The global transition in energy structure and the pursuit of “dual carbon” targets have propelled the rapid expansion of energy storage systems, which serve as critical technological pillars for modern power grids. As a key enabler for integrating and accommodating renewable energy generation while enhancing grid flexibility, the demand for these systems is experiencing exponential growth. Among various technologies, electrochemical energy storage, particularly lithium-ion batteries, dominates the market due to its advantages in energy density, efficiency, cost, and deployment speed. The battery energy storage system, packaged typically within robust enclosures or cabinets, forms the core of such installations. The structural integrity of these battery cabinets is paramount, directly influencing the overall operational safety and longevity of the battery energy storage system.
With the globalization of the energy storage industry, the transportation of large-scale battery energy storage systems over long distances has become commonplace. Consequently, ensuring structural safety during transit has emerged as a significant concern. These cabinets exhibit particular sensitivity to vibration due to their design characteristics: the high energy density translates into substantial mass per unit, resulting in significant inertial forces that challenge structural integrity under dynamic loads far more than traditional electrical equipment. Transportation-induced vibrations can cause damage, ranging from bolt preload loss to severe structural failure. Of particular concern is the potential for invisible, cumulative damage that may not be immediately detectable but severely compromises the lifespan and safety of the battery energy storage system.
Existing research on the structural response of battery enclosures under transport conditions is relatively scarce, with most studies focused on vehicular battery packs. These investigations typically analyze dynamic responses under various conditions such as modal analysis, random vibration, shock, and drop/collision, followed by structural optimization. A common simplification in modeling involves representing battery cells as concentrated mass points or rigid bodies, with further simplifications applied to inter-cell connections. Some studies model the battery block as a homogeneous solid to create more refined pack models. While informative for automotive applications, the structural details, scale, and mounting boundaries of a stationary battery energy storage system cabinet differ significantly. Therefore, direct application of these models to large battery cabinets is not sufficiently validated.
To comprehensively understand the vibrational characteristics and dynamic structural response of a battery cabinet under simulated transport conditions, this study combines experimental testing with detailed numerical simulation. We conducted shaker table tests on a representative battery cabinet to extract its fundamental dynamic properties and observe its response to random vibration. Concurrently, we developed a high-fidelity finite element model of the cabinet. This model was subjected to modal analysis, sinusoidal sweep simulation, and random vibration simulation under conditions mirroring the experiments. The close agreement between simulation predictions and experimental measurements validates the model’s accuracy. This validated model provides a powerful tool for predicting vibrational behavior, assessing structural performance, and guiding the design optimization of future battery energy storage system cabinets, ultimately enhancing their transportation safety and reliability.
1 Theoretical Background of Vibration Testing
1.1 Response of Damped Systems to Harmonic Excitation
The equation of motion for a single-degree-of-freedom system with viscous damping under a harmonic load is given by:
$$m\ddot{u} + c\dot{u} + ku = p_0 \sin(\omega t)$$
with initial conditions \( u|_{t=0} = u(0) \) and \( \dot{u}|_{t=0} = \dot{u}(0) \). Here, \( m \) is mass, \( c \) is the damping coefficient, \( k \) is stiffness, \( u \) is displacement, \( \dot{u} \) is velocity, \( \ddot{u} \) is acceleration, \( p_0 \) is the amplitude of the harmonic force, and \( \omega \) is the excitation circular frequency.
The complete solution consists of transient and steady-state components:
$$u(t) = u_c + u_p = e^{-\zeta \omega_n t}(A \cos \omega_D t + B \sin \omega_D t) + C \sin \omega t + D \cos \omega t$$
where \( \omega_n = \sqrt{k/m} \) is the undamped natural frequency, \( \zeta = c / (2m\omega_n) \) is the damping ratio, and \( \omega_D = \omega_n \sqrt{1-\zeta^2} \) is the damped natural frequency. The constants A, B, C, and D are determined from initial conditions and system parameters. The transient component, vibrating at the system’s damped natural frequency \( \omega_D \), decays exponentially due to damping. The steady-state response persists and is given by:
$$u(t) = C\sin(\omega t) + D\cos(\omega t) = u_0 \sin(\omega t – \phi)$$
where \( u_0 \) is the amplitude of steady-state vibration and \( \phi \) is the phase angle.
The dynamic magnification factor \( R_d \), defined as the ratio of the dynamic displacement amplitude to the static displacement \( u_{st} = p_0/k \), is a key parameter:
$$R_d = \frac{u_0}{u_{st}} = \frac{1}{\sqrt{ [1 – (\omega/\omega_n)^2]^2 + [2\zeta (\omega/\omega_n)]^2 }}$$
Resonance occurs when the excitation frequency approaches the system’s natural frequency, leading to a peak in the response. For lightly damped systems typical in engineering, the resonance frequency is very close to \( \omega_n \). In our tests, the natural frequencies of the battery energy storage system cabinet are identified from the peak response amplitudes during sinusoidal sweep tests.
The damping ratio \( \zeta \) can be estimated from the frequency response curve using the half-power bandwidth method. The bandwidth \( \Delta f = f_2 – f_1 \) is measured at the response amplitude level of \( 1/\sqrt{2} \) (approximately 0.707) times the peak amplitude. The damping ratio is then calculated as:
$$\zeta \approx \frac{f_2 – f_1}{2 f_0}$$
where \( f_0 \) is the resonance (peak) frequency. This method is employed in this study to determine the damping characteristics of the battery cabinet.
1.2 Fundamentals of Random Vibration
Transportation vibration is a random process. Unlike deterministic vibration, random vibration cannot be described by a precise function of time but must be characterized using statistical methods. A random process \( X(t) \) is described by its probability distribution functions. For a stationary random process, the statistical properties are invariant with time shift. A crucial descriptor is the auto-correlation function \( R_X(\tau) \), which measures the correlation of the signal with itself at different time lags:
$$R_X(\tau) = E[X(t) X(t+\tau)]$$
where \( E[ ] \) denotes the expected value. To understand the frequency content, the Power Spectral Density (PSD) function \( S_{XX}(f) \) is used. It describes how the power (or mean square value) of the random process is distributed over frequency and is obtained via the Fourier Transform of the auto-correlation function. The transportation vibration environment for a battery energy storage system is typically specified by a PSD profile. This study employs the ISTA 3E spectrum, a standard test protocol by the International Safe Transit Association for packaged-products in transit, which is well-suited for simulating the vertical vibration experienced by large, heavy units like battery cabinets during long-haul truck transport.
2 Experimental Design and Setup
2.1 Test Specimen and Equipment
The test specimen was a commercial battery energy storage system cabinet. For safety, the coolant was drained, and the battery cells were discharged to a 5% State of Charge (SOC) prior to testing. Functional checks confirmed the cabinet was operational. The test was conducted using an electrodynamic vibration system with a high-performance digital controller. The system’s specifications ensured it could accommodate the large mass and deliver the required vibration profiles.
2.2 Fixturing and Boundary Conditions
To simulate the cabinet’s mounting within a larger energy storage system, a custom test fixture was designed and fabricated. The cabinet was mounted on a baseplate fixed to the shaker table. The connection between the cabinet and the baseplate differed at the front and rear: bolt connections were used near the front, while dowel pins provided locating and shear transfer near the rear. A top crossbeam, attached to the cabinet via bolts, was connected to triangular support structures. The orientation of these support structures was changed depending on the test direction: they were placed at the front and rear for excitation in the cabinet’s longitudinal (X) direction and on the sides for excitation in the lateral (Y) and vertical (Z) directions.
2.3 Instrumentation and Data Acquisition
Piezoelectric accelerometers were deployed at critical locations to capture the dynamic response. Control accelerometers were mounted on the shaker table platen. Response accelerometers were placed on the battery module end plates, the top crossbeam, beam connectors, and on internal components like the Battery Management System (BMS) and sensor modules. A triaxial accelerometer was mounted inside the cabinet. Throughout the tests, the cabinet’s internal parameters (e.g., cell voltages) were monitored via a dedicated data acquisition system to ensure operational safety.
| Parameter | Value |
|---|---|
| External Dimensions (L×W×H) | 1130 mm × 1203 mm × 2521 mm |
| Total Mass | 3920 kg |
| Battery Chemistry | Lithium-ion |
| Pre-test SOC | 5% |
2.4 Test Procedures
Sinusoidal Sweep Vibration Tests: Sweep tests were conducted separately in the longitudinal (X) and lateral (Y) directions. The frequency range was 5 Hz to 200 Hz with a constant acceleration amplitude of 0.1 g, using a logarithmic sweep rate of 1 octave per minute.
Random Vibration Test: A vertical (Z-direction) random vibration test was performed using the ISTA 3E PSD profile. The total test duration was 2 hours, with a mid-point inspection after 1 hour.
3 Experimental Results and Analysis
3.1 Sinusoidal Sweep Vibration Results
The acceleration frequency response curves from various sensor locations were analyzed. A representative curve from a sensor on the top crossbeam clearly showed a dominant peak, indicating the fundamental resonant frequency of the cabinet-fixture system in that direction.
Using the half-power bandwidth method on the response curves from multiple sensors, the fundamental natural frequencies and associated damping ratios for the cabinet in both directions were calculated. The results from several key sensors are summarized below. The final reported values are the statistical averages, providing a robust characterization of the battery energy storage system cabinet’s dynamic properties.
| Excitation Direction | Average Natural Frequency, \( f_0 \) (Hz) | Average Damping Ratio, \( \zeta \) |
|---|---|---|
| Longitudinal (X) | 7.485 | 0.051 |
| Lateral (Y) | 8.431 | 0.070 |
3.2 Random Vibration Test Results
During the 2-hour vertical random vibration test, the cabinet exhibited moderate shaking with audible but not severe impact noises. The acceleration time histories and PSDs from all sensors were recorded. Post-test visual inspection revealed no obvious structural damage or deformation to the cabinet itself, indicating good inherent robustness of the battery energy storage system enclosure for this specific transport simulation.
However, inspection of the test fixture revealed a notable finding: the bolts connecting the top crossbeam to the triangular supports had loosened. Furthermore, torque measurement of the cabinet’s mounting bolts to the baseplate showed a significant preload loss in one of the bolts (from 50 N·m to 23 N·m, a 54% reduction). This bolt, which is part of the cabinet’s installation system, experiences high stress due to the cabinet’s large mass under sustained random vibration, leading to preload relaxation. While this specific connection can be reinforced in practice, its loosening during the test highlights the importance of considering connection integrity in the design and assessment of battery energy storage system installations subjected to dynamic environments.
4 Finite Element Modeling and Simulation
4.1 Model Development
A detailed finite element model of the battery cabinet, including the internal module structure, external frame, and the test fixture, was developed. Non-structural components like cooling lines and small electronic boxes were omitted as the focus was on primary load-bearing structures. The model consisted of shell and solid elements, with a total of over 500,000 elements and 400,000 nodes. Material properties for steel (SPCC, Q355) and aluminum components were assigned based on standard values.
| Material | Density (t/mm³) | Young’s Modulus (MPa) | Poisson’s Ratio | Yield Strength (MPa) |
|---|---|---|---|---|
| SPCC Steel | 7.85e-9 | 210,000 | 0.3 | 200 |
| Q355 Steel | 7.85e-9 | 210,000 | 0.3 | 355 |
| Aluminum Plate | 2.77e-9 | 69,000 | 0.3 | – |
4.2 Boundary Conditions and Connections
The model’s boundary conditions replicated the experimental setup. The base of the fixture was fully constrained. The connections between the cabinet and the fixture base were modeled in detail: bolt connections were simulated using beam elements with appropriate pre-tension, while the dowel pin connections were modeled using surface-to-surface contact interactions with “hard” normal contact and “penalty” tangential friction.
4.3 Modal Analysis Results
A modal analysis was performed, incorporating the effects of contact and preload from the bolted connections. The first ten natural frequencies and mode shapes were extracted. The fundamental frequencies in the X and Y directions were identified and compared directly with the experimental results.
| Direction | Experimental \( f_0 \) (Hz) | Simulated \( f_0 \) (Hz) | Mode Order in FE Model | Error |
|---|---|---|---|---|
| Longitudinal (X) | 7.485 | 7.727 | 2 | +3.23% |
| Lateral (Y) | 8.431 | 8.746 | 2 | +3.74% |
The simulation results show excellent agreement with the experimental data, with errors less than 5%. This validates the accuracy of the finite element model in capturing the essential stiffness and mass distribution of the battery energy storage system cabinet and its fixture.
4.4 Sinusoidal Sweep Simulation
Using the validated FE model, forced harmonic response analyses were conducted. The same 0.1 g acceleration sweep from 5-200 Hz was applied as base excitation. The damping ratios obtained from the experiments (0.051 for X, 0.070 for Y) were assigned to the model. The acceleration frequency response at nodes corresponding to the physical sensor locations was extracted. The comparison for a representative sensor location shows a high degree of correlation between the simulated and experimental response curves in both magnitude and frequency characteristics, further confirming the model’s fidelity.
4.5 Random Vibration Simulation
A frequency-domain random response analysis was performed for the vertical (Z-direction) test condition. The ISTA 3E PSD profile was applied as the base input. The analysis provided statistical root-mean-square (RMS) values of stress and displacement.
The results indicated that the maximum displacement occurred at the top of the cabinet, with a value on the order of a few millimeters. The stress distribution showed elevated stress in the top crossbeam connection area of the fixture and at the lower rear crossbeam connection inside the cabinet. Crucially, the maximum calculated stress values were below the yield strength of the respective materials (159 MPa in the fixture, 114 MPa in the cabinet vs. 200+ MPa yield strengths). This linear elastic prediction aligns with the experimental observation of no permanent structural deformation. The simulation successfully identified the high-stress zones, which correlated with the locations where bolt loosening was observed (fixture top connections) or where high preload loss occurred (cabinet base connections), demonstrating the model’s utility in pinpointing potential weak links in a battery energy storage system under transport loads.
5 Conclusions
This study presents a comprehensive experimental and numerical investigation into the vibration characteristics of a large-scale battery energy storage system cabinet, focusing on its dynamic response under simulated transportation conditions. The key findings and contributions are as follows:
- Dynamic Characterization: Through controlled sinusoidal sweep vibration tests, the fundamental natural frequencies and damping ratios of the cabinet-fixture system were accurately determined. These parameters are essential for any dynamic analysis or design of a battery energy storage system.
- Transport Simulation Resilience: The vertical random vibration test based on the ISTA 3E spectrum demonstrated the inherent robustness of the cabinet’s primary structure, as no visible damage occurred. However, the test revealed vulnerability in bolted connections, highlighting a critical area for design attention in battery energy storage system packaging to prevent preload loss during transit.
- Validated High-Fidelity Model: A detailed finite element model was developed and rigorously validated against experimental data. The modal analysis showed less than 5% error in predicting the fundamental frequencies. The harmonic response simulations closely matched the measured acceleration frequency response curves. This validated model accurately represents the dynamic behavior of the battery energy storage system cabinet.
- Predictive Simulation Capability: The random vibration simulation, while based on linear frequency-domain methods, effectively identified areas of high stress concentration. These areas corresponded well with locations of observed connection issues (bolt loosening), proving the model’s value as a predictive design and assessment tool. It is acknowledged that nonlinear time-history analysis would be required to directly predict phenomena like bolt loosening or plastic damage.
The integrated approach of physical testing and correlated numerical simulation provides a robust methodology for evaluating and improving the structural integrity of battery energy storage system cabinets. The validated finite element model serves as a reliable digital twin, enabling engineers to predict vibrational responses, assess structural safety under various loading scenarios, and perform virtual design optimization for future generations of energy storage products. This work contributes directly to enhancing the transportation safety and long-term reliability of battery energy storage systems, supporting their secure integration into the global energy infrastructure.