Full-Scale Experimental Investigation of Temperature Distribution and Smoke Flow in a Road Tunnel with a Novel Water Mist Fire Fighting System

Abstract

This study presents a novel water mist fire fighting system that integrates water mist sprays and water mist curtains, designed to achieve simultaneous fire suppression, thermal insulation, and smoke control. Three full-scale experiments were conducted under various fire scenarios, and the changes in fire behavior and heat release rate were examined to evaluate the effectiveness of the water mist system in extinguishing fires. Additionally, the spatiotemporal changes in ceiling temperature were monitored to assess the cooling and protective effects of the water mist. The thermal insulation capability of the system was also investigated by detecting the temperature distribution inside the tunnel. Moreover, the smoke conditions upstream and downstream of the tunnel were analyzed to evaluate the smoke-blocking performance of the water mist system. The findings demonstrate that the water mist fire fighting system is highly efficient in attenuating the fire and restricting its progression. Within the water mist spray section, the average ceiling temperature decreased exponentially during both the initial and steady burning phases across all tested fire scenarios. Nonetheless, the smoke-carrying capacity of the water mist spray is limited. Fortunately, the dispersed smoke was diluted by water mist, markedly enhancing visibility and mitigating the impact of smoke on tunnel illumination.

1. Introduction

Road tunnels are vital components of transportation infrastructure, engineered to navigate complex terrain and environmental constraints. They play an irreplaceable role in enhancing traffic efficiency, optimizing space utilization, and contributing to ecological sustainability [1,2]. Currently, China is transitioning to a transportation power, with road tunnels shifting from large-scale construction to operational and management phases [3,4]. However, tunnel-related disasters can lead to severe loss of life and property [5]. Hence, safety assurance remains a paramount concern in managing road tunnels.

Fire incidents rank among the most frequent hazards in tunnels, posing serious threats to occupants, vehicles, and tunnel structures [6,7]. When a fire occurs, the temperature inside the tunnel rises rapidly, leading to severe burns to the skin and respiratory tract [8,9]. Tunnel fires also generate large amounts of smoke and toxic gases, such as carbon monoxide and hydrogen cyanide [10]. Due to the enclosed nature and limited ventilation of tunnels, smoke tends to accumulate, drastically reducing visibility and exacerbating respiratory distress [11]. The inhalation of these toxic gases can quickly lead to poisoning, unconsciousness, or even death [12]. Over 20 years ago, major tunnel fire accidents caused significant casualties worldwide. As a case in point, the Mont Blanc Tunnel fire in 1999, which occurred between France and Italy, claimed 39 lives, and the Gotthard Tunnel fire in Austria in 2001 resulted in 11 fatalities [13]. In China, data from the police reported 256 fire incidents in road tunnels between 2010 and 2021, with the number of tunnel fires increasing annually [3]. These alarming events have prompted tunnel operators and researchers to pursue more effective fire protection strategies and emergency response measures.

Water-based fixed fire fighting systems (WFFFS) were proposed as an effective supplementary measure for controlling tunnel fires as early as 60 years ago [14]. The water mist fire fighting system, a type of WFFFS, is a clean and efficient solution [15,16]. Utilizing only water, this system avoids ecological contamination while safeguarding both critical infrastructure and personnel within designated zones [17,18]. Water mist effectually reduces the flame temperature and slows the fire development in the early stages by absorbing heat through rapid evaporation [19,20]. Owing to the high evaporation efficiency of the water mist, the system can quickly absorb large amounts of heat, significantly lowering the ambient temperature. This thermal control helps prevent structural instability or collapse, protects vehicles and equipment, and reduces overall property damage. Additionally, the fine water droplets in the mist are small enough to absorb tiny particles and harmful gases in the smoke. This process significantly lowers the smoke concentration in the air, improves visibility within the tunnel, and facilitates fire rescue operations and personnel evacuation. Notably, the water mist system consumes only 1% to 5% of the water required by conventional sprinkler systems [21].

Water mist is primarily employed to suppress fires [22]. Many researchers have conducted large-scale fire tests with WFFFS, mainly to evaluate their effectiveness prior to implementation in real tunnels. For instance, Ingason et al. [23] conducted fire tests to compare the fire suppression effectiveness of various nozzles in the Runehamar Tunnel, Chang et al. [24] performed heptane pool combustion tests with water mist, and Li et al. [25] conducted a diesel pool combustion test with WFFFS in an urban tunnel. Additionally, some researchers have carried out model-scale tunnel fire tests to examine the effectiveness of WFFFS. For instance, Ingason [26] conducted a 1:23 scale tunnel fire test with water mist, Li and Ingason [27] performed a 1:10 scale fire test with an automatic spray system, and Blanchard et al. [28] performed a fire suppression test in a medium-scale tunnel. These investigations have contributed to the development of fire suppression strategies tailored to tunnel environments. The findings from these studies demonstrate that WFFFS can effectively limit fire growth and prevent flame spread to adjacent vehicles. Furthermore, by continuously absorbing heat through evaporation, water mist also provides effective thermal protection for tunnel structures.

On the other hand, except for extinguishing fires, water mist can also function as a water curtain for fire partitioning [29,30]. While most research has focused on the effectiveness of WFFFS in directly suppressing tunnel fires, relatively few studies have explored water-based fixed compartmentation systems (WFCS) that utilize water curtains [31]. The primary function of WFCS is not to directly extinguish the fire but to use a water curtain or water screen to segregate the fire source from the rest of the tunnel, ensuring the safety of personnel in the unaffected areas. Several researchers have explored water mist systems designed to prevent smoke spread and separate fire zones. For example, Ingason [26] conducted a 1:23 scale model tunnel fire test using the water curtain installed either upstream or downstream of the fire source to evaluate its effectiveness in inhibiting the diffusion of fire. Sun et al. [29], Li et al. [32], and Chen et al. [33] investigated the effectiveness of water mist systems in limiting smoke. Nishino et al. [34] executed a large eddy simulation of spray resistance in tunnel fires, examining the impact of water sprays on smoke flow characteristics. Their findings indicated that the water spray diverted ceiling smoke downward, creating airflow patterns that channeled combustion smoke toward the floor. They also observed that recirculating smoke could re-enter the combustion zone, potentially suppressing flames through localized oxygen depletion.

In this study, a new water mist fire fighting system, consisting of water mist sprays and a water mist curtain, was designed based on the concepts of WFFFS and WFCS. This system uniquely combines these two principles into a unified design. The system aims to achieve simultaneous fire suppression, heat insulation, and smoke barrier effects. Specifically, the spray section is primarily responsible for cooling and suppressing the fire plume, while the curtain section functions as a vertical barrier that inhibits smoke and heat propagation along the tunnel. This dual-action mechanism offers a comprehensive protective effect not typically found in conventional systems. Three different fire scenarios were set up, and the variations in the fire behavior and heat release rate under these scenarios were recorded to assess the performance of water mist in fire suppression. Additionally, the spatiotemporal changes in the tunnel ceiling temperature were monitored to assess the cooling protection provided by the water mist. The effectiveness of the thermal insulation was further explored by recording the temperature distribution changes during the tunnel space. Furthermore, the smoke conditions upstream and downstream of the tunnel were captured using high-definition cameras to evaluate the capability of the water mist system to block smoke. Figure 1 illustrates the test flow diagram indicating the experimental procedures.

2. Fire Test Description

2.1. Full-Scale Test Tunnel

The combustion experiments were conducted in a full-scale test tunnel made of steel, with a total length of 160 m. As displayed in Figure 2, the net height of the tunnel is 7.5 m, and its width is 10.4 m. The tunnel comprises a dual-carriageway configuration providing a clear lane width of 7.9 m. The enclosure surfaces, including walls and ceiling, feature light-gray aluminum cladding, with 3 m high yellow painted sections flanking both sides. The tunnel floor is paved with concrete, and LED luminaires are symmetrically mounted along both sides for illumination. Additionally, a black baffle at a height of 2 m from the ground is positioned at the tunnel exit to block the fire smoke. This baffle helps to more clearly observe the amount of smoke downstream of the fire source.

2.2. Water Mist Fire Fighting System (WMFFS)

Building on our previous research [35] and drawing upon the principles of both water-based fixed fire suppression and compartmentation systems, we propose a novel water mist fire fighting system. This system comprises a combination of water mist spray and water mist curtain, designed to simultaneously suppress the fire source and block the spread of smoke. The design concept of this water mist fire protection system is illustrated in Figure 3.

The WMFFS utilized in this investigation features a water mist spray section composed of two kinds of nozzles, with their definite arrangement and positioning illustrated in Figure 4. The primary objectives of this water mist spray section are to suppress flame combustion and block the high temperature and smoke generated by the fire source. The longitudinal coverage of the water curtain measures 29 m, with distribution lines running parallel to the vault curvature. For the cooling and extinguishing function in the central section, the pipe spacing is 3 m, while for heat and smoke insulation on both sides, the pipe spacing is reduced to 2 m. Two kinds of nozzles, referred to as Nozzle I and Nozzle II, are employed in the system, with their designs and arrangement patterns shown in Figure 4. Nozzle I emits a conical mist dispersion pattern to achieve 120° projection angles, while Nozzle II produces a vertical water curtain. Both nozzles operate at the same working pressure of 0.8 MPa and a water flow rate of 12.25 L/min. Based on particle size distribution tests, 99% of the water mist droplets reach a diameter (D_v,0.99) of 390 μm or smaller, while 50% of the droplets (D_v,0.50) measure 193 μm or less.

2.3. Fire Source

According to previous studies [36,37,38,39], small five-seater vehicles make up a significant proportion of vehicles in highway tunnels. Therefore, the fire scenario in this study focuses on fires involving small cars traveling through tunnels. PIARC [40] suggests that the heat release rate (HRR) of a small car fire typically ranges between 2.5 and 8 MW.

In this experiment, the fire source is designed as an oil pan fire. Diesel fuel is chosen because it generates a large amount of smoke upon ignition, which facilitates observation [8,41,42,43]. The diesel is contained within a combustion pan constructed from welded steel plates. The HRR of the oil pan fire primarily depends on the combustion area, allowing the fire intensity to be controlled by adjusting the size of the fuel pan. Diesel fuel was selected for the combustion source primarily because of its high smoke yield, which facilitates the observation of smoke movement and stratification during tunnel fires. This makes it especially suitable for evaluating the smoke-blocking capability of the water mist system. Although real tunnel fires may involve various fuels, including gasoline or battery materials, diesel is widely used in full-scale fire experiments due to its stable combustion characteristics and prevalence in both passenger and commercial vehicles. The size of the fuel pans was carefully designed to generate HRRs within the typical range (2.5–8 MW) for small passenger vehicle fires, ensuring that the fire conditions reflect realistic tunnel fire scenarios. The formula for calculating the HRR (Q) of the fuel pool [39,44] is presented in Equation (1):

$$Q=\chi m^{\prime }\left(1-e^{-k\beta D}\right)A_f\Delta H_c$$

(1)

where χ stands for the combustion efficiency, which is assigned a value of 0.75 [45]; $m^{\prime }$ denotes the maximum mass loss rate, with a value of 0.045 kg/(m²∙s) for diesel fuel [46]; kβ = 2.1 m⁻¹ [46]; D stands for the diameter of the burning area; A_f corresponds to the burning area; ΔH_c is the calorific value of combustion, with a value of 43 MJ/kg for diesel fuel [42].

As displayed in Figure 5a, the oil pan used in this test is a square oil pan with dimensions of 1.5 m × 1.5 m × 0.2 m, and its combustion diameter can be approximated using Equation (2) [45]. Based on Equation (1), the HRR of a single oil pan is approximately 3.2 MW. To generate varying HRRs, different numbers of oil pans (one or two) were used in the fire tests, as depicted in Figure 5a,b.

$$D=2\sqrt{{\displaystyle \frac{A_f}{\pi }}}$$

(2)

In addition, as indicated by Equation (1), the HRR is impacted by the pool area, while there is no information regarding the fuel volume. In fact, the fuel volume affects the period of the fire development. Additionally, by default, Equation (1) is valid for a pool fire with a large fuel depth. The HRR is significantly lower for a pool fire with a fuel depth of 2–3 mm than for a pool fire with deep fuel [45]. In the present study, the fuel depth was about 8 mm, which was sufficient to generate a steady HRR.

2.4. Test Methods

Two high-definition cameras captured combustion dynamics throughout the thermal exposure sequence. As displayed in Figure 6, one camera was positioned outside the range of the WMFFS to capture the movement of smoke during the test. The other camera was placed near the fire source to document the combustion process and flame characteristics of fire. Additionally, after each fire test, the amount of smoke deposited on the black baffle located downstream of the fire source was recorded via a camera.

A weighing system was used to monitor quasi-static variations in the fuel mass within the pan. This system comprised four redundant load cells with a precision of 0.1 kg, collecting data in 1 s intervals. Specifically, the weighing system had a total estimated uncertainty of ±0.2 kg, considering sensor precision and structural stability under thermal influence. As depicted in Figure 6, the weighing system was installed underlying the oil pan, with a rack placed between the pressure sensors and the oil pan to provide thermal insulation.

The temperature inside the tunnel was measured via Type K armored thermocouples, model WRNT-35. The thermocouple has a maximum measurable temperature of 600 °C and a precision of 0.1 °C, with a rated measurement uncertainty of ±0.75% of the reading. Temperature data were collected in 1 s intervals. The thermocouple probes were initially arranged longitudinally along the tunnel vault, positioned 0.05 m vertically below the vault, with a spacing of 2.5 m between each probe. To gather additional temperature data, probes were also placed at descending heights from the vault. Twelve thermocouples were fixtured at heights of 2 m, 4 m, and 6 m above the floor, respectively. Figure 6 illustrates the placement of the thermocouple probes, with a total of 67 thermocouples distributed throughout the tunnel.

2.5. Process of Tests

A series of three combustion tests were conducted in the full-size tunnel with operational water mist suppression.

In the first fire test (T1), a single oil pan with dimensions of 1.5 m × 1.5 m × 0.2 m was used as the fire source to simulate a car fire scenario. Approximately 19 L of diesel fuel was poured into the pan, and 150 mL of 99% pure alcohol was added to aid ignition. Based on Equation (1), the HRR of the ignition source was calculated to be 3.2 MW. As illustrated in Figure 7a, the fire source was placed in the midpoint of the water mist spray section, located 64.5 m from the tunnel entrance. The installation was aligned with the midline of one traffic lane, maintaining a 2 m lateral offset from the central axis of the tunnel.

In the second fire test (T2), two oil pans were used as the fire source to simulate the combustion of two vehicles following a collision. Approximately 17.5 L of diesel fuel was poured into each oil pan, along with 150 mL of 99% pure alcohol for ignition. Both oil pans were ignited simultaneously, resulting in a combined HRR of 6.4 MW. As exhibited in Figure 7b, the fire source was positioned in the middle of the water mist spray section, 64.5 m from the tunnel entrance, and located on the centerline of one lane, 2 m away from the centerline of the tunnel cross-section.

In the third fire test (T3), a single oil pan was used as the ignition source to simulate the burning scenario of a small vehicle. Approximately 18 L of diesel fuel was poured into the pan, with 150 mL of 99% pure alcohol added for ignition. The HRR of this fire source was 3.2 MW. As exhibited in Figure 7c, the fire source was positioned at the edge of the water mist spray section, within the water curtain, 73.5 m from the tunnel entrance. It was located on the centerline of one lane, 2 m away from the centerline of the tunnel cross-section.

The three fire tests considered variations in the HRR and the locations of the fire source. In each test, the WMFFS was activated after the fire source had burned for a specified period. According to the Chinese standard “Technical Requirements for Highway Tunnel Fire Alarm Equipment (JTT 610-2016)” [47], the fire alarm time in a tunnel should not exceed 60 s. Specifically, the 60 s activation delay was designed to simulate a realistic response time that allows for occupant detection and early evacuation initiation, in line with typical human reaction and alarm response timelines in tunnel fire scenarios. Therefore, in all three fire tests, the WMFFS was activated 60 s after the fire source was ignited.

In this study, all fire tests were conducted under the operation of the water mist system. Although a free-burning test (i.e., without fire suppression) is commonly used as a reference in fire protection experiments, it was not included in this work because of the significant safety concerns and structural constraints of the full-scale tunnel.

The designed fire scenarios in this study involved heat release rates (HRRs) up to 6.4 MW. Without the cooling effect of the water mist, such high thermal loads would lead to continuous accumulation of heat beneath the tunnel ceiling, posing a serious threat to the structural integrity of the tunnel lining. Given that the tunnel was constructed for repeated fire testing and long-term use, the risk of thermal-induced damage or partial collapse rendered it unfeasible to perform unsuppressed combustion tests.

3. Results

3.1. Process of Fire Source Suppression by WMFFS

Figure 8 illustrates the entire process, from ignition to extinguishment, of a single oil pan fire during the first fire test under the action of water mist. As exhibited in Figure 8a,b, the fire developed rapidly after ignition, with intense flames visible by 60 s. In Figure 8c–f, the fire did not grow fiercer after 120 s of water mist application, and the flame morphology remained largely unchanged. At this stage, the water mist particles rapidly evaporated and absorbed a significant amount of thermal by contacting the high-temperature fire source, thus preventing further fire growth. As the water mist continued to suppress the fire, the flame began to weaken, as displayed in Figure 8g–k. During this phase, the continuous ejection of water mist created turbulence, disrupting the convection currents of the fire and causing the flames to become irregular or even partially extinguished. After 360 s, the fire source appeared to reignite briefly, becoming more intense after a long period of suppression, as displayed in Figure 8l. However, this resurgence was short-lived, as the fire was quickly suppressed by the fine mist, as depicted in Figure 8m. As the local flames gradually subsided, the ignition source became more localized, as exhibited in Figure 8n–r. Finally, the fire was completely extinguished after 589 s. During this stage, as the water mist continued to evaporate and absorb heat, the temperature of the fire source decreased. More mist droplets adhered to the fuel surface, leading to a reduced fuel supply to the fire source. Consequently, the flame gradually diminished until it was fully extinguished.

Similarly, Figure 9 depicted the entire process, from ignition to extinguishment, of two oil pan fires during the second fire test under the action of water mist. As exhibited in Figure 9a,b, the fires burned rapidly, with the ignition source reaching intense flames within 60 s of ignition. In Figure 9c–g, the water mist quickly disturbed the flame morphology once activated, differing from the observations in the first fire test. This discrepancy may be attributed to the larger burning area of the fire sources and the increased actionable area of the water mist droplets in the second test. As shown in Figure 9h–m, the flames became increasingly irregular, and the fire diminished as the water mist continued to interact with the flames. Subsequently, Figure 9n–r illustrates the progressive extinguishment of localized flames within the oil pans resulting from the evaporation and heat absorption of the water mist. Based on Figure 9s–v, after 570 s of combustion, the fire in the left oil pan was nearly extinguished, leaving only faint flames. As shown in Figure 9w,x, at 690 s, the fire in the right oil pan briefly flared up after suppression, becoming intense for a short period before weakening again. Ultimately, the fires were completely extinguished after 739 s of burning. Although both oil pans were ignited simultaneously, the fires in the second test burned for a longer duration. This extended burning time may be due to the higher power of the second fire source and its elevated flame temperature, which required more water mist droplets to evaporate and absorb heat to reduce the ignition temperature sufficiently.

Unfortunately, during the third fire test, water ingress caused a short circuit in the camera positioned near the fire source, preventing the recording of flame morphology changes.

Combined with the fire combustion process of the first and second fire tests in Figure 8 and Figure 9, the inhibition effect of water mist on the fire source can be described as a confrontation between the water mist and the fire. This expression systematically highlighted the synergistic effects of various physical and chemical mechanisms involved in water mist fire suppression and extinguishing. This adversarial process represented an evolution of dynamic equilibrium, where the fire source attempted to maintain the heat and chemical reaction conditions necessary for combustion, while the water mist disrupted these conditions through several mechanisms. These included heat absorption, disturbance of chemical reaction chains, turbulence and gas dynamics, and interruption of the fuel supply.

3.2. HRR

In the actual fire scenario, the power of the fire source typically increased over time before stabilizing at a constant level. During the initial growth phase of the fire, the HRR of the fire source generally follows a time-based t² law, as expressed in Equation (3):

$$Q=\alpha t^2$$

(3)

In this equation, α represents the growth coefficient, and the combustion rate of diesel is very fast, with a value of 0.19 kW/s² [43], and t denotes the time elapsed after ignition (s). According to Equation (3), the time required for the three fire tests in this study to reach a stable combustion state under free combustion conditions is approximately 129 s.

As displayed in Figure 10, the weighing system recorded the variation in fuel mass in the oil pan from ignition to extinguishment across the three fire tests. In Figure 10a, during T1, the single oil pan fire continued to burn for 589 s under the action of water mist. The mass loss rate of the fuel initially increased, then decreased, and ultimately showed a negative growth as more water entered the oil pan. In Figure 10b, during T2, the two oil pan fires burned for approximately 739 s under the impact of water mist. The fuel mass in each oil pan in T2 was about 14.5 kg (17.5 L), while the mass in a single pan in T1 was about 16 kg (19 L). Despite the initial fuel mass difference, the continuous combustion time in T2 was longer than in T1. This is because the HRR in T2 is higher, generating more heat and requiring more water mist for heat absorption and cooling. Additionally, after 300 s, the fuel mass in the oil pan began to increase. This may be due to the larger oil pan area in T2, which caused more water mist droplets to fall into the oil pan. As shown in Figure 10c, during T3, a single oil pan fire burned for about 640 s under the impact of fine water mist. Compared with T1, the fire source in T3 was positioned less favorably, and the oil pan received a lower density of water mist. In consequence, the fire source in T3 burned longer than in T1 and consumed more fuel.

In general, the real-time HRR of the fire source during the fire process can be calculated using mass change data collected by the weighing system. The calculation of the actual HRR is shown in Equation (4):

$$\dot{Q}=\chi \dot{m}∆H_c$$

(4)

In which, χ represents the combustion efficiency, with a value of 0.75 (consistent with Equation (1)), $\dot{m}$ is the real-time mass loss rate (kg/s), and ΔH_c is the calorific value of combustion, which for diesel fuel is 43 MJ/kg (consistent with Equation (1)).

Following the fuel mass change curve and Equation (4), the actual HRR curve of the fire source during the fire test is displayed in Figure 10. As exhibited in Figure 10, all three fire tests passed through the initial development stage and reached the stable combustion stage. In T1 and T3, the fire source reached stable combustion at approximately 110 s after ignition, which was slightly earlier than the theoretical calculated value. In T2, the ignition source reached stable combustion at about 95 s, also earlier than the theoretical value. However, in the middle and later stages of the fire tests, the increasing disturbance of the fire source by the water mist, along with the rising volume of water entering the oil pan, caused the mass change recorded by the weighing system to no longer accurately represent the fuel mass change in the oil pan. For instance, the HRR $\dot{Q}$ of the fire source in T2 increased rapidly between 270 s and 300 s after ignition, but, as presented in Figure 9i,j, the fire intensity did not change during this period. This is likely a result of the rapid evaporation of water mist droplets entering the pan under the influence of the high-temperature flame, causing a brief surge in the mass loss rate $\dot{m}$. Additionally, when the mass measured by the weighing system increases, it is not possible to accurately calculate the real-time HRR $\dot{Q}$ of the fire source. Therefore, during the action of water mist, the real-time HRR cannot be calculated based on the mass loss rate, as this would yield inaccurate results.

It should be noted that the accuracy of HRR estimation using the fuel mass loss rate is affected by several factors during the mist suppression stage. Although the weighing system provides reliable readings in the early and quasi-steady combustion stages, the activation of the water mist system introduces external disturbances, such as the accumulation of water droplets in the oil pan and the rapid evaporation of water under high-temperature flames. These factors may lead to abnormal fluctuations or even increases in the measured mass, which do not correspond to actual fuel consumption. Consequently, the real-time HRR values calculated based on Equation (4) during this phase may deviate from the true combustion intensity.

In contrast, it is more accurate to estimate the HRR for each fire test with the average fuel mass loss rate $\overline{\dot{m}}$ during the steady combustion period (quasi-steady-state), and the average HRR $\overline{\dot{Q}}$, as calculated by Equation (5) [19,48]. This is because, during the stable combustion phase, the fuel mass loss occurred at a relatively stable rate. At this stage, the ignition source was minimally disturbed by the water mist, and the mass loss rate monitored by the weighing system can be approximated as the real fuel mass loss rate. The $\overline{\dot{Q}}$ values for the three fire tests, calculated using Equation (5), were 1.4 MW, 2.3 MW, and 1.4 MW, respectively. These values were significantly lower than the theoretical value of Q obtained from Equation (1). This discrepancy can be attributed to the fact that the fire source did not reach its maximum intensity before the water mist fire protection system was activated, and the water mist effectively inhibited the further growth of the fire after its activation.

$$\overline{\dot{Q}}=\chi \overline{\dot{m}}∆H$$

(5)

It is worth noting that under water mist suppression, the estimation of HRR based solely on mass loss may be significantly biased because of the cooling and smothering effects that inhibit combustion and promote incomplete burning. Therefore, in real fire scenarios involving water mist systems, engineers should consider using a multi-indicator approach to improve the accuracy of HRR estimation. In addition to mass loss, temperature rise profiles (especially ceiling temperature trends), gas concentration measurements (e.g., O₂ and CO), and flame radiation characteristics can provide more comprehensive insights into the combustion intensity. When feasible, applying calorimetric methods such as the oxygen consumption principle (widely regarded as reliable for HRR measurement) can offer a more robust estimation, especially in controlled or instrumented tunnel environments. This integrated strategy helps account for the complex fire dynamics under active suppression and enhances the reliability of fire size assessment in engineering applications.

3.3. Tunnel Ceiling Temperature

3.3.1. Spatiotemporal Distribution of Temperature Rise Under the Tunnel Ceiling with Water Mist

Figure 11 displays the spatiotemporal distribution of temperature rise near the tunnel ceiling during the three fire tests. As depicted in Figure 11a, the tunnel ceiling temperature in T1 initially rose, then decreased. The installation range of the WMFFS in T1 spans from −14.5 m to 14.5 m, with the fire source center at 0 m. Within this range, the sections from −14.5 m to −10.5 m and from 10.5 m to 14.5 m are water walls formed by water mist sprays and curtains designed for smoke and heat insulation. The fine water droplets demonstrate significant attenuation of radiative heat originating from combustion processes, and a distinct temperature dividing line is observed in Figure 11a. Water mist activation occurred 60 s after ignition. Between 60 s and 100 s, no clear temperature dividing line was observed because the fire was still in its early stages of development, and the tunnel ceiling temperature remained relatively low. After 100 s, a clear temperature dividing line became visible as the fire entered the stable combustion phase, and significant smoke accumulated beneath the ceiling. Under the influence of the WMFFS, the ceiling temperature within the water wall initially rose, then fluctuated and decreased, with the temperature distribution decreasing along the longitudinal direction. During the period from 100 s to 300 s after ignition, the ceiling temperature in the water wall remained high, as the fire source was undergoing intense combustion at this time, as displayed in Figure 8d–h. Additionally, between 100 s and 200 s, the fire source was in the stable combustion phase, with a high HRR. After 300 s of ignition, the impact of the fine water mist on the flame increased, with more droplets entering the oil pan and the burning area of the flame gradually shrinking. As a result, the ceiling temperature within the water wall began to decrease after 300 s. However, as shown in Figure 8m,p, as the flame was suppressed by the water mist, it no longer burned vertically upward. This resulted in brief high temperatures above the oil pan at approximately 400 s and 500 s. On the other hand, the ceiling temperature outside the water wall remained below 55 °C, and after 300 s, the temperature dropped further, falling below 40 °C.

Similarly, as exhibited in Figure 11b, the ceiling temperature in T2 also rose initially and then decreased, but the peak temperature at the ceiling in T2 was higher. The installation range of the WMFFS in T2 was the same as in T1, and a clear temperature cut-off line was observed in Figure 11b. For T2, the ceiling temperature within the water wall remained higher between 100 s and 300 s after ignition, as the HRR of the fire source was high, and the flame was in an intense combustion state, as displayed in Figure 9d–h. As the water mist continued to suppress the fire source, the flames from the two oil pans no longer merged, and the two fires burned separately, as seen in Figure 9k–m. After 300 s, as shown in Figure 11b, two new peaks appeared in the ceiling temperature distribution. Later, with continued water mist application, the ceiling temperature inside the water wall gradually decreased. On the other hand, the ceiling temperature outside the water wall remained below 60 °C, and after 300 s, the ceiling temperature outside the water wall fell below 45 °C.

In contrast, due to the change in the fire source location in T3, the installation range of the WMFFS in T3 was adjusted to align with the fire source, spanning from −23.5 m to 5.5 m, with the fire source center positioned at 0 m. The sections from −23.5 m to −19.5 m and from 1.5 m to 5.5 m functioned as water walls for smoke and heat insulation. A distinct temperature demarcation line was observed at the water wall. During the period from 100 s to 300 s after ignition, the ceiling temperature inside the water wall was higher, corresponding to the stable combustion phase of the fire source and its increased heat release rate. However, after 300 s, the ceiling temperature inside the water wall showed a sharp rise. This increase was likely due to the location of the oil pan in an area with low water mist density, where the suppression effect of the fine water mist was insufficient. Fortunately, as the fine water mist system continued to operate, the ceiling temperature within the water wall began to decrease and stabilize. On the other hand, a notable difference was observed in the ceiling temperatures outside the water wall, depending on their proximity to the fire source. The ceiling temperature outside the water wall, farther from the fire source, remained consistently below 35 °C. In contrast, the ceiling temperature outside the water wall closer to the fire source was higher, ranging between 50 °C and 55 °C during the period from 100 s to 450 s. After 450 s, this temperature gradually decreased to below 45 °C.

3.3.2. Longitudinal Temperature Distribution of the Tunnel Ceiling

Figure 12 demonstrates the distribution of average ceiling temperature at various longitudinal positions relative to the fire source. The temperatures shown represent the average ceiling temperature over the 300 s following ignition (i.e., from 0 s to 300 s). For this analysis, the average ceiling temperature at each longitudinal position in the tunnel was calculated separately, as depicted in Figure 12. For both T1 and T2, thermal patterns on either side of the fire source showed a mirrored distribution relative to the longitudinal axis. Within the water wall, as the distance from the fire source increased, the average ceiling temperature diminished. Outside the water wall, the average ceiling temperature at the first measurement point was notably lower, and it continued to decrease gradually as the distance to the fire source increased. Additionally, a higher HRR from the fire source corresponded to a higher average ceiling temperature. In T3, the average ceiling temperature on either side of the fire source was not symmetrical because of the thermal insulation effect of the water mist fire protection system. However, a similar trend was noticed in which the average ceiling temperature within the water wall declined as the distance to the fire source rose, and the same applied to the average ceiling temperature outside the water wall. Notably, the farther the water wall was located from the fire source, the more effective its insulation became, and the cooling effect of the water mist inside the water wall was more pronounced in reducing ceiling temperatures.

In the development of tunnel fire experiments in this research, the fire plume generated by the oil pan fire underwent a constant temperature decay process as it diffused longitudinally within the area affected by the water wall. This temperature attenuation was caused not only by the evaporation of water mist absorbing heat from the smoke layer but also by continuous heat exchange between the smoke layer and the tunnel wall. Prior studies have systematically investigated the spatial evolution of thermal plumes along road tunnels during fire incidents. Hu et al. [49] derived the ceiling temperature distribution of tunnel fire smoke, showing exponential attenuation along the longitudinal direction, based on full-scale experimental studies and theoretical methods. Delihatsios [50] demonstrated that the ceiling temperature in a corridor also decreases exponentially along the longitudinal direction. Based on the experimental results shown in Figure 12, the ceiling temperature distribution is associated with the distance from the fire source. Therefore, this study proposes using Equation (6) to describe the relationship between ceiling temperature and distance:

$$∆T=A{e}^{-Bx}$$

(6)

where ΔT represents the divergence between the ceiling temperature and the ambient temperature, x denotes the distance to the fire source, and A and B are constants.

According to the fire plume theory [51,52], the temperature decay behavior of the tunnel ceiling can be better described using the dimensionless expression provided in Equation (7).

$${\displaystyle \frac{∆T}{∆T_{max}}}=A{e}^{-B{\displaystyle \frac{x-x_0}{H_c}}}$$

(7)

where ΔT_max represents the difference between the maximum ceiling temperature and the ambient temperature, x − x₀ denotes the distance to the fire source, and H_c is the clear height of the tunnel.

Figure 13 depicts the non-dimensional scaling of mean ceiling temperature as a function of spatial separation from the combustion zone. Clearly, within the range of the water wall for all three fire tests, the average ceiling temperature demonstrates a strong exponential correlation with the distance from the fire source. For T1 and T2, where the ignition source center is located at the same position, this relationship is described by Equation (8). For T3, the relationship is represented by Equation (9). It should be noted that the exponential decay relationship observed between the ceiling temperature and the distance from the fire source is derived from a limited number of full-scale fire tests (three scenarios). While the consistency of this pattern across different fire conditions in the present study provides preliminary support for its validity, a larger set of repeated tests under varied conditions would be required to statistically confirm the repeatability and robustness of this relationship. Therefore, the exponential decay correlation proposed here should be interpreted as an empirical observation specific to the tested configurations rather than a universally generalized law. Future work will focus on expanding the dataset and performing uncertainty quantification to validate this trend further.

$${\displaystyle \frac{∆T}{∆T_{max}}}=0.96e^{-0.56{\displaystyle \frac{x-x_0}{H_c}}}$$

(8)

$${\displaystyle \frac{∆T}{∆T_{max}}}=0.94e^{-0.79{\displaystyle \frac{x-x_0}{H_c}}}$$

(9)

According to Figure 13 and Equation (8), within the scope of the WMFFS, when the fire source location stays unchanged, the temperature distribution of the tunnel ceiling caused by oil pan fires with varying power can be characterized based on the distance to the fire source. Outside the coverage of the WMFFS, the ceiling temperature of the tunnel was generally low because of the thermal insulation effect of the water wall. The ceiling temperature outside the water wall decreased in a stepwise manner with a growing distance from the water wall. For T3, the change in the fire source location increased the coverage spectrum of the WMFFS on the smoke layer on one fire source side. As described by Equation (9), the exponential decay of the ceiling temperature in T3 along the longitudinal direction was faster compared with T1 and T2. Additionally, the ceiling temperature outside the water wall remained very low. This indicated that as the range of the WMFFS increased, the cooling effect of the water mist on the ceiling and the thermal insulation effect of the water wall on the fire was enhanced.

3.4. Temperature Distribution Within the Tunnel

Figure 14, Figure 15, and Figure 16 demonstrate the progressive spatial variation of heat dispersion patterns during distinct fire development phases in T1, T2, and T3 tests, respectively. These temperature distribution maps, drawn from data collected by 67 thermocouple probes, provided a clear visualization of the cooling and thermal insulation effects of the WMFFS. In tunnel fires, the temperature propagation mechanism involves multiple physical processes, including heat convection, heat conduction, heat radiation, and smoke diffusion. Consequently, the temperature distribution within the tunnel space served as an indicator of the fire smoke flow dynamics to some extent.

As shown in Figure 14, in T1, the WMFFS demonstrates effective thermal insulation. 60 s after ignition, the temperature in the upper space near the fire source began to rise gradually, as illustrated in Figure 14a. At this point, the WMFFS has just been activated. By 90 s, the temperature within the tunnel continued to rise. The temperature increased both inside the water mist zone and outside the water wall, as presented in Figure 14b. At this stage, the cooling and thermal insulation effects of the system were not yet significant, likely because of the response time required after activation. After 120 s, the thermal insulation effect became evident, with a noticeable temperature difference at the water wall, as displayed in Figure 14c–e. During this period, the fire source reached a stable combustion phase, resulting in a steady thermal stratification within the tunnel. Within the range of the water mist system, the temperature rose gradually, peaking at 189 s. Meanwhile, the temperature outside the water wall began to decrease. As shown in Figure 14f, at 240 s, the continued operation of the water mist system weakened the fire source, resulting in a temperature decrease within the range of WMFFS. After 300 s, the temperature distribution in the active zone of WMFFS changed, and the temperature peak was no longer directly above the oil pan. As shown in Figure 14g–k, during this stage, the water mist increasingly disrupted the fire source. While there were no significant changes in the temperature distribution within the range of WMFFS, the overall trend showed fluctuations followed by a gradual decline. By 589 s, the fire source was extinguished. Residual low temperatures remained within the range of WMFFS, as shown in Figure 14l. Outside the water wall, residual heat persisted at the tunnel ceiling, likely due to lingering fire fumes.

Similarly, as exhibited in Figure 15a,b, in T2, the temperature within the tunnel rose rapidly after the fire source was ignited. In the initial stage of activation, WMFFS did not exhibit a significant thermal insulation effect. Compared with T1, the tunnel temperature within the range of the water mist system was higher, and the peak temperature (171 s) was reached earlier, as illustrated in Figure 15c,d. After the peak temperature, the temperature outside the water wall remained stable for a period rather than dropping significantly, as shown in Figure 15e–h. This suggests that the thermal insulation capacity of the WMFFS has limitations. Additionally, it indicates that the system may have a restricted capacity for smoke management. When the fire source entered a phase of significant attenuation, the temperature outside the tunnel began to decrease markedly, as depicted in Figure 15i–k. Furthermore, as shown in Figure 15l–o, the fire source in T2 continued to burn for a longer period after the decay phase compared with T1. This prolonged burning required more water for continuous cooling and insulation.

As shown in Figure 16, similar to T1, the WMFFS in T3 did not exhibit significant cooling or thermal insulation effects during the initial stage of activation. Nevertheless, with the continuous operation of the water mist system, the temperature within its range began to decrease after reaching a peak at 166 s. For the tunnel space outside the range of the water mist system, a distinction in temperature distribution was observed between the left side (away from the fire source) and the right side (near the fire source) of the water wall. The temperature on the left side, away from the fire source, gradually decreased over time, while the temperature on the right side, near the fire source, remained higher for a period before eventually decreasing. Combined with Figure 11c, the performance of the system under low mist density conditions in T3 can be summarized that reduced mist coverage led to slower temperature decay.

3.5. Smoke Conditions Inside the Tunnel

Figure 17, Figure 18 and Figure 19 illustrate the smoke conditions outside the range of WMFFS, upstream of the fire source in the tunnel, as recorded by a high-definition camera during the three fire tests. As exhibited in Figure 17a, fire smoke spread rapidly along the tunnel ceiling toward both ends after ignition. During the intense combustion phase of the fire source, the WMFFS acted as a barrier to the smoke, as seen in Figure 17b,c. However, with the rapid accumulation of smoke in the sprinkler section, the smoke-handling capacity of the water mist system appeared to reach its limit. Gradually, smoke began to overflow along the tunnel ceiling, as exhibited in Figure 17d–f. Notably, the spilled smoke transitioned from black to white after passing through the water mist system. This may occur because the water mist droplets capture and adsorb carbon particles and other suspended solids, reducing the concentration of black carbon in the smoke and thereby lightening its color. This phenomenon can improve visibility in the tunnel during a fire. When the power of the fire source increases, more smoke is generated by combustion. Consequently, the amount of smoke overflowing from the water mist system also increased, as illustrated in Figure 18.

Notably, the spilled smoke transitions from black to white after passing through the water mist system, as observed in all three tests. This phenomenon is presumed to occur due to the interaction between water mist droplets and soot particles. The sub-millimeter droplets generated by the high-pressure nozzles can effectively capture and agglomerate fine carbonaceous particles through mechanisms such as inertial impaction, Brownian diffusion, and interception. As a result, the concentration of black soot is reduced, and the optical characteristics of the smoke shift toward lighter, whiter hues dominated by water vapor and condensed aerosols. Similar effects have been documented in laboratory-scale studies where laser-based diagnostics revealed a significant decrease in soot volume fraction after water mist application [53]. Although such microscopic data were not collected in the present full-scale tunnel tests because of sensor limitations under high-temperature and humid conditions, these mechanisms provide a reasonable explanation for the observed smoke-whitening effect. Future research will aim to include real-time particle sizing and laser extinction measurements to validate this inference further.

A more pronounced smoke-blocking effect and smoke overflow are observed in Figure 18. As shown in Figure 19b, the smoke was initially trapped by the WMFFS. However, as the amount of smoke increased, it gradually overflowed. The spilled smoke appeared to sink near the water wall, which may be due to a decrease in the smoke temperature after interacting with the water mist. This cooling effect weakened the thermal buoyancy force, causing the smoke to sink. Additionally, it could be that the downward driving force of the water mist contributes to the settling of the smoke.

To more intuitively observe the amount of spilled smoke, the smoke downstream of the fire source was recorded at the tunnel exit after the fire source was extinguished in each fire test, as shown in Figure 20. Less smoke was observed in T1 and T3, with minimal impact on brightness and visibility within the tunnel. In contrast, the amount of smoke spilled out in T2 was more due to the larger scale of the fire source. This suggests that the smoke-carrying capacity of WMFFS has an upper limit over a certain length. Fortunately, the dilution and scrubbing of the spilled smoke within the tunnel did not distinctly influence its brightness or visibility.

4. Discussion

Based on the changes in the fire source and HRR observed in Figure 8, Figure 9 and Figure 10, the timely activation of WMFFS can effectively suppress fire development. Through mechanisms such as heat competition, chemical reaction interference, turbulent disturbance, and fuel supply blockage, the water mist disrupts the thermal and chemical conditions required to sustain combustion, thereby weakening the fire and inhibiting its progression. When the burning area of the fire source is larger, the water mist exhibits a stronger suppression effect. Similarly, a higher water mist density enhances the fire suppression performance of WMFFS. From the temperature variations at the tunnel ceiling shown in Figure 11, Figure 12 and Figure 13, it can be seen that the water mist absorbs heat through continuous evaporation. This results in only a brief period of high temperatures above the fire source at the tunnel ceiling. Within the coverage area of WMFFS, the average ceiling temperature declines exponentially along the longitudinal direction during both the ignition and stable combustion phases. The greater the system’s coverage of the smoke layer, the faster the ceiling temperature attenuation along the tunnel’s longitudinal axis. From the thermal and smoke insulation effects depicted in Figure 14, Figure 15, Figure 16, Figure 17, Figure 18, Figure 19 and Figure 20, the water mist system isolates the thermal in the fire zone effectively. However, waste heat outside the overage of WMFFS primarily originates from escaping smoke. A limitation of the system is its finite capacity to manage smoke within a specific range. As the fire continues to burn, smoke gradually accumulates beneath the ceiling and then escapes beyond the coverage of WMFFS. Fortunately, the escaping smoke is whitened because of the dilution and scrubbing effect of the water mist, which reduces its impact on brightness and visibility within the tunnel. It is important to note that, despite the water mist system being activated 60 s into the test, there was a noticeable “lag” in its fire suppression and heat and smoke insulation effects. This delay is attributed to the time required for water to travel through the pipes and the initial loss of mist due to interaction with the smoke layer.

In general, the WMFFS proposed in this paper demonstrates effective fire suppression, heat insulation, and smoke insulation capabilities. In anticipation of a tunnel fire, the water mist acts directly on the burning vehicle, adjacent vehicles, and nearby cargo. Owing to the small particle size and large total surface area of water mist particles, it exhibits excellent heat absorption, which helps inhibit fire development, reduce the power of the fire source, and lower temperature and radiation intensity. Additionally, the water mist particles have a wide coverage area and high spray velocity, enabling them to wet adjacent vehicles and goods quickly. This prevents fire spread and minimizes the risk of escalation. The shielding water curtain further acts as a barrier, limiting the spread of smoke and fire, thereby reducing the hazards posed by both and creating favorable conditions for personnel evacuation and escape. The combined effects of the water mist and the shielding water curtain effectively suppress fire progression, prevent smoke spread, eliminate smoke particles, improve visibility within the tunnel, and reduce smoke temperature. These factors together provide a safer environment for evacuating trapped individuals and afford rescuers more time to reach the scene.

However, for practical tunnel applications, it is important to consider implementing a longer water mist section with a wider coverage area to achieve better heat and smoke insulation. Additionally, this study focused solely on a fire scenario involving diesel pool combustion, while various other types of fires can occur in real-life situations. The effectiveness of the water mist system against different types of fires in the tunnel environment remains uncertain. In particular, the rapid development and increasing use of new energy vehicles equipped with lithium-ion batteries pose new challenges for tunnel fire safety. Given the instability of lithium-ion batteries, it is crucial to explore the potential of using water mist systems to suppress the thermal runaway process in new energy vehicles in future research.

5. Conclusions

In this paper, a series of full-scale fire tests were carried out to evaluate the performance of the proposed new WMFFS under various fire scenarios. The system’s effectiveness in fire suppression, heat insulation, and smoke insulation was analyzed by examining the temporal and spatial variations of key parameters, including the fire source, HRR, ceiling temperature, temperature distribution within the tunnel space, and smoke behavior during the fire tests under the influence of the water mist. The main conclusions are summarized as follows:

(1) The WMFFS, when activated promptly, can effectively suppress the fire and inhibit its progression. The suppression effect becomes more pronounced as the burning area of the fire source increases. For instance, under the influence of the water mist, fire sources with designed HRRs of 3.2 MW and 6.4 MW were reduced to 1.4 MW and 2.3 MW, respectively. Furthermore, when the spray density of the water mist is higher, the suppression rate of the water mist system on the fire is greater.

(2) Under the action of WMFFS, high temperatures occur briefly only above the fire source on the tunnel ceiling. Within the water mist section, the average ceiling temperature declines exponentially along the longitudinal direction after ignition and during stable combustion. A longer water mist section, which increases the system’s range of action on the smoke layer, results in a faster decay of the average ceiling temperature along the longitudinal direction.

(3) At the initial stage of activation, the cooling and thermal insulation effects of WMFFS on the fire area are less significant. This “lag” is primarily caused by the time required to transport water through the pipes and the loss of freshly sprayed mist due to obstruction by the smoke layer. Once the water mist system stabilizes, most of the high temperature generated by the fire is contained within the fire section protected by the water mist. Beyond this section, residual heat from spilled smoke is generally confined to areas near the tunnel ceiling.

(4) The smoke-carrying capacity of a given length of the water mist section has an upper limit. As the fire continues to burn, smoke eventually spills out of the protected section and mixes with the water mist. However, the fine water mist dilutes and cleanses the smoke, changing it from black to white, which mitigates its impact on visibility and brightness within the tunnel.

(5) In future work, additional tests involving other types of fuels will be conducted to further evaluate the adaptability and effectiveness of the proposed water mist fire extinguishing system under various fire scenarios. Comparative studies with other fire suppression systems—considering differences in operating pressure, nozzle configurations, and spray characteristics—will also be carried out to better assess relative performance. Furthermore, potential design adaptations will be explored to enhance the system’s coverage area and ensure effective protection in larger or more complex tunnel geometries.