The global imperative to achieve carbon peak and carbon neutrality has propelled the development and integration of new energy and renewable energy sources. Enhancing energy utilization efficiency is a paramount concern worldwide. For a major energy producer and consumer like China, this dual challenge necessitates a robust and safe energy storage industry to balance decarbonization goals with the energy demands of economic growth. Electrochemical energy storage systems are pivotal in this transition, facilitating grid stability and renewable energy integration. However, the increasing frequency of fire and explosion incidents in electrochemical battery energy storage system installations poses significant risks to life and property, raising public concern and potentially hindering widespread adoption. Ensuring fire safety is, therefore, a critical prerequisite for the healthy and secure development of the energy storage sector. This article focuses on the fire safety design for a 220 kWh small-scale energy storage battery cabinet, presenting a detailed protection scheme. By addressing the unique fire risks of lithium-ion batteries through a systematic, layered approach, this design aims to provide a reliable reference for fire prevention and control in modern energy storage applications.
1. Introduction and Project Definition
The core objective is to design a reliable fire protection solution for a specific 220 kWh containerized battery energy storage system. The system under protection utilizes lithium iron phosphate (LiFePO4) cells with a nominal voltage of 3.2V and capacity of 163Ah. The full energy storage unit comprises a single battery string (or “cluster”). This cluster is composed of 18 battery modules connected in series, resulting in a configuration of 216S2P (216 cells in series, 2 in parallel). The total system parameters are approximately 691.2V, 326Ah, delivering 225.3 kWh. Each individual module has a configuration of 12S2P (38.4V, 326Ah, 12.5 kWh).
These 18 modules are housed within three identical battery cabinets, each with dimensions of 600mm (W) × 800mm (D) × 1800mm (H). Each cabinet contains 6 modules. The DC power from the battery cluster is fed to a Power Conversion System (PCS) cabinet rated at 75.1 kW, which interfaces with the AC grid. The PCS is a bidirectional converter responsible for transforming DC power from the batteries to AC for the grid (discharge) and converting AC grid power to DC for charging the batteries. The fire protection system’s scope encompasses the PCS cabinet, the three battery cabinets, and, most critically, the battery modules inside those cabinets. The primary fire hazard originates from the potential thermal runaway of lithium-ion cells within the modules.
2. Overall System Architecture and Design Philosophy
The fire safety system is designed based on the specific environmental conditions and hazard characteristics of the protected battery energy storage system. The core design philosophy employs a multi-layered, “defense-in-depth” strategy. It integrates detection and suppression into a cohesive, automatic system using clean agent fire suppression technology. The scheme features a dual deployment strategy: internally within each battery module and externally within the cabinet housing.
Each individual battery module is equipped with a dedicated, multi-parameter fire detection probe and a point-source, clean agent fire suppression unit. This configuration enables continuous monitoring and early warning of incipient thermal runaway, allowing for targeted, point-to-point fire suppression at the very origin of the hazard. Simultaneously, each battery cabinet and the PCS cabinet are protected as separate compartments by independent clean agent fire suppression units and area smoke/heat detectors. This creates an effective secondary protection layer within the cabinet’s internal volume, addressing fires that may originate from electrical connections, wiring, or the PCS components themselves.
The system architecture is designed for simplicity, high reliability, and long-term maintenance-free operation. The selected extinguishing agent is non-conductive, non-corrosive, leaves no residue, and causes no secondary damage to sensitive electronic equipment, thereby minimizing potential losses. A critical redundancy feature is incorporated: a mechanical temperature-activated backup initiation device is installed in each cabinet. This ensures the fire suppression system can operate even in a complete power loss scenario, guaranteeing reliable fire control under all foreseeable failure conditions.
To maximize response efficiency and isolate incidents, each cabinet is designated as an independent protection zone. The entire battery energy storage system is thus divided into four distinct zones: three for the battery cabinets and one for the PCS cabinet. Each zone can be activated independently based on local detection, or all can be coordinated and managed centrally by a master fire control panel.
Table 1: Overall Fire Protection System Component Summary
| Subsystem / Location | Primary Components | Quantity | Key Function |
|---|---|---|---|
| Battery Module (Internal Layer) | Multi-parameter Detector, Miniaturized Fire Suppression Unit, Nozzle, Tubing | 18 sets (one per module) | Earliest detection & point-source suppression of cell thermal runaway. |
| Battery Cabinet (External Layer) | Area Detector, Cabinet Fire Suppression Unit (integrated nozzle), Mechanical Actuator | 3 sets (one per cabinet) | Compartment-level detection & total flooding suppression for cabinet interior. |
| PCS Cabinet (External Layer) | Area Detector, Cabinet Fire Suppression Unit (integrated nozzle), Mechanical Actuator | 1 set | Detection & total flooding suppression for PCS and electrical components. |
| Control & Coordination | Integrated Control Box (per battery cabinet), Master Fire Control Panel | 3 + 1 | Signal processing, decision logic, alarm indication, and system activation. |
3. Zoned Protection Strategy and Rationale
The protection strategy is explicitly divided into two primary layers, each addressing a specific scale and stage of a potential fire event within the battery energy storage system.
3.1 Layer 1: Individual Battery Module Protection Zone
The protection object is the lithium-ion cells inside a single battery module. The primary threat is thermal runaway initiated in one or more cells. Upon detection of characteristic signatures (gas, temperature, smoke) by the internal probe, a signal is sent to the Integrated Control Box. The control logic then activates only the fire suppression unit directly assigned to that specific module. The agent, Perfluoro(2-methyl-3-pentanone) (commonly called Novec 1230 or similar “clean agent”), is discharged through a spray nozzle mounted on the module’s enclosure. It rapidly vaporizes, creating a total flooding environment within the module. This action serves two critical functions: 1) Fire Suppression: Removing heat and interrupting the chemical chain reaction. 2) Temperature Control: Cooling adjacent cells to prevent thermal propagation and cascading failure, which is the most significant risk in a battery energy storage system.
3.2 Layer 2: Cabinet Enclosure Protection Zone
The protection objects are all electrical and non-battery components within the cabinet volume: cabling, busbars, connectors, fuses, and the PCS equipment itself. The fire hazard here is a conventional electrical or equipment fire. Detection relies on area-type detectors mounted at the top of the cabinet. Upon alarm, the dedicated cabinet suppression unit activates, flooding the entire enclosure volume with clean agent to extinguish the fire. This layer operates independently but can also function as a critical backup or supplementary action if a module-level event escalates or if the primary fire originates from electrical sources.
The system’s response can be modeled as a state machine. Let $S_{module}(t)$ represent the state of a battery module (0=Normal, 1=Alarm, 2=Discharge). Let $S_{cabinet}(t)$ represent the state of its host cabinet. The system’s transition is governed by logical conditions based on detector signals $D_{module}$ and $D_{cabinet}$.
$$
\text{Activation Condition for Module } i: \quad S_{module_i}(t+1) = 2 \quad \text{if} \quad D_{module_i}(t) > \tau_{alarm}
$$
$$
\text{Activation Condition for Cabinet } j: \quad S_{cabinet_j}(t+1) = 2 \quad \text{if} \quad \left( \sum_{i \in j} D_{module_i}(t) > \tau_{cabinet} \right) \lor \left( D_{cabinet_j}(t) > \tau_{cabinet} \right)
$$
Where $\tau_{alarm}$ and $\tau_{cabinet}$ are predefined alarm thresholds.
4. Detailed System Design and Component Selection
4.1 Fire Detection System Design
Early and accurate detection is paramount for mitigating battery energy storage system fires. A single-parameter detector is insufficient due to the complex behavior of thermal runaway.
4.1.1 Detector Selection for Battery Modules
The characteristic signatures of Li-ion cell thermal runaway are well-documented: rapid temperature rise ($\frac{dT}{dt}$), emission of volatile organic compounds (VOCs) like hydrogen, ethylene, and methane, generation of carbon monoxide (CO), and eventually, smoke from vented electrolyte. Therefore, a multi-parameter composite detector is selected for installation inside each module. This detector integrates sensors for Temperature, VOC, CO, and Smoke (photoelectric). A fusion algorithm processes these signals to provide a highly reliable alarm while minimizing nuisance alarms from benign environmental changes. The alarm logic can be represented as a weighted function:
$$
\text{Alarm Index } A = w_T \cdot f(T, \frac{dT}{dt}) + w_{VOC} \cdot g(C_{VOC}) + w_{CO} \cdot h(C_{CO}) + w_S \cdot k(S_{obs})
$$
An alarm is declared when $A \geq A_{critical}$. The detector is installed within the module’s wiring compartment or a dedicated cavity, connected via a CAN bus network for reliable digital communication and individual addressing.
Table 2: Typical Fire Signature Parameters for Li-ion Thermal Runaway
| Parameter | Typical Pre-alarm Range | Alarm Threshold (Example) |
|---|---|---|
| Temperature Rate-of-Rise | > 1°C/min | > 5°C/min sustained for 30s |
| VOC Concentration (LEL%) | 0 – 10% LEL | > 15% LEL |
| CO Concentration (ppm) | 10 – 50 ppm | > 100 ppm |
| Smoke Obscuration (%/m) | 0 – 2 %/m | > 3 %/m |
4.1.2 Detector Selection for Cabinets
For the cabinet-level protection zone, a similar but strategically placed composite detector is used. It is installed at the highest point inside the cabinet (typically the top), where smoke and heat accumulate. If a cabinet has internal partitions that obstruct smoke flow, additional detectors are required in each sub-section to ensure adequate coverage.
4.2 Fire Suppression System Design
4.2.1 Suppression Unit for Battery Modules
Each battery module (~12.5 kWh, volume < 0.03 m³) is protected by a compact, self-contained fire suppression unit. This unit utilizes a solid-state gas generator to pressurize and atomize its stored liquid agent upon receiving an electrical signal from the detector/controller. For this application, a unit containing ≥1.6 kg of Perfluoro(2-methyl-3-pentanone) (Novec 1230 or equivalent) is specified. Key advantages include:
- Zero Pressure Storage: Safe during transport and handling.
- Rapid Discharge: Achieves effective concentration within the small module volume in seconds.
- Maintenance-Free: Hermetically sealed with a long service life.
- Point-Source Application: Minimizes agent usage and targets the hazard directly.
The required mass of agent $m_{agent}$ for a module can be estimated based on the volume $V_{module}$ and the design concentration $C_d$ required to suppress a Li-ion fire (typically 6-8% v/v for this agent), considering the material factor for the enclosure $k_m$:
$$
m_{agent} = \frac{V_{module} \cdot C_d \cdot \rho_{air}}{k_m \cdot (1 – C_d)} \quad \text{(simplified volume-based estimate)}
$$
Given the small, sealed volume of a module, a fixed unit size proven by testing is selected rather than a calculated value.
4.2.2 Suppression Unit for Cabinets
For the cabinet-level total flooding protection, a larger, similar technology unit with an integrated directional spray nozzle is selected. The agent mass is calculated based on the net cabinet volume $V_{cabinet}$ (subtracting solid obstructions). The design must achieve and maintain the minimum design concentration $C_{min}$ for a specified holding time (e.g., 10 minutes). The calculation follows the standard total flooding formula for clean agents:
$$
m_{cabinet} = \frac{V_{cabinet} \cdot C_d}{ (1 – C_d) } \cdot \frac{1}{s} \cdot \left( \frac{T_{amb}}{T_{ref}} \right)^{1/2}
$$
Where $s$ is the specific volume of the agent vapor at reference temperature $T_{ref}$, and $T_{amb}$ is the ambient temperature. The nozzle is positioned at the cabinet’s top to ensure optimal dispersion.
Table 3: Key Properties of Perfluoro(2-methyl-3-pentanone) Agent
| Property | Value / Characteristic |
|---|---|
| Chemical Formula | CF₃CF₂C(O)CF(CF₃)₂ |
| Design Concentration (for Class A, C fires) | 4 – 6% (v/v) |
| No Observed Adverse Effect Level (NOAEL) | > 10% (v/v) |
| Ozone Depletion Potential (ODP) | 0 |
| Atmospheric Lifetime | ~5 days (0.014 years) |
| Global Warming Potential (100-yr) | < 1 |
| Dielectric Strength | > 110 kV (excellent electrical insulator) |
| Residue | None (clean, evaporates completely) |
4.2.3 Nozzle and Piping
For module-level units, a flat-spray atomizing nozzle with a K-factor ≥ 2.0 (where flow $Q = K \cdot \sqrt{P}$) is connected via a short, flexible, braided hose (length ≤ 2m). For cabinet units, an integrated nozzle is preferred to simplify installation.
4.3 Redundant Activation: Mechanical Temperature Actuator
To address the failure scenario of a complete system power loss, a purely mechanical backup is installed in each cabinet. This device contains a fusible element (eutectic alloy) designed to melt at a specific temperature (e.g., 100°C). Upon melting, it completes an electrical circuit, directly initiating the fire suppression units within that cabinet (both cabinet unit and all module units). This provides a final, fail-safe layer of protection, ensuring the battery energy storage system has a passive defense mechanism. This initiation also sends a dry-contact signal to the control panel to indicate activation.
4.4 Control System Architecture
The control system is hierarchical for efficiency and clarity.
- Integrated Control Box (ICB): One is installed in each battery cabinet. It acts as a local controller, continuously polling the multi-parameter detectors of its six assigned modules via CAN bus. Upon confirming an alarm from a specific detector (using the fusion algorithm), the ICB sends the firing command to the corresponding module’s suppression unit. It also relays status and alarm information to the Master Control Panel.
- Master Fire Control Panel (FCP): This is the central nervous system. It displays real-time status (Normal, Alarm, Discharge, Fault) for all four cabinets (3 battery, 1 PCS) and all 18 modules. It receives alarms from cabinet-level detectors and from the ICBs. It can be configured for manual override, system test, and alarm history logging. It also provides primary power and supervises all wiring for open/short circuits.
Table 4: System Response Logic Matrix
| Event Location | Module Detector Alarm | Cabinet Detector Alarm | Mechanical Actuator Triggers | System Response |
|---|---|---|---|---|
| Single Battery Module | YES | NO | NO | ICB activates only the affected module’s suppression unit. Local & central alarm. |
| Battery Cabinet (Electrical Fire) | NO | YES | NO | FCP/ICB activates the cabinet’s suppression unit. Local & central alarm. |
| Multiple Modules / Severe Event | YES (Multiple) | YES (Possible) | NO | ICB activates affected module units. FCP may also activate cabinet unit. Local & central alarm. |
| Any Cabinet (Power Loss Scenario) | N/A (No Power) | N/A (No Power) | YES | Mechanical actuator directly initiates all suppression units (cabinet + modules) in that cabinet. |
5. Design Verification and Performance Validation
The proposed fire safety design for the 220 kWh battery energy storage system was validated through a full-scale live fire test conducted in collaboration with a major energy storage provider. A representative battery module was instrumented and intentionally driven into thermal runaway within a test cabinet configured with the described protection system.
Key Test Results:
- Detection Performance: The internal multi-parameter detector provided an early alarm during the incipient gas generation phase, several minutes before open flame and significant smoke production.
- Suppression Performance: Upon automatic activation, the module-level suppression unit discharged its agent. Open flame was extinguished within 5 seconds of agent application.
- Temperature Control Performance: The most critical metric for a battery energy storage system fire is preventing thermal propagation. The system successfully controlled the temperature. The peak temperature within the faulted module was rapidly reduced, and the temperature of adjacent modules remained well below the thermal runaway threshold. The target temperature of the affected module was brought below 85°C within 10 minutes.
- Post-Suppression Stability: No re-ignition occurred during a prolonged observation period (1 hour), confirming the effectiveness of the agent in both suppression and cooling.
This validation confirmed that the layered, point-source design can effectively detect, suppress, and critically, contain a lithium-ion battery fire at its earliest stage, preventing a single module failure from escalating into a catastrophic cabinet or system-wide event.
6. Conclusion and Future Outlook
The fire safety design presented for a 220 kWh containerized battery energy storage system demonstrates a practical and effective solution to a critical industry challenge. By implementing a dual-layer (module-level and cabinet-level) protection strategy using integrated multi-parameter detection and rapid, targeted clean agent suppression, the system addresses the unique hazards of lithium-ion battery thermal runaway. The inclusion of a redundant mechanical activation mechanism ensures reliability even under power failure conditions. This design philosophy—emphasizing early detection, point-source intervention, and compartmentalization—is essential for managing the inherent risks of electrochemical energy storage.
The successful experimental validation underscores the system’s capability to provide early warning, rapid fire suppression, and, most importantly, thermal management to halt cascading cell failures. This approach not only protects valuable assets but also enhances public confidence in the safety of deployed battery energy storage system installations. Future developments may involve integrating these protection systems with broader facility Building Management Systems (BMS) and Energy Management Systems (EMS) for coordinated response, as well as continued research into even earlier detection signatures (e.g., pressure, acoustic emissions) and advanced cooling techniques for larger-scale battery energy storage system applications. Standardization of such safety architectures will be crucial for the sustainable growth of the global energy storage industry.