With the global push towards carbon neutrality and peak carbon emissions, the green energy industry has experienced rapid growth. New energy vehicles, as crucial carriers in this transition, hold a strategic position in transportation. According to statistics, the number of new energy vehicles in China reached 13.1 million by the end of 2022, reflecting a year-on-year growth of 67.13%, indicating a high-speed expansion trend. Lithium iron phosphate (LFP) batteries, serving as efficient energy storage devices for electric vehicles, offer advantages such as high energy density, long service life, and strong temperature adaptability. However, after prolonged and frequent use, LFP batteries inevitably face issues like increased internal resistance and capacity degradation due to aging, which constrain the lifespan, safety, and power performance of the EV battery pack, thereby impacting consumer experience. Typically, automotive manufacturers warranty the EV battery pack based on national regulations, requiring that the battery’s state of health (SOH) remains above 80% after 8 years or 200,000 kilometers of use. Thus, both vehicle manufacturers and battery producers need to assess the warranty lifespan during the product design and development phase.
Extensive research has been conducted worldwide on evaluating the warranty lifespan of EV battery packs, focusing on cell cycle aging, calendar aging mechanisms, and vehicle-level durability validation under driving conditions. Mainstream aging models include mechanistic models and empirical models. Mechanistic models emphasize modeling internal side reactions, incorporating factors such as lithium-ion intercalation/deintercalation, reaction heat, and charge transfer, resulting in complex multi-parameter models with intertwined reactions. Due to significant differences in electrochemical reactions among various lithium-ion battery types, parameter identification for these models is challenging, limiting their generalizability in practical engineering applications. Among empirical models, the equivalent circuit model is most common. It estimates battery aging by deriving parameters like internal resistance and state of charge (SOC) through equivalent circuits, combined with the Arrhenius equation from chemical kinetics to fit the relationship between battery capacity and time. Additionally, to accurately describe the aging state and parameter changes during aging, correction factors can be added to the Arrhenius equation. This model features simple parameter identification and low computational load, making it a primary method for assessing battery warranty lifespan. With advancements in digital technology, battery aging simulation software based on the Arrhenius equation has rapidly developed and found engineering applications. During product development, AMESim simulation software, which couples data analysis with empirical aging models, plays a significant role in battery life prediction. This software offers practical physical model modules, enabling the establishment of dynamic simulation models based on test data to effectively reflect real vehicle driving conditions. In this study, we leverage AMESim to build a one-dimensional electrochemical-thermal coupling model for predicting the warranty lifespan of an LFP EV battery pack. By analyzing test data from 15 Ah LFP cells and pack-level measurements, we extract key parameters, validate model accuracy, and simulate long-term performance under realistic driving scenarios.
Our research begins with characterizing the fundamental parameters of the cells and the EV battery pack. The cells used are cylindrical LFP cells with a nominal capacity of 15 Ah, produced in the same batch. The EV battery pack is assembled from four modules, each consisting of 90 cells connected in a 10-parallel 9-series configuration, resulting in a total nominal capacity of 150 Ah and a voltage of 115.2 V. The assembly process involves grouping cells into modules and then integrating modules into the final pack structure. The basic specifications are summarized in Table 1.
| Parameter | Cell Value | Battery Pack Value |
|---|---|---|
| Nominal Capacity | 15 Ah | 150 Ah |
| Nominal Voltage | 3.2 V | 115.2 V |
| Configuration | — | 10P36S |
| Material System | LFP/Graphite | |
| Low-Temperature Heating | — | Heating Film |
| Operating Temperature | 0 to 55 °C | |
| Maximum Charge Rate | 0.68 C | |
| Shape | Cylindrical | — |
To capture the electrochemical behavior, we performed hybrid pulse power characterization (HPPC) tests on individual cells at temperatures of -20, -10, 0, 25, 45, and 55 °C. These tests provide data on internal resistance and open-circuit voltage (OCV) as functions of SOC and temperature. For aging characterization, cycle life tests were conducted at 25, 45, and 55 °C with a voltage range of 2.5 to 3.65 V, using steel clamps to simulate mechanical constraints within the pack. Calendar aging tests were performed at 25 °C at various SOC levels (100%, 80%, 50%, 30%, 5%). Additionally, the assembled EV battery pack underwent cycle testing at 25 °C with a 1 C rate charge-discharge protocol.
The aging of the EV battery pack is governed by both calendar and cycle aging mechanisms. Calendar aging refers to irreversible capacity loss during storage due to side reactions like self-discharge, primarily influenced by storage temperature T and SOC. The capacity fade due to calendar aging is modeled using an Arrhenius-based equation:
$$ Q_{\text{loss-cal}} = A(\text{SOC}) \times \exp \left( -\frac{E_a}{RT} \right) \times t^z $$
where \( Q_{\text{loss-cal}} \) is the percentage capacity loss, \( A(\text{SOC}) \) is the pre-exponential factor dependent on SOC, \( E_a \) is the activation energy (J/mol), \( R \) is the gas constant (8.314 J/(mol·K)), \( T \) is the absolute temperature (K), \( t \) is the aging time in days, and \( z \) is the power factor.
Cycle aging results from charge-discharge processes during vehicle operation, with key factors being charge rate \( C_{\text{rate}} \) and pack temperature \( T \). The capacity fade due to cycle aging is expressed as:
$$ Q_{\text{loss-cyc}} = B(I) \times \exp \left( -\frac{E_a + \alpha C_{\text{rate}}}{RT} \right) \times (N \times \text{DOD} \times Q)^z $$
where \( Q_{\text{loss-cyc}} \) is the percentage capacity loss, \( B(I) \) is the pre-exponential factor related to current \( I \), \( \alpha \) is an aging factor, \( C_{\text{rate}} \) is the charge rate, \( N \) is the number of cycles, DOD is the depth of discharge (%), \( Q \) is the nominal capacity, and \( z \) is the power factor.
For simulation, we adopted a second-order RC equivalent circuit model to represent the dynamic behavior of the LFP battery. This model includes ohmic resistance and polarization resistance, capturing voltage drops during transient operations. Using AMESim software, we constructed an electrochemical-thermal coupling model for the EV battery pack, integrating the electrical model with thermal dynamics. The main simulation model comprises several modules: battery module 1 (representing a well-cooled section), battery module 2 (representing a poorly cooled section), heating module, charging module, driving profile module, and heat exchange module. This setup allows us to simulate real-world conditions, including temperature variations across the pack.
The internal resistance of the battery cell is highly temperature-dependent. From HPPC data, we extracted ohmic and polarization resistances at different temperatures and SOC levels. For instance, at 25 °C, the simulated terminal voltage based on the second-order RC model closely matched experimental data, with a root mean square error of only 46.86 mV. The resistance trends showed that both ohmic and polarization resistances increase as temperature decreases, due to reduced electrochemical reaction rates and higher electrolyte viscosity at low temperatures. Moreover, polarization resistance rises significantly at low SOC during high-rate discharge, leading to pronounced voltage drops. These resistance characteristics are critical for accurately modeling the EV battery pack’s performance under varying operating conditions.
The open-circuit voltage (OCV) also varies with temperature. By analyzing HPPC data, we obtained OCV-SOC relationships at different temperatures. Simulations showed that OCV decreases slightly with increasing temperature, and the simulated values aligned with measured OCV data within 1% error. This precision ensures reliable SOC estimation in the model, which is vital for lifespan prediction.
Fitting the aging models to experimental data, we derived parameters for calendar and cycle aging. For calendar aging at 25 °C, the pre-exponential factor \( A(\text{SOC}) \) varied with SOC, as listed in Table 2, with \( E_a = 31,700 \) J/mol and \( z = 0.466 \). The model effectively captured capacity fade over storage time.
| SOC (%) | \( A(\text{SOC}) \) | \( E_a \) (J/mol) | \( z \) |
|---|---|---|---|
| 100 | 310 | 31,700 | 0.466 |
| 80 | 240 | ||
| 50 | 210 | ||
| 30 | 195 | ||
| 5 | 150 |
For cycle aging, tests at 25, 45, and 55 °C revealed accelerated capacity fade at higher temperatures, consistent with SEI layer growth dominant degradation. The fitted parameters were \( B(I) = 470 \), \( z = 0.92 \), and \( \alpha \) incorporated into the Arrhenius term. These parameters enabled the model to predict capacity loss under various cycling conditions.
Thermal characteristics of the EV battery pack are crucial for lifespan simulation. We calibrated the convective heat transfer coefficients in the model by comparing simulated temperature profiles with experimental data from different operating scenarios: low-temperature slow charging, room-temperature discharge, high-temperature discharge, and low-temperature standby. Temperature sensors placed at the hottest (sensor 8#) and coldest (sensor 3#) locations within the pack provided validation data. The simulation accurately reproduced temperature trends, with maximum errors below 3 °C, confirming the model’s ability to predict internal temperature gradients in the EV battery pack.
To validate the overall model accuracy, we conducted an accelerated cycle test on the EV battery pack at 25 °C, measuring capacity retention over cycles. After 229 cycles, the experimental capacity retention was 95.84%, while the simulation predicted 95.63%, an error of only 0.3%. Temperature profiles during charge and discharge also matched closely, with simulated temperatures deviating by less than 2 °C from measurements. This high precision underscores the reliability of our electrochemical-thermal coupling model for EV battery pack lifespan assessment.
Further simulation over 1,200 cycles examined the impact of increasing internal resistance on pack temperature. Results indicated that as resistance grows with aging, the temperature rise during discharge increases slightly (e.g., from 14.2 °C to 16.5 °C for a 20% resistance increase), but the overall temperature distribution remains relatively stable. This suggests that resistance growth alone does not drastically alter the thermal behavior of the EV battery pack, but temperature non-uniformities can exacerbate aging.
For warranty lifespan prediction, we simulated the EV battery pack under the CLTC-P (China Light-duty Vehicle Test Cycle-Passenger) driving cycle, which includes urban, suburban, and highway segments with a total duration of 1,800 s and distance of 14.48 km. The simulation incorporated real-world usage patterns: daily commuting of 57.92 km (round trip), charging every three days with a 2:1 ratio of fast to slow charging, and ambient temperature conditions representative of Hefei city (minimum -1 °C, annual variation 27 °C, daily variation 7 °C). The thermal management strategy was also integrated: when the minimum temperature \( \theta_{\text{min}} \leq 5 \) °C and temperature difference \( \theta_{\text{diff}} < 18 \) °C, a PTC heating film activates using external power; heating stops when \( \theta_{\text{min}} > 10 \) °C or \( \theta_{\text{diff}} > 20 \) °C, after which charging proceeds. This strategy ensures the EV battery pack operates within a safe temperature range.
The simulation projected capacity and internal resistance changes over 8 years or 200,000 kilometers. Results showed that the EV battery pack experienced a capacity fade of 17.4%, meeting the warranty target of SOH > 80%. However, module 2 (poorer cooling) exhibited faster degradation than module 1 (better cooling), attributed to temperature differences of up to 5 °C during discharge and 4 °C during charge within the pack. This non-uniformity contributed to an additional 1.9% capacity loss, highlighting the importance of thermal management design in minimizing temperature gradients for extending the lifespan of the EV battery pack.
In summary, we developed a high-fidelity electrochemical-thermal coupling model using AMESim for predicting the lifespan of an LFP EV battery pack. By integrating cell-level test data, aging models, and pack thermal dynamics, the model accurately simulated performance under realistic conditions. Validation against experimental data confirmed its precision, with capacity prediction errors below 0.5%. The warranty simulation under CLTC-P cycling demonstrated that the EV battery pack can retain over 82.6% capacity after 8 years or 200,000 km, complying with industry standards. Key findings emphasize that internal temperature disparities accelerate aging by increasing resistance, underscoring the need for effective thermal management in EV battery pack design. Future work could explore optimizing cooling strategies and incorporating more detailed aging mechanisms to further enhance prediction accuracy for diverse operating environments.