1. Introduction
With the continuous development of society and economy, the chemical industry has risen rapidly and played an important role in promoting social progress and urban construction [1]. However, this booming industry is also accompanied by substantial safety hazards. Chemical accidents such as fires, explosions, and toxic gas diffusion not only threaten the safety of facilities and personnel within the plant but may also affect the surrounding environment and residents [2,3]. As the demand for chemicals continues to grow, China has more than 5000 types of hazardous chemicals, and the number of major sources of danger continues to climb [4]. During the production, transport, and storage of chemicals, the slightest mishap may cause an accident, resulting in serious casualties and economic losses [5]. This high-risk characteristic makes the study of chemical accident prevention and emergency response particularly important [6,7].
Recently, with the increasing number of chemical plants, accidents caused by toxic gas leaks have continuously posed threats to people’s production and daily life [8]. For example, a toxic gas leak occurred in the southern Jordanian port city of Aqaba, resulting in 14 deaths and 265 injuries. In Santos Port, São Paulo, Brazil, a toxic gas leak incident caused 52 people to be hospitalized for poisoning. In China, the major explosion and fire accident at Liaocheng Lushi Chemical, where a hydrogen peroxide unit caught fire and exploded, led to 10 fatalities, 1 injury, and direct economic losses of CNY 54.45 million. Against this backdrop, efforts should be made to prevent chemical accidents as much as possible; additionally [9,10,11], reasonable research and guidance should be provided for emergency response after such accidents occur [12,13].
The issue of evacuation has become a major challenge that we must urgently address [14,15]. It is necessary to predict the potential spread of toxic gases, fire, or explosion ranges before an accident occurs, providing scientific guidance for emergency response [16,17]. The goal is to evacuate people to a safe location in the shortest possible time while minimizing the risk of poisoning or injury [18]. Previous studies have focused on predicting hazard zones for individual accidents, often neglecting the impact of climate change on these predictions [19]. In terms of evacuation routes, although algorithms are used to find the optimal and safest paths, there is still a lack of simulation modeling of evacuee behavior during the evacuation process [17]. This study aimed to combine annual meteorological data to predict the potential hazard zones of chemical accidents and conduct computer simulation of large-scale chemical accident evacuations.
In recent years, many experts and scholars have conducted relevant research in this field. For example, some studies have combined disaster scenarios with psychological behavior, using AnyLogic software to establish a system dynamics model of panic spreading, and conducted simulation experiments based on the level of disaster and crowd behavior [20]. In toxic gas diffusion simulations, previous research has used ALOHA software to model accident scenarios and delineate impact areas based on different wind speeds, using ERPGs (Emergency Response Planning Guidelines) to determine the extent of gas dispersion and evacuation locations [19]. However, these studies lack consideration of the changes in hazard zones caused by varying meteorological conditions, making their findings more targeted but less universal [21]. In emergency evacuation simulations, Pathfinder software is mainly used for fire accident simulations, typically in single buildings, small-scale areas, and high-density environments. For example, in campus evacuation simulations, different population capacities can be set, zones can be divided, results can be analyzed, and strategies can be optimized [22,23,24]. For crowded places such as shopping malls and large theaters, Pathfinder can be used to simulate dense crowd evacuation events and analyze evacuation strategies by considering crowd types and alarm response times [25,26]. However, most studies have overlooked the need for evacuation guidance for members in large-scale scenarios. Some scholars have attempted to use Pathfinder to simulate fire evacuations in informal settlements and refugee camps, studying the effects of crowd density, delays, route selection, and restricted paths on evacuation efficiency and have achieved good results, verifying Pathfinder’s applicability in such scenarios.
In summary, the current application of Pathfinder software is still mainly focused on evacuation simulations for single buildings [27,28]. This is because Pathfinder can visualize congestion, path selection, and flow characteristics during the evacuation process, thereby providing support for adjusting and optimizing evacuation strategies [29]. However, research on large-scale scenarios remains limited [30,31]. Therefore, this study applied Pathfinder to simulate large-scale evacuations in chemical industrial parks. Building upon this, the study further considered the similarities and differences between fire and chemical accidents, conducting simulations of hazardous chemical accidents. This paper assumes that various factors in different scenarios may influence the evacuation outcomes [32,33]. Psychological factors, evacuation strategies, and site geographical information, among other aspects, can all affect the final evacuation outcomes [34,35]. This study analyzed the impact of these factors on overall evacuation results in different evacuation scenarios, further extending the traditional application scope of Pathfinder software. By simulating the evacuation process in a chemical industrial park, this research verifies the feasibility of Pathfinder in the field of hazardous chemical accidents and provides theoretical support and practical reference for its application in more complex emergency scenarios.
2. Methodology
The entire hazardous chemical accident simulation flowchart of this study is shown in Figure 1 and is divided into three main parts:
Part I: This study selected a case of an accident in a chemical plant. Based on the equipment and meteorological parameters, simulations were conducted to determine the impact range of the incident. The study further assessed the level of hazard posed to evacuees in different areas of the plant.
Part 2: An evacuation model was developed, incorporating building areas, chemical facility zones, and road networks. First, road network data and building contours were organized with ArcGIS to obtain vector data. This data was then imported into CAD for drawing and processing. Finally, the processed data was imported into Pathfinder software to establish the evacuation model. The creation of the evacuation model involved tasks such as the arrangement of evacuees, setting of behavioral parameters, and placement of triggers, ensuring that the simulation results were both realistic and reliable.
Part 3: Multiple evacuation simulations were conducted using Pathfinder software, with various parameters set to compare and analyze the impact of different factors on the evacuation process. The study identified key areas that require particular attention during hazardous chemical accident evacuations, as well as less critical aspects, to effectively improve evacuation efficiency, protect the lives and property of personnel, and minimize overall losses.
2.1. Study Sites
In the accident scenario constructed for this study, a real incident was referenced. According to the accident investigation report of the accident, an explosion and fire occurred during pressurized sealing operations at the inlet pipeline of the alkylation unit’s water wash tank in a chemical plant. The geographical information of the accident area is shown in Figure 2. The plant is surrounded by walls and guardrails, with three exits located to the north for both pedestrian and vehicle access. The total area of the plant is approximately 1.312 .
The study location was based on Panjin City, Liaoning Province, with local geographical and meteorological parameters being utilized. The coordinates are 122.18° N, 41.36° E, and it is situated in a warm, temperate, continental semi-humid monsoon climate zone. According to the China Meteorological Administration (CMA), the average annual temperature in the region is 10.5 °C. The climate is characterized by long cold periods, strong winds on the plains, higher humidity in the east, drier conditions in the west, concentrated rainfall, abundant sunshine, and four distinct seasons. The province experiences clear seasonal variations, with each season exhibiting unique climatic features.
According to the accident investigation report, the chemical facility involved in the incident was an alkylation unit, with a designed capacity of 16 × 104 t/a, which was later upgraded to 20 × 104 t/a. The alkylation unit area measures 108 m in length (east–west) and 70 m in width (north–south), covering an area of 7560 . This study uses this facility as a prototype for toxic gas leakage simulation. The primary leaked substances were isobutane and n-butane.
2.2. Case Study on Impact Range Prediction
During an accident, accurately and rapidly predicting the affected range plays a crucial role in emergency response and rescue efforts. Various dispersion models differ in precision and advantages, with several well-established models currently in use, including the Gaussian plume model [36], integrated models, and computational fluid dynamics (CFD) models [37,38,39]. ALOHA software employs the Gaussian dispersion model to predict gas concentrations by modeling common leakage scenarios. The building infiltration time constant is approximated using ALOHA’s built-in default method.
Before the evacuation simulation is run, the required evacuation range must be determined. ALOHA software, developed by the U.S. EPA, is used to predict the impact area of chemical accidents. The affected area is divided into three hazard zones, including red, orange, and yellow, based on the Emergency Response Planning Guidelines (ERPG), which are defined as follows: ERPG-3 is the maximum airborne concentration below which nearly all individuals can be exposed for up to 1 h without experiencing or developing life-threatening health effects. ERPG-2 is the maximum airborne concentration below which nearly all individuals can be exposed for up to 1 h without experiencing or developing irreversible or other serious health effects or symptoms that could impair an individual’s ability to take protective action. ERPG-1 is the maximum airborne concentration below which nearly all individuals can be exposed for up to 1 h without experiencing more than mild, transient adverse health effects or without perceiving a clearly defined objectionable odor. The primary chemical responsible for the explosion in the alkylation unit was butane. The three ERPG values for butane are 5500 ppm, 17,000 ppm, and 53,000 ppm, respectively [19].
Next, ALOHA software was used for range simulation, with the continuous leakage and dispersion of n-butane following a fire in the alkylation unit being modeled. Table 1 lists the relevant parameters, with leakage data obtained from the incident’s alkylation unit and meteorological parameters based on local weather data for the day of the accident.
In addition to the primary parameters mentioned in Table 1, there were several other considerations in the simulation. The vapor cloud ignition time was set to 0 s; according to the accident report, the explosion was triggered by an ignition source, so the vapor cloud ignition type in the software was set to “ignited by spark” or “ignited by flame.” The congestion level was set to “uncongested, easily passable.” Additionally, the toxic zone’s yellow area was set to 500 ppm rather than AEGL-1, as the study aimed to evacuate as many people as possible to prevent secondary explosions, which are common in hazardous chemical accidents. The final impact area of the accident is shown in Figure 3, with the toxic gas dispersion zone, flammable zone, and explosion impact zone from left to right, respectively. In the toxic gas dispersion zone, members in the yellow area will first evacuate this zone before proceeding to the exit, while evacuees outside this zone are prohibited from entering.
2.3. Identification of Hazardous Zones
Since chemical accidents may occur under varying weather conditions, it is essential to assess the impact range across different seasons. In other words, evacuation planning should not be limited to individual case studies but should account for seasonal variations in accident outcomes. Therefore, this study integrated the impact ranges from different seasonal scenarios to generate an annual comprehensive risk zone, providing a more holistic perspective for evacuation strategy development.
2.3.1. Meteorological Analyses
To assess the impact range of toxic gas dispersion following a leakage accident, meteorological data should include annual average temperature, wind speed, and wind direction. Given that wind speed and direction vary across seasons and significantly influence dispersion patterns, this study conducted four separate simulations based on the predominant wind conditions in spring, summer, autumn, and winter. The results were then integrated to generate an annual comprehensive meteorological risk zone for the chemical industrial park.
The primary focus of this study was not the collection or statistical processing of meteorological data but rather the use of such data as a basis for predicting seasonal variations in gas dispersion patterns. If a particular area is consistently affected across all seasons, it should be prioritized for protective measures or evacuation planning. This study utilized 2019 meteorological data from Panjin, Liaoning Province, as the foundational dataset for analyzing annual climatic conditions. The meteorological data in the manuscript was provided by the Changchun Meteorological Bureau. The main contents included the monthly average temperature and wind speed shown in Figure 4 and the frequency of various wind directions across different seasons illustrated in Figure 5.
2.3.2. Annual Consolidated Risk Area
According to Figure 5, the four wind directions with the highest number of occurrences throughout the year were N, NNE, SSW, and NNW. The main wind direction in spring was SW, the main wind direction in summer was SSW, the main wind direction in autumn was SSW, and the main wind direction in winter was N. Because of the large temperature difference between the four seasons in Panjin, the wind speed in autumn and winter was significantly lower than that in spring and summer, and the change of the meteorological factors would have a significant impact on the scope of the diffusion of gases. The annual comprehensive meteorological risk zone of CIP was mapped out according to the predictions of the four seasons. Predictions were made to draw the annual comprehensive meteorological risk area of the chemical industrial park. Figure 4 and Figure 5 show the main wind direction and average temperature and wind speed in each season, which were used to distinguish the four different simulations, and the other parameters were set the same.
Panjin, Liaoning Province, is located in the northern hemisphere, and the meteorological data of this study were from February to April in the spring, from May to July in the summer, from August to October in the autumn, and from November to January in the winter. The average daily temperature, main wind direction, and the average wind speed of the main wind direction for each season are summarized in Table 2.
In the process of hazardous chemical accident evacuation, understanding the extent of the affected area is a prerequisite for making informed and effective evacuation decisions. This knowledge helps minimize the risks faced by evacuees, ensuring their safety and reducing potential health impacts. Therefore, this study conducted a detailed analysis of toxic gas dispersion across different seasons, providing critical insights for optimizing evacuation strategies.
This study utilized ALOHA software, which employs the Gaussian plume model to predict the dispersion range of gases, determining the spread of different concentrations. The research primarily focused on the toxic gas dispersion range, overlaying the fatal concentration ranges and tolerable concentration ranges from all four seasons. The leakage source point was superimposed with the location of the alkylation unit in the chemical plant to generate a comprehensive risk zone map based on annual meteorological conditions. This zone represents the area most likely to cause casualties during the evacuation process and should be given special attention.
Figure 6 shows the risk zone of the annual integrated meteorological tolerable range, which is AEGL-1, where the concentration of toxic gases is greater than 5500 ppm 60 min after the accident and where the exposed person does not feel any discomfort, or only mild discomfort, for 60 min. Although 5500 ppm is still a high concentration, it is not so high as to pose an immediate threat to health, and evacuees from the area should leave the area first and then find a suitable path to the exit of the chemical plant.
Figure 7 Annual integrated meteorological lethal concentration range risk area. This area is AEGL-3, where the toxic gas concentration value is greater than 53,000 ppm 60 min after of the accident, at which concentration, i.e., after 60 min of exposure, almost everyone will have a fatal reaction, which may result in the death of most exposed persons. Therefore, the area is considered inaccessible, and the two roads of the plant through the area are considered unusable, The unusable roads are marked in green, and if the risk of year-round leakage is considered, they are considered closed when an accident occurs to avoid more serious injuries or deaths. However, due to the uncertainty of the location at the time of the real accident, as well as the uncertainty of the meteorological conditions, the later simulations did not set the two roads as closed but rather took a specific case as a study to observe which factors in the evacuation campaign would have a significant impact on the overall evacuation.
Within 60 min after the leak, lower wind speeds in autumn and winter resulted in slower gas dispersion and prolonged presence of toxic concentrations. Consequently, it is crucial to minimize the duration of exposure within the plant and evacuate as quickly as possible. In the spring, summer, and autumn, the gas dispersion predominantly extends northward or northeastward, whereas in winter, it shifts southeastward. Therefore, enhanced management of ammonia transportation and storage is necessary during autumn and winter to mitigate associated risks.
Under annual comprehensive meteorological conditions, high-risk dispersion areas are primarily concentrated to the north and south of the chemical plant. Areas beyond 161 m north and 240 m south of the facility exhibit lower risk. Based on this analysis, emergency shelters can be strategically positioned to the east and west to ensure their safety and facilitate the rapid evacuation of individuals from the most hazardous zones.
3. Modeling
3.1. Construction of Geometric Shapes
The creation of the chemical plant geometry in Pathfinder consisted of the following steps. Figure 8 illustrates the key stages, from obtaining the initial model data to generating the final Pathfinder-compatible model. These steps were specifically designed for the case study plant but can also serve as a reference for larger areas. Figure 8 outlines the five main steps in model construction: Selecting the study area and determining its geographic coordinates. Importing the vector map of the selected area into ArcGIS for processing. Converting the processed data into CAD format. Exporting the 3D model into Pathfinder. Creating road networks, rooms, doors, and exits in Pathfinder.
Since some chemical facilities were not fully represented in the downloaded map, their boundaries had to be manually delineated in CAD. Given the potential for secondary hazards during chemical plant evacuations, evacuees are restricted from moving through gaps between chemical installations or approaching them too closely. Instead, movement is constrained to designated evacuation routes.
To ensure the general applicability of this study in evaluating various factors affecting chemical plant evacuations, certain structural adjustments were made to the buildings. Consequently, the evacuation simulation includes single-story buildings, multi-story buildings, chemical facility zones, and road networks to comprehensively represent the evacuation environment.
Previous studies have primarily focused on Pathfinder’s modeling capabilities in indoor building evacuations. However, the software has also been applied to large-scale crowd movement simulations in settings such as subways, train stations, and stadiums.
To further explore the versatility of Pathfinder and demonstrate its reliability in chemical plant evacuations, additional efforts were made to extend its application. This study incorporated less commonly used terrain types to model hazardous chemical accident evacuations, representing various real-world environments and assessing their impact on evacuation dynamics.
Since the accident occurs within a chemical plant, certain chemical installations are treated as obstacles with defined boundaries during the evacuation process. As a result, the usable road area is limited to the central portion of pathways, ensuring evacuees maintain a safe distance from hazardous facilities. To account for safety margins, the modeled boundaries of chemical installations are slightly larger than their actual physical dimensions.
In the initial simulation 1, the plant’s northern section includes three exits. Given the facility’s daily operational requirements, the initial exit width was set to 10 m. In this context, an “exit” refers to a designated point where evacuees leave the plant, marking the completion of the evacuation. Evacuees were recorded as having reached safety once they moved along the designated roadways to an area outside the assumed accident zone. The pathway width was estimated based on map resolution and spatial constraints.
3.2. Configuration Model
The scenario design aims to (1) compare the differences between conventional evacuation and evacuation under hazardous chemical accident conditions and (2) evaluate the applicability and effectiveness of Pathfinder software in hazardous chemical evacuation modeling. Each simulation modifies specific parameters within the model to observe the impact of various factors on evacuation outcomes and to draw meaningful conclusions. A total of eight evacuation scenarios were developed. Simulations 1–7 were designed to assess the influence of different factors on chemical evacuation, while Simulation 8 was specifically aimed at analyzing the effect of individual behavior on evacuation performance in chemical accident scenarios.
Simulations 1–7 were generated based on five key parameters from the evacuation analysis: population size, exit width, the addition or reduction of exits in various directions, and the establishment of emergency shelters. Different Pathfinder models were created based on the varying data used in each simulation. Table 3 summarizes the simulation results and the specific factors reviewed in each scenario.
It is important to note that the combination of ALOHA and Pathfinder is used for predicting and guiding evacuations. This study aimed to demonstrate that simulations can provide effective guidance for chemical evacuations. However, the results are not universally applicable. For different locations and hazardous chemical incidents, targeted simulations should be conducted, as the conclusions drawn here cannot be directly applied. The importance of various factors should be reanalyzed based on the specific conditions of each case.
The process of evacuation is always accompanied by the issue of evacuation instructions and the transmission of information and other issues, which is often referred to as the pre-evacuation time. Different scenarios of the crowd’s acceptance rate of the pre-evacuation alarm also vary after the evacuation alarm through different channels, for when to evacuate will also produce different reaction times, which is related to the time beginning from the decision to take action to the start of the evacuation time. Therefore, for all 7 scenarios, we set the pre-evacuation time to be normally distributed, with the mean being set to the average of 15 min and the standard deviation being set to 5 min.
As there were no elderly or children within the evacuation scenarios, only young and middle-aged people, the populations for the seven scenarios used the default occupant profile in Pathfinder, which assigns a lossless movement speed of 1.19 m/s to each individual. The occurrence of chemical accidents is accompanied by fire, corrosion, and vapor cloud explosions, which can also bring about a domino effect, and the prediction of its consequences is an important task to prevent the spread of the equipment unit to unit. With the development of the chemical industry, the desire to increase the raw material reserves of the chemical plant and the plant footprint has also grown, increasing the possibility of the domino effect occurring. During the evacuation process, crowds should not approach the vicinity of the chemical plant or pass through the center.
Scenarios 1 and 2 compare standard evacuation with hazardous chemical accident evacuation. In the latter, the simulation incorporates the dispersion range of toxic substances, requiring evacuees to first ensure their safety by exiting the affected area before proceeding toward an evacuation route outside the contaminated zone.
In Pathfinder, this is implemented by defining the n-butane dispersion area as an independent “room” (represented in gray in Figure 9). Evacuees are assigned the action “go to rooms” to leave this area, after which they are restricted from re-entering it. This is enforced by configuring the room’s exit door as a one-way door, preventing re-entry.
For buildings located at the boundary between the toxic gas dispersion zone and unaffected areas, certain doors within the affected region may be closed, requiring evacuees to use alternative exits. Once evacuees successfully leave the n-butane dispersion zone, they proceed to the nearest available exit to complete the evacuation process.
Simulations in Scenarios 3, 4, 5, 6, and 7 analyzed the influence of different factors on hazardous chemical accident evacuation. Scenarios 3 and 4 examined the impact of population size variations, with the factory’s standard workforce set at 2000 people, while Scenario 3 reduced this number to 1500, and Scenario 4 increased it to 2500.
Figure 10 shows the location of emergency shelters; yellow areas indicate Emergency Shelter 1, and gray areas indicate Emergency Shelter 2. As shown in Figure 11 and Figure 12, the toxic gas concentration within and around the two shelters remains below the minimum threshold after 1 h of continuous leakage, confirming their feasibility as emergency shelters.
Scenario 5 assessed the effect of exit width expansion, increasing the exit width from 10 m to 12 m. The factory originally had only three northern exits, while Scenario 6 introduced additional exits in three other directions to facilitate evacuation. Scenario 7 incorporated emergency shelters outside the gas dispersion zone.
During the evacuation process, evacuees first exit the gas dispersion zone. After this initial action, they then select the nearest available exit or emergency shelter. Reaching an emergency shelter was considered to be a successful evacuation outcome in the simulation.
Scenario 8 simulated individual behaviors during chemical plant evacuations, aiming to validate Pathfinder’s capability in realistically representing hazardous chemical accident evacuations. Based on the distance of individuals from the accident site and the varying concentration levels of toxic gases in their surroundings, evacuees were assigned different behavioral responses. Individuals autonomously decided to evacuate based on their perceived level of danger, while also influencing the behavior of others through peer interactions.
These behavioral dynamics were implemented using Pathfinder’s built-in trigger function. The simulation first divided the affected area into three zones based on toxic gas concentration levels and then assigned different evacuation behaviors to individuals in each zone accordingly.
In Zone 3, where toxic gas concentration is highest, individuals are closest to the leakage point and explosion/fire center, causing gas levels to escalate rapidly within a short period. Due to the shockwave and loud explosion noise, all individuals in this zone initiate self-evacuation within 5–20 s after the accident occurs.
In Zone 2, the perceived danger is lower. Here, evacuees have a 75% probability of self-initiating evacuation within 10–35 s after the accident.
In Zone 1, where the perceived threat is minimal but still present, 50% of individuals begin self-evacuation within 20–60 s after the incident.
Additionally, self-evacuating individuals influence others within a 2 m radius during the evacuation process. This influence is modeled as a 50% probability of prompting nearby individuals to evacuate if they are informed about the danger (e.g., through verbal communication). Once an individual initiates evacuation, their behavior remains unchanged and is no longer affected by other triggers.
Individuals who do not self-evacuate will evacuate upon the official alarm announcement, which follows the same normal distribution pattern as that in Scenario 2. Table 4 presents the detailed trigger settings for each zone.
In all scenarios, all simulation members made the following additional assumptions.
Individuals evacuate on foot only.
Individuals evacuate and begin moving directly to the exits once their delay time had expired.
At least one way out of the affected area is known.
No physical injuries that could impede evacuation movements occur.
Individuals act alone (i.e., did not belong to a group).
Individuals do not change their exit options once the simulation starts.
4. Results
Table A1 (Appendix A) presents the total evacuation time, total number of evacuees, and the time taken to reach each exit or shelter across the seven simulation scenarios. The comparison between Simulation 1 and Simulation 2 reflects the difference between a conventional evacuation and one under a hazardous chemical accident. The total evacuation time increased by 288 s in the latter. This increase is attributed to the need for evacuees to first exit high-risk zones—such as toxic gas dispersion areas, explosion sites, and fire-affected regions—before proceeding toward safe exits. This additional step, intended to avoid secondary injuries, inevitably prolongs the overall evacuation process.
In this study, the baseline population of 2000 was adjusted upward and downward by 500 in Simulations 4 and 3, respectively. The total evacuation time remained largely unchanged, which differs from the results of some previous studies [40] and indicates that the plant has high personnel carrying capacity under the given spatial and structural conditions. This suggests that, within a certain range, changes in population size do not significantly impact evacuation efficiency. Future studies may further investigate the critical population threshold beyond which congestion occurs and evacuation performance begins to degrade.
In Simulation 5, the exit width was increased from 10 m to 20 m based on Simulation 2. However, this expansion did not yield significant improvement in evacuation outcomes. Some previous studies have shown that exit width affects the overall efficiency of evacuation. The difference from earlier studies is due to the size of the area and the population density [41]. In chemical plants, evacuation exits often serve as vehicle access points, and are therefore relatively wide by default, reducing the likelihood of large-scale congestion during evacuation. Nevertheless, the prolonged presence of large vehicles near exits can pose potential hazards and should be strictly avoided. For narrower exits, however, increasing the width can substantially reduce evacuation time and alleviate crowding. Thus, exit width remains a critical factor that must be carefully considered in evacuation planning.
In Simulation 6, an additional exit was added on the eastern, southern, and western sides of the chemical plant, based on the configuration in Simulation 2. This adjustment proved effective, reducing the total evacuation time by 356 s. It is important to note that due to the uncertainty in the dispersion range of toxic gases during hazardous chemical accidents, evacuees must first move away from contaminated zones. As a result, the initial direction of evacuation may be different from, or even opposite to, the direction of the nearest exit. This highlights the need to provide evacuation exits in as many directions as possible.
Moreover, the evacuation capacity of each directional exit varies. Figure 13 illustrates the number of evacuees at each exit over time in Simulation 6. Exit 3 served only 70 individuals and exhibited relatively low efficiency, as indicated by the gentle slope of its curve. Figure 14 shows the movement trajectories of evacuees, with blue lines representing their paths. Exit 3 is located in the northeastern corner of the plant, surrounded by densely distributed chemical installations, with fewer nearby residences and narrower access roads, all of which contribute to its low usage.
For the scenario in Simulation 6, it is recommended that alternative directional exits be considered, particularly on the eastern side of the plant. Exit 3, primarily used for routine cargo transport and vehicle access, should not be designated as an emergency evacuation route in the event of a hazardous chemical accident.
In Simulation 7, two emergency shelters were added based on the setup in Simulation 2, reducing the total evacuation time to 870 s—a decrease of 489 s. A total of 1103 evacuees chose to move toward the shelters, which is similar to previous studies, demonstrating that the inclusion of emergency shelters significantly enhanced evacuation performance [42,43]. This suggests that within the context of this case, establishing shelters is the most effective improvement strategy.
As shown in Figure 10, although Emergency Shelter 2 is located farther from the plant and away from densely populated areas, its presence is crucial due to the “leave the toxic gas dispersion zone” action. This requirement causes some evacuees to move in a direction opposite to the nearest designated exits, making Shelter 2 a strategically valuable facility. Similarly, Emergency Shelter 1 offers a more efficient option for evacuees on the western side of the plant and helps reduce the flow burden on Exit 1.
Due to the large area and low population density of the chemical plant, the simulations did not encounter path capacity limitations or congestion. As a result, Simulations 3, 4, and 5 showed no significant difference in final evacuation time compared to Simulation 2. However, this does not imply that factors such as total number of evacuees or exit width are irrelevant in hazardous chemical accident scenarios. These parameters must still be carefully assessed on a case-by-case basis, especially in settings with higher density or spatial constraints.
The time at which the last evacuee passed through each exit or reached a shelter across the seven simulations is presented in Table 5, reflecting the maximum utilization duration for each exit. In Simulation 1, which represents a conventional evacuation scenario, exit 1 took the longest time to clear, with the final evacuee exiting at 1359.03 s. After the “leave the toxic gas dispersion zone” action was introduced in Simulation 2, the longest evacuation time shifted to Exit 3.
As shown in Figure 15, under normal conditions, the southwestern area of the plant is the farthest from any exit, leading residents in that zone to evacuate via Exit 1. Due to the extended distance, Exit 1 becomes the last to complete evacuation. However, in Simulation 2, the toxic gas dispersion area is concentrated in the central and eastern portions of the plant. Evacuees from these zones must take longer detours to avoid contaminated areas, resulting in Exit 3 becoming the last to clear. This illustrates that evacuation patterns under hazardous chemical accidents can significantly differ from those under normal conditions.
Figure 16 shows the total number of evacuees over time, and since the pre-evacuation time is normally distributed, all seven evacuations show or are similar to an “s” curve. The pre-evacuation simulation 1 and simulation 2 are not very different, and the curve is very close because the location near the exit does not belong to the toxic gas diffusion range, so the region can be quickly evacuated when the danger comes, so it is important to establish the personnel-intensive buildings such as office buildings and control centers in the location near the exit of the plant. Position curvature synchronization occurs in each of Simulation 2, Simulation 3, and Simulation 4; the curve slope of Simulation 4 is greater than that of Simulation 2, which is greater than that of Simulation 3, indicating that with the number of people change does not affect the quality of the evacuation; that is, there is no congestion and other situations occur; otherwise, the curve slope will change. There is no wasted time due to queuing during the whole process of evacuation, indicating that the road capacity in this case can fully meet the requirements of the personnel. The curves of Simulation 2 and Simulation overlap, and the widening of exits does not significantly help evacuation.
Simulations 6 and 7, which achieved earlier evacuation completion times and more evacuees within the same time frame, represent the two most effective improvements identified in this study. Although the pre-evacuation phase in Simulation 6 is similar to that in other scenarios, evacuation between 400 and 800 s is notably more efficient due to the addition of directional exits. Simulation 7 demonstrates higher evacuation efficiency at earlier stages, enabling some evacuees to reach shelters sooner. As the impact of toxic gases escalates over time following a chemical accident, reaching safe areas earlier is critically important.
Simulation 8 investigated the effect of spontaneous group behavior on evacuation in a hazardous chemical accident evacuation, with Simulation 2 being used as a basis for arranging triggers in areas with greater concentrations of toxic gases. Figure 17 shows a plot of the total number of evacuees over time in Simulation 2 and Simulation 8. The difference in the number of evacuees confirms the influence of individual behavior on the overall evacuation. In Pathfinder software, it is possible to simulate the behavior of the crowd by arranging the triggers to simulate the unique behavior in chemical accidents, which makes the simulation results more realistic and reliable, and demonstrates the feasibility and development prospects of the use of Pathfinder software in the evacuation of hazardous chemical accidents.
The evacuation speed of Simulation 2 was greater than that of Simulation 8 and was within 520 s after the accident. Due to the inclusion of the trigger in Simulation 8, the evacuating members will spontaneously evacuate in advance before the evacuation alarm is issued, so the evacuation speed should be greater than that of Simulation 2. The result is the opposite, indicating that in the pre-evacuation period, due to the inclusion of individual behaviors, although the members in the area will evacuate in advance, they are accompanied by crowding, confusion, and a blind escape leading to irrational routes being chosen. The inclusion of individual behaviors in the pre-evacuation period reduces the evacuation efficiency. Therefore, after the accident, the evacuation warning should be issued as soon as possible to calm down the evacuees so that the crowd can remain calm and evacuate according to reasonable routes to avoid congestion, panic, and blind evacuation.
The curve of Simulation 8 after 520 s is located above that of Simulation 2, which indicates that more people are evacuated in Simulation 8 in the same time, and the addition of individual behavior has a positive effect on evacuation in general, which suggests that individual behavior should be added into the computer simulation of the evacuation of hazardous chemical accidents. This phenomenon also suggests that in order to reduce the hazards of accidents, evacuation simulations should be carried out in advance in areas prone to accidents, such as chemical plants and warehouses, in order to raise people’s awareness of self-protection.
In addition to the impact on the overall evacuation, the inclusion of individual behaviors has also led to changes in people’s evacuation choices. Figure 18 shows the number of people at different exits in Simulation 2 and Simulation 8, with evacuating members in Simulation 2 preferring to evacuate towards Exit 1 and Exit 2 and those in Simulation 8 preferring to evacuate towards Exit 3. The fact that the accident point is closer to Exit 3 indicates that people are more likely to choose the routes they often use during spontaneous evacuation, ignoring the possible harm caused by hazardous gases. This calls for guidance on evacuation and the issuance of evacuation warnings as soon as possible.
5. Discussion
The following directions based on this study can be explored in future research:
① This paper focuses more on the macro aspects of overall evacuation, including the overall evacuation path and evacuation time. However, in hazardous chemical evacuations, many micro aspects are also worth noting, such as the group factor of a group, whether a leader will emerge and others will follow during evacuation, or whether there will be a person who needs rescue and if others will need to assist in getting them to a safe location. To address these issues, more “actions” need to be set up in the Pathfinder software to simulate the various scenarios.
② The construction of chemical plants is often required to be located far from populated areas, but with the continuous development of the chemical industry, it is inevitable that pedestrians and even residential areas will appear near chemical plants. Due to the uncertainty of accidents, they may spread to larger areas and affect surrounding residents. This will lead to a dramatic increase in population density, making evacuation much more challenging, as additional factors must be considered. The simulation and prediction method proposed in this paper and how to improve and adapt it will be a key issue to address in future work.
③ Future research could refine the selection of evacuation paths by planning more suitable routes for evacuees in different areas of the simulation, thereby mitigating the impact of the accident. This could be achieved with the help of computational algorithms to identify the optimal path. However, when the evacuation area is large, ensuring the practical feasibility of the selected paths will present a significant challenge.
6. Conclusions
The evacuation of a hazardous chemical accident has, on the one hand, the universality of conventional evacuation, with the need to evacuate as quickly and safely as possible to ensure the safety of evacuating members’ lives and property, which takes into account the establishment of optimal paths, the flow of paths, congestion, and consideration of the behavior of individual or groups of evacuees [44,45]; on the other hand, it involves numerous contingencies, with the day’s weather, temperature, humidity, and wind direction affecting the eventual diffusion of the poisonous gases, fire zones, and explosion zones, which are to be avoided as much as possible by evacuating people. This paper provides a series of systematic methods to simulate the formation of an accident, illustrates the differences between evacuation in the case of a hazardous chemical accident and a conventional evacuation, gives an indication of which factors are more helpful in evacuation in order to avoid similar accidents, and reminds decision makers to take more and more effective safety precautions before an accident occurs.
The results of the simulations show that the use of ALOHA software and Pathfinder software together can analyze the different factors involved in the evacuation of a hazardous chemical accident, and the following results were obtained:
① Compared with conventional evacuation, the uncertainty in the dispersion range of toxic gases means that evacuees in hazardous chemical accidents often cannot follow the shortest path to safety. To ensure their safety, evacuation routes may deviate significantly from a straight line between the starting point and the nearest exit [46,47]. In some cases, evacuees may initially move in a direction opposite to the exit. However, this change in route is not random but is based on hazard zone modeling using ALOHA software to avoid the most dangerous areas in the shortest possible time. The hazard range predictions take into account critical factors such as climate conditions, building types, and the properties of the leaked substance, enabling scientifically informed route planning that maximizes the distance from danger.
② The prediction of annual risk zones provides guidance for the establishment of emergency shelters and the selection of evacuation routes, making the division of hazardous areas more universal and enhancing the realism and reference value of the simulation results. It further enriches the research experience in this area [42].
③ For the case of this paper, the location of the accident is very empty, no congestion occurs after many simulations with different conditions, the road flow is sufficient, and the exit width is sufficient to meet the evacuation requirements; these findings build upon previous research to provide new insights [41].
④ When exits are established in all four directions, the evacuation can be finished 544 s earlier; when there are two emergency shelters, the evacuation can be finished 778 s earlier. These two factors provide evacuees with more route options. With increased path choices, evacuees can complete evacuation through alternative routes after leaving the danger zone, without having to take detours to reach distant exits. This allows them to complete the evacuation more quickly while staying away from the most hazardous areas. It is worth noting that for evacuees located near emergency shelters, the distance to the shelters is significantly shorter than is the distance to designated exits. The presence of shelters enables some individuals to complete evacuation earlier, thereby reducing their exposure to risk and positively influencing the overall evacuation outcome.
⑤ The deployment of triggers in Pathfinder enhances the realism of hazardous chemical accident evacuation simulations. By incorporating dynamic behavioral responses, the movement of evacuees in the simulation better reflects real-world evacuation patterns. This approach provides valuable insights into evacuation dynamics and serves as a reference for optimizing the timing and effectiveness of evacuation alarm announcements.
The research in this paper can be used to make hazard predictions for locations where chemical accidents may occur, to derive the extent of hazard coverage, and to determine what approach to take and what factors to look out for, which can make evacuation after an accident more efficient. It provides insights for large-scale evacuation and enriches future research directions.
Conceptualization, Y.S. and B.W.; methodology, Y.S. and B.W.; software, Y.S. and Y.W.; data curation, J.Z., Y.Z., and Y.W.; writing—original draft preparation, Y.S.; writing—review and editing, B.W., J.Z., Y.Z., and X.W.; supervision, B.W. and J.Z. All authors have read and agreed to the published version of the manuscript.
Not applicable.
Not applicable.
All relevant data are available in the article.
The authors declare no conflicts of interest.
Footnotes
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.
Figure 1 Flowchart for the simulation of hazardous chemical accidents.
Figure 2 Geographic information map of the accident area.
Figure 3 Software simulation area.
Figure 4 Temperature and wind speed in Panjin City by month in 2019.
Figure 5 Monthly average wind speed in Panjin in 2019.
Figure 6 Annual combined meteorological tolerable range risk zone.
Figure 7 Annual combined meteorological lethal concentration range risk zone.
Figure 8 Processes involved in building the Pathfinder model.
Figure 9 n-Butane diffusion range and one-way gate setting.
Figure 10 Shelter names and locations.
Figure 11 Indoor and outdoor concentrations in Shelter 1 during a 1 h period.
Figure 12 Indoor and outdoor concentrations in Shelter 2 over 1 h period.
Figure 13 Cumulative number of occupants evacuating through selected exits in Simulation 6.
Figure 14 Evacuee movement trajectory in Simulation 6.
Figure 15 Simulation 2 evacuation path diagram.
Figure 16 Variation in the total number of evacuees by simulation.
Figure 17 Comparison of the number of evacuees over time between Simulation 2 and Simulation 8.
Figure 18 Comparison of the number of evacuees at each exit between Simulation 2 and Simulation 8.
Table of leakage and meteorological parameters.
| Leakage Parameter | Input | Diffusion Parameter | Input |
|---|---|---|---|
| Type of leak | Open sources | Average temperature (°C) | 10.5 |
| Leakage | Based on simulation | Average wind speed (m/s) | 1.25 |
| Leakage (t/min) | 1 | Prevailing wind direction | South-easterly wind |
| Leakage source height (m) | 1 | Measuring height | Ground level |
| Relative humidity | 66 per cent | ||
| Surroundings | Unobstructed trees |
Meteorological table by season.
| Temperature (°C) | Wind Speed (m/s) | Wind Direction | |
|---|---|---|---|
| Spr. | 6 | 1.7 | SW |
| Sum. | 22 | 1.8 | SSW |
| Fal. | 19 | 1.3 | SSW |
| Win. | −2 | 1.4 | N |
The seven scenarios in Pathfinder.
| Simulation | Number | Pre-Evacuation Time | Existence of Emergency Shelters | Increase in Exports | Outlet Width | Whether to Consider Leaving the Dispersal Zone First | Consideration |
|---|---|---|---|---|---|---|---|
| 1 | 2000 | Normal distribution μ = 15 min, σ = 5 min | Nonexistent | No increase | 10 m | No | Changes in total evacuation time against hazardous chemical accidents |
| 2 | 2000 | Normal distribution μ = 15 min, σ = 5 min | Nonexistent | No increase | 10 m | Yes | |
| 3 | 1500 | Normal distribution μ = 15 min, σ = 5 min | Nonexistent | No increase | 10 m | Yes | Decrease in the number of people |
| 4 | 2500 | Normal distribution μ = 15 min, σ = 5 min | Nonexistent | No increase | 10 m | Yes | Increase in the number of people |
| 5 | 2000 | Normal distribution μ = 15 min, σ = 5 min | Nonexistent | No increase | 20 m | Yes | Increased exit width |
| 6 | 2000 | Normal distribution μ = 15 min, σ = 5 min | Nonexistent | Increasing exports in other directions | 10 m | Yes | Increase in exports in different directions |
| 7 | 2000 | Normal distribution μ = 15 min, σ = 5 min | Remain | No increase | 10 m | Yes | Presence of emergency shelters |
Trigger settings.
| Trigger Number | Typology | Active Position | Simulation Content |
|---|---|---|---|
| Trigger 1 | Specified rooms | Region 1 | Evacuation of 100 per cent of members commenced after 5–20 s |
| Trigger 2 | Specified rooms | Region 2 | 75 per cent of members started evacuation after 10–35 s |
| Trigger 3 | Specified rooms | Region 3 | 55 per cent of members started evacuation after 20–60 s |
| Trigger 4 | Line of sight | Moving with evacuating members | Interaction between members |
| Trigger 5 | Global | Entire evacuation zone | Evacuation alert issued |
Time of departure of last person at exits and shelters.
| Exit 1 Final Time | Exit 2 Final Time | Exit 3 Final Time | Exit 4 Final Time | Exit 5 Final Time | Exit 6 Final Time | Travel Time to the Shelter | |
|---|---|---|---|---|---|---|---|
| Simulation 1 | 1359.03 | 1172 | 1000 | Nonexistent | Nonexistent | Nonexistent | Nonexistent |
| Simulation 2 | 1393 | 851 | 1647.28 | Nonexistent | Nonexistent | Nonexistent | Nonexistent |
| Simulation 3 | 1441 | 821 | 1635.28 | Nonexistent | Nonexistent | Nonexistent | Nonexistent |
| Simulation 4 | 1261 | 854 | 1647.28 | Nonexistent | Nonexistent | Nonexistent | Nonexistent |
| Simulation 5 | 1393 | 851 | 1647.28 | Nonexistent | Nonexistent | Nonexistent | Nonexistent |
| Simulation 6 | 955 | 572 | 509 | 825 | 1103.78 | 838 | Nonexistent |
| Simulation 7 | 643 | 857 | 799 | Nonexistent | Nonexistent | Nonexistent | 869.52 |
Appendix A
Number of evacuees at each exit and total evacuation time and number of evacuees.
| Number of Evacuees at Exit 1 | Number of Evacuees at Exit 2 | Number of Evacuees at Exit 3 | Number of Evacuees at Exit 4 | Number of Evacuees at Exit 5 | Number of Evacuees at Exit 6 | Number of Evacuees at Shelters | Total Evacuation Time | Total Number of Evacuees at Each Exit | |
|---|---|---|---|---|---|---|---|---|---|
| Simulation 1 | 1141 | 636 | 223 | Nonexistent | Nonexistent | Nonexistent | Nonexistent | 1359.03 | 2000 |
| Simulation 2 | 1014 | 429 | 557 | Nonexistent | Nonexistent | Nonexistent | Nonexistent | 1647.3 | 2000 |
| Simulation 3 | 777 | 329 | 421 | Nonexistent | Nonexistent | Nonexistent | Nonexistent | 1635.3 | 1527 |
| Simulation 4 | 1261 | 516 | 703 | Nonexistent | Nonexistent | Nonexistent | Nonexistent | 1647.3 | 2500 |
| Simulation 5 | 1014 | 429 | 557 | Nonexistent | Nonexistent | Nonexistent | Nonexistent | 1645.3 | 2000 |
| Simulation 6 | 478 | 320 | 70 | 432 | 397 | 303 | Nonexistent | 1103.8 | 2000 |
| Simulation 7 | 269 | 429 | 199 | Nonexistent | Nonexistent | Nonexistent | 1103 | 869.5 | 2000 |
1. Zeng, T.; Chen, G.; Yang, Y.; Reniers, G.; Zhao, Y.; Liu, X. A systematic literature review on safety research related to chemical industrial parks. Sustainability; 2020; 12, 5753. [DOI: https://dx.doi.org/10.3390/su12145753]
2. Seo, S.-K.; Yoon, Y.-G.; Lee, J.-S.; Na, J.; Lee, C.-J. Deep neural network-based optimization framework for safety evacuation route during toxic gas leak incidents. Reliab. Eng. Syst. Saf.; 2022; 218, 108102. [DOI: https://dx.doi.org/10.1016/j.ress.2021.108102]
3. He, Z.; Weng, W. Synergic effects in the assessment of multi-hazard coupling disasters: Fires, explosions, and toxicant leaks. J. Hazard. Mater.; 2020; 388, 121813. [DOI: https://dx.doi.org/10.1016/j.jhazmat.2019.121813] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/31843413]
4. Hou, J.; Gai, W.-M.; Cheng, W.-Y.; Deng, Y.-F. Hazardous chemical leakage accidents and emergency evacuation response from 2009 to 2018 in China: A review. Saf. Sci.; 2021; 135, 105101. [DOI: https://dx.doi.org/10.1016/j.ssci.2020.105101]
5. Wang, R.; Xu, K.; Xu, Y.; Wu, Y. Study on prediction model of hazardous chemical accidents. J. Loss Prev. Process Ind.; 2020; 66, 104183. [DOI: https://dx.doi.org/10.1016/j.jlp.2020.104183]
6. Sierra, D.; Montecchi, L.; Mura, I. Stochastic modeling and analysis of vapor cloud explosions domino effects in chemical plants. J. Braz. Comput. Soc.; 2019; 25, 11. [DOI: https://dx.doi.org/10.1186/s13173-019-0092-8]
7. Zhou, J.; Reniers, G. Petri net simulation of multi-department emergency response to avert domino effects in chemical industry accidents. Process Saf. Environ. Prot.; 2021; 146, pp. 916-926. [DOI: https://dx.doi.org/10.1016/j.psep.2020.12.004]
8. Feng, J.R.; Gai, W.-M.; Yan, Y.-B. Emergency evacuation risk assessment and mitigation strategy for a toxic gas leak in an underground space: The case of a subway station in Guangzhou, China. Saf. Sci.; 2021; 134, 105039. [DOI: https://dx.doi.org/10.1016/j.ssci.2020.105039]
9. Wang, B.; Li, D.; Wu, C. Characteristics of hazardous chemical accidents during hot season in China from 1989 to 2019: A statistical investigation. Saf. Sci.; 2020; 129, 104788. [DOI: https://dx.doi.org/10.1016/j.ssci.2020.104788]
10. Cui, W.; Chen, X.; Li, W.; Li, K.; Liu, K.; Feng, Z.; Chen, J.; Tian, Y.; Chen, B.; Chen, X. Simulation of a Hazardous Chemical Cascading Accident Using the Graph Neural Network. Sustainability; 2024; 16, 7880. [DOI: https://dx.doi.org/10.3390/su16187880]
11. Zhou, K.; Xiao, L.; Lin, Y.; Yuan, D.; Wang, J. A statistical analysis of hazardous chemical fatalities (HCFs) in China between 2015 and 2021. Sustainability; 2022; 14, 2435. [DOI: https://dx.doi.org/10.3390/su14042435]
12. Neto, J.; Morais, A.J.; Gonçalves, R.; Coelho, A.L. Guiding Evacuees to Improve Fire Building Evacuation Efficiency: Hazard and Congestion Models to Support Decision Making by a Context-Aware Recommender System. Buildings; 2023; 13, 3038. [DOI: https://dx.doi.org/10.3390/buildings13123038]
13. Deng, Q.; Zhang, B.; Zhou, Z.; Deng, H.; Zhou, L.; Zhou, Z.; Jiang, H. Evacuation time estimation model in large buildings based on individual characteristics and real-time congestion situation of evacuation exit. Fire; 2022; 5, 204. [DOI: https://dx.doi.org/10.3390/fire5060204]
14. Hou, J.; Gai, W.; Cheng, W.; Deng, Y. Statistical analysis of evacuation warning diffusion in major chemical accidents based on real evacuation cases. Process Saf. Environ. Prot.; 2020; 138, pp. 90-98. [DOI: https://dx.doi.org/10.1016/j.psep.2020.02.028]
15. Juntao, L.; Wei, W.; Qi, Z.; Ruiping, Y. Multi-Stage Emergency Materials Scheduling Based on Multimodal Transportation Under Uncertainty. Oper. Res. Manag. Sci.; 2023; 32, pp. 33-39.
16. Li, X.; Chen, C.; Hong, Y.; Yang, F. Exploring hazardous chemical explosion accidents with association rules and Bayesian networks. Reliab. Eng. Syst. Saf.; 2023; 233, 109099. [DOI: https://dx.doi.org/10.1016/j.ress.2023.109099]
17. Gai, W.; Deng, Y. Dynamic Emergency Route Planning for Major Chemical Accidents: Models and Application. Emergency Guidance Methods and Strategies for Major Chemical Accidents; Gai, W.; Deng, Y. Springer Nature Singapore: Singapore, 2022; pp. 73-103.
18. Du, L.; Feng, Y.; Tang, L.; Lu, W.; Kang, W. Time dynamics of emergency response network for hazardous chemical accidents: A case study in China. J. Clean. Prod.; 2020; 248, 119239. [DOI: https://dx.doi.org/10.1016/j.jclepro.2019.119239]
19. Jiang, Y.; Xu, K.; Gai, W.; Salhi, S. Emergency response for tackling major accidental toxic gas releases: What should be done and when?. Saf. Sci.; 2022; 154, 105819. [DOI: https://dx.doi.org/10.1016/j.ssci.2022.105819]
20. Xu, S.; Wang, J.; Li, J.; Wang, Y.; Wang, Z. System dynamics research of non-adaptive evacuation psychology in toxic gas leakage emergencies of chemical park. J. Loss Prev. Process Ind.; 2021; 72, 104556. [DOI: https://dx.doi.org/10.1016/j.jlp.2021.104556]
21. Ye, R.; Xu, S.; Liu, J.; Wang, J.; Wang, Y.; Jiang, J. Investigation and simulation of non-adaptive psychology of crowd evacuation under toxic gas leakage accident in chemical park. J. Loss Prev. Process Ind.; 2023; 85, 105160. [DOI: https://dx.doi.org/10.1016/j.jlp.2023.105160]
22. Syed Abdul Rahman, S.A.F.; Abdul Maulud, K.N.; Pradhan, B.; Syed Mustorpha, S.N.A.; Che Ani, A.I. Impact of evacuation design parameter on users’ evacuation time using a multi-agent simulation. Ain Shams Eng. J.; 2021; 12, pp. 2355-2369. [DOI: https://dx.doi.org/10.1016/j.asej.2020.12.001]
23. Gao, S.; Chang, C.; Liu, Q.; Zhang, M.; Yu, F. Study on the optimization for emergency evacuation scheme under fire in university building complex. Heliyon; 2023; 9, e14277. [DOI: https://dx.doi.org/10.1016/j.heliyon.2023.e14277] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/36942225]
24. Xu, M.; Peng, D. Pyrosim-based numerical simulation of fire safety and evacuation behaviour of college buildings. Int. J. Saf. Secur. Eng.; 2020; 10, pp. 293-299. [DOI: https://dx.doi.org/10.18280/ijsse.100218]
25. Snohová, A.; Kučera, P. Comparative Study of SFPE and Steering Modes in Pathfinder to Optimise Evacuation Routes. Buildings; 2024; 14, 2361. [DOI: https://dx.doi.org/10.3390/buildings14082361]
26. Spearpoint, M.; Arnott, M.; Xie, H.; Gwynne, S.; Templeton, A. Comparative analysis of two evacuation simulation tools when applied to high-rise residential buildings. Saf. Sci.; 2024; 175, 106515. [DOI: https://dx.doi.org/10.1016/j.ssci.2024.106515]
27. Zhang, Z.; Kong, H. Study on the Safety Evacuation of University Dormitory Occupants Based on Pathfinder. Int. J. Adv. Smart Converg.; 2024; 13, pp. 328-334.
28. Ivanov, M.L.; Chow, W.-K. Experimental and numerical evacuation study in tall office building. J. Build. Eng.; 2023; 76, 107103. [DOI: https://dx.doi.org/10.1016/j.jobe.2023.107103]
29. Liu, X.; Kong, H. Safety Evacuation Simulation for a Hospital Outpatient Building Based on Pathfinder. Int. J. Adv. Cult. Technol.; 2024; 12, pp. 587-594.
30. Liu, J.; Li, S. Pathfinder-Based Simulation and Optimization of Evacuation of Large Commercial Complexes. J. Build. Constr. Plan. Res.; 2023; 11, pp. 27-35. [DOI: https://dx.doi.org/10.4236/jbcpr.2023.112002]
31. Zhang, Z.; Lee, J.; Kong, H. Evaluation of Evacuation Safety in University Libraries Based on Pathfinder. Int. J. Internet Broadcast. Commun.; 2024; 16, pp. 237-246.
32. Wu, Y.; Xia, J.; Haiyirete, X. Enhancing Disaster Resilience in Hospitals Through Flow Space-Optimized Evacuation Routes. Sustainability; 2025; 17, 5419. [DOI: https://dx.doi.org/10.3390/su17125419]
33. Wang, L.; Zhang, Z.; Lu, S.; Wang, J. Behavior selection models of fire evacuations with the consideration of adaptive evacuation psychologies. Sustainability; 2024; 16, 3607. [DOI: https://dx.doi.org/10.3390/su16093607]
34. Wei, Y.; Liu, J.; Jin, L.; Wang, S.; Deng, F.; Ou, S.; Pan, S.; Wu, J. Individual Behavior and Attention Distribution During WayFinding for Emergency Shelter: An Eye-Tracking Study. Sustainability; 2023; 15, 11880. [DOI: https://dx.doi.org/10.3390/su151511880]
35. Hosseini, O.; Maghrebi, M.; Maghrebi, M.F. Determining optimum staged-evacuation schedule considering total evacuation time, congestion severity and fire threats. Saf. Sci.; 2021; 139, 105211. [DOI: https://dx.doi.org/10.1016/j.ssci.2021.105211]
36. Ni, J.; Yang, H.; Yao, J.; Li, Z.; Qin, P. Toxic gas dispersion prediction for point source emission using deep learning method. Hum. Ecol. Risk Assess. An Int. J.; 2020; 26, pp. 557-570. [DOI: https://dx.doi.org/10.1080/10807039.2018.1526632]
37. Hanna, S.R.; Hansen, O.R.; Dharmavaram, S. FLACS CFD air quality model performance evaluation with Kit Fox, MUST, Prairie Grass, and EMU observations. Atmos. Environ.; 2004; 38, pp. 4675-4687. [DOI: https://dx.doi.org/10.1016/j.atmosenv.2004.05.041]
38. Regens, J.L.; Gunter, J.T.; Amin, M.; Nowakowski, A.; Navaz, H. Parameterizing Potential Exposure to Sulfur Mustard (HD) Using Mixed Model Regression. Hum. Ecol. Risk Assess. An Int. J.; 2011; 17, pp. 1229-1246. [DOI: https://dx.doi.org/10.1080/10807039.2011.618387]
39. Scargiali, F.; Di Rienzo, E.; Ciofalo, M.; Grisafi, F.; Brucato, A. Heavy Gas Dispersion Modelling Over a Topographically Complex Mesoscale: A CFD Based Approach. Process Saf. Environ. Prot.; 2005; 83, pp. 242-256. [DOI: https://dx.doi.org/10.1205/psep.04073]
40. Kuligowski, E.D.; Gwynne, S.M.V.; Xie, H.; Westbury, A.; Antonellis, D.; Pongratz, C. Simulating Evacuation of Humanitarian Settlements. Fire Technol.; 2023; 61, pp. 1327-1357. [DOI: https://dx.doi.org/10.1007/s10694-023-01431-6]
41. Cai, Z.; Zhou, R.; Cui, Y.; Wang, Y.; Jiang, J. Influencing factors for exit selection in subway station evacuation. Tunn. Undergr. Space Technol.; 2022; 125, 104498. [DOI: https://dx.doi.org/10.1016/j.tust.2022.104498]
42. Zhang, D.; Mao, Z.; Gong, M.; Ren, J.; Zuo, S.; Chen, X. Study on optimization of shelter locations and evacuation routes of gas leakage accidents in chemical industrial park. Process Saf. Environ. Prot.; 2023; 177, pp. 556-567. [DOI: https://dx.doi.org/10.1016/j.psep.2023.07.029]
43. Chen, W.; Shi, Y.; Wang, W.; Li, W.; Wu, C. The spatial optimization of emergency shelters based on an urban-scale evacuation simulation. Appl. Sci.; 2021; 11, 11909. [DOI: https://dx.doi.org/10.3390/app112411909]
44. Yoshitsugu, N.; Abe, S.; Yamamoto, K. Method for deriving evacuation routes considering disaster risk. J. Inf. Process.; 2020; 28, pp. 1065-1072. [DOI: https://dx.doi.org/10.2197/ipsjjip.28.1065]
45. Huang, P.; Lin, X.; Liu, C.; Fu, L.; Yu, L. A real-time automatic fire emergency evacuation route selection model based on decision-making processes of pedestrians. Saf. Sci.; 2024; 169, 106332. [DOI: https://dx.doi.org/10.1016/j.ssci.2023.106332]
46. He, Z.; Shen, K.; Lan, M.; Weng, W. The effects of dynamic multi-hazard risk assessment on evacuation strategies in chemical accidents. Reliab. Eng. Syst. Saf.; 2024; 246, 110044. [DOI: https://dx.doi.org/10.1016/j.ress.2024.110044]
47. Zhang, W.; Li, C.; Gai, W. How does evacuation risk change over time? Influences on evacuation strategies during accidental toxic gas releases. Int. J. Disaster Risk Reduct.; 2024; 108, 104531. [DOI: https://dx.doi.org/10.1016/j.ijdrr.2024.104531]
© 2025 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.
Details
; Wang, Yilin 1 1 Changchun Institute of Technology, College of Jilin Emergency Management, Changchun 130012, China; [email protected] (Y.S.); [email protected] (X.W.); [email protected] (Y.Z.); [email protected] (Y.W.)
2 Institute of Natural Disaster Research, School of Environment, Northeast Normal University, Changchun 130117, China; [email protected]