In the pursuit of global “dual carbon” strategic goals, lithium-ion energy storage batteries have emerged as the core energy carrier for large-scale energy storage applications due to their high energy density and long cycle life. However, the safety incidents caused by thermal runaway (TR) and its rapid propagation have become a critical bottleneck that restricts the high-quality development of the energy storage industry. The rational selection of thermal insulation materials and the optimization of their thickness are key technical methods to block the propagation path of thermal runaway within a battery pack. In this study, we focus on suppressing TR propagation by constructing a systematic research framework of “experimental validation – simulation exploration – optimization iteration” to investigate the selection of insulation materials and the optimization of their thickness.
We selected eight typical insulation materials, including aerogel, ceramic fiber mat, and mica sheet, and constructed an experimental platform for a 314 Ah Lithium Iron Phosphate (LFP) battery pack. Critical thermal runaway propagation data were obtained through single-cell overcharge triggering experiments. Based on the experimental data, a coupled thermal-electrical-chemical simulation model was calibrated, and the influence of material thermophysical parameters and thickness gradients on the heat transfer path was systematically analyzed. By combining the critical experimental data with quantitative simulation results, a “material performance – thickness – suppression effect” correlation model was established, leading to the proposal of an optimal insulation scheme. The results show that with a 1.2 mm thick aerogel interlayer, the peak temperature of the adjacent cell is reduced to 99.96°C, the thermal runaway triggering delay time extends to over 30 minutes, and the heat flux density attenuation rate reaches 89%, thereby completely suppressing thermal runaway propagation. The deviation between the simulation predictions and the experimental data is less than 5%, verifying the reliability of the model. This study provides an experimental basis and a simulation-based optimization method for the safe design of energy storage battery packs, offering direct reference value for their thermal safety optimization and engineering applications.
Keywords: energy storage battery pack; thermal runaway propagation; thermal insulation material; thickness optimization; multi-physics simulation; thermal management
1. Introduction
Driven by the global strategic goals of “carbon peak and carbon neutrality”, renewable energy sources such as wind and solar energy, with their advantages of being clean and low-carbon, are continuously increasing their grid integration ratio, becoming a core force in the transformation of the energy structure. However, the output of renewable energy is characterized by strong volatility and intermittency, which severely restricts the frequency and voltage regulation capabilities of the power grid. Energy storage systems, as key equipment for smoothing output fluctuations and enhancing grid flexibility and stability, have experienced explosive growth in recent years. According to mainstream industry reports, the global energy storage market scale has continued to rise in recent years, surpassing the $300 billion mark. Among these, lithium-ion batteries, with their advantages of high energy density, long cycle life, and high charge-discharge efficiency, account for more than 85% of the technical application share in the energy storage market. They have become the mainstream energy storage carrier in the field of large-scale energy storage.
A battery pack, as the core energy unit of an energy storage system, integrates tens to thousands of individual cells in series and parallel. Its energy density directly determines the installed capacity and cost-effectiveness of the energy storage system. In a typical 20-foot containerized energy storage system, the number of integrated packs can exceed 200 units, with a total energy of over 5 MWh. In pursuit of high energy density, individual cells are arranged closely, with a minimum spacing between cells of 2-5 mm, resulting in a very strong thermal coupling effect and creating an inherent structural hazard for thermal runaway propagation. As energy storage projects are upgraded towards “high energy density and high integration”, the risk of heat accumulation inside the pack increases significantly, further exacerbating the difficulty of thermal management. The prevention and control of thermal runaway propagation has become a core technical bottleneck restricting the high-quality development of the energy storage industry.
Although LFP batteries currently widely used in the energy storage field have higher thermal decomposition temperatures and lower heat generation rates during thermal runaway compared to ternary lithium batteries, offering significant advantages in safety and cost, they can still undergo violent reactions such as positive electrode material decomposition and electrolyte vaporization under abusive conditions like overcharging, external short circuits, and mechanical impact, leading to thermal runaway. After a single cell undergoes thermal runaway, the heat released instantaneously is transferred to adjacent cells through three pathways: thermal conduction (accounting for 60%-70%), thermal radiation (accounting for 20%-30%), and thermal convection (accounting for 10%). This creates a domino-like chain reaction, ultimately leading to the thermal runaway of the entire pack. The national standard GB/T 36276, “Lithium-ion battery for electric energy storage”, explicitly lists battery thermal runaway as a core safety test item. Its core purpose is to evaluate the safety protection capability of energy storage packs under extreme conditions, fundamentally preventing catastrophic accidents such as fires and explosions.
Currently, commonly used engineering methods for suppressing thermal runaway propagation include optimizing the cell arrangement, improving the efficiency of the cooling system, and selecting high-performance insulation materials. Among these, the rational selection of insulation materials and the optimization of their thickness have become the most valuable technical paths for engineering applications due to their controllable cost and convenient implementation. Design engineers typically evaluate the effectiveness of insulation schemes through experimental validation. However, the traditional trial-and-error experimental method has prominent issues such as high cost, long cycle times, and high safety risks. Furthermore, faced with a wide variety of insulation materials, design engineers lack a systematic selection basis and methodology, making it difficult to achieve an optimal balance between material thermal conductivity, thickness, cost, and process feasibility.
Numerical simulation technology can accurately simulate the complex thermal field distribution and propagation dynamics during thermal runaway, providing a quantitative basis for material selection and structural design, effectively compensating for the shortcomings of experimental research. Based on this, this paper constructs a systematic research framework of “experimental validation – simulation exploration – optimization iteration”. By selecting a variety of typical insulation materials, we obtain critical thermal runaway propagation data through experiments, calibrate a coupled thermal-electrical-chemical simulation model, systematically explore the suppression law of material thermal conductivity and thickness on thermal runaway propagation, establish a “material performance – thickness – suppression effect” correlation model, and propose an optimal insulation scheme, providing technical support for the safe design of energy storage battery packs.
2. Experimental Platform Setup and Scheme Design
2.1 Experimental Object
The experimental object is a 314 Ah LFP single cell. The core technical parameters are shown in the table below.
| No. | Parameter | Value | Unit/Note |
|---|---|---|---|
| 1 | Battery Type | LFP | |
| 2 | Nominal Capacity | 314 Ah | @ 25°C, 0.5P |
| 3 | Nominal Voltage | 3.2 V | @ 25°C, 0.5P |
| 4 | Cell Length (X) | 71.75±0.8 mm | At 40% SOC |
| 5 | Cell Length (Y) | 174.0±0.8 mm | / |
| 6 | Cell Length (Z) | 204.4±0.8 mm | Excluding terminals |
| 7 | Density | 2.292 kg/L | |
| 8 | Cell Specific Heat Capacity | 999 J/(kg·K) | |
| 9 | Cell In-plane Thermal Conductivity | 7.6439 W/(m·K) | |
| 10 | Cell Through-plane Thermal Conductivity | 18.5747 W/(m·K) |
The energy storage battery pack used in the experiment consists of 104 cells of this type connected in series, using a dual 52-series module symmetrical design. Each module contains 13 cells. Metal end plates are configured at both ends of the module, fixed by two steel straps. Thermal insulation material is laid between the cells, and a liquid cooling plate is arranged at the bottom of the module. A 1.5 mm thick thermally conductive silicone pad is added between the module and the liquid cooling plate to balance heat dissipation efficiency and cushioning/vibration damping.
2.2 Experimental Scheme Design and Results
To verify the effectiveness of different insulation schemes, we designed two sets of comparative experiments. The initial scheme used a 1.2 mm thick mica sheet as the inter-cell insulation material. The optimized scheme replaced the insulation material with an aerogel of the same thickness but with lower thermal conductivity. The experiments were conducted according to the standard of GB/T 36276-2023. A single cell located at the center of the middle module of the pack was selected as the target for triggering thermal runaway, using overcharging. The temperature acquisition cycle was set to 1 second. The test was stopped if any of the following conditions occurred: three consecutive temperature rise rates were ≥ 3°C/s, the triggered cell caught fire or exploded, the surface temperature of the triggered cell reached 300°C, or the test time reached 4 hours. Throughout the experiment, the temperature and voltage of the triggered cell and its adjacent cells were recorded.
Comparing the temperature and voltage curves and the thermal runaway propagation situation of the two groups of experiments allowed us to evaluate the suppression effect of the insulation material selection. The experimental results showed that the initial scheme using a 1.2 mm mica sheet failed to prevent the propagation of thermal runaway. The thermal runaway of the triggered cell caused a chain reaction, leading to the subsequent failure of multiple surrounding cells. In contrast, the optimized scheme using a 1.2 mm aerogel successfully suppressed the thermal propagation. In the mica sheet scheme, the triggered cell’s voltage dropped sharply to 0 V at 1412 s after the start of the experiment, and three consecutive temperature rise rates were ≥ 3°C/s, confirming the triggering of thermal runaway. Its temperature reached 112.1°C at the first second after triggering, and 122.2°C at the third second, ultimately causing a chain thermal runaway in adjacent cells. In the aerogel scheme, the triggered cell showed thermal runaway characteristics at 1444 s, with the voltage dropping to zero and three consecutive temperature rise rates ≥ 3°C/s. The cell temperatures at the first and third seconds after triggering were 113.1°C and 121.8°C, respectively. However, the thermal runaway of this cell did not cause thermal propagation, indicating that the aerogel insulation scheme can effectively block the cross-cell propagation of thermal runaway. The maximum temperature of the adjacent cells in the experiment was only 100.5°C.
By comparing the test data of the triggered thermal runaway cell and the temperature data of the adjacent cells from the two experiments, we observed that when the cell temperature exceeded 110°C, the rate of internal side reactions accelerated dramatically, and the risk of thermal runaway increased significantly. Therefore, we set 110°C as the critical criterion for cross-cell thermal runaway propagation in this study: if the temperature of the adjacent cell reaches or exceeds 110°C during the triggered thermal runaway test, it is determined that thermal runaway will propagate across cells; if the adjacent cell temperature does not reach this threshold, it is determined that the thermal runaway propagation is effectively suppressed.
3. Simulation Model Establishment and Validation
3.1 Geometric Model Establishment and Simplification
To ensure simulation accuracy while improving computational efficiency, we simplified the original 104-series pack model. The thermal runaway propagation follows a progressive path: triggered cell → adjacent cell → same module cells → adjacent module → entire pack. Its core mechanism mainly depends on local heat transfer characteristics. The simplified 52-series model fully retains the core thermal correlation and chain reaction mechanism of “single cell-adjacent cell-module”, without altering the critical heat transfer laws and insulation/thermal conduction characteristics. Therefore, it can be used to derive the thermal safety performance of the original pack.
To further optimize mesh quality and computational efficiency, a secondary simplification was performed during preprocessing: minor components with very low sensitivity to thermal simulation, such as bolts and wiring harnesses, were removed; non-critical features like fillets and chamfers were also removed.
3.2 Mesh Generation and Independence Validation
The simplified model was meshed, and grid independence validation was performed using the 1.2 mm aerogel condition as the baseline. The basic mesh sizes were set to 14 mm, 12 mm, 10 mm, 8 mm, and 6 mm, yielding the total number of mesh elements and the temperatures at key sampling points on the adjacent cell, as shown in the table below. The results indicate that when the total number of mesh elements increased from 838 million (basic size 8 mm) to 1442 million (basic size 6 mm), the critical temperature indicator only increased slightly by 0.15°C (change rate < 0.2%), indicating that the results were essentially convergent. Considering both computational accuracy and efficiency, the scheme with a total of 838 million elements (basic size 8 mm) was selected for subsequent simulations.
| Basic Size (mm) | Total Mesh Elements (Million) | Simulated Adjacent Cell Temperature (°C) |
|---|---|---|
| 14 | 430 | 100.05 |
| 12 | 500 | 100.48 |
| 10 | 625 | 100.73 |
| 8 | 838 | 100.96 |
| 6 | 1442 | 101.11 |
3.3 Material Parameters
The thermophysical parameters for the components of the energy storage battery system are listed in the table below.
| No. | Component | Material | Density (kg/m3) | Specific Heat Capacity (J·kg⁻¹·K⁻¹) | Thermal Conductivity (W·m⁻¹·K⁻¹) |
|---|---|---|---|---|---|
| 1 | Battery Cap | PP | 1183 | 1581 | 0.02 |
| 2 | Silicone Thermal Pad | Silicone | 800 | 966 | 2 |
| 3 | End Plate | Aluminum Alloy | 2700 | 900 | 200 |
| 4 | PC Sheet | PC | 1200 | 1256 | 0.2 |
| 5 | Liquid Cooling Plate | Aluminum Alloy | 2700 | 900 | 200 |
| 6 | Bracket | Aluminum Alloy | 2700 | 900 | 200 |
The thermophysical parameters of the insulation materials are shown in the table below.
| No. | Material | Density (kg/m3) | Specific Heat Capacity (J·kg⁻¹·K⁻¹) | Thermal Conductivity (W·m⁻¹·K⁻¹) |
|---|---|---|---|---|
| 1 | Aerogel | 200 | 600 | 0.024 |
| 2 | Ceramic Fiber Mat | 220 | 1580 | 0.035 |
| 3 | Aluminum Silicate Wool | 200 | 1000 | 0.042 |
| 4 | Glass Wool | 80 | 1000 | 0.05 |
| 5 | Nano-porous Insulation | 200 | 1100 | 0.06 |
| 6 | Expanded Perlite Board | 220 | 1180 | 0.07 |
| 7 | Rigid PVC Foam | 60 | 1300 | 0.08 |
| 8 | Mica Sheet | 2800 | 960 | 0.11 |
3.4 Thermal Runaway Heat Generation Model and Boundary Conditions
To accurately simulate the complex heat generation and heat transfer behavior during battery thermal runaway, we employed a mechanism-based thermal-electrical-chemical coupling model. This model considers the internal heat generation of the battery as a superposition of multiple physical and chemical processes. The total heat generation rate \(Q_b\) can be expressed by the following formula.
$$ Q_b = Q_r + Q_s + Q_J + Q_P $$
In this formula, \(Q_r\) is the reaction heat generated by reversible chemical reactions within the battery itself. \(Q_s\) is the side reaction heat caused by the decomposition of the electrolyte and separator inside the battery; its value is small and can be neglected. \(Q_J\) is the Joule heat generated by the current passing through the internal resistance of the battery. \(Q_P\) is the polarization heat generated by the polarization internal resistance when the battery undergoes polarization.
To establish a reliable simulation model, the \(Q_b\) in the above heat generation model was first inverted and calibrated based on the experimental data of the triggered thermal runaway from section 1.2, ensuring that the simulation could reproduce the temperature evolution curve of the triggered cell.
The boundary conditions were set as follows:
- Thermal Boundary: The calibrated heat generation \(Q_b\) is applied to the battery as a volumetric heat source. Heat is primarily dissipated through the thermal conduction path. Thermal radiation is neglected; only thermal conduction and natural convection are considered.
- Convection Boundary: The exterior of the pack is subject to natural convection with air; the convective heat transfer coefficient is set to 10 W/(m²·K).
- Initial Condition: The initial temperature of the entire system is set to 23.2°C.
The simulation model was validated using the experimental data from the non-propagated thermal runaway condition (aerogel scheme) from section 1.2. A comparison of the results shows that the maximum temperature of the triggered cell was 398.1°C (occurring at 1531 s) in the experiment, compared to 397.72°C (occurring at 1526 s) in the simulation. The peak temperature of the adjacent cell was 99.5°C (occurring at 1843 s) in the experiment, compared to 99.96°C (occurring at 1850 s) in the simulation. The errors between the simulation and experimental values for all key indicators are within 5%, validating that the established model has high accuracy and reliability, and can effectively characterize the temperature evolution process during battery thermal runaway.
4. Exploration of the Influence Laws of Insulation Performance
Based on the validated simulation model, this section conducts parametric simulation studies from three dimensions: material thermal conductivity, thickness gradient, and the coupling effect of the two. By quantifying key thermal response indicators such as the adjacent cell peak temperature (\(T_{adj}\)), thermal runaway triggering delay time (\(\tau\)), and heat flux density attenuation rate (\(\eta\)), we reveal their influence laws.
Here, \(T_{adj}\) is the peak temperature of the adjacent cell. The thermal runaway triggering delay time \(\tau\) is defined as the time from the onset of thermal runaway in the triggered cell (temperature rise rate > 3°C/s) until the adjacent cell reaches the critical temperature of 110°C; if it is not reached, it is recorded as >1800 s. The heat flux density attenuation rate \(\eta\) is defined as:
$$ \eta = \left(1 – \frac{q_{ins}}{q_{0}}\right) \times 100\% $$
Here, \(q_{0}\) is the steady-state heat flux density between cells without insulation material, and \(q_{ins}\) is the steady-state heat flux density after laying the insulation material.
4.1 Suppression Effect of Material Thermal Conductivity on Thermal Runaway Propagation
Fixing the insulation layer thickness at 1.2 mm, we simulated eight materials with thermal conductivities (\(\lambda\)) ranging from 0.024 to 0.11 W·m⁻¹·K⁻¹. We focused on monitoring the transient process of heat transfer to the adjacent cell after the triggered cell undergoes thermal runaway. The key results are summarized in the table below.
| Insulation Material | \(\lambda\) (W·m⁻¹·K⁻¹) | \(T_{adj}\) (°C) | \(\eta\) (%) | \(\tau\) (s) | Blocked Propagation |
|---|---|---|---|---|---|
| 1_Aerogel | 0.024 | 100.96 | 89 | >1800 | Yes |
| 2_Ceramic Fiber Mat | 0.035 | 109.22 | 82 | >1800 | Yes |
| 3_Aluminum Silicate Wool | 0.042 | 113.70 | 75 | 320 | No |
| 4_Glass Wool | 0.05 | 118.89 | 67 | 248 | No |
| 5_Nano-porous Insulation | 0.06 | 124.49 | 60 | 196 | No |
| 6_Expanded Perlite Board | 0.07 | 129.49 | 58 | 151 | No |
| 7_Rigid PVC Foam | 0.08 | 134.06 | 53 | 124 | No |
| 8_Mica Sheet | 0.11 | 142.87 | 42 | 102 | No |
The simulation results reveal a clear critical effect of material thermal conductivity on the inhibition of thermal propagation. As shown in the table, when the thermal conductivity \(\lambda\) is lower than 0.035 W·m⁻¹·K⁻¹, the peak temperature of the adjacent cell is effectively controlled below the critical value of 110°C, and thermal propagation is successfully blocked. When \(\lambda\) exceeds this critical value, the adjacent cell temperature exceeds 110°C, breaching the thermal runaway trigger threshold and causing a chain thermal runaway reaction.
Further analysis shows a significant non-linear negative correlation between thermal conductivity and the suppression effect. As \(\lambda\) decreases from 0.11 W·m⁻¹·K⁻¹ to 0.024 W·m⁻¹·K⁻¹, the thermal runaway triggering delay time \(\tau\) increases from 102 s to over 1800 s, and the heat flux density attenuation rate \(\eta\) increases from 42% to 89%. The mechanism is that the low thermal conductivity material weakens the dominant heat transfer path of thermal conduction, forming an efficient thermal resistance barrier between cells, which greatly delays the heat accumulation process in the adjacent cell.
4.2 Suppression Effect of Thickness Gradient on Thermal Runaway Propagation
To investigate the influence law of thickness, we selected aerogel (\(\lambda\)=0.024 W·m⁻¹·K⁻¹) as the research object for its excellent insulation performance, and set up 10 thickness gradients from 0.6 mm to 2.4 mm for simulation. The results are shown in the table below.
| Thickness \(d\) (mm) | \(T_{adj}\) (°C) | \(\tau\) (s) | \(\eta\) (%) | Volumetric Energy Density \(E_v\) (Wh/L) | Blocked Propagation |
|---|---|---|---|---|---|
| 0.6 | 117.78 | 260 | 78 | 395.93 | No |
| 0.8 | 109.61 | >1800 | 82 | 394.73 | Yes |
| 1.0 | 104.03 | >1800 | 86 | 393.74 | Yes |
| 1.2 | 99.96 | >1800 | 89 | 392.77 | Yes |
| 1.4 | 96.84 | >1800 | 91 | 391.78 | Yes |
| 1.6 | 94.98 | >1800 | 92 | 390.79 | Yes |
| 1.8 | 93.27 | >1800 | 93 | 389.81 | Yes |
| 2.0 | 91.92 | >1800 | 94 | 388.83 | Yes |
| 2.2 | 90.70 | >1800 | 95 | 387.86 | Yes |
| 2.4 | 89.75 | >1800 | 96 | 386.89 | Yes |
The research indicates that the insulation layer thickness \(d\) is positively correlated with the thermal resistance \(R\) (\(R = d/\lambda\)), but its effect on improving safety exhibits a significant diminishing marginal return. When the thickness increases from 0.6 mm to 1.0 mm, \(T_{adj}\) rapidly decreases from 117.8°C to 104.0°C, successfully crossing the critical threshold, with a significant suppression effect. In the thickness range of 1.0 mm to 1.6 mm, each 0.2 mm increase in thickness reduces \(T_{adj}\) by about 3-5°C, and the protective effect is still evident. However, when the thickness exceeds 1.6 mm, the benefit of further increasing thickness in reducing temperature decreases sharply (from 1.6 mm to 2.4 mm, \(T_{adj}\) only decreases by about 5.2°C), with a marginal improvement rate of less than 0.7°C/mm.
At the same time, an increase in thickness directly reduces the effective space within the pack for accommodating cells, leading to a linear decrease in the system’s volumetric energy density \(E_v\) (about 1 Wh/L per 0.2 mm). Therefore, the essence of thickness optimization is to find a balance between safety, system energy density, and cost. Overall, for the aerogel material, the 1.2 mm to 1.6 mm range is the “best cost-performance” thickness range that balances efficient thermal blocking with minimal space occupation.
4.3 Coupled Influence Law of Thermal Conductivity and Thickness
To clarify the synergistic mechanism of thermal conductivity \(\lambda\) and thickness \(d\), we constructed a simulation matrix combining 8 values of \(\lambda\) and 10 values of \(d\) (a total of 80 working conditions). The matrix data for the adjacent cell peak temperature \(T_{adj}\) and the thermal runaway triggering delay time \(\tau\) are shown in the tables below.
| \(d\) (mm) \ \(\lambda\) (W·m⁻¹·K⁻¹) | 0.024 | 0.035 | 0.042 | 0.05 | 0.06 | 0.07 | 0.08 | 0.11 |
|---|---|---|---|---|---|---|---|---|
| 0.6 | 118.96 | 127.08 | 135.78 | 141.73 | 147.81 | 152.83 | 157.08 | 165.13 |
| 0.8 | 110.70 | 120.35 | 126.18 | 131.97 | 138.05 | 143.25 | 147.83 | 156.59 |
| 1.0 | 105.07 | 113.68 | 119.11 | 124.63 | 130.50 | 135.64 | 140.27 | 149.22 |
| 1.2 | 100.96 | 109.22 | 113.70 | 118.89 | 124.49 | 129.49 | 134.06 | 142.87 |
| 1.4 | 97.80 | 104.75 | 109.42 | 114.29 | 119.60 | 124.40 | 129.52 | 137.36 |
| 1.6 | 95.93 | 101.63 | 105.97 | 110.54 | 115.55 | 120.14 | 125.15 | 132.55 |
| 1.8 | 94.20 | 99.06 | 103.11 | 107.41 | 110.32 | 116.52 | 121.15 | 128.30 |
| 2.0 | 92.83 | 97.15 | 100.71 | 104.76 | 109.22 | 110.71 | 117.10 | 124.53 |
| 2.2 | 91.61 | 95.55 | 98.69 | 102.51 | 106.73 | 108.34 | 113.20 | 121.18 |
| 2.4 | 90.64 | 94.13 | 97.17 | 100.54 | 104.54 | 105.97 | 109.50 | 118.15 |
| \(d\) (mm) \ \(\lambda\) (W·m⁻¹·K⁻¹) | 0.024 | 0.035 | 0.042 | 0.05 | 0.06 | 0.07 | 0.08 | 0.11 |
|---|---|---|---|---|---|---|---|---|
| 0.6 | 260 | 149 | 116 | 95 | 81 | 71 | 65 | 60 |
| 0.8 | 404 | 235 | 177 | 134 | 108 | 92 | 80 | 71 |
| 1.0 | >1800 | 320 | 246 | 193 | 144 | 118 | 100 | 85 |
| 1.2 | >1800 | >1800 | 350 | 248 | 196 | 151 | 124 | 102 |
| 1.4 | >1800 | >1800 | >1800 | 309 | 242 | 196 | 154 | 123 |
| 1.6 | >1800 | >1800 | >1800 | 360 | 291 | 237 | 194 | 147 |
| 1.8 | >1800 | >1800 | >1800 | >1800 | >1800 | 278 | 231 | 177 |
| 2.0 | >1800 | >1800 | >1800 | >1800 | >1800 | 328 | 264 | 213 |
| 2.2 | >1800 | >1800 | >1800 | >1800 | >1800 | >1800 | 335 | 248 |
| 2.4 | >1800 | >1800 | >1800 | >1800 | >1800 | >1800 | >1800 | 273 |
Based on the data, the three-dimensional surface plots clearly show that \(T_{adj}\) decreases as \(\lambda\) decreases or \(d\) increases. To achieve effective thermal propagation blocking (\(T_{adj} < 110°C\)), the material must satisfy a specific \(\lambda\)-\(d\) combination condition. For example, for high-performance aerogel with \(\lambda = 0.024\), only about 1.0 mm thickness is required; for the slightly higher conductivity ceramic fiber with \(\lambda = 0.035\), about 1.2 mm thickness is needed; if mica sheet with \(\lambda = 0.11\) is used, even if the thickness is increased to 2.4 mm, the \(T_{adj}\) is still as high as 118.2°C, which cannot meet the blocking requirement.
To quantitatively describe the relationship between thermal resistance (\(d/\lambda\)) and the thermal runaway triggering delay time \(\tau\), we performed a nonlinear regression analysis on the valid data points from the simulation where \(\tau \le 1800\) s. The results show that \(\tau\) and \(d/\lambda\) follow an exponential asymptotic growth relationship. The established empirical model is:
$$ \tau = 1800 \times \left[1 – \exp\left(-0.034 \times \frac{d}{\lambda}\right)\right] $$
The model’s coefficient of determination R² = 0.902, indicating that \(\tau\) approaches an upper limit (1800 s) as the thermal resistance increases. This reflects the fact that when the thermal resistance of the insulation layer is large enough, the triggering time for thermal runaway will far exceed the simulation duration, thereby achieving effective suppression. From an engineering selection perspective, a quick assessment can be made based on the thermal resistance threshold.
- If complete suppression of thermal runaway within the simulation period is required (\(\tau > 1800\) s), the condition \(d/\lambda \ge 35\) mm·m·K·W⁻¹ must be met.
- If a significant delay of thermal runaway is required (\(\tau \ge 150\) s), the condition \(d/\lambda \ge 12\) mm·m·K·W⁻¹ must be satisfied.
This model and the threshold criteria provide a clear quantitative basis for the selection and thickness optimization of insulation materials, aiding in achieving the best balance among safety, system energy density, and cost.
5. Conclusions
Focusing on the 314 Ah LFP energy storage battery, this paper systematically conducted research on the selection and thickness optimization of insulation materials for suppressing thermal runaway propagation in battery packs. The main conclusions are as follows:
(1) Critical Role of Thermal Conductivity: Material thermal conductivity is a key factor influencing thermal runaway propagation, with a clear critical threshold. When the thermal conductivity is less than or equal to 0.035 W·m⁻¹·K⁻¹, the material can form an effective thermal resistance barrier between cells, blocking thermal propagation. As \(\lambda\) decreased from 0.11 W·m⁻¹·K⁻¹ to 0.024 W·m⁻¹·K⁻¹, the thermal runaway triggering delay time for adjacent cells increased from 102 s to over 1800 s, and the heat flux density attenuation rate increased from 42% to 89%.
(2) Positive Correlation with Diminishing Returns for Thickness: The thickness of the insulation material is positively correlated with the suppression effect, but the benefit has a clear diminishing marginal return. For the aerogel material, the 1.2 mm to 1.6 mm range is the optimal cost-performance thickness range that balances safety, system energy density, and cost. Within this range, the peak temperature of the adjacent cell can be kept below 100°C, the heat flux density attenuation rate is \(\ge 89\%\), and the loss in volumetric energy density is controlled within 3 Wh/L.
(3) Coupling Synergy and Engineering Selection Criteria: There is a significant coupled synergistic effect between thermal conductivity and thickness on the suppression of thermal runaway. Based on simulation data, an empirical model for the thermal runaway triggering delay time \(\tau = 1800 \times [1 – \exp(-0.034 \times (d/\lambda))]\) (R²=0.902) was established, revealing the law of \(\tau\) approaching an upper limit as thermal resistance increases. The derived engineering selection criteria are: for complete suppression, \(d/\lambda \ge 35\) mm·m·K·W⁻¹; for significant delay (\(\tau \ge 150\) s), \(d/\lambda \ge 12\) mm·m·K·W⁻¹. This provides a quantitative basis for the rapid assessment and optimized design of insulation schemes.
It should be noted that these conclusions are based on experiments and simulations for a specific battery model (314 Ah LFP). Different battery types, capacities, manufacturers, and designs may lead to changes in parameters such as the thermal runaway trigger temperature and heat generation characteristics. Therefore, the specific numerical values proposed in this paper (such as critical thermal conductivity, thickness thresholds, and thermal resistance criteria) are applicable to batteries of similar specifications. However, the more important contribution is that the “experimental validation – simulation exploration – optimization iteration” framework, the experimental methods, and the simulation optimization approach established in this work are generally applicable. They can serve as a methodological reference for the thermal safety design of different energy storage battery packs, assisting in achieving a balanced optimization between safety and energy density in engineering applications.