The rapid expansion of renewable energy integration necessitates advanced battery energy storage solutions to stabilize grid operations. Lithium iron phosphate (LFP) batteries dominate this sector due to their cost-effectiveness and thermal stability. However, thermal runaway (TR) remains a critical safety concern, as it generates explosive gases and triggers catastrophic failures. This study evaluates multi-parameter composite sensors for TR detection in a 40-foot battery energy storage compartment, focusing on gas evolution, sensor responsiveness, spatial distribution, and ignition impacts.
1. Experimental Framework
1.1 Sensor Design and Configuration
Five multi-parameter composite sensors (A–E) were developed, integrating distinct detection principles for hydrogen (H₂), carbon monoxide (CO), carbon dioxide (CO₂), volatile organic compounds (VOCs), smoke density, pressure, and temperature (Table 1).
Table 1: Sensor configurations and detection principles
| Sensor | H₂ Detection | VOC Detection | CO Detection | CO₂ Detection | Smoke | Temperature | Pressure |
|---|---|---|---|---|---|---|---|
| A | Electrochemical | Solid Polymer Electrochemical | Electrochemical | – | Semiconductor | Thermocouple | – |
| B | Semiconductor | Solid Polymer Electrochemical | Electrochemical | – | Semiconductor | Thermocouple | – |
| C | Electrochemical | Solid Polymer Electrochemical | Electrochemical | – | Semiconductor | Thermocouple | Piezoresistive |
| D | Catalytic Combustion | Photoionization | Electrochemical | Infrared | Semiconductor | Thermocouple | – |
| E | Semiconductor | Solid Polymer Electrochemical | Electrochemical | Infrared | Semiconductor | Thermocouple | – |
Key parameters:
- H₂: Catalytic combustion (D) vs. electrochemical (A, C) vs. semiconductor (B, E).
- VOC: Photoionization (D) vs. solid polymer electrochemical (A, B, C, E).
- CO/CO₂: Electrochemical (A–E) and infrared (D, E).
1.2 Test Setup
A 280 Ah LFP battery (3.2 V, SOC = 100%) was subjected to thermal abuse via a 950 W heating plate. Experiments included:
- Non-ignition tests: TR triggered by heating until safety valve rupture.
- Ignition tests: Post-TR ignition using a 20 J spark igniter.
- Auxiliary tests: Heating battery blue film and adhesive tape to isolate VOC sources.
Sensors were distributed at 2.4 m intervals across the compartment (edge, intermediate, and center positions). Temperature profiles (T1–T4) monitored surface and ambient thermal gradients.
2. Key Findings
2.1 Thermal Runaway Gas Evolution Dynamics
TR progresses through three phases:
- Pre-venting (0–1,088 s): SEI decomposition releases CO₂. VOC emerges first (10⁻⁵–7.3×10⁻⁵) due to electrolyte volatilization and blue film pyrolysis.
- Post-venting (1,089–1,682 s): Safety valve rupture discharges combustible gases. Catalytic H₂ sensors (D) detected H₂ at 1,462 s, 300 s earlier than electrochemical sensors. VOC (photoionization) reached 2×10⁻⁴ at 1,493 s.
- Thermal Escalation (1,683–2,850 s): Cell rupture and smoke dispersion. CO₂ peaked at 9.59×10⁻⁴ (1,935 s), lagging behind H₂/CO due to slower reaction kinetics.
Figure 1: Gas concentration profiles during TR (non-ignition)
| Parameter | Detection Time (s) | Peak Concentration |
|---|---|---|
| VOC | 750 | 6×10⁻³ |
| H₂ | 1,462 | 2,000 ppm |
| CO | 1,765 | 684 ppm |
| Smoke | 1,665 | 4,801 mg/m³ |
| CO₂ | 1,758 | 1,296 ppm |
2.2 Sensor Performance Comparison
- H₂ Detection: Catalytic combustion (D) outperformed electrochemical and semiconductor types with faster response (Δ𝑡 = 100 s) and stability.
- VOC Detection: Photoionization (D) achieved earlier detection (Δ𝑡 = 600 s) than solid polymer electrochemical sensors.
- Temperature/Pressure: Compartment roof sensors showed negligible changes (<1°C, no pressure shift) during non-ignition TR. Proximity sensors (T4) recorded 313°C, validating localized thermal monitoring.
Equation 1: Fick’s diffusion law for gas propagation
ρDABdxdYA+ρDBAdxdYB=0
where DAB∝θ03/2P0−1 (diffusivity increases with temperature).
2.3 Impact of Ignition
Ignition altered gas composition and sensor dynamics (Table 2):
- CO and smoke surged (≥1,000 ppm and 10,000 mg/m³).
- VOC dropped by 50% due to combustion consumption.
- Temperature rise rate escalated from 0.10°C/min to 0.78°C/min.
Table 2: Parameter comparison (ignition vs. non-ignition)
| Parameter | Non-Ignition | Ignition |
|---|---|---|
| H₂ | ≥2,000 ppm | ≥2,000 ppm |
| CO | 684 ppm | ≥1,000 ppm |
| VOC | ≥10,000 ppm | 5,778 ppm |
| Smoke | 4,801 mg/m³ | ≥10,000 mg/m³ |
| CO₂ | 1,296 ppm | ≥5,000 ppm |
| Temp. Rise Rate | 0.10°C/min | 0.78°C/min |
2.4 Spatial Propagation Rates
Gas dispersion velocities depended on position and ignition (Table 3):
- Edge heating: Faster propagation (H₂: 28.26 mm/s; CO: 29.57 mm/s) than center heating.
- Ignition: Thermal buoyancy amplified velocities by 4–8× (e.g., VOC: 132.53 mm/s).
Equation 2: Propagation velocity calculation
V=∣tn,ε−tn+1,ε∣2400
where n = sensor index, ε = detection threshold.
Table 3: Propagation velocities under different conditions
| Parameter | Center Heating (mm/s) | Edge Heating (mm/s) | Ignition (mm/s) |
|---|---|---|---|
| H₂ | 23.48 | 28.26 | 97.61 |
| CO | 27.21 | 29.57 | 132.76 |
| VOC | 15.41 | 21.30 | 132.53 |
| Smoke | 21.75 | 26.15 | 129.95 |
| CO₂ | 35.71 | 51.90 | 194.80 |
3. Optimization Strategies for Battery Energy Storage Safety
3.1 Sensor Selection and Placement
- Priority sensors: Catalytic H₂ and photoionization VOC detectors for early TR warning.
- Layout:
- Edge zones: 0.64–1.28 m spacing.
- Central zones: 0.92–1.85 m spacing.
- Supplementary metrics: Localized temperature (module-level) and post-ignition CO₂/smoke for fire confirmation.
3.2 Multi-Stage Alert Protocol
- Stage 1 (VOC > 10⁻⁴): Pre-TR warning (electrolyte leakage/blue膜 pyrolysis).
- Stage 2 (H₂ > 500 ppm): TR confirmation.
- Stage 3 (CO > 300 ppm + smoke > 1,000 mg/m³): Fire risk alert.
Equation 3: Thermal runaway energy release
QTR=∑micp,iΔT+ΔHrxn
where mi = mass of cell components, cp,i = specific heat, ΔHrxn = reaction enthalpy.
4. Conclusion
This study validates the superiority of catalytic H₂ and photoionization VOC sensors in battery energy storage compartments, achieving 600 s earlier TR detection than conventional methods. Edge-mounted sensors and ignition-aware protocols enhance response reliability. Future work will explore sensor durability under variable environmental conditions and multi-parameter fusion algorithms.