In the development of electric vehicles (EVs), the battery pack is a critical component that stores energy and powers the vehicle. During operation, the EV battery pack is subjected to random vibrations induced by road irregularities, which are transmitted through the chassis. These vibrations can lead to fatigue damage, potentially causing structural failures, electrical disconnections, or thermal runaway, compromising safety and performance. Therefore, it is essential to conduct vibration fatigue analysis during the design phase to ensure the durability and reliability of the EV battery pack. In this article, I explore the vibration fatigue characteristics of an EV battery pack through finite element simulation, focusing on modal analysis, frequency response analysis, and combined multi-axis vibration fatigue assessment. The goal is to provide insights into simulation methodologies that can help optimize the design of EV battery packs, reduce development time, and cut costs while meeting regulatory standards.
The importance of vibration fatigue analysis for EV battery packs cannot be overstated. As EVs become more prevalent, their battery systems must withstand harsh environmental conditions, including mechanical vibrations from diverse road surfaces. Traditional physical testing is time-consuming and expensive, making simulation-based approaches like finite element analysis (FEA) invaluable. By using FEA, engineers can predict dynamic responses and fatigue life early in the design process, allowing for iterative improvements before prototyping. This article details a comprehensive simulation workflow, from model creation to fatigue evaluation, emphasizing the use of frequency-domain techniques and linear damage accumulation theory. The EV battery pack under study is modeled with detailed components, excluding non-critical parts like wiring harnesses, to balance accuracy and computational efficiency. Through this analysis, I aim to demonstrate how simulation tools can effectively assess the vibration fatigue performance of EV battery packs, ensuring they meet safety requirements such as those outlined in standards like GB 38031-2020.
Vibration fatigue analysis involves studying how structures respond to dynamic loads over time, leading to cumulative damage. For an EV battery pack, this requires understanding its dynamic characteristics under random excitation. The theoretical foundation includes frequency response analysis, power spectral density (PSD), and fatigue life prediction methods. Frequency response analysis, or sweep frequency analysis, computes the system’s response to harmonic excitation, revealing resonance frequencies and stress amplitudes. This is crucial for identifying critical modes that could amplify vibrations and accelerate fatigue. The equation of motion for a multi-degree-of-freedom damped system is given by:
$$M \ddot{x} + C \dot{x} + Kx = f(t)$$
where \(M\) is the mass matrix, \(C\) is the damping matrix, \(K\) is the stiffness matrix, \(\ddot{x}\), \(\dot{x}\), and \(x\) are acceleration, velocity, and displacement vectors, respectively, and \(f(t)\) is the external force vector. To solve this efficiently for large-scale models like an EV battery pack, the modal superposition method is often employed. This method transforms the physical coordinates into modal coordinates using the system’s natural modes, simplifying the equations. The transformation is expressed as:
$$x(t) = [\Phi] \{y\}$$
where \([\Phi]\) is the matrix of modal shapes and \(\{y\}\) is the modal coordinate vector. Substituting into the equation of motion and pre-multiplying by \([\Phi]^T\) yields decoupled equations:
$$[\Phi]^T M [\Phi] \{\ddot{y}\} + [\Phi]^T C [\Phi] \{\dot{y}\} + [\Phi]^T K [\Phi] \{y\} = [\Phi]^T f(t)$$
Assuming natural modes are orthogonal, \([\Phi]^T M [\Phi] = I\), \([\Phi]^T K [\Phi] = \omega^2\), and \([\Phi]^T C [\Phi] = 2\xi\omega\), where \(\omega\) is the angular frequency and \(\xi\) is the damping ratio. This reduces to a set of single-degree-of-freedom equations:
$$\ddot{y} + 2\xi\omega \dot{y} + \omega^2 y = [\Phi]^T f(t)$$
For frequency response analysis, a unit harmonic excitation is applied, and the system’s transfer function \(H(\omega)\) is obtained. The response in the frequency domain can then be used for fatigue assessment. Power spectral density (PSD) is a key concept in random vibration analysis, representing the distribution of vibration energy across frequencies. For a stationary random process \(X(t)\), the PSD \(S(\omega)\) is derived from the autocorrelation function \(R(\tau)\) via Fourier transform:
$$S(\omega) = \frac{1}{2\pi} \int_{-\infty}^{+\infty} R(\tau) e^{-j\omega\tau} d\tau$$
In practice, PSD profiles are defined based on standards like GB 38031-2020, which specify acceleration PSD levels for different frequency ranges to simulate real-world road conditions. For the EV battery pack, these profiles are applied in three orthogonal directions (X, Y, Z) to account for multi-axis loading.
Fatigue life prediction uses the stress responses from frequency analysis to estimate damage accumulation. The nCode software, employed in this study, utilizes the Dirlik method for PSD counting and the Goodman correction for mean stress effects. The stress response \(X_{\text{resp}}\) due to acceleration PSD is calculated as:
$$X_{\text{resp}} = |H(\omega)|^2 S(\omega)$$
The cumulative fatigue damage \(D\) is evaluated using Miner’s linear damage rule:
$$D = \sum_{i=1}^{n} \frac{n_i}{N_i}$$
where \(n_i\) is the number of cycles at stress amplitude \(\sigma_i\), and \(N_i\) is the number of cycles to failure at that amplitude from the material’s S-N curve. If \(D \geq 1\), failure is predicted. This approach allows for efficient fatigue simulation in the frequency domain, avoiding time-consuming time-history analysis.
To conduct the vibration fatigue analysis, a detailed finite element model of the EV battery pack was created using ANSA software. The model includes key components such as the battery cells, enclosure, cooling plates, and structural supports, while omitting minor parts like cables to simplify computation. The mesh was generated with an appropriate element size to ensure accuracy, and connections between parts were modeled using rigid elements and contacts in OptiStruct. The material properties for major components are summarized in Table 1, which highlights the use of aluminum alloys for their lightweight and strength characteristics, crucial for EV battery pack design.
| Component Name | Material Name | Yield Strength (MPa) | Tensile Strength (MPa) |
|---|---|---|---|
| Enclosure and Brackets | AL6061 | 240 | 290 |
| Cell Housings | AL3003 | 125 | 140 |
| Cooling Plates, Module End Plates, Side Plates | AL5083 | 150 | 320 |
The finite element model consists of shell and solid elements, with a total of over 500,000 nodes and elements to capture detailed stress distributions. The EV battery pack is constrained at the mounting points where it attaches to the vehicle chassis, simulating real boundary conditions. Modal analysis was performed to extract natural frequencies and mode shapes up to 200 Hz, as higher frequencies are less critical for road-induced vibrations. The results, shown in Table 2, indicate several modes within this range, with the first global mode occurring around 45 Hz, involving significant mass participation in Z-direction translation and Y-axis rotation. This information is vital for understanding potential resonance risks in the EV battery pack.
| Mode Number | Frequency (Hz) | X-Translation Mass Participation | Y-Translation Mass Participation | Z-Translation Mass Participation | X-Rotation Mass Participation | Y-Rotation Mass Participation | Z-Rotation Mass Participation |
|---|---|---|---|---|---|---|---|
| 1 | 30.5 | 2.31e-11 | 1.19e-8 | 2.61e-8 | 5.56e-4 | 1.88e-8 | 3.05e-7 |
| 2 | 32.1 | 1.01e-10 | 3.27e-7 | 3.58e-7 | 6.82e-3 | 3.01e-7 | 1.84e-7 |
| 3 | 34.1 | 8.01e-5 | 2.39e-11 | 6.81e-3 | 2.96e-8 | 2.60e-3 | 7.56e-12 |
| 4 | 35.7 | 1.45e-5 | 1.29e-10 | 4.28e-2 | 3.94e-8 | 4.48e-2 | 1.44e-10 |
| 5 | 45.0 | 1.54e-5 | 4.78e-9 | 8.11e-1 | 4.67e-7 | 7.88e-1 | 1.07e-8 |
| 6 | 58.3 | 2.97e-7 | 5.38e-5 | 1.41e-5 | 1.96e-3 | 2.07e-5 | 4.85e-5 |
| 7 | 58.7 | 1.16e-4 | 1.92e-7 | 4.48e-3 | 6.45e-6 | 6.79e-3 | 1.80e-7 |
| 8 | 62.3 | 1.16e-2 | 1.47e-7 | 4.20e-4 | 1.82e-5 | 3.12e-2 | 2.31e-7 |
| 9 | 63.8 | 1.05e-7 | 8.55e-5 | 7.33e-9 | 2.07e-3 | 4.98e-7 | 9.23e-5 |
| 10 | 64.3 | 5.39e-7 | 5.02e-5 | 3.60e-8 | 1.09e-4 | 1.46e-6 | 1.21e-4 |
| 11 | 68.1 | 2.29e-8 | 9.65e-4 | 7.11e-9 | 2.58e-2 | 4.83e-8 | 8.87e-4 |
| 12 | 68.4 | 9.66e-6 | 5.90e-9 | 6.78e-5 | 1.55e-7 | 1.30e-4 | 8.88e-9 |
| 13 | 71.2 | 8.36e-11 | 1.74e-6 | 2.14e-11 | 8.72e-6 | 5.60e-9 | 3.07e-6 |
| 14 | 71.4 | 3.38e-4 | 1.93e-7 | 6.53e-6 | 4.39e-6 | 8.27e-7 | 2.14e-7 |
| 15 | 71.9 | 1.45e-5 | 1.97e-6 | 4.09e-4 | 4.23e-5 | 1.75e-3 | 1.96e-6 |
Frequency response analysis was conducted next, applying a unit sinusoidal acceleration excitation from 0 to 200 Hz in each direction (X, Y, Z) at the mounting points. The damping ratio was set to 0.02, a typical value for battery systems, to avoid unrealistic peaks at resonance. The transfer functions for stress were computed using modal superposition, and results for a sample element on the cover plate are plotted in Figures 1-3, showing stress amplitude versus frequency. These plots reveal distinct peaks at frequencies corresponding to modal values, such as 45 Hz for Z-direction excitation, aligning with the global mode. This correlation confirms that resonance effects dominate the dynamic response of the EV battery pack, underscoring the need for careful design to mitigate fatigue risks.
For multi-axis vibration fatigue analysis, the frequency response results were imported into nCode along with PSD profiles from GB 38031-2020. The standard specifies combined random and constant-frequency vibration tests, as detailed in Tables 3 and 4. Random vibration involves PSD levels over 12 hours per direction, while constant-frequency vibration applies a fixed amplitude at 24 Hz for 2 hours per direction. These conditions simulate long-term road exposure and specific resonant excitations, respectively, providing a comprehensive fatigue assessment for the EV battery pack.
| Frequency (Hz) | Z-axis PSD (g²/Hz) | Y-axis PSD (g²/Hz) | X-axis PSD (g²/Hz) |
|---|---|---|---|
| 5 | 0.0150 | 0.00200 | 0.00600 |
| 10 | — | 0.00500 | — |
| 15 | 0.0150 | — | — |
| 20 | — | 0.00500 | — |
| 30 | — | — | 0.00600 |
| 65 | 0.0010 | — | — |
| 100 | 0.0010 | — | — |
| 200 | 0.0001 | 0.00015 | 0.00003 |
| Frequency (Hz) | Z-axis Amplitude (g) | Y-axis Amplitude (g) | X-axis Amplitude (g) |
|---|---|---|---|
| 24 | ±1.5 | ±1.0 | ±1.0 |
In nCode, a duty cycle load spectrum was created to combine these two vibration types, and the Dirlik method was used to estimate cycle counts from the PSD. The material S-N curves were generated automatically based on tensile strength inputs, with Goodman correction applied. The fatigue damage contour plot, shown in Figure 4, indicates maximum damage values near structural discontinuities or connection points, but all are below 0.0357, well under the Miner failure threshold of 1. This suggests that the EV battery pack design has sufficient fatigue resistance under the specified conditions. However, it is important to note that simulation accuracy depends on factors like mesh quality, material models, and loading assumptions; thus, physical testing is recommended for validation.
The results demonstrate the effectiveness of simulation-based vibration fatigue analysis for EV battery packs. By integrating modal, frequency response, and fatigue simulations, engineers can identify weak points and optimize designs early. For instance, the global mode at 45 Hz could be shifted by stiffening the enclosure or adjusting mass distribution, reducing resonance effects. Additionally, the use of multi-axis loading accounts for real-world vibration complexity, offering a more realistic assessment than single-axis tests. This approach aligns with industry trends toward digital twin and virtual prototyping, which are crucial for accelerating EV development. Future work could include nonlinear effects, thermal-mechanical coupling, or correlation with experimental data to further refine the models.
In conclusion, vibration fatigue analysis is a vital step in ensuring the durability and safety of EV battery packs. Through detailed finite element modeling and frequency-domain simulation, this study has shown that the EV battery pack meets regulatory fatigue requirements under combined random and constant-frequency vibrations. The methodologies outlined here—including modal analysis, frequency response, and nCode fatigue simulation—provide a robust framework for evaluating and improving EV battery pack designs. As EV technology advances, such simulation tools will play an increasingly important role in delivering reliable and high-performance energy storage systems, contributing to the broader adoption of electric vehicles worldwide.