In the pursuit of global carbon neutrality, lithium-ion batteries have become the predominant energy storage medium for large-scale applications due to their high energy density and long cycle life. However, the safety risk associated with thermal runaway and its catastrophic propagation remains a critical bottleneck for the advancement of the industry. The objective of this study is to systematically investigate the selection of interlayer insulation materials and the optimization of their thickness to suppress thermal runaway propagation within an energy storage cell pack. We established a comprehensive research framework integrating experimental validation, numerical simulation, and iterative optimization. A 314 Ah lithium iron phosphate (LFP) energy storage cell pack was constructed, and a controlled overcharge-induced thermal runaway experiment was conducted to capture critical data. Subsequently, a coupled thermal-electrical-chemical simulation model was calibrated and validated against the experimental results. This model was then employed to explore the influence of material thermal conductivity and thickness on the heat transfer dynamics between cells. Finally, we developed a correlation model linking material performance and thickness to suppression efficacy, leading to an optimal insulation strategy.
The core challenge in ensuring the safety of an energy storage cell pack lies in preventing the Domino effect, where the thermal runaway of one cell triggers its neighbors. This heat transfer is primarily driven by conduction. Our investigation focused on eight typical insulation materials: aerogel, ceramic fiber mat, aluminum silicate wool, glass wool, nano-porous insulation material, expanded perlite board, rigid PVC foam, and mica sheet. These materials have thermal conductivities ranging from 0.024 to 0.11 W·m⁻¹·K⁻¹.
Experimental Platform and Critical Data
An experimental platform was built using a 314 Ah LFP energy storage cell. The key parameters of the cell are presented in the table below.
| Parameter | Value | Unit |
| :— | :— | :— |
| Battery Type | Lithium Iron Phosphate (LFP) | – |
| Nominal Capacity | 314 | Ah (25°C, 0.5P) |
| Nominal Voltage | 3.2 | V (25°C, 0.5P) |
| Cell Length (X) | 71.75 ± 0.8 | mm |
| Cell Width (Y) | 174.0 ± 0.8 | mm |
| Cell Height (Z) | 204.4 ± 0.8 | mm (excluding posts) |
| Density | 2.292 | kg/L |
| Mass | 5.67 ± 0.2 | kg |
| Specific Heat Capacity | 999 | J/(kg·K) |
| In-plane Thermal Conductivity | 7.6439 | W/(m·K) |
| Through-plane Thermal Conductivity | 18.5747 | W/(m·K) |
Two comparative tests were conducted using a 1.2 mm thick layer of either mica sheet or aerogel between the cells. In the test with the mica sheet (λ=0.11 W·m⁻¹·K⁻¹), the triggered cell failed after 1412 seconds. Its temperature rose to 112.1°C in the first second and 122.2°C in the third second, subsequently triggering a chain reaction of thermal runaway in the adjacent energy storage cell.
Conversely, the test with the aerogel interlayer (λ=0.024 W·m⁻¹·K⁻¹) showed a different outcome. The triggered cell experienced thermal runaway at 1444 seconds, but the adjacent cell remained stable. The peak temperature of the adjacent energy storage cell was only 100.5°C, demonstrating that thermal runaway propagation was successfully suppressed. Based on these experiments, we identified a critical temperature threshold of 110°C: if the adjacent cell temperature exceeds this value, propagation is likely to occur.
Simulation Model and Validation
A three-dimensional simulation model of the energy storage cell pack was developed and simplified to improve computational efficiency. The geometry included the cells, insulation layers, end plates, and cooling plate. A grid independence study was performed, and a mesh with approximately 8.38 million elements was selected as a balance between accuracy and speed.
The thermal runaway heat generation model is described by the following formula:
$$Q_b = Q_r + Q_s + Q_J + Q_p$$
where \(Q_b\) is the total heat generation rate, \(Q_r\) is the reversible reaction heat, \(Q_s\) is the side reaction heat (negligible here), \(Q_J\) is the Joule heat from internal resistance, and \(Q_p\) is the polarization heat. The model was calibrated using the experimental data from the successful aerogel test.
The material properties used in the model are listed below.
| Component | Material | Density (kg/m³) | Specific Heat (J·kg⁻¹·K⁻¹) | Thermal Conductivity (W·m⁻¹·K⁻¹) |
| :— | :— | :— | :— | :— |
| Battery Cover | PP | 1183 | 1581 | 0.02 |
| Thermal Pad | Silicone | 800 | 966 | 2 |
| End Plate | Aluminum Alloy | 2700 | 900 | 200 |
| PC Sheet | PC | 1200 | 1256 | 0.2 |
| Cooling Plate | Aluminum Alloy | 2700 | 900 | 200 |
| Support Frame | Aluminum Alloy | 2700 | 900 | 200 |
The validation results showed excellent agreement. The simulated peak temperature of the triggered cell was 397.72°C (experimental: 398.1°C), and the adjacent cell reached 99.96°C (experimental: 99.5°C). The deviation was less than 5% for all key metrics, confirming the reliability of the simulation model.
Influence of Material Thermal Conductivity
Using the validated model, we first investigated the effect of material thermal conductivity (\(\lambda\)) on thermal runaway suppression. The insulation thickness was fixed at 1.2 mm.
| Insulation Material | λ (W·m⁻¹·K⁻¹) | Adjacent Cell Peak Temp. \(T_{adj}\) (°C) | Heat Flux Attenuation Rate \(\eta\) (%) | Trigger Delay Time \(\tau\) (s) | Propagation Suppressed? |
| :— | :— | :— | :— | :— | :— |
| Aerogel | 0.024 | 100.96 | 89 | >1800 | Yes |
| Ceramic Fiber Mat | 0.035 | 109.22 | 82 | >1800 | Yes |
| Aluminum Silicate Wool | 0.042 | 113.70 | 75 | 320 | No |
| Glass Wool | 0.05 | 118.89 | 67 | 248 | No |
| Nano-porous Material | 0.06 | 124.49 | 60 | 196 | No |
| Expanded Perlite Board | 0.07 | 129.49 | 58 | 151 | No |
| Rigid PVC Foam | 0.08 | 134.06 | 53 | 124 | No |
| Mica Sheet | 0.11 | 142.87 | 42 | 102 | No |
The results indicate a critical threshold. Materials with a thermal conductivity of 0.035 W·m⁻¹·K⁻¹ or lower effectively prevented propagation, keeping the adjacent energy storage cell temperature below 110°C. As \(\lambda\) increased, the trigger delay time (\(\tau\)) decreased sharply, and the heat flux attenuation rate (\(\eta\)) diminished.
Influence of Insulation Thickness
We next investigated the effect of aerogel thickness (d) on suppression performance. The results are summarized below.
| Thickness d (mm) | \(T_{adj}\) (°C) | \(\tau\) (s) | \(\eta\) (%) | Volumetric Energy Density \(E_v\) (Wh/L) | Propagation Suppressed? |
| :— | :— | :— | :— | :— | :— |
| 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 |
A thickness of 0.8 mm was sufficient to suppress propagation. However, a clear diminishing return is observed. The safety benefit from increasing thickness from 0.6 to 1.0 mm is dramatic, but beyond 1.6 mm, the reduction in \(T_{adj}\) becomes marginal while the loss in volumetric energy density is linear. A thickness of 1.2 to 1.6 mm represents an optimal balance for this specific energy storage cell.
Coupled Influence of Conductivity and Thickness
A comprehensive matrix of 80 simulations was performed to study the coupled influence of \(\lambda\) and d. The following table shows the peak temperature of the adjacent energy storage cell (\(T_{adj}\)) for various combinations.
| d (mm) \ λ (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 |
The data shows that multiple (\(\lambda\), d) pairs can achieve the <110°C target. A key finding is the relationship between thermal resistance (\(R = d/\lambda\)) and the trigger delay time (\(\tau\)). By analyzing scenarios where \(\tau\) was finite (<1800 s), we derived an empirical model:
$$\tau = 1800 \times [1 – \exp(-0.034 \times \frac{d}{\lambda})]$$
This model (\(R^2=0.902\)) shows that the trigger delay time approaches an asymptote as thermal resistance increases. From this, we propose the following engineering selection rules:
– To fully suppress propagation for the duration of the simulation (\(\tau > 1800\)), the thermal resistance must satisfy: \(d/\lambda \geq 35\) mm·m·K·W⁻¹.
– To significantly delay propagation (\(\tau \geq 150\) s), the thermal resistance must satisfy: \(d/\lambda \geq 12\) mm·m·K·W⁻¹.
Conclusion
This study provides a systematic framework for optimizing interlayer insulation in an energy storage cell pack to suppress thermal runaway propagation. The key conclusions are:
1. Material thermal conductivity has a critical threshold. For a 1.2 mm thick layer, materials with \(\lambda \leq 0.035\) W·m⁻¹·K⁻¹ are required to prevent propagation from a single cell failure.
2. Increasing insulation thickness improves safety, but with diminishing returns. For aerogel, a thickness of 1.2 to 1.6 mm offers an optimal trade-off between safety, energy density, and cost.
3. The coupled effect of thermal conductivity and thickness can be summarized by the thermal resistance (\(d/\lambda\)). The empirical model \(\tau = 1800 \times [1 – \exp(-0.034 \times d/\lambda)]\) provides a powerful tool for quick evaluation of candidate materials.
4. An optimal solution was validated: a 1.2 mm aerogel layer reduces the adjacent energy storage cell peak temperature to 99.96°C, achieving a heat flux attenuation rate of 89% and fully suppressing thermal runaway propagation. This research provides both experimental validation and a robust simulation-based method for enhancing the thermal safety of large-scale energy storage systems.