Assessing Human Exposure to Fire Smoke in Underground Spaces: Challenges and Prospects for Protective Technologies

Abstract

Urban underground spaces, including tunnels, subways, and underground commercial buildings, have grown quickly as urbanization has progressed. Fires frequently break out following industrial accidents and multi-hazard natural disasters, and they can severely damage human health. Fire smoke is a major contributor and a major hazard to public safety. The flow patterns of fire smoke in underground spaces, the risks to human casualties, and engineering and personal protective technologies are all thoroughly reviewed in this work. First, it analyzes the diffusion characteristics of fire smoke in underground spaces and summarizes the coupling effects between human behavior and smoke spread. Then, it examines the risks of casualties caused by toxic gases, particulate matter, and thermal effects in fire smoke from both macroscopic case studies and microscopic toxicological viewpoints. It summarizes engineering protection strategies, such as optimizing ventilation systems, intelligent monitoring and early warning systems, and advances in the application of new materials in personal respiratory protective equipment. Future studies should concentrate on interdisciplinary collaboration, creating more precise models of the interactions between people and fire smoke and putting life-cycle management of underground fires into practice. This review aims to provide theoretical and technical support for improving human safety in urban underground space fires, thereby promoting sustainable urban development.

1. Introduction

With the acceleration of urbanization, underground spaces have been increasingly developed and utilized as an important means to alleviate land resource constraints and expand urban functions, including subways, underground shopping malls, parking garages, tunnels, and utility tunnels. Underground spaces have become an indispensable component supporting sustainable, high-quality, and stable urban development [1]. However, in the aftermath of Multi-Hazard Natural Disasters and Industrial Accidents, fires often occur and may cause severe harm to human health. The unique environment of underground spaces also presents significant fire safety challenges [2]. Unlike above-ground buildings, underground spaces—due to their inherent enclosed nature, high occupant density, limited evacuation routes, and difficulties in rescue operations—pose far greater dangers and potential catastrophic consequences in the event of a fire. In recent years, the global number of fires and associated casualties has remained high [3].

In 1987, a fire on an escalator at King’s Cross St. Pancras station on the London Underground in the UK resulted in 31 fatalities. In 2000, a fire at the Dongdu Commercial Building in Luoyang, China, caused 309 deaths due to poisoning and suffocation, with 7 others injured. In 2022, a fire in an underground cold storage facility at the Xinchangxing Market in Dalian led to nine deaths. In 2024, a fire occurred on the underground floor of a street-level shop in Xinyu, resulting in 39 fatalities and 9 injuries. From January to October 2024, a total of 770,000 fires were reported in China, causing 1559 deaths and 2202 injuries, all showing an increase compared to the previous year.

Table 1. Some underground-space fire accidents.

Time	Location	Reason	Consequence
18 November 1987	King’s Cross St. Pancras Station, London Underground, UK	Escalator fire	31 fatalities
25 December 2000	Dongdu Commercial Building, Luoyang, China	Illegal hot work (welding) operations	309 fatalities, 7 injured
18 February 2003	Jungangno Station, Daegu Metro, South Korea	Arson	198 fatalities, 298 missing
5 June 2013	Moscow Metro, Russia	Cable fire	66 injured
12 January 2015	L’Enfant Plaza Station, Washington D.C. Metro, USA	Train carriage fire	1 fatality, multiple injuries
26 January 2016	Ginza Station, Tokyo Metro, Japan	Combustion of unknown substance at station ventilation shaft	68,000 passengers affected
10 February 2017	Tsim Sha Tsui Station, Hong Kong, China	Arson	18 injured
27 March 2020	110th Street Station, New York City Subway, USA	Train carriage fire	1 fatality, 16 injured
2 January 2022	Xinchangxing Market, Dalian, China	Unauthorized welding ignited polyurethane foam in underground cold storage during construction	8 fatalities, 5 injured, 1 firefighter fatality
23 January 2023	Barranca del Muerto Station, Mexico City Metro, Mexico	Electrical short circuit	26 cases of respiratory discomfort
24 January 2024	Jialeyuan Street Shops, Xinyu, China	Violation of hot work regulations during cold storage construction	39 fatalities, 9 injured

summarizes some representative fire incidents in underground spaces and their severe consequences.

In underground spaces, there are multiple hazards from fire smoke. First, smoke obscures light, significantly reducing visibility and making it difficult for occupants to identify correct escape routes, thereby sharply decreasing evacuation efficiency and potentially triggering panic [4]. Second, smoke contains a large amount of toxic and harmful gases, directly threatening human life and safety. Finally, the heat carried by the smoke can cause burns or respiratory tract scalds. Thermal radiation from flames, hypoxia, and fire smoke are the three primary sources of fire hazards, with respiratory system damage caused by inhaling fire smoke being a leading cause of fatalities [5]. Statistics from the U.S. National Fire Protection Association indicate that deaths due to inhalation of fire smoke account for 60–75% of all fire-related fatalities [6]. When individuals inhale fire smoke particles through the nasal and oral cavities, these particles deposit on respiratory surfaces, enter the lungs, and may even enter the bloodstream, leading to increasing incidence and mortality rates of respiratory diseases, cardiovascular diseases, and lung cancer. Therefore, studying the exposure interaction characteristics between occupants and smoke in underground fire scenarios is of great practical importance. The findings can contribute to developing engineering and personal protection technologies. This work is crucial for ensuring underground space safety and formulating emergency plans. After a fire occurs in an underground space, most occupants will choose to evacuate. During evacuation in such environments, people need to overcome gravity to move upward, which is generally more complex and strenuous than evacuating downward from a high-rise building [7,8]. Furthermore, extensive human movement occurs during escape, which can influence the interaction characteristics between fire smoke and the immediate smoke environment around individuals, as well as alter the exposure levels and injury risks associated with inhaled fire smoke.

This paper aims to provide a comprehensive review of research progress in this field, with the main content covering the smoke movement patterns in underground spaces, the interaction between smoke and dynamic occupants, exposure characteristics and injury mechanisms of the human respiratory system, and advances in engineering and individual protection technologies for underground space fire scenarios. It will elaborate on the exposure risks and injury mechanisms of smoke to humans through both macro-level typical fire case analysis and micro-level research on respiratory system injury mechanisms. This review adopts a structured narrative approach to synthesize the diverse and conceptually broad literature in the field. The literature search was performed using core academic databases, including Web of Science, Scopus, and Google Scholar, with keywords such as “underground fire smoke,” “human exposure risk,” “evacuation simulation,” and “smoke control”. The selection of publications was guided by explicit criteria, prioritizing seminal works, high-impact studies, and recent advancements pertinent to the core themes of smoke dynamics, human risk assessment, and protective technologies for fire smoke exposure in urban underground spaces. It will discuss the current state of smoke protection technologies in underground space fires, including advanced engineering protection systems and novel personal protective equipment, and provide an outlook on future development trends. It is to offer robust theoretical support and technical references for the safety design, emergency management, and technological innovation of underground spaces.

2. Smoke Diffusion Patterns in Underground Space Fires

In underground space fires, combustion characteristics and smoke movement patterns differ significantly from those in above-ground environments. Moreover, underground spaces are densely populated, and in the event of a fire, the evacuation path for personnel is long and may intersect with smoke diffusion routes. Complex smoke diffusion patterns will further threaten personnel safety. This is also the reason why most fatal fire accidents are caused by smoke. Therefore, the process of smoke spread in underground space fires deserves special attention.

Smoke spread is influenced by the complex coupling of spatial structure, ventilation systems, and dynamic human behavior [9,10,11]. The enclosed nature of underground spaces is a critical factor in the rapid spread of fire smoke. After a fire ignites, high-temperature smoke, due to its lower density, rises rapidly and forms a stable smoke layer beneath the ceiling or roof, spreading quickly horizontally. When smoke encounters vertical channels such as stairwells or shafts, it accelerates its spread to upper levels due to the “stack effect,” posing a lethal threat to occupants on higher floors. Over the past few decades, researchers domestically and internationally have conducted a series of studies focusing on typical underground space scenarios, providing important references for establishing more accurate models of fire smoke diffusion [12]. The King’s Cross St. Pancras station fire on the London Underground serves as a typical case study. In 1987, a fire originating at the base of a wooden escalator was channeled and accelerated within the enclosed escalator incline, creating a distinct “trench effect.” This ultimately led to a flashover, projecting flames and dense smoke into the ticket hall above, resulting in 31 fatalities. This incident highlights the amplifying effect that narrow, elongated passages in underground spaces can have on smoke spread.

Focusing on representative scenarios such as subways and tunnels, scholars have conducted research on the smoke diffusion characteristics of fires within these environments. Key aspects investigated include temperature distribution beneath the ceiling, smoke stratification behavior, and smoke overflow patterns. Regarding maximum temperature, previous researchers revised classical predictive models for maximum smoke temperature, as shown in

Table 2. Maximum temperature models under different fire scenarios.

Previous Studies	Maximum Temperature Models	Fire Scene Characteristics
Alpert [13]	$∆T_{max}=16.9{\displaystyle \frac{Q^{2/3}}{H^{5/3}}}$	Ceiling without boundary
Ji et al. [30]	$∆T_{max}=17.9{\displaystyle \frac{Q^{2/3}}{H^{5/3}}}$	Restriction of side wall
Yao et al. [31]	$∆T_{max}=22.7{\displaystyle \frac{Q^{2/3}}{H^{5/3}}}$	Enclosed channel
Tang et al. [32]	$\Delta T_{max}=\left\{\begin{array}{c}\left(1.37-0.37{\displaystyle \frac{D}{\left(\frac{w}{2}\right)}}\right){\displaystyle \frac{\dot{Q}}{u{r}^{1/3}{H}_d^{5/3}}},u^{\prime }>0.19\\ 17.5\left(1.33-0.33{\displaystyle \frac{D}{\left(\frac{w}{2}\right)}}\right){\displaystyle \frac{{\dot{Q}}^{2/3}}{H_d^{5/3}}},u^{\prime }\le 0.19\end{array}\right.$	Effect of transverse fire position
Li et al. [33]	$\Delta T_{max}=\left\{\begin{array}{c}{\displaystyle \frac{\dot{Q}}{V{r}^{1/3}{H}_d^{5/3}}},V^{\prime }>0.19\\ 17.5{\displaystyle \frac{{\dot{Q}}^{2/3}}{H_d^{5/3}}},V^{\prime }\le 0.19\end{array}\right.$	Effect of wind speed

. By comparing and analyzing these models, it can be concluded that the maximum smoke temperature is directly related to the fire source power and the ceiling height, but its correlation coefficient is influenced by factors such as sidewall limitations, opening conditions, fire source location, and wind speed conditions. Therefore, for specific fire scenarios, the prediction model for maximum smoke temperature can be improved by incorporating key considerations based on Alpert’s modeling approach [13]. Furthermore, the influence of factors such as structural dimensions, fire source location, top vents, and blockage conditions has been analyzed, leading to further refinements of temperature prediction models [14,15,16,17]. In terms of temperature decay, Delichatsios [18] proposed a model for the temperature distribution of smoke flow along ceiling beams. He et al. [19] developed an exponential decay model for smoke temperature in corridor fire scenarios. Hu et al. [20] and Fan et al. [21] conducted theoretical analyses of fire smoke spread in long and narrow spaces, establishing exponential temperature distribution functions and lateral temperature distribution models based on full-scale experimental data. The influence of the slope of the main tunnels and connecting tunnels on longitudinal temperature distribution has been studied [22]. The effects of blockages and slope on longitudinal temperature distribution have also been analyzed [23]. Scholars have additionally considered the impact of structural materials, smoke barriers, ambient pressure, and ventilation and smoke extraction on temperature distribution [24,25,26,27,28]. It is noted that the structures involved in the aforementioned studies on fire smoke temperature distribution are primarily narrow spaces, where the temperature along the longitudinal centerline can be considered representative. However, Long et al. [29] found that in underground spaces with relatively small aspect ratios, such as double-island subway stations, the smoke temperature distribution beneath the ceiling cannot be simply represented by the temperature along the central longitudinal axis. Similarly, it is necessary to consider more refined scenarios, describe temperature distribution based on a universal model, and modify the model by obtaining a large amount of experimental and testing data to allow for its practical applicability.

The stratification state of fire smoke is a crucial indicator for assessing fire safety [34,35,36]. Smoke stratification characteristics are governed by the combined effects of thermal buoyancy and inertial forces, so the Froude number (Fr) and Richardson number (Ri) are frequently used to analyze the stability of smoke stratification in tunnels and other underground space fires [37,38]. A “stratification velocity” theory and critical Froude number have been established, and the coupling effect between longitudinal airflow and upstream blockage near the fire source has been analyzed [39]. Research indicates that obstacles can disrupt smoke stratification, but a stable smoke layer can still be maintained under appropriate longitudinal ventilation velocities. Numerous other factors significantly influencing thermal stratification have also been considered, including the vent location, ceiling extraction, outdoor wind, and water sprays [40,41,42,43,44]. Calculation methods for smoke layer height under different structural scenarios have been proposed. These include a semi-empirical model for estimating smoke layer thickness under unconfined ceilings under steady-state conditions [45], and models accounting for the influence of tunnel height, tunnel length, and ventilation velocity on smoke layer height distribution [46,47,48,49]. Long et al. [50] proposed a calculation model for smoke layer height in subway station fires under natural ventilation conditions.

After understanding the smoke diffusion pattern in the space of the floor where the fire is located, it is necessary to further analyze the diffusion mode of smoke to the upper layer caused by the chimney effect. At this time, the smoke overflow is a significant phenomenon, which needs to be emphasized as the connection point for smoke transmission in the upper and lower spaces. For this issue, scholars have conducted analytical studies on smoke overflow characteristics under various scenarios [51,52]. In subway station fires, smoke primarily spreads upward through stairways and escalators, forming vertical opening overflow at stairwell openings. Reynolds [53] analyzed mass and energy changes between the upper and lower spaces connected by stairs using dimensionless parameters. Ergin-Ozkan et al. [54] investigated the influence of opening size on airflow within stairwells. Chen et al. [55,56] analyzed the structural characteristics of the airflow field at critical stairwell openings, explored the normal distribution characteristics of airflow velocity, and employed the virtual fire source method to predict critical states. They also characterized temperature distributions above and behind obstacles and in the vertical direction, quantifying the impact of obstacles on subway station fire safety design.

In summary, scholars have conducted extensive theoretical analyses, full-scale experiments, scaled model tests, and numerical simulations focusing on the fire smoke diffusion characteristics in typical underground spaces such as tunnels, subways, and commercial buildings, yielding substantial research outcomes. These studies, considering fire incidents in scenarios like subways and tunnels, have investigated the influence of factors including ventilation conditions, fire intensity, source location, and material properties on fire smoke diffusion patterns. They have established classical models for ceiling temperature distribution and introduced modifications to these models accounting for factors such as building structure, obstructions, slope, and ventilation conditions. The characteristics of smoke stratification and the smoke overflow patterns in different scenarios have also been analyzed. These research efforts collectively enrich the theoretical understanding of fire smoke diffusion in underground spaces, and provide a fundamental reference for accurately assessing human exposure and injury processes within fire smoke environments.

3. Human Exposure Risk Assessments in Underground Space Fires

3.1. Risk Factors for Human Casualties in Underground Space Fire Cases

Casualty statistics from underground fire accidents reveal the prevalence of smoke inhalation as a cause of death. Integrating macroscopic fire case analyses with microscopic toxicological studies enables a comprehensive understanding of the lethality of smoke and its associated risks to humans. For instance, the 1987 King’s Cross London Underground fire resulted in 31 fatalities, the 1995 Baku Metro fire caused 289 deaths, and the 2003 Daegu subway fire in South Korea led to 198 fatalities, many of whom succumbed to asphyxiation from inhaling toxic smoke. In these disasters, smoke was a common dominant factor in the mass casualties. Furthermore, post-disaster analyses indicate that, beyond the fire itself, a series of design and management deficiencies exacerbated the hazards posed by smoke. After the fire erupted, the power system automatically shut down, plunging the station into darkness as emergency lighting and exit signs failed to operate, leaving passengers to grope for escape routes in the dark. More critically, train doors failed to open due to the power outage, trapping numerous passengers inside the carriages. The station’s ventilation system also proved severely inadequate in capacity, unable to exhaust the massive amounts of thick smoke in a timely manner. This allowed toxic smoke to rapidly spread and even travel through ventilation ducts to connected underground shopping areas at the surface level. Consequently, many victims were not burned to death but were asphyxiated by toxic smoke after escaping the trains but failing to find exits. Additionally, the corners in pedestrian passages may further impede crowd evacuation efficiency [57]. In summary, these accidents exposed multiple systemic failures: malfunctioning emergency systems (power outage triggered by the fire caused critical emergency lighting and guidance systems to fail), obstructed evacuation (train doors failed to open, trapping passengers), and ventilation system defects (inadequate smoke exhaust capacity, and even facilitation of smoke spread to other areas). This section examines key methodological shifts, beginning with foundational computational models, progressing to techniques designed to handle data ambiguity, and culminating in dynamic frameworks capable of simulating real-time accident evolution. The chosen methods’ crucial contribution to the development of the field’s analytical capacities is examined.

Scholars have developed risk assessment models for fire evacuation scenarios to quantify these risks and consider various failure factors as comprehensively as possible [58]. Furthermore, by integrating computational fluid dynamics (CFD) simulation and the Dijkstra algorithm, these approaches quantify individuals’ inhalation exposure risk to toxic gases, thereby enabling evacuation route optimization. Related methods have been adapted and improved for application in fire smoke scenarios [59,60]. Yuan et al. [61] proposed a method integrating Event Tree Analysis (ETA), CFD simulation, and evacuation modeling for risk assessment in toxic gas leakage accidents, wherein dynamic evacuation is determined based on a cellular automata (CA) model. Their study analyzed the synergy between technical safety barriers and emergency evacuation, exploring its effectiveness in reducing fatality risks. Si et al. [62] investigated the diffusion patterns of large amounts of toxic and harmful smoke in a mine network during coal mine roadway fires. They found that the resistance of high-temperature airflow correlates with the ventilation resistance of the roadway under normal airflow conditions and is proportional to the airflow temperature. This finding provides a basis for selecting emergency evacuation zones and implementing safety measures.

As mentioned, underground spaces possess inherent limitations, such as inadequate natural lighting and poor ventilation, which can easily induce panic among evacuees during a fire incident [63]. Escape routes are typically confined to limited staircases and evacuation passages, and these narrow exits further increase the difficulty of escape. Moreover, underground spaces often accumulate substantial combustible materials. Once a fire occurs, flames can spread rapidly, while smoke spreads along the same direction as evacuating occupants. Compared to above-ground spaces and other types of underground environments, occupant evacuation and emergency rescue operations in underground spaces are considerably more challenging. Therefore, to prevent fire accidents in underground spaces and reduce casualty risks, it is essential to conduct a comprehensive risk analysis of fire-related casualties and quantify both the fire risks and their potential consequences.

The lack of detailed data and uncertainties are primary factors limiting the effectiveness and accuracy of traditional risk assessment methods. With the popularization and application of fuzzy set theory, traditional risk assessment methods have been significantly expanded and improved. Masalegooyan et al. [64] proposed an integrated framework using fuzzy fault tree analysis to assess the probability of major fire risks in landfills. Kong et al. [65] used ETA and fuzzy numbers to characterize the uncertainty regarding fire risk and analyzed potential fire scenarios caused by fire protection system failures. Fuzzy set theory has adapted traditional risk assessment methods to diverse risk assessment needs; however, the structure of these studies is static and incapable of deductive reasoning. Furthermore, many studies failed to represent conditional dependencies between variables. Bayesian Networks (BN) provide a tool to address the aforementioned issues [66,67,68]. Yuan et al. [69] proposed a fire risk assessment method for underground engineering, enabling comprehensive analysis of causes, occurrences, hazards, and other information. Chen et al. [70] established a dynamic risk assessment model for cotton warehouse fires through fuzzy comprehensive evaluation, case analysis, and BN. Wu et al. [71] studied the evolution of subway station fire incidents, finding that complex structures, high passenger flow, peak hours, and fires inside trains could lead to severe economic losses and casualties. These studies provide new perspectives for fire risk assessment in underground spaces and have driven continuous progress.

Recently, research on fire risk assessment for underground buildings has gradually increased, paying more attention to the safety and evacuation of occupants during a fire [72,73,74,75]. However, most studies lack in-depth analysis of fire risk factors from the perspective of actual fire accidents, and the related risk factors often struggle to be comprehensive and complete. Fire is a dynamic and complex process, and the assessment of casualty risk requires comprehensive consideration of the interactions among multiple links, such as fire ignition, emergency response, and occupant evacuation. Fu et al. [58] addressed the challenges faced by underground commercial buildings in fire safety management, constructing a fire casualty risk assessment model combining fuzzy FTA and BN, aiming to quantitatively analyze the dynamic risk of casualties caused by fires in underground commercial buildings. This model comprises accidental fire ignition, uncontrolled fire development, and failed evacuation. It can comprehensively identify risk factors, reflect the impact of different emergency actions on accident evolution, and aid in selecting the optimal emergency response plan. These models can guide the formulation of fire prevention and control strategies and emergency response efforts for underground spaces, providing an innovative tool for enhancing the safety of underground spaces and reducing fire accidents and casualties.

3.2. Interaction Characteristics Between Human Respiratory and Fire Smoke

Fire smoke is the primary factor causing casualties in underground space fires. From a microscopic perspective, fire smoke composition is complex, and its toxicity varies depending on the burning materials and combustion efficiency. Fire smoke primarily consists of a large volume of air heated by the flame and high-temperature vapors released from burning materials (hot air), toxic gases, and smoke, comprising decomposed and condensed unburned substances. Hence, its injury mechanisms can be primarily categorized into three types: thermal damage, toxic damage, and particle-induced obstruction damage (Figure 1). The main and most lethal toxic component in smoke is carbon monoxide (CO). Its affinity for hemoglobin molecules is far greater than that of oxygen, allowing it to rapidly displace oxygen, leading to acute carbon monoxide poisoning and tissue hypoxia. Furthermore, smoke contains irritant gases such as sulfur dioxide (SO₂) and nitrogen dioxide (NO₂), which can cause severe irritation to the eyes and respiratory tract. Particulate matter (PM) in fire smoke, especially fine particulate matter with a diameter of less than 2.5 μm (PM2.5), can easily bypass the upper respiratory tract and deposit deep within the lungs, even entering the bloodstream. PM2.5 may also adsorb carcinogens such as polycyclic aromatic hydrocarbons. Inhalation of significant amounts of PM2.5 can trigger systemic inflammation, exacerbate cardio-pulmonary diseases, and significantly increase mortality risk [5].

Investigating the injury mechanisms of heat and smoke particles on the human respiratory tract and analyzing the resultant damage are essential for accurate risk assessment and targeted protection research, representing a forefront challenge in urban public security [76]. Farahmand, Goodarzi, and Phuong [77,78,79] have conducted research on the injury mechanisms of the human respiratory tract exposed to fire environments. Their work has preliminarily elucidated fundamental relationships, such as the characteristics of damage in different regions of the human respiratory tract caused by inhaling fire smoke at varying temperatures, and the correlation between exposure duration and casualty risk. A consensus is emerging that by monitoring the flow velocity and deposition of particles in various regions of a simulated human respiratory system model under different respiratory flow rates, the impact of fire smoke on the respiratory system can be quantified, thereby revealing the injury mechanisms. The transport of two major fire smoke hazard components (heat and smoke particles) within the human respiratory tract has been investigated by Weng et al. [80,81,82,83]. They have obtained data on the temperature field and the distribution patterns of smoke particle deposition within the respiratory tract, and established models for heat transfer and smoke particle transport and deposition in fire smoke environments. Xu et al. [83] establish a set of three-dimensional realistic respiratory tract models considering gender, age, and occupation, providing a foundational resource for studying the transport laws of fire smoke in the human respiratory tract. Utilizing Particle Image Velocimetry (PIV) technology, they revealed the distribution patterns of airflow fields within the respiratory tract models, precisely measuring the characteristics of airflow movement in the tracheal region during both inhalation and exhalation states. Cui et al. and Li et al. have also employed CFD methods to solve for the flow field changes within the respiratory tract [84,85,86,87]. They analyzed the influence of breathing state, intensity, and pattern on the distribution of the flow field within the respiratory tract, providing foundational support based on transport laws of flow fields for accurately predicting fire smoke injury mechanisms and assessing the degree of damage. Xu et al. [80] established a respiratory tract heat transfer model that considers characteristics such as convective heat transfer, evaporative heat transfer, and the physiological response of human tissues. They determined the distribution patterns of the temperature field within the realistic human respiratory tract [81], systematically studying the mechanisms of thermal damage therein. Furthermore, they analyzed the influence of physical size and density of smoke particles in typical fire scenarios, along with breathing parameters, on their transport and deposition behavior, obtaining the deposition distribution patterns of smoke particles in local regions of the respiratory tract [82]. Then, they established precise assessment methods and procedures for evaluating thermal damage, particle-induced obstruction, and toxic damage in local regions of the respiratory tract. These included analyzing the risk levels of thermal burns, inhaled particle dosage, and the carcinogenic and non-carcinogenic risks from toxic particles in the human respiratory tract under fire smoke scenarios. These works provide methodological support for assessing the safety and health of trapped individuals and rescue personnel in fire smoke environments, offer foundational models for medical personnel in diagnosing and treating respiratory tract injuries, and provide a scientific basis for developing respiratory protective equipment and evaluating its performance in fires.

In addition to analyzing the transport patterns of smoke particles within the human respiratory tract, it is also crucial to consider the interaction characteristics between fire smoke and evacuating occupants in the environment. In underground space fire scenarios, the evacuation behavior of occupants and the spread of smoke form a complex, coupled process with both factors influencing each other [88,89]. Smoke first exerts direct physiological and psychological impacts on occupants. Dense smoke obscures visibility, preventing individuals from identifying correct escape routes and often causing them to blindly head towards seemingly brighter but potentially more dangerous areas due to “phototaxis.” Studies have shown that judgment and movement speed significantly decrease in smoke-filled environments, directly affecting evacuation efficiency. Furthermore, the dynamic evacuation behavior of occupants can, in turn, influence smoke movement and personal exposure. Activities such as walking generate wake flows [90,91]. These movement-induced airflows disturb the local flow field and enhance the dispersion of particulate matter [92]. Although these studies primarily focus on pollutants at ambient temperature, the underlying physical mechanisms they reveal are also applicable to fire smoke. The movement of multiple individuals may increase the inhalation exposure risk for others through airflow mixing effects [93]. In summary, the collective findings from these studies suggest that quantifying exposure risk in fire smoke environments must incorporate the movement behaviors of occupants. However, a significant gap remains in accurately describing the interaction characteristics between the movement of personnel and fire smoke.

Scholars have investigated the impact of human activities such as walking, and limb and torso movements on the dispersion of pollutants. However, these studies have primarily focused on ambient-temperature inhalable particles, volatile organic compounds, and infectious disease droplets [94,95,96], with relatively few addressing fire smoke specifically. Han et al. indicated that the movement of personnel creates thermal convection effects with high-temperature toxic gases, significantly altering the distribution of hazards within risk zones by promoting mixing between gases of different temperatures [97]. Luo et al., studying air pollution from wildfires in California, USA, investigated the impact of activities like walking and window-opening on individual respiratory exposure, demonstrating that human activity can lead to increased particulate matter concentrations in the breathing zone [98]. Benabed et al. found that airflows generated by human movement can cause particles previously settled on floors and object surfaces to become resuspended, entering the human breathing zone [99,100]. Wu et al. further demonstrated that occupant movement can significantly alter an individual’s transient inhalation exposure and increase the respiratory exposure risk for others in the environment, an effect that is more pronounced in multi-occupant movement scenarios [101,102]. These results highlight the necessity of considering the influence of human behaviors when quantifying their exposure and injury in underground fire.

Typical human postures and movement behaviors involved in fire smoke evacuation scenarios primarily include running, upright walking, crouching/walking bent forward, and crawling. It is noted that changes in an individual’s position directly affect their instantaneous inhalation exposure characteristics. Airflows induced by movement can also alter the inhalation patterns for smoke particles of different sizes. Furthermore, the psychological stress commonly experienced during evacuation can lead to an increased breathing rate, thereby elevating exposure risks. Therefore, it is essential to investigate the interaction dynamics between evacuating individuals and fire smoke under various conditions, including different movement postures and activity intensities. Building on this, integrating the positional data of evacuees with the patterns of smoke spread (as disturbed by movement) and the spatiotemporal distribution of smoke concentration is crucial for accurately assessing respiratory system exposure and injury risks among individuals evacuating through fire smoke environments. This integrated approach will subsequently inform the optimization of evacuation strategies and the improvement of protective equipment.

In summary, macroscopic fire case studies and microscopic scientific research form a complete causal chain. The macroscopic cases reveal that smoke asphyxiation is the primary cause of fatalities, while microscopic studies, through component analysis and CFD simulations, explain at the particle level how fire smoke specifically causes damage to the human respiratory system. These two approaches, from different scales, confirm the lethality of smoke, creating a closed loop between theory and practice. The complexity of fire smoke composition in underground spaces poses severe challenges to protection technologies. For instance, with the recent development of new energy sources, fires related to such incidents have also become increasingly common [103,104]. The wide variety of toxic substances in smoke, which constantly change with combustion conditions, requires protective equipment to possess broad-spectrum protective capabilities, rather than being targeted only at a single harmful gas. This complexity drives the advancement of protection technology from simple filtration or isolation towards multi-functional and composite development.

In conclusion, researching the mechanisms of respiratory system exposure and injury in fire smoke environments holds significant practical importance. It is essential to consider the interaction characteristics between typical human movements, such as evacuation and rescue operations, and fire smoke, quantify the patterns of human smoke inhalation, and subsequently establish methods for assessing personnel risks in fire smoke scenarios. This will enable accurate assessment of human respiratory exposure risks and provide fundamental references for the development of protective equipment and the formulation of evacuation strategies.

4. Fire Smoke Protective Technologies

The protection system for underground space fires is an integrated framework comprising engineering protection technologies, emergency evacuation strategies, and personal protective equipment, and its development is advancing towards greater intelligence and precision [105,106]. In underground fire smoke control, it is generally necessary to apply air pressure to confine the smoke to specific areas and reduce its spread. The minimum required air velocity or air volume that just satisfies these control conditions is defined as the critical ventilation velocity or critical air volume [107,108,109]. For different underground space structures and fire scenarios, studying appropriate ventilation modes is essential for effective smoke control.

Longitudinal ventilation remains the dominant strategy for tunnel fires; its primary mechanics involve creating a critical velocity to prevent smoke back-layering. The relationship between critical velocity and factors like heat release rate has been extensively modeled, with key studies proposing dimensionless correlations such as critical velocity [110,111,112]. However, the presence of a train blockage introduces a complex perturbation to this flow, effectively reducing the cross-sectional area and increasing the required critical velocity depending on blockage ratio [113]. This necessitates design adjustments that go beyond standard formulas, incorporating Computational Fluid Dynamics (CFD) simulations to account for such real-world obstructions. The influence of ambient pressure has also been studied, revealing that as ambient pressure decreases, the reduced inertial force leads to an increase in smoke back-layering length [114]. Furthermore, factors such as the cross-sectional shape [115,116], slope [117,118,119], fire source location [120], and shaft ventilation [121,122] have been investigated.

For fires in multi-level underground structures, the primary objective is to confine the smoke to the fire-originating floor and prevent its diffusion to upper levels [123,124]. The operational mode of ventilation and smoke extraction during a fire is another critical engineering technique for fire smoke control requiring in-depth study. Subway stations are typically equipped with ventilation and smoke exhaust systems. Under normal circumstances, these systems operate in air supply and return modes, functioning to introduce fresh air and regulate the ambient temperature in public areas. During a fire, the air supply and return systems in the affected area must be switched to a smoke exhaust mode, ensuring the timely extraction of fire smoke and safeguarding the safe evacuation of occupants. When evaluating ventilation performance, factors such as vent locations, fan operation timing, air curtains, and piston effects are also considered [125,126,127].

Beyond traditional ventilation and smoke exhaust engineering design, a new generation of fire monitoring systems is enabling a shift from passive response to active prediction through artificial intelligence technologies. These systems integrate various sensors such as fire signal detectors and charge-coupled device (CCD) cameras, utilizing multi-source information fusion to enhance the accuracy of fire situation identification [128,129,130]. The core of these intelligent systems lies in specific machine learning architectures. For instance, Convolutional Neural Networks (CNNs) are increasingly applied to process video feeds from CCD cameras for early flame and smoke visual recognition, while Long Short-Term Memory (LSTM) networks are used to analyze temporal data from gas and temperature sensors for predicting fire development trends. This represents a clear integration of computer vision and environmental sensing—an interdisciplinary fusion that allows for faster and more reliable fire detection compared to single-source systems. Unlike traditional systems that rely on fixed thresholds, intelligent systems can continuously refine their decision rules through self-learning algorithms. Some intelligent early warning devices employ fuzzy neural networks and nonlinear fire models to predict the severity and development trends of fire situations and can automatically integrate with firefighting systems to achieve precise rescue operations.

Ventilation systems remain crucial for fire smoke control in underground spaces. However, future ventilation systems should be driven by artificial intelligence and big data, possessing significantly enhanced adaptive capabilities. The use of artificial intelligence and machine learning can further extend to tasks such as real-time occupant tracking and the adaptive, dynamic control of ventilation systems. Meanwhile, the application of digital twin technology enables the creation of dynamic, real-time fire scenario simulations, which holds significant potential for optimizing both ventilation strategies and evacuation management. Intelligent algorithms can dynamically optimize ventilation strategies based on real-time environmental data—such as temperature and smoke concentration—coupled with fire simulation data, enabling rapid response and highly efficient smoke extraction [131]. Such systems effectively mitigate the negative impacts associated with traditional ventilation strategies. By dividing the underground space into independent smoke compartments, the intelligent system can activate exhaust vents only within the compartment affected by fire, thereby concentrating efforts on smoke removal and creating favorable conditions for occupant evacuation and rescue operations.

Beyond engineering systems, personal protective equipment (PPE) serves as the last line of defense for self-rescue and mutual aid in fire scenarios [132,133]. Filtering self-rescue respiratory protective devices are crucial tools for fire escape. Their core principle involves processing inhaled air through a filtration unit that utilizes adsorption, absorption, catalysis, and direct filtration mechanisms to remove carbon monoxide, hydrogen cyanide, toxic smoke, and hazardous particulates from the smoke, thereby providing breathable air [134]. National standards, such as GB/T 38451-2019 [135], establish strict technical requirements for PPE, including minimum protection times against various harmful gases like carbon monoxide and specifications for breathing resistance. Innovations in materials science represent a key breakthrough for enhancing the performance of personal protection. Traditional filter media face limitations when confronting high temperatures and complex smoke compositions. Nanofiber filter media show promising application potential. Characterized by extremely small fiber interstices, nanofiber media can maintain high filtration precision while reducing breathing resistance [136,137]. Research indicates that nanofiber membranes fabricated via electrospinning technology can effectively filter fine particulate matter, bacteria, and viruses from the air, offering the advantages of high filtration efficiency and a low pressure drop. Certain nanofiber materials can even be functionalized to remove specific gases, showing potential for broad-spectrum protection. Some researchers have developed novel heat-resistant “sponge ceramics” that are ultra-lightweight, heat-tolerant, and compressible, showing promise for applications in firefighting garments and smoke filtration. Unlike the brittleness of traditional ceramics, these sponge ceramics can withstand high temperatures while maintaining elasticity, offering a new solution for respiratory protection in high-heat environments. Furthermore, artificial intelligence is transforming the research and development paradigm for new materials used in PPE [138]. Researchers can leverage AI simulation systems to rapidly identify optimal solutions from vast arrays of material formulations, significantly shortening R&D cycles and accelerating the transition of new protective materials from the laboratory to practical application.

Overall, human protection in fire smoke scenarios within underground spaces is a systematic endeavor, and its future development will focus on multi-disciplinary technology integration, intelligent systems, and full life-cycle safety management. One such development is interdisciplinary research integration. Future fire safety in underground spaces will be the product of deep integration across multiple disciplines, including fluid dynamics, thermodynamics, materials science, and computer science. This entails more accurately incorporating human psychology and behavior into fire evacuation models, and investigating the complex micro-scale coupling between human movement-induced airflows and fire smoke, thereby enabling the more precise assessment of occupant exposure risks. Second is the further application of intelligent technologies. The Internet of Things, big data, and artificial intelligence will comprehensively empower underground space fire protection systems, facilitating a shift from passive response to active prediction, and from imprecise rescue to precise rescue. For instance, AI algorithm-based dynamic path planning methods can utilize real-time data (like temperature and carbon monoxide levels) extracted from fire simulation software to dynamically determine optimal evacuation routes and maximum allowable escape times. This technology should be deeply integrated with intelligent ventilation systems and personal protective equipment, forming a real-time responsive, self-optimizing intelligent emergency network. Furthermore, continuous innovation in new protection technologies is crucial. Materials science still holds immense potential for innovation in the field of personal protective equipment. Future research should develop multifunctional composite filter media capable of simultaneously filtering particulate matter and adsorbing toxic gases, enhancing the durability and reliability of protective equipment under extreme high temperatures, and leveraging AI-assisted design to accelerate the commercial application of a new generation of protective equipment. Through these methods, fire safety in underground spaces will advance to a higher level, providing a solid guarantee for the sustainable development of cities.

5. Conclusions

This paper offers a systematic review of the smoke movement patterns, related casualty risk characteristics, and recent developments in engineering protection technologies and personal protective equipment in underground spaces fires. This review attempts to provide theoretical and technical assistance for improving human safety in fire scenarios and promote sustainable urban development.

(1) Considerable research efforts have been devoted to comprehending fire smoke diffusion characteristics, smoke control theory, and coordinated smoke management strategies in common underground spaces such as subways and tunnels. This research includes a theoretical analysis, experimental studies, and numerical simulations. The impact of variables like ventilation, fire intensity, source location, and material characteristics on smoke dispersion patterns has been methodically examined. Building structure, obstacles, slope, and ventilation conditions are some of the important characteristics that have been incorporated into classical models for ceiling temperature distribution. Smoke stratification behavior and overflow patterns in various scenarios have been thoroughly analyzed.

(2) Respiratory system injury caused by inhaling fire smoke constitutes a primary cause of fire-related fatalities. This review analyzes the risks posed by toxic gases, particulate matter, and thermal effects in fire smoke from both macroscopic case study and microscopic toxicological perspectives, forming a closed loop connecting practical observation with theoretical mechanisms. Future research needs to focus on the dynamic interactions between characteristic human movements and fire smoke, quantify the patterns of human inhalation, and establish precise personnel risk assessment methodologies to enable accurate risk prevention and control.

(3) Human protection in underground space fire scenarios is a systematic endeavor, the future development of which should concentrate on multi-disciplinary technology integration, intelligent system application, and full life-cycle safety management. Advancing accurate models of human–fire–smoke interaction through interdisciplinary research and leveraging AI for holistic fire management are crucial. Simultaneously, the application of intelligent technologies must be deepened to optimize ventilation and enhance monitoring and early-warning systems. Specific breakthroughs in new material development for advanced personal protective equipment and the implementation of intelligent, adaptive evacuation systems represent critical technological paths for future research.