes in charge-discharge cycles, environmental conditions, and load profiles. First-life data, including cycle count, depth of discharge (DoD), temperature exposure, and state-of-charge (SoC) windows, provides essential inputs for remapping capacity fade and impedance growth in second-life applications.
Capacity remapping techniques adjust first-life degradation curves to reflect second-life conditions. One approach involves using empirical models that correlate historical stress factors with capacity loss. For example, if a battery experienced high-temperature operation during its first life, its Arrhenius-based aging rate can be recalibrated for stationary storage, where thermal management may be more controlled. Physics-based models, such as those incorporating porous electrode theory, can also be adapted by updating boundary conditions to reflect second-life usage patterns.
Machine learning techniques enhance residual capacity estimation by training on first-life datasets to predict second-life performance. Neural networks and Gaussian process regression have been applied to model nonlinear aging behaviors, particularly when dealing with heterogeneous battery populations. These data-driven methods excel in scenarios where precise historical data is available, enabling accurate state-of-health (SoH) predictions even with varying usage histories.
Performance forecasting in second-life batteries must account for application-specific demands. Stationary storage systems, for instance, often operate at partial DoD and lower C-rates compared to EVs. Aging models must integrate these operational profiles to project cycle life and calendar aging. Hybrid models combining empirical, physical, and data-driven approaches offer robust forecasts by balancing mechanistic insights with real-world variability.
Economic considerations heavily influence the development of second-life aging models. Accurate predictions are necessary to determine the remaining useful life (RUL) and optimize financial returns. Overestimating RUL can lead to premature system failures, while underestimating may result in unnecessary retirement of viable batteries. Levelized cost of storage (LCOS) calculations rely on these models to assess whether second-life batteries are cost-competitive with new systems.
Safety considerations are equally critical. Second-life batteries may exhibit unpredictable failure modes due to prior degradation. Aging models must identify risks such as lithium plating, separator degradation, or thermal runaway precursors. Incorporating safety margins into performance forecasts ensures that repurposed systems operate within reliable thresholds. Abuse testing data from first-life stages can inform these safety assessments, highlighting potential weak points under second-life conditions.
Validation of second-life aging models requires real-world testing under representative conditions. Accelerated aging tests alone may not capture the long-term effects of repurposing, necessitating field data from pilot deployments. Cross-validation between laboratory and operational datasets improves model accuracy and reliability.
In summary, aging models for second-life batteries bridge the gap between first-life degradation and repurposed performance. By integrating historical data, remapping capacity fade, and incorporating economic and safety constraints, these models enable the sustainable and profitable reuse of retired EV batteries. Continued refinement through machine learning and hybrid modeling approaches will further enhance their predictive power, supporting the growing market for second-life energy storage solutions.