Full-Scale Experimental Study on the Combustion Characteristics of a Fuel Island in a High-Speed Railway Station

Abstract

This study aims to provide a reference for the fire protection design and fire emergency response strategies for fuel islands in high-speed railway stations and other transportation buildings. By using an industrial calorimeter, this paper analyzes the combustion characteristics of a fuel island. For the fuel island setup in this test, the fuel island fire development cycle was relatively long, and the maximum fire source heat release rate reached 4615 kW. Before the fire source heat release rate reaches the maximum peak, the HRR curve slowly fluctuates and grows within the first 260 s after ignition. Within the time range of 260 s to 440 s, the fire growth rate resembled that of a t² medium-speed fire, and within the time range of 400 s to 619 s, it more closely aligned with a t² fast fire. It is generally suggested that the growth curve of t² fast fire could be used for the numerical simulation of fuel island fires. The 1 h fire separation method adopted in this paper demonstrated a good fire barrier effect throughout the combustion process.

1. Introduction

High-speed rail has become the main mode of transportation for Chinese residents traveling long distances. By the beginning of 2024, the operating mileage of China’s high-speed rail was expected to reach 45,000 km [1,2]. High-speed rail stations are characterized by high passenger density and dense population. In order to better serve passengers and improve the comfort of those waiting for trains, some commodity retail outlets, referred to as “fuel islands”, are usually set up in high-speed railway stations, as shown in Figure 1. The fuel island refers to a small, roofless display and retail service facility located in a large open space. These facilities are required to be limited to an area of 6–20 m². Such commodity retail outlets present certain fire hazards, and the development and spread of fire in these structures is unstable, which increases the overall fire risk in high-speed railway stations [3,4,5]. Therefore, it is very necessary to carry out full-scale tests on fuel island fire spread in high-speed railway stations to better understand their fire characteristics and development behavior.

The study of fire hazards in high-speed railway stations has gradually attracted the attention of scholars. In the early stage, Li et al. [6] proposed a fire scenario design method for transportation buildings. Zhang et al. [7] analyzed a calculation model for determining the safe distance of fuel island fires in large spaces and recommended using the point source and line source models in engineering applications. Bi [8] discussed the application of performance-based design concepts for fuel islands in room-in-room structures in large spaces, based on engineering examples. Zhang [9] examined fire safety strategies for commercial shops in transportation buildings by analyzing the architectural characteristics and fire hazards of a certain transportation hub project. Liu et al. [10] conducted full-scale experiments in large-scale transportation buildings and commercial shops, obtaining data on the fire heat release rate and temperature field variation. Ma et al. [11] used numerical simulation to study the variation of heat release rate during electrical fires of seats in high-speed railway stations. Yang et al. [12] used FDS software to analyze the fire development process of massage chair fires in high-speed railway stations and determined the critical ignition distance. Peng et al. [13] proposed a fire protection design solution for fuel islands in the public distribution spaces of public transportation projects. Chen et al. [14] used numerical simulation methods to analyze the variation law of thermal radiation in fuel island fire scenarios within the waiting hall of a high-speed railway station. Summarizing the above studies, existing research on fire hazards in high-speed railway stations mainly relies on the performance-based construction of fire scenarios and numerical simulation of the fire development process. However, numerical simulation work inevitably requires simplifications of models and boundary conditions, which may cause deviations from actual fire scenarios.

This paper conducts a full-scale experimental investigation into the combustion and spread characteristics of fuel island fires in high-speed railway stations. It measures and analyzes the evolution of fuel island fire scale, heat release rate, and temperature distribution, aiming to provide a basis and reference for fire protection design and emergency response o in such high-speed railway stations and transportation buildings. Moreover, the full-scale data could serve as a valuable baseline for supporting AI model training [15], validating CFD simulations [16], or benchmarking future experimental and theoretical studies [17].

2. Experimental Setup

2.1. Industrial Calorimeter Introduction

This test was conducted at the Hebei Research Base of the Institute of China Academy of Building Research Fire Institute. This test was conducted under an industrial calorimeter (range of 5 MW). The length of the smoke hood is 8.0 m, and the width is 6.0 m. The manufacturer of this equipment is Tasitek Testing Instrument Technology Co., Ltd., which is located in Suzhou, China. The measurement principle of the industrial calorimeter is as follows.

In the combustion test, the heat release rate $q$ of the sample is in kilowatts (kW) and could be calculated according to Equation (1):

$$q=E{V}_{298}{x}_{{\mathrm{O}}_2}^{\mathrm{a}}\left({\displaystyle \frac{\phi }{\phi \left(\alpha -1\right)+1}}\right)-{\displaystyle \frac{E}{E_{{\mathrm{C}}_3{\mathrm{H}}_8}}}{q}_b$$

(1)

where

$q$ is the heat release rate of the sample, in kilowatts (kW);

$E$ is the energy released by the unit volume of oxygen consumed by the sample during combustion, and the unit is kilojoule per cubic meter (kJ/m³); for the sample, when the ambient temperature is 25 °C, $E$ is 17.2 × 10³ kJ/m³;

$V_{298}$ is the volume flow rate in the exhaust duct at 25 °C under atmospheric pressure, in cubic meters per second (m³/s);

$x_{{\mathrm{O}}_2}^{\mathrm{a}}$ is the mole fraction of oxygen;

$\phi$ is the oxygen consumption coefficient;

$\alpha$ is the oxygen consumption expansion coefficient for combustion reactions, which is taken as 1.105;

$E_{{\mathrm{C}}_3{\mathrm{H}}_8}$ is the energy released by the unit volume of oxygen consumed by the propane during combustion, and the unit is kilojoule per cubic meter (kJ/m³); for propane, when the ambient temperature is 25 °C, $E_{{\mathrm{C}}_3{\mathrm{H}}_8}$ is 16.8 × 10³ kJ/m³;

$q_b$ is the heat release rate of the ignition source, in kilowatts (kW).

The oxygen consumption coefficient $\phi$ could be calculated according to Equation (2):

$$\phi ={\displaystyle \frac{x_{{\mathrm{O}}_2}^0\left(1-x_{{\mathrm{CO}}_2}\right)-x_{{\mathrm{O}}_2}\left(1-x_{{\mathrm{CO}}_2}^0\right)}{x_{{\mathrm{O}}_2}^0\left(1-x_{{\mathrm{CO}}_2}-x_{{\mathrm{O}}_2}\right)}}$$

(2)

where

$\phi$ is the oxygen consumption coefficient;

$x_{{\mathrm{O}}_2}^0$ is the initial mole fraction of oxygen in the exhaust duct;

$x_{{\mathrm{CO}}_2}$ is the mole fraction of carbon dioxide in the exhaust duct during the test;

$x_{{\mathrm{O}}_2}$ is the mole fraction of oxygen in the exhaust duct during the test;

$x_{{\mathrm{CO}}_2}^0$ is the initial mole fraction of carbon dioxide in the exhaust duct.

The mole fraction of oxygen $x_{{\mathrm{O}}_2}^{\mathrm{a}}$ could be calculated according to Equation (3):

$$x_{{\mathrm{O}}_2}^{\mathrm{a}}=x_{{\mathrm{O}}_2}^0\left(1-x_{{\mathrm{H}}_2\mathrm{O}}^{\mathrm{a}}\right)$$

(3)

where

$x_{{\mathrm{H}}_2\mathrm{O}}^{\mathrm{a}}$ is the mole fraction of water vapor in the environment.

The mole fraction of water vapor $x_{{\mathrm{H}}_2\mathrm{O}}^{\mathrm{a}}$ could be calculated according to Equation (4):

$$x_{{\mathrm{H}}_2\mathrm{O}}^{\mathrm{a}}={\displaystyle \frac{H}{100 p}}\times \exp \left[23.2-\left({\displaystyle \frac{3816}{T_s}}\right)\right]$$

(4)

where

$H$ is the initial relative humidity in the laboratory before the test, in %;

$p$ is the atmospheric pressure in the laboratory before the test, in pascal (Pa);

$T_s$ is the gas temperature in the exhaust duct, in Kelvin (K).

Note: Equations (1)–(4) are obtained based on the following constraints:

(a)

Ignoring the amount of carbon monoxide produced during combustion;

(b)

The effect of water vapor produced during combustion on the measurement is partially considered;

(c)

$E$ = 17.2 × 10³ kJ/m³ is the average value of most combustion products.

2.2. Full Scale Fuel Island

This test was conducted under an industrial calorimeter so that the spread of the fuel island fire could be observed, and the heat release of the fuel island fire could be understood. As shown in Figure 2, the fuel island is arranged in the center of the smoke hood, ensuring that the entire fuel island is within the range of the smoke hood. The items in the fuel island are mainly snacks, gifts, and general merchandise commonly used by daily passengers.

Taking reference from the “Code for Design of Fire Prevention for Railway Engineering” [18], which stipulates that the area of dispersed commercial installations should be controlled to 20 m², the fuel islands in this test are all assembled with shelves, with a width of 4.2 m, a length of 5.0 m, and a total area of 21 m². Compared with the code, it is more conservative. According to the relevant stipulation of the “Code for Fire Protection Design of Buildings” [19], the width of one-way stream of people is 0.55 m, so the design channel width of the fuel island is not less than 0.55 m. The test also tested the effect of a 1 h fire separation method applied to the fuel island. The fire separation method uses red bricks arranged into an I-shaped wall with a thickness of 0.12 m, and a layer of 0.02 m thick cement board is covered on the top of the wall. The schematic diagram of the fuel island test plan layout is shown in Figure 3a, and the schematic diagram of the one-hour fire separation method is shown in Figure 3b.

At the beginning of the experiment, use a torch dipped in a little alcohol to ignite the pillows on the second layer of the shelf at the ignition position. Immediately after ignition, remove the torch from the test area and extinguish it to ensure that the HRR of the torch does not affect the experimental measurement results.

2.3. Measurement Layout

HRR and THR were measured by the industrial calorimeter, and the data acquisition system of the industrial calorimeter should record and store all instrument input data at a frequency of no more than 3 s, and the cumulative measurement error should be less than 10% [20]. Eight thermocouple trees with a total of 56 K-type thermocouples were arranged at different locations to analyze the changes in temperature field in the fire scene, as shown in Figure 4. Generally, the maximum uncertainty error of the thermocouple is less than 6% [21]. Seven thermocouples were arranged on each thermocouple tree with a spacing of 0.5 m. Cameras were arranged in four directions of the fuel island to record the test phenomena.

3. Results and Analysis

3.1. Experiment Phenomenon

As shown in Figure 5, where panels (a–e) are taken by camera 3, panel (f) is taken by camera 1, and panels (g–i) are taken by camera 4. When the combustion progressed to 80 s, the flame began to spread to both sides. At 107 s, the goods on the second shelf were ignited and obvious smoke began to appear. At 334 s, the goods on the fourth shelf were ignited, and the entire shelf entered the combustion state. At 602 s, as the combustion progressed, a large amount of hot smoke accumulated in the space, and the shelf in the middle of the fuel island was ignited. At 787 s, the back of the middle shelf of the fuel island was ignited. Subsequently, the right half of the fuel island was ignited, and the left half entered the decay stage. At 1548 s, the entire fuel island entered the combustion state and then entered the decay stage. The flame was low throughout the combustion process, and the vertical direction was within 1 m of the height of the fuel island. The shelf structure remains relatively good, and there is less melting droplet of combustible materials such as goods, so the probability of fire spreading due to melting droplets is also relatively small. There is no smoke leakage during the measurement process, which would not affect the accuracy of the results.

As shown in Figure 6, which was taken by Camera 2, adjacent shelves could well verify the effectiveness of the 1 h fire separation method. During the entire combustion process of the fuel island, although the flame was very fierce, due to the existence of fire separation measures, the flame did not spread to the adjacent shelf area separated by the wall. This fully demonstrates that the fire separation method effectively played a role in creating the fire barrier and successfully prevented the spread of fire.

3.2. Distribution of Temperature Field

Thermocouple trees T2, T6, and T8 reflect the temperature field changes in burning fuel islands, as shown in Figure 7. The timing of the peak appearing on each thermocouple tree is different, indicating that the fire first spreads along the shelf to position T2 of the thermocouple tree, and then spreads to positions T6 and T8. It can be observed that the temperature of each thermocouple tree has two peaks. This is consistent with the phenomenon that the right half of the fuel island was ignited, and the left half entered the decay stage.

Thermocouple trees T3 and T4 reflect the temperature field changes in the adjacent shelf areas separated by walls, as shown in Figure 8. The temperature field distribution in the area where the fuel island fire occurred showed the characteristics of high temperature in the lower part and low temperature in the upper part. However, in the adjacent shelf areas separated by walls, the temperature field distribution showed the opposite trend, that is, the high temperature was in the upper part and the low temperature was in the lower part, with a maximum temperature of 260 °C. The reasonable explanation for this phenomenon is that the heat convected from the top of the fuel island to the adjacent shelf is much greater than the heat conducted from the wall to the adjacent shelf. This phenomenon further confirms the effectiveness of the 1 h fire separation method for fire barrier.

Thermocouple trees T1, T5, and T7 reflect the temperature field changes in burning fuel islands, as shown in Figure 9. After ignition, T1 rapidly increases in temperature due to its proximity to the fire source, reaching its maximum value between 550 and 750 s, with the highest combustion temperature exceeding 800 °C. The temperature changes in the T5 and T7 are relatively small in the early stage. When the temperature at T1 begins to decrease, the temperature at T5 and T7 begins to rise. T5 reaches its maximum value around 1400 s, while T7 reaches its maximum value around 1700 s.

Thermocouple trees T1, T5, and T7 are plotted as contours, as shown in Figure 10b, and the value plane of the contours is shown in Figure 10a. On the same thermocouple tree, the bottom thermocouple is 3 m away from the top thermocouple, and the spacing between the thermocouple trees is 1.6 m, so the contours depict the temperature field of a plane of size 3.2 m × 3 m. The contours could intuitively reflect the distribution characteristics and changing patterns of fire scene temperature [22]. The high-temperature area gradually shifted from the left half of the fuel island to the right half. The temperature of the bottom of the T1 thermocouple tree rose rapidly due to its proximity to the fire source, and the highest temperature exceeded 800 °C. The temperature of the T5 and T7 thermocouple trees did not change much due to the obstruction of the shelves. As the fire spread from the left half of the fuel island to the right half, the fire entered the secondary development stage, and the temperature of the T5 and T7 thermocouple trees rose.

3.3. Distribution of Radiation Heat Flux

The HRR (heat release rate) refers to the heat released by the combustion of a material per unit time under specified test conditions [23]. The THR (total heat release) refers to the total amount of heat released by a material from the time it is ignited to the time the flame is extinguished under a preset incident heat flux intensity. Measuring and analyzing the HRR and THR of the fuel island could provide a basis and reference for the numerical simulation setting, fire protection design and emergency response of fire accidents in such high-speed railway stations and transportation buildings. As shown in Figure 11, the HRR of the fuel island has two peaks, which also verifies the observed phenomenon that the right half of the fuel island was ignited, and the left half entered the decay stage. The first peak is 4615 kW at 615 s, and the THR is 645 MJ at this time. The second peak is 4277 kW at 1533 s, and the THR is 3433 MJ at this time. Comparing the test curve with the t² fire, it is found that the HRR curve of the fuel island increases slowly within 260 s of ignition. In the time range of 260 s to 440 s, the growth rate of this test is similar to that of medium-speed fire, and in the time range of 400 s to 619 s, the growth rate of this test is similar to that of fast fire.

4. Conclusions

This paper uses an industrial calorimeter to analyze the fire spread behavior of fuel islands and draws the following conclusions:

• For the fuel island setup in this test, the fuel island fire development cycle is relatively long, and the maximum fire source heat release rate is 4615 kW. The relevant experimental analysis and data results have been applied in the special fire protection design of high-speed railway stations such as Chongqing East Station and Nanjing North Station in China.
• Before the fire source heat release rate reaches the peak, the HRR curve slowly fluctuates and grows within the first 260 s of ignition. Within the time range of 260 s to 440 s, the growth rate of this test is similar to that of t² medium-speed fire, and within the time range of 400 s to 619 s, the growth rate of this test is similar to that of t² fast fire. It is generally suggested that the growth curve of t² fast fire could be used for the numerical simulation of fuel island fires.
• The 1 h fire separation method adopted in this paper demonstrates a good fire barrier effect throughout the combustion process.

This paper aims to investigate the combustion and spread characteristics of fuel island fires in high-speed railway stations in depth and explore them through full-scale experimentation. During the research, we accurately measured and analyzed the scale of fuel island fires, heat release rate, and temperature distribution. Through the collection and research of these data, we hope to provide a scientific basis for the fire protection design of fuel islands in high-speed railway stations and transportation buildings, as well as an effective reference for emergency response planning. This study not only helps improve the safety of such buildings but also provides valuable information to firefighters in related fields so that they could respond more effectively to possible fire accidents.