The Simulation-Based Analysis Focusing on Street Obstruction of Evacuee Mobility to Mitigate Disaster Risk: Chiang Mai Historic City

Abstract

While urban historic areas are most vulnerable to disasters, they offer insights into leveraging their features to mitigate risk. This study analyzes scientific approaches to evacuation simulations to assess the tolerance of historic areas. Using a heritage-led disaster risk reduction approach, this study uses a heritage site as a case study for evacuation. This study uses a GIS-based methodology to define various blockage risks, categorizing them as no-obstruction, rubble-obstruction, on-street vehicle obstruction, and combined obstruction. The input parameters were transferred from a GIS to a simulation application, with combined obstruction representing the worst-case scenario. No-obstruction served as a baseline for measuring historic area vulnerability. Statistical analysis evaluated time usage and the number of evacuees, while GIS identified vulnerable places and street congestion. Obstructions significantly increase evacuation risks, with combined obstructions posing a 3.8 times higher risk than the no obstruction scenario (2638 s compared to 683 s). Vehicle obstruction causes a vulnerability of 1404 s, while building collapse-related rubble obstruction causes a vulnerability of 1073.1 s, despite creating dead-end streets. The strategy of reinventing heritage sites as temporary evacuation sites appears viable. This approach can support evacuees during and after disaster responses and expand options for ensuring urban heritage resilience.

1. Introduction

Community-based Disaster Risk Reduction (DRR) conforms to and is consistent with local wisdom and traditional knowledge [1,2]. Traditional knowledge has been usually overlooked; with a collaborative and appropriate approach to both indigenous and scientific knowledge, it can help enlighten disaster research and strengthen the DRR strategies of individuals and their communities to cope with and be resilient against disasters [2,3,4,5]. To balance indigenous and scientific knowledge in DRR, the most important element is finding how these existing conditions work together to reduce disaster losses from high-level hazards [2,3,6,7,8]. Given implementation of DRR measure, the most important aspect to consider in efficiently minimizing risks is preparedness. Evacuation planning is one of the elements that can strengthen preparedness process. Unfortunately, it has been focusing on theoretical or mathematical algorithms rather than an application/practical approach [9,10]. The qualitative expression for disaster risk is described by the types of hazards, vulnerability, and exposure. Practical application, on the one hand, involves assessing components that induce risks such as the vulnerability of physical structures and infrastructure toward hazardous events and, on the other hand, measuring the intensity of risk sources [11,12,13]. However, less focus may have been placed on aspects that mitigate adverse impacts such as human mobility during evacuation/incidents [14,15]. The mechanisms through which vulnerable individuals evade and/or seek refuge in shelters can be thoroughly investigated and enhanced through spatial and temporal analysis, such as evacuation simulations [16].

To reduce disaster vulnerability, many researchers have proposed simulations and/or predictive data to represent the real situation, especially in urban areas [17,18,19]. By properly establishing environmental parameters, an evacuation simulation can use and test different potential hazard scenarios. Several discoveries have been made in evacuation simulation studies and techniques focusing on evacuation calculation models [17,18,20,21,22,23]. Many scenarios have been presented to analyze the complexity of evacuees and predict some managerial problems such as total evacuation time and flow conservation of street/transport. Different modes of transportation are associated with evacuation scale and distance [24,25,26,27,28,29]. According to our knowledge, simulation in managerial/operational is thus understudied, while disaster study applicability is at its lowest. In 2006, Altay and Green reviewed disaster operations management research in the 1990s and compared it to related disaster studies. Despite the expansion of disaster publications since the 1980s and their progressive increase in the 2000s, the study of simulations as a methodological approach in disaster research, particularly in implication/application, has been neglected. Furthermore, this study found that recognizing the sources and types of essential information is critical [30]. A continuation study from Altay and Green, published in 2013 by Galindo and Batta, which investigated scholarly work in the 2000s, confirmed the underutilization of disaster research in the mitigation phase [31]. Simulation remains understudied, while disaster study applicability is at its lowest. The study’s findings revealed no significant change in disaster research; the research output is more focused on the model, algorithm, and driven approach than on the application approach. Their results indicated that future research must thoroughly investigate disruptions using statistical analysis to reflect real-world conditions, which may imply appropriate mitigation solutions. In 1998, Tufekci and Wallace recommended highlighting the need for pre- and post-disaster event tactics as well as new technology, analytical techniques, and scientific models [32].

In evacuation planning, chances of survival improve when it is easier and faster to reach an emergency shelter. Evacuation routes and zones should be established to ensure residents’ safety and mitigate disaster hazards as an application-oriented effort [9,33]. Another aspect is evacuee behavior because the occurrence of natural disasters may cause a chaotic situation [34]. According to wayfinding research, the presence of an easily recognizable/memorable object as an important place or prominent landmark at a decision point or area that links to nodes and places can facilitate human orientation in the wayfinding process [35] and help people spend shorter time and reduce navigation error to reach the destination, including in an unfamiliar environment [36,37]. However, people have many reasons for selecting shelters and for evacuation routes. For example, they may select a particular place over others for evacuation destination depending on their perception of familiarity and sense of support [38]. In some cases, the decision to evacuate is based on social cues or spontaneous behavioral responses because of the impact of other evacuees [39]. Concurrently, they may select escape routes because some aspects represent an affordance for evacuation, such as proximity to the shelter and outflow ability such as clear path and visibility [40,41,42]. As a result, escape areas may have a high crowd density, which can cause longer evacuation time and impede movement flow. To the best of our knowledge, limitations persist in the investigation of human movement in real stampede incidents or during disaster evacuations. In this case, computer models help simulate crowd movement in evacuation scenarios. Findings of an experimental laboratory study showed that the layout of the escape area influences the characteristics of crowd movement, and that obstacle placement could affect evacuation flow and mobilization time [43]. Whereas some research underlined that evacuation behavior in the real situation could differ and suggested an emphasis on arrival time to the desired shelter [44,45]. In addition, an examination of how the obstacle affects the street such as the debris of collapsed buildings or parked vehicles to mitigate disaster risks is required [46,47,48,49]. Findings revealed that a collapsed building may worsen the flow of movement. Ito et al. (2020) found that collapsed buildings that block the road can lead to delays and that enhancing the seismic-resistance reinforcement of all houses can increase the number of people who reach the evacuation shelter by 3–4% [48]. Castro et al. (2019), in an experiment on different ground motion intensities in earthquake scenarios, also found that evacuation time significantly increases because of building debris blocking the road with parked cars [47].

In this regard, a required evacuation model, which should realistically simulate a certain situation, must also be examined to identify a city’s vulnerable areas, such as ABM capacity and congestion [50]. The agent-based model (ABM) can replicate the movement of an agent, commonly referred to as an evacuee. The model is a powerful and effective instrument for determining human behavior in relation to the environment [23,50]. The utility of evacuation appears useful for DRR policy, particularly in a micro-scale city [51]. In 2019, Esposito et al. evaluated and identified the major issues in evacuation shelter and routing operations with a focus on disaster operations management. They described certain concerns and recommended the implementation of ABM in evacuation simulation with various contexts of human (agent) behavior and environmental elements. Furthermore, their findings sought to advance shelter placement optimization as the evacuation takes place [9].

An urban historic landscape may become more vulnerable than a modern developed area [11]. The Ideology of Historic urban landscape formed in 1976 mentioned the role of “historic areas” [52]. These areas encompass a collection of buildings, structures, and open spaces situated within their respective natural and ecological contexts. Since the Charter of Venice in 1964, conservation efforts have focused on both theoretical and practical aspects, with a particular emphasis on monuments. This led to the establishment of the Charter for Urban Conservation in 1987 through the Washington Charter [53,54]. This approach facilitated the advancement of historic city conservation by expanding its scope from monuments to their surrounding context, including urban streets, urban spaces, and natural settings [55]. The ideology of historic cities as heritage has gradually improved through the application of these charters [56,57,58,59]. Subsequently, it evolved into the concept of Historic Urban Landscape, which UNESCO recommended as a viable approach to conserving urban heritage rather than focusing on individual sites [60,61]. This concept presents an opportunity to address contemporary global challenges by utilizing indigenous knowledge to sustain cities. As a change management strategy, the Historic Urban Landscape approach may contribute to mitigating disasters [61]. The Historic Urban Landscape and Historic City Conservation have recently been defined as the management of change rather than attempting to freeze it in place. While the protection of human life and (urban) livelihoods has been overlooked and understudied, this approach emphasizes the importance of adapting to change [62,63]. In 2021, Ripp et al. proposed the concept of heritage-based urban development, which involves remodeling or reinventing cultural heritage as a potential resource for urban development, particularly for enhancing urban resilience [64,65].

Data on massive evacuations with a lack of short warning periods are more difficult to obtain than those on buildings or human behavior [34,66,67]. A historic area with a deformed grid is one of the most extremely vulnerable areas to various types of disaster hazards [68,69] because the variations in the length of their sightlines and width of spaces may cause difficulties for the continuous flow of evacuee movement. Furthermore, historic areas’ character, labyrinthine nature, and narrowness, combined with the clustered nature of settlements, make them highly problematic for evacuation planning because of potential loss of orientation among evacuees. These streets, however, are unmodifiable elements because they are permanent structures that usually hold historical and cultural significance. Therefore, they could easily be blocked by building debris, vehicles, and street bottlenecks due to evacuees. Moreover, vulnerable buildings in historical areas are close to the aforementioned narrow streets and pose a higher risk in evacuation planning [70,71,72]. Regarding historical areas, Okubo (2016) suggested the use of their indigenous elements as traditional wisdom to cope with disasters [5] and/or enhance the resistance of existing elements [13,73] to reduce vulnerability. In 2024 Ripp et al. proposed the integrated usage of urban heritage as a resource for heritage-led resilience. The idea is to use their strength such as robustness, adaptive capacity, and self-organization which have been tested and survive the multi-hazard for centuries [65,74]. Moreover, elements of historical areas such as religious spaces have a high potential of transforming into evacuation shelters [75]. In Asian cities, temples are religious spaces that contribute a positive character such as emergency/temporary shelters, as they can provide basic needs such as food, a place to stay, and mental support resources for anyone who needs help. Also, the layout of this type of religious place is designed in scale to serve residents in a more localized portion of the city such as a district or a neighborhood. It was also intended to provide people a place for certain activities, particularly cultural and social gatherings, so its location is easily accessible for a great number of people (i.e., near the road, at the city or community center). As well as in Thailand, the Buddhist temples (or Wat) have served as the sacred community centers such as social/cultural, educational, social support, recreational space, etc [76,77]. With the idea of heritage as a resource for urban resilience, the site of study refers to these sacred landscapes where it may be able to reinvent itself as one of the evacuation plans.

This study simulates an evacuation route with various obstacles such as building debris/parking vehicles and presents a decisive solution in the environment to establish an optimum combination of evacuation routes and shelter/site. Religious places are determined as evacuation destinations. It also compares obstructions and determines the optimization of shelter location/overcapacity as well as its potential solution scenarios. The simulation will be adapted from a building evacuation application. However, evacuee conduct is universally applicable and has a solid foundation. The simulation will be analyzed in terms of various blockage circumstances. The scale metrics of obstruction will be simultaneously measured by total time, time to depart, and accumulation of evacuation. Based on Altay and Green (2006) and Esposito et al. (2019), this study is expected to partly fill a literature gap by requiring the improvisation of an unusual event in evacuation simulation as part of disaster preparedness [78], which emphasizes the definition of disaster management to contain the degree of control [9,30]. This inquiry will integrate GIS with simulation applications as its primary toolkit to better analyze and recommend public policy utilizing risk maps as its focal point and issues.

2. Materials and Methods

2.1. Background of Study Site

The historical city of Chiang Mai in northern Thailand is the setting for this research. Chiang Mai features a historical district that is currently used as the city’s center. Founded in 1296, Chiang Mai is located on the nearest plain to a river and a mountain, which makes the area fertile for agriculture. The brick wall and moat surrounding the city’s core were kept as visible examples of the integrity of the city’s form. The city retains its labyrinthine street network, which is intricately layered from historical passageways to modern streets. Furthermore, residents of Chiang Mai used to worship the ruling dynasty as semi-divine by representing its hierarchy and limits in its shape [79]. In the 12th to 14th centuries, Chiang Mai was the capital of the “Lan Na region,” which represented Tai-Dai ethnicity [80]. From the beginning, Chiang Mai has been designated as Lan Na’s capital. Its founders were consulted to design and lay the concept of the city shape, which signifies Lan Na’s symbolic and spiritual core [81,82]. Since its founding, Chiang Mai has steadily transformed into a Buddhist temple serving as the heart of the community. These religious places are associated with sociocultural activity. Currently, there are 38 active temples in the historical district of Chiang Mai, where distinctive practices continue to be part of Lan Na’s cultural identity.

During Thailand’s counter colonial period and early modernization in the 18th and 19th centuries, some significant locations in the historical area were replaced by central government offices or quarters. Examples include the city hall, post office, city court, and even the jailhouse [83]. Religious buildings, such as temples, were exempt from replacement because of their role as places of worship and have mostly remained in place since the founding era. This movement is still visible in modern times; thus, it is not surprising that such religious buildings have been retained. To accommodate automobile transportation, modern urban streets were built on top of traditional streets as a result of development [84] (See Figure 1).

2.2. Conceptual Framework

This study’s simulation test will concentrate on potential obstructions in the old city. The vulnerable area will be identified, and comparative data on different obstructions will be retrieved. The research technique will integrate a GIS tool and a simulation application, and the outcomes will include maps and statistical descriptions. The components of the investigation will include the egress point, which is the origin of agents representing evacuees from historical buildings. Temples will be used as the evacuation destination because they are prevalent in the Chiang Mai historical area. Furthermore, these temples have required open spaces that function as temporary shelters for evacuees, which can be repurposed to address aspects of historical city disaster resilience.

This study uses three types of obstruction to create a simulation in an area of the old city. These obstruction scenarios will then be examined and compared in the evacuation simulation. Building debris blocking the street will be created. The simulation will employ building collapse formulas to calculate the risk of roadway blockage using the GIS toolkit. The secondary obstacle object will be an on-street parked vehicle, which will imply a temporary street blockage. Vehicle obstruction will be estimated using the space syntax theory of network integration. Finally, the vehicle obstruction will merge with the building rubble obstruction to form a situation of combined obstruction. All obstruction scenarios will compare and discuss the vulnerability of evacuation routes and the overcapacity of designated shelters (see Figure 2). The primary focus of this study will be the time of evacuation, specifically the maximum time required for evacuation. The study will compare different scenarios to identify the critical factors that influence evacuation. The congestion that arises during the simulation will determine the location of evacuation vulnerability. The study will use the number of agents that successfully reach the evacuation destination as a measure of the site’s capacity. The maximum number of agents will be displayed as an evacuation capacity map, and comparisons will be made within and between scenarios to identify vulnerable areas and determine whether the site is overcapacity.

2.3. Determination of Origin and Destination

The main goal of this analysis is to simulate an evacuation scenario considering religious entities as designated evacuation destinations. Evacuees’ origin is the egress point, which connects the building entrance and the street, while their destination is the religious places. Egress from a building, which is the agent’s origin from their residential space, is where they first appear in the simulation process. Information about the building’s exit routes was gathered through a field survey and then entered into the GIS program. Buildings and streets are connected by an area designated as egress. The field survey classified egress point types into normal, large, service, emergency escape, and unused groups based on their purpose and size. Normal access refers to an egress point from a building with a width of 1–5 m. Large access pertains to an egress point with a width greater than 5 m. Service access is defined as an egress point that originates from the service access or secondary access to the building. A fire exit is an egress point that is designated as emergency access. Unused access refers to a building’s unutilized access points. From the field survey, the number of egress points are as follows: normal access: 2793 (66.58%), large access: 1357 (32.35%), service access: 23 (0.55%), fire exit: 3 (0.07%), and unused access: 19 (0.45%) (see Figure 2).

The agent’s destination is the religious space. This concept uses a historical element, such as a religious space, as a supportive factor for a temporary evacuation shelter during an emergency in the city. In Asian culture, religious temples have substantial space and adequate facilities to serve as evacuation hubs [75]. With complicated space and street narrowness, a historical city remains ill-equipped to respond to any hazard [1,2]. The use of the religious space as an evacuation space means increasing the choices, capacity, and resilience of emergency shelters, which aim to decrease vulnerability and casualties from extreme hazards to the highest extent [85]. The largest type of public space in Chiang Mai is the religious space, which is distributed throughout the city. As a living historical city, Chiang Mai has preserved the relation between the community and sacred space as a spirit of place. Local communities continue to use these venues to represent their culture and tradition; this condition is more ideal for proposing religious facilities as evacuation shelters because of the habitual from–to journey and their natural function as community centers.

2.4. Input Parameter of Evacuees (Agent)

This study uses Pathfinder version 2022.3 (Educational license), a multiscale agent-based simulation application that has been successfully used in evacuation scenarios. However, most Pathfinder simulations focus on a certain building or city area [86]. The primary outcomes to examine in this simulation are time and congestion.

Regarding the flow of agents, this analysis refers to the critical study of people in panic, which are the main problems during disaster and/or post-disaster situations and are difficult to predict [17]. In most situations, severe earthquakes are followed by fire hazards and building collapse. Thus, the evacuation model’s combination data should be used to generate the optimal reaction to the worst-case scenario. To represent empirical studies of crowd movement, this study adopts the National Fire Prevention Association and Ezel Kendik’s mean of egress [87] to determine how long it would take to exit a building. The egress flow of evacuees is included as an origin in Pathfinder as a step egression flow (See Supplementary Materials for input parameter). Step egression flow rate is calculated considering consecutive egress to the outside of the building. Meanwhile, the speed of agents will be determined using the standard range of pedestrian walking speed from 0.8–1.3 m/s. The speed range will represent people with a wide range of abilities [88,89]. Using this model, we defined people with impaired mobility and older adults as moving at an average speed rather than dividing people into the agent groups to provide an overview of the evacuees.

The controlled factors are the flow, location, and number of egression points, as well as the behavior profile of the evacuees, for a reliable comparison analysis. The evacuee profile has a shoulder width of 45.6 cm and a height of 165 cm, based on average of the Thai human scale [90]. Speed refers to the maximum travel velocity when referring to the agent speed parameter, but it also depends on the extent of congestion during the evacuation.

Evacuee destination has been configured to be random, which means that each evacuee can choose any location independently. The randomness of travel choice will represent the options of evacuation of agents if they encounter the obstruction and/or congestion. The simulation’s movement mode is “steering,” in which evacuees move and interact with others via a steering system or turning. For example, if one evacuee is traveling with another, the application algorithm will group the evacuees. This mode was chosen for its ability to mimic human behavior and natural movement. The calculation of this algorithm, however, is for an individual evacuee and excludes social aspects such as family grouping relationships.

2.5. Creation of Street Obstruction Scenarios

Obstruction is a crucial issue to consider when planning an evacuation. In an urban setting, an obstruction could be something as simple as a moving object on the street, such as a trash can, lamppost, or damaged signs, or as significant as an immovable object that blocks a street or roadway, such as vehicles and building debris. These increase the risk of evacuation and may produce street blockage. The following scenarios are constructed based on significant street blockage situations for retrieving detailed simulations.

2.5.1. No Obstruction Scenario

This is the fundamental situation in which only egress from the building to the entry gate of the evacuation destination was simulated. This scenario tests street bottlenecks and time usage in terms of evacuation. Evacuee congestion is assessed to determine the potential risk of the metropolitan network itself.

2.5.2. Vehicle Obstruction Scenario

A vehicle obstruction simulation is performed using abandoned vehicles on the street, which may result in street blockage and an increased evacuation risk. Integration analysis of space syntax is conducted to determine the number of automobiles on the street. This measurement evaluates the extent of connectivity when moving across the area of study. The most integrated street is the one with the fewest turns to another street in the network and is regarded as having the most options for travel [91]. This assessment explains urban network accessibility issues associated with ease of access and street choice by determining a integration value, consisting of global and local integration, through the limit of angular turning as R = n and R = 3, respectively [92,93,94].

The outcomes of the space syntax integration are employed to estimate the number of abandoned vehicles on the street, utilizing global integration as the estimation method. The objective of this method is to estimate the number of parked vehicles on the street that will appear on-street in higher integration values compared to streets with lower integration values. The worst-case scenario is chosen to represent the greatest danger to evaluate the congestion due to vehicle obstruction. The expected number of cars is derived using the following equation. (see Equation (1))

$$Vehicle(s) on street=\prod _i{\displaystyle \frac{Global integration\left(Length\right)}{10}}$$

(1)

A GIS algorithm is then used to calculate the number of vehicles on the street and simulate vehicle obstruction. Vehicle dimension is the average dimension of a compact car or a sedan from the Dimension website, which is 1.72 × 4.5 m.

2.5.3. Building Rubble Obstruction Scenario

According to Equation (2), the study adopted building debris radius as BV according to Santarelli et al.’s assessment in 2018 and critical post-event scenarios to create a risk map by Quagliarini et al. in 2016, who proposed the criteria for the estimation of a building’s collapse radius in a historical city [95,96]. In this case, the collapse of a structure resulted in a street blockage. The structure’s rubble could completely or partially obstruct the street. Building height is the most important factor in evaluating a building collapse scenario. V stands for a building’s vulnerability, which in this case is the type of building material. The magnitude ratio Rm, which refers to the ratio of the magnitude of the event to the highest magnitude, will be 4.1:6.2, with reference to the most severe seismic events in the Chiang Mai area. In this study, “h/W” stands for the building height and façade facing street width. (see Equation (2)). For example, the common buildings in the site study is 12 m with façade street width of 4 m, the BV will be 9.3 by using highest magnitude of 6.2 and Vulnerable of building (V) is low because its material. The BV will represent as the buffer value from the building and overlay into the street.

$$BV=V·R_m·{\displaystyle \raisebox{1ex}{$h$}\!\left/ \!\raisebox{-1ex}{$W$}\right.}$$

(2)

The previous equation classifies the obstruction circumstances into two. The first is complete obstruction, which affects full blockage and results in a cut-off or dead-end street. Complete blockage also includes obstacles caused by two or more buildings on the opposite sides of the street. Second, the degree of the partial blockage is determined by debris volume, which is then determined by GIS. Partial blockage is caused by a 50% to 80% street obstruction, making evacuation difficult.

2.5.4. Combined Obstruction Scenario

The combination of these obstacles poses the greatest risk of a vulnerable scenario. In the worst-case evacuation scenario, the combined materials of blockage, which includes both building rubble and obstructing vehicles, are used.

3. Results and Discussion

3.1. Overview of Comparative Results

By comparing different scenarios, one can quantify vulnerability from the lowest risk (no obstruction) to the highest risk (combined obstruction) scenarios. In these scenarios, the total number of evacuees ranged from 34,751 to 34,967 at maximum. Even though evacuees released at the egress points were controlled in this simulation, the fluctuations in the total number of evacuees were caused by congestion in front of some egress points. In every scenario, no one was trapped in the models. From no obstruction to combined obstruction, the total evacuation time (maximum time usage) ranged from 628.8 to 2638.0 s. The shortest time required to evacuate to a religious place was 3.9 s. The minimum distance for all evacuees was 3.1 m, and the largest distance varied from 609.6 to 790.7 m. More than distance, vulnerability was caused by the length of time required for evacuation. Significant differences in total evacuation time reflect the risk in the obstruction scenarios. With regard to distance and time ratios, the combined obstruction ratio is the highest at 3.34, followed by vehicle obstruction at 2.26, and building rubble obstruction at 1.36, which is slightly higher than the no obstruction scenario at 0.89. Therefore, despite the greater distance of building rubble obstruction, the flow and congestion of evacuees are substantially less than those in a vehicle obstruction scenario. These findings agreed with the previous studies [47,48], which suggested that the building debris on the street could impair evacuees’ mobility, and even increase vulnerability when combine with vehicle obstruction.

Table 1. Comparison Descriptive Statistic.

	No Obstruction (n = 34,967)		Vehicle Obstruction (n = 34,751)		Building Rubble Obstruction (n = 34,961)		Combined Obstruction (n = 34,793)
	Time (s)	Distance (m)	Time (s)	Distance (m)	Time (s)	Distance (m)	Time (s)	Distance (m)
Min	3.9	3.1	3.9	3.1	3.9	3.1	3.9	3.1
Max	682.8	609.6	1404.3	619.6	1073.1	785.2	2638	790.7
Average	112.4	165.6	261.2	176.2	183.5	176.2	343.4	127.4

n = number of evacuees (agents), s = second(s), m = meter(s).

summarizes details regarding evacuation time and distance in these scenarios. By comparing the maximum evacuation times between scenarios, the combined obstruction presents the most significant risk, with a total of 2638 s (44 min). This is approximately 3.86 times higher than the scenario with no obstruction (baseline scenario), which has a maximum evacuation time of 682.8 s (11 min). Surprisingly, the vehicle obstruction time for evacuation is secondary in terms of risk. The vehicle obstruction scenario takes 1404.3 s (23.5 min) to complete the evacuation, which is 331.2 s (5.5 min) longer than the scenario with building rubble obstruction. The comparison of evacuation times suggests that it is crucial to control vehicle-on-street parking or any temporary obstruction to mitigate the risk of congestion during evacuation.

All scenarios peaked between 18 and 19 s. Immediately after the peak is reached, the curve of the gradient slope began to significantly drop. During this time, 602–765 individuals were able to evacuate to the temples. The curve line started to differentiate after the peak. Building rubble obstructions were equivalent to no obstructions, while vehicle and combined obstructions were similar. All scenario curves began to fall when 10,000 people remained and nearly flattened when 3000 people were left. Significantly, variations in travel time affect evacuation time. Furthermore, vehicle obstruction is more vulnerable than building obstruction, as it has a nearly identical curved slope to combined obstruction (see Figure 3).

3.2. Simulation

At 18–19 s, all streets within the models are used from an instantaneous evacuee snapshot. From 0 to 20 s, no congestion was observed; each evacuee was able to egress from buildings in 19.1 s. Depending on the obstruction, every evacuee could be seen moving toward the evacuation sites in 120–180 s; congestion, however, was difficult to observe.

Evacuee congestion began to emerge in the same area, such as the western and northeastern parts of the city, 200–400 s after they remained in the system. Evacuation vulnerability was determined between 450 and 800 s, and every obstruction scenario identified the city’s northeastern region as the area of risk. The simulated comparison demonstrates the risk associated with the temple’s gate congestion; as a result, the risk varied depending on the obstructions that affected evacuation time, but the vulnerable area remained in the same location (see Figure 4).

As anticipated, congestion in the evacuation simulation is evident at the entrance of designated evacuation sites. While the combined obstruction scenario presents the most significant vulnerability, resulting in numerous congestion locations and extended evacuation times, the vehicle obstruction scenario demonstrates a more pronounced distribution of congestion locations compared to the building rubble obstruction scenario.

Figure 4 shows the time usage comparison based on the remaining number of evacuees (persons). Notably, the last 1000 evacuees were still present after 780.9 s (approximately 13 min) in the vehicle obstruction scenario, whereas the Combined obstruction scenario (the worst-case) resulted in the evacuation of all evacuees within 800 s. Despite a mere 20 s difference in the number of remaining evacuees, most of them were congested near the entrance of the evacuation site.

This phenomenon in the simulation undermines the necessity of vehicle control in historic cities and the potential risks associated with on-foot evacuation, which may be worsened by on-vehicle evacuation.

In addition to the evacuation time usage in Table 1, the simulation results reveal the locations of vulnerable congestion points that necessitate appropriate measures such as vehicle parking restrictions and building structural reinforcement. These measures are essential for minimizing evacuation time and ensuring the safety and efficiency of the evacuation.

3.3. Time to Evacuation Destination and Accumulated Route

For comparison, this study explored the same time interval; at 20 s, the simulation results revealed risk arising from distance and evacuee congestion. The distance and number of nearby evacuation sites substantially affected evacuation time. The city’s northeast and the outer rim of the western part, which have many evacuees but fewer evacuation sites than other areas, were the same locations for all scenarios. In these areas, the travel time to an evacuation site is greater than 500 s, and in the case of combined blockage, it is greater than 1000 s. These are the most vulnerable locations.

From the comparison of obstruction scenarios, accumulated routes emerged on long-distance routes originating from the same heavy-congestion area in the northeast. The northeastern part of the city, which leads to the east gate of the Lamchang temple, is the most vulnerable. The longest accumulated routes were concentrated in these areas. Longer queues developed from the evacuees’ increased use of time. This indicated that this particular area required resolution or spatial-risk mitigation.

3.4. Evacuation Capacity and Gate Location

The evacuation location capacity and gate flow of the obstruction scenarios are compared as follows. Initially, the vulnerability of the gate in the same location was shared by the first and last evacuees. In terms of the flow of last-in evacuees into the temple, the Lamchang temple was the most vulnerable gate. In each scenario, the east gate of the Lamchang temple produced a distinctively high flow rate, which is affected by the abovementioned indicator display. This gate undoubtedly poses the highest risk of congestion and hazard during an evacuation. According to this comparison, no evacuees entered the north gate of the Dubphai temple. Another important finding was the total number of evacuees in place; many chose to evacuate in certain temples. The temples with the highest evacuee counts were Muen Ngen Kong, Pha Khaw, and Lamchang, with more than 2000 evacuees each. This was another shared feature in all scenarios (dark brown color in Figure 5). According to the findings support Michon and Denis (2001)’s theory that people are likely to choose the place at a decision point for their evacuation destination [35]. The outcomes showed that these temples are situated close to labyrinth urban networks. Evacuees have no choice but to evacuate to those temples in the scenario. Furthermore, these temples also have small areas to accommodate evacuees. Thus, to reduce risk and cope with community evacuation, priority in terms of temple capacity and risk from congestion must be examined and resolved further.

The study’s findings corroborate the availability of historic urban landscape elements as a resource for urban resilience. The simulation demonstrates the absence of immobilized agents or casualties within the environment. This implies that temples and urban streets, as relevant networks, can serve as evacuation routes during disasters. Rather than relying solely on vehicles for evacuation plans in historic cities, the simulation suggests that walking may provide an alternative option. Given the nature of historic cities, streets are often narrow and challenging to navigate for vehicles as an evacuation mode. The use of temples as evacuation sites/destinations plays a crucial role in this study. It highlights the potential of heritage sites as evacuation sites and their dual function during disasters, rather than being merely deadweight that requires support. Local temples as evacuation sites pave the way for community-based disaster risk reduction, fostering collaboration with their environment (cultural and natural/man-made), which is essential for addressing the uncertainties associated with climate change and can be integrated into urban planning [60,97,98]. During the mitigation phase of disaster risk reduction management, the results indicate the risk of overcapacity and congestion in certain streets. The mitigation plan may emphasize the preparation of temple spaces, such as the provision of emergency spaces and/or equipment. Critical streets, such as those near temple entrances, may regulate parking based on vehicle obstructions and reduce congestion at temple entrances. During the preparation phase, the community must actively participate in determining the evacuation plan by including temples as evacuation sites. Evacuation drills involving temples as evacuation sites/shelters should be conducted and understood to address potential challenges. During the recovery phase, the temple can serve as an evacuation site, functioning as a command center for recovery management or as a semi-temporary shelter for evacuees, providing livelihood support during the ongoing recovery and rebuilding phase of the city. The evacuation plan can be adapted for municipal-level implementation as an official evacuation plan, incorporating the disaster risk reduction cycle management theme as previously mentioned. Community-based management also plays a crucial role in the successful implementation of this plan. The labyrinthine urban network of the historic city possesses its own logical structure [68,99], with most streets leading to the temple’s entrance due to the close relationship between the surrounding residents and the temple.

4. Conclusions

The investigation confirmed vulnerable locations and routes, demonstrating the risk through various obstruction scenarios. Similar hazardous locations were identified, and the complex deformed grid network may be at risk during an evacuation. The agent-based simulation highlighted the problem’s major details, necessitating additional solutions. The reinvention of religious places like Buddhist temples has great potential as evacuation sites. Despite the worst obstruction, including a vehicle on the street and building rubble, no individuals were stranded. However, the seclusion of the temples in the labyrinth area poses a risk of congestion and overcapacity, as shown by the obstruction scenarios. These risks can be mitigated by specific solutions based on their location and morphology.

The agent-based simulation proved effective in defining evacuation risks. While the actual situation may be more dangerous due to uncontrollable factors, these simulations highlight potential risks that could be mitigated by using repurposed evacuation locations. They considered alternative evacuation route plans and selected locations based on capacity.

The combined obstruction scenario was clearly the most time-consuming in the simulation results at 386% higher than the no obstruction scenario. The vehicle obstacle illustrated the temporary but immovable obstruction scenario resulting from these notable findings. Travel time for evacuation increased by 205% compared with obstruction due to building rubble, which increased by only 157%. These results highlight the significance of temporary obstructions such as leftover vehicles, which may affect evacuees’ path to the shelters, even though the building rubble obstruction created a dead-end and partial blockage in the evacuation route. Particularly in a historical area, where narrowing streets are one of its characteristics, necessary countermeasures are urgently needed to mitigate the impact of temporary obstruction. Furthermore, while the building alone poses a lower risk during an evacuation, when considering combined obstruction, building density and height restrictions are critical not only for the esthetic value of the historical city but also for its safety aspect. Priority must be placed on the temples’ evacuee capacity to fulfill the mitigation strategy of using temples as shelters. To accommodate the large density of temporary evacuees, temples that can house more than 2000 evacuees, such as Muen Ngen Kong, Pha Khaw, and Lamchang, need supplementary and add-on measures. The reinvention of temples will succeed if the density and additional measures at crucial areas such as routes and evacuation locations are prioritized. According to concept of urban heritage resilience by Ripp et al. [74], to open the approach of using heritage as a crucial resource for disaster reduction especially in emergency response and recovery phases was proven to be possible by using these temples open space as one of the historic area evacuation sites [65]. The utilization of heritage sites as resources for disaster risk reduction aligns with the development of the heritage triple helix model, which civil society is highly essential for the heritage pillar of urban social and cultural life [11]. In response to the European Union’s encouragement to develop a knowledge-based society through collaboration between universities, industry, and government [100,101]. The model incorporates urban heritage as a key element. The model’s questions and goals revolve around how heritage plays a role in enhancing social development and promoting urban innovations while serving as an arena for these processes. Heritage policy-making and public–private partnerships are emphasized as crucial for supporting the model, with heritage being recognized as a valuable resource. The government actively promotes the involvement of civil society through cultural rights as a central framework [11]. To overcome the paradigm of viewing heritage as an obstacle to resilience [102], the model incorporates urban heritage elements into the integration of urban resilience adaptation. This partially manifests in the study, which demonstrates the potential to transform Buddhist temples into evacuation sites or shelters within the historic city of Chiang Mai. This concept can be applied to any religious spaces, particularly in Asian cities characterized by sparse density, such as Chiang Mai. Utilizing the community hub of religious places can positively impact urban resilience adaptation by reducing the cost of constructing evacuation sites or shelters while simultaneously preserving cultural heritage areas.

To obtain further information, future studies may perform a simulation of solutions and provide appropriate measures to mitigate sensitive places through ABS. The simulation parameter could be expanded to include additional types of obstruction, particularly those that pose temporary threats such as electric poles, high fences alongside the route, billboards, and others. An obstruction test using these innovative settings will provide a better overview of evacuation science, which we anticipate would safeguard residents and/or reduce casualties during a crisis. The agent parameter should include various types of mobility-impaired individuals. It could be integrated with the social logic of families living together, which may form clusters of evacuees moving from their homes to evacuation sites.