Applying Systems Thinking Concepts to Major Casualty Fires: Lessons Learned from Taiwan

Abstract

At 2:54 A.M. on 14 October 2021, a devastating fire erupted in a high-rise building in Kaohsiung City, Taiwan, involving 12 floors above ground and a basement level, resulting in 46 fatalities and 41 injuries. The official investigation pinpointed regulatory deficiencies and negligence among relevant department officials. The persistence of major fires globally underscores that merely relying on post-incident investigation reports is insufficient to fully uncover the underlying problems, highlighting the complexity of fire-related systemic challenges. This study adopts a systems thinking approach and synthesizes findings from various sources, including the investigation reports of this fire and the Grenfell Tower fire, research on fatal fires, and literature on high-rise building fires. It examined the systemic issues related to fires from three angles: resident characteristics, building factors, and situational factors. The analysis exposes the deep complexity of fire-related systemic problems and the interconnections among various contributing elements. Comprehensive initiatives that span educational, legislative, policy, and economic domains must be launched to reduce the frequency of fires and enhance survival rates. The insights from this study offer a profound understanding of the fundamental problems associated with fires and aim to inform strategies to prevent similar tragedies in the future.

1. Introduction

On 14 October 2021, at 2:54 A.M., a 13-story structure with 12 stories above ground and one below called the Cheng Chung Cheng Building in Kaohsiung, Taiwan, suffered the deadliest residential fire in Taiwan’s post-war history and the second most severe building fire overall. This catastrophic event claimed 46 fatalities and 41 injuries, drawing significant public attention and prompting widespread discussion across Taiwan [1]. The Kaohsiung City government swiftly established a special investigation team comprising fire protection, construction, and law experts. This team conducted a thorough review over 15 days and, on 29 October 2021, published a detailed 68-page report. The report assessed pre-existing fire safety and building management measures, evaluated emergency response and rescue efforts, and identified critical deficiencies. Furthermore, it proposed amendments to relevant laws and called for the legal and administrative accountability of involved public officials [2].

1.1. The Cheng Chung Cheng Building Fire Disaster

1.1.1. Background of the Cheng Chung Cheng Building

The Cheng Chung Cheng Building, which has been operational since 25 October 1981, comprises 12 floors above ground and one below ground. Initially, its layout included department stores on floors 1 through 4, cinemas on floors 5 and 6, small offices on floors 7 to 11, and a restaurant on the 12th floor (see

Table 1. Original use and the actual use before the fire (each floor).

Floor	Original Legal Use Type (with the Permit by the Kaohsiung Government)	Actual Use Type Before the Fire
12F	Restaurant	Vacant (not for business or residential use)
11F	Office	Divided into suites, the empty house was uninherited and was illegally occupied after old veterans passed away
10F
9F
8F
7F
6F	Movie theater and settlement place of the theater owner	Settlement place of the theater owner
5F	Movie theater	Vacant (not for business or residential use)
4F	Shopping mall
3F	Shopping mall
2F	Shopping mall
1F	Shopping mall	An electric appliance store, antique tea set store; some stores closed and vacant; some wooden compartment-type stores
B1F	Shopping mall	Vacant (not for business or residential use)

). Over time, shifts in population dynamics and commercial centers, along with changing consumer behaviors, led to the closure of the cinema in 1995. Subsequently, the retail spaces frequently changed ownership, and the office levels were progressively converted for residential use. By 2002, the commercial floors of the building ceased their original operations, with floors 7 to 11 repurposed as residential areas. The resulting decline in living conditions on these floors led to lower property values and rents, attracting predominantly economically disadvantaged groups. The ownership structure of the building is notably intricate, consisting of 377 individual owners and 117 co-owners of the land. This complexity in ownership, coupled with most residents not holding property rights, complicates the building’s management and maintenance. Consequently, public facilities within the building have deteriorated significantly, reflecting the challenging dynamics of its property rights and management [2].

1.1.2. Cause of the Fire

On the evening of 13 October 2021, leading into the early hours of 14 October, a critical fire incident unfolded in a rental suite on the first floor of the Cheng Chung Cheng Building. The sequence of events began with two heated disputes between a woman and her boyfriend, after which the boyfriend left the premises. The woman exited the suite at 2:16 a.m., carelessly discarding unextinguished agarwood powder into a trash can. This act ignited the combustible materials within the bin, and the flames quickly spread to nearby furniture and other flammable items. The building’s design exacerbated the situation, as floors 1 to 6, previously commercial spaces, contained numerous flammable materials. Additionally, the ground floor served as a parking area for 59 motorcycles, further fueling the fire. The abundant combustible items intensified the fire’s rapid spread, leading to extremely fierce flames. Compounding the hazard, the smoke from the fire ascended through the building’s three emergency staircases, where missing or dysfunctional safety doors on various floors allowed the smoke to permeate throughout the building. This severe impediment to visibility and air quality drastically hindered the residents’ ability to escape, culminating in a devastating tragedy with numerous casualties (see Figure 1 and Figure 2).

1.1.3. Details of the Investigation Report

The administrative investigation report that was issued by the Kaohsiung government two weeks after the fire identified several factors contributing to the tragedy:

➢

Negligence with unextinguished agarwood: The initial cause of the fire was the careless handling of unextinguished agarwood, which was used as a mosquito repellent. The individual responsible left the building without extinguishing the agarwood, igniting nearby flammable materials.

➢

Inadequate Maintenance of Fire Safety Equipment: The building’s fragmented ownership led to non-compliance with legal standards for using and maintaining fire safety equipment, specifically failing to adhere to Article 9 of Taiwan’s Fire Service Act.

➢

Deficient Management Organization: The building lacked a proper management organization for overseeing fire safety, with existing functions limited to maintaining cleanliness, neglecting necessary fire safety prevention, preparedness, and emergency response as required under the Condominium Administration Act.

➢

Need for Regulatory Amendments: The investigation highlighted gaps in the Fire Service Act, particularly Article 9, which does not require maintenance declarations for fire equipment in vacant areas of composite buildings; and Article 37, which does not ensure timely fire safety inspections in buildings without a management committee.

➢

Oversight by Authority with Jurisdictions: The complexity of property rights within the building hindered adequate fire safety inspections. Fire inspectors did not engage with public works or police to address these challenges, and the Public Works Bureau in Kaohsiung has not promoted public safety in smaller buildings nor facilitated necessary inter-agency coordination with the fire department.

1.2. Literature Review

Numerous studies have examined the fatal mechanisms of residential fires in recent years, highlighting strong correlations with individual vulnerabilities, building characteristics, and socioeconomic conditions. For example, Xiong et al. compared fire survivors with fatal cases and found that the risk of death was significantly higher among individuals who were elderly, lived alone, were under the influence of alcohol or psychotropic substances, were asleep at the time of ignition, or were in the room where the fire originated [3]. Similarly, Brennan emphasized that many fire victims could not escape due to age-related frailty, health limitations, or the absence of smoke alarms [4]. Spearpoint and Hopkin, through spatiotemporal analysis, identified that residential fires frequently begin in kitchens or bedrooms and are more likely to occur at night. These findings underscore the importance of considering risk configurations, particularly in high-density dwellings such as apartment buildings [5]. A 20-year analysis of fire fatalities in Sweden further classified deaths by typical causes, including clothing ignition, fires involving beds or sofas, and technical malfunctions—information that informs targeted prevention strategies [6]. In Taipei, studies have shown that the primary causes of residential fires were electrical failures and discarded cigarette butts, with most fatalities involving older men who succumbed to smoke inhalation [7]. Istre et al., in their study of fire incidents, reported significantly higher casualty rates in low-income neighborhoods, attributing this to the age of the housing stock and the absence of functioning smoke detectors [8,9].

Fires in high-rise buildings present additional complexities due to factors such as fire spread patterns, smoke movement, structural deformation under heat, and evacuation dynamics. Vertical shafts in high-rise structures can accelerate the spread of smoke and flames through chimney and piston effects [10], while combustible exterior insulation materials contribute to rapid facade fire propagation, leading to extensive casualties and property damage. Traditional prescriptive fire codes are insufficient for addressing these scenarios, prompting scholars to advocate for performance-based fire protection designs and comprehensive risk assessments using fire dynamics and structural simulations [11,12]. Evacuation behavior is another critical concern. Ronchi and Nilsson found that age, physical and mental condition, and familiarity with the building significantly influence evacuation decisions and timing [13]. The current “stay-put” strategy, designed to prevent occupants from entering smoke-filled environments, may inadvertently increase mortality when fire and smoke spread rapidly [14]. Although the overall mortality rate in high-rise fires is lower than in low-rise buildings, individual high-rise incidents often result in concentrated fatalities, necessitating an integrated, cross-disciplinary response [15].

Despite growing literature on high-rise fires and associated risks, most research remains focused on risk factors. There is a noticeable gap in understanding such disasters’ systemic and multi-layered origins. Larsen argued that large-scale disasters, as low-probability but high-consequence events, cannot be fully understood through linear cause analysis alone, as this approach overlooks more profound systemic vulnerabilities within environmental, organizational, and human interaction contexts [16]. The Cheng Chung Cheng Building fire in Kaohsiung is a critical example of such a complex incident. It illustrates how multiple contributing factors, though individually recognized in prior literature, interact in a tightly coupled system as both a high-rise fire and a mass-casualty disaster. These factors, including building design, occupant vulnerability, fire spread dynamics, and evacuation behavior, cannot be treated in isolation. Their interdependence underscores the need for a systems-based analytical approach to understand and mitigate such disasters fully. As Cook suggests, institutionalizing reflective learning from past disasters is crucial to converting experience into actionable knowledge [17]. Only through such mechanisms can societies enhance their capacity for prevention and resilience. Moving from linear learning to a systems-based approach represents a critical shift for future disaster risk reduction and policy development.

2. Methodology and Materials

2.1. System Dynamics and System Thinking

System dynamics is a modeling approach rooted in system thinking, which focuses on understanding and addressing problems within the broader context of interconnected systems rather than as isolated events. While systems are a familiar concept—evident in natural ecosystems and human organizations alike—decision makers in fields such as business, policy, and diplomacy often overlook the complex interdependencies between components. As a result, they may respond to surface-level symptoms without considering the broader systemic consequences their actions may trigger. The value of system dynamics lies not in precise numerical forecasting, but in the insights it offers into causal relationships and feedback loops within complex systems. A well-known example is the World3 model, which demonstrates that even with technological progress, the Earth’s resource capacity remains finite and cannot sustain unlimited growth in population and consumption. Sterman argues that the actual utility of such models is not predictive accuracy, but their ability to reshape users’ mental models, fostering more holistic and informed decision-making [18]. Consequently, system dynamics has been widely applied across diverse fields, including supply chain management, education, public health, water resource planning, power systems, and climate change [19].

This study contends that disasters often result not from a single cause, but from systemic failures involving multiple breakdowns across layers of defense, as illustrated by the “Swiss Cheese Model” [20]. When vulnerabilities in different protective layers align, they can lead to catastrophic outcomes. Addressing only the immediate symptoms may offer temporary relief, but without tackling the underlying structural issues, such efforts may inadvertently exacerbate systemic risks [21]. Therefore, this research advocates adopting systems thinking as a foundational analytical framework to explore the interrelated elements within disaster contexts. By identifying root causes and systemic interactions, this approach aims to inform the development of sustainable risk management and prevention strategies.

Systems thinking is a methodical approach that explores the dynamics and interrelationships of complex events through the analysis of varied causal interactions and their transformations. This perspective highlights the interactions among elements within a system and approaches issues from a structural viewpoint. Core to systems thinking are three fundamental components: reinforcing causal loops, balancing causal loops, and time delays. These elements provide a theoretical framework for dissecting complex events [21].

2.1.1. Reinforcing Causal Loops

A reinforcing causal loop drives behavior in a system to consistently strengthen or weaken, establishing a unidirectional and perpetual trend of behavioral changes. As this loop persists, it amplifies the existing behavior, thus creating a self-reinforcing cycle. If left uncontrolled, the cycle can cause behavior to grow exponentially or decline rapidly [22]. For instance, increased investment in safety can lead to better safety consequences, boosting further investment in safety measures (see Figure 3).

2.1.2. Balancing Causal Loops

A balancing causal loop functions differently from a reinforcing causal loop. In this loop, when behavior within a system expands or contracts to a certain threshold, a regulatory force activates to prevent further escalation or decline, stabilizing the system [21]. This type of loop often involves a control mechanism or a strategy for stabilization. An illustrative example is the temperature control system of a home air conditioner. If the indoor temperature strays from a preset value, the air conditioner automatically adjusts its operation to maintain a consistent temperature (see Figure 4).

2.1.3. Time Delays

Time delay is a prevalent phenomenon in systems, representing the lag between an action and its consequences. This delay, often occurring in the interaction among variables, is typically symbolized by a double bar in system diagrams [21]. Time delays can obscure the immediate impacts of decisions or actions, complicating the timely adjustment of strategies [22]. This characteristic is ubiquitous in policymaking and corporate management. For instance, the benefits of investments in security may not be evident until after a significant period has elapsed (see Figure 3).

By employing the systems thinking framework, researchers and decision-makers can effectively discern and comprehend the structure, dynamics, and interrelationships within complex systems. This understanding enables them to accurately intervene and adjust key elements within the system, enhancing its stability and predictability.

2.2. Critical Factors

Learning from disasters through a systems thinking approach typically involves thoroughly analyzing various data sources, including accident investigation reports, relevant records, and media coverage, to pinpoint critical factors. However, given that the investigation process might yield conflicting data, the final official investigation report is often prioritized, while other sources are utilized as supplementary information [24]. In this study, the official report is integrated with relevant scholarly articles on fatal and high-rise building fires, notably including the inquiry report on the Grenfell Tower fire, which shares contextual similarities with the Cheng Chung Cheng Building fire. By comparing and analyzing these two incidents, critical factors are identified. These factors are categorized into three groups: characteristics of the residents, the fire situations, and building-related issues, all summarized in

Table 2. Critical factors related to fire-related fatalities in high-rise buildings.

Group	Critical Factors	Impacts	Reference
Residents’ Characteristics	Gender	Judgment Capacity of Residents	[4,7,8,9,10,11,12,13,14,15,23,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49]
	Income
	Age
	Education
	Socioeconomic Status
	Mental State
	Physiological Condition
	Substance
	Occupation
	Illness
Building-Related	Compartmentation	Flame and Smoke Spread Rate
	Fire Safety Equipment
	Fire Service Act
	Building Technical Regulations
	Property Value
	Building’s Age
	Number of Households
	Number of Floors
	Structure Strength
	Building Height
	Wall Material Combustibility
	Ventilation System
	Decoration Materials Combustibility
	Quantity of Interior Items
	Building Usage Complexity
Fire Situation	Condominium Administration Act	Behavioral Responses of Residents
	Fire Service Act
	Management Committee
	Investment in Fire Management
	Building Maintenance
	Area of Fire Occurrence
	Cause of Fire
	Routine Fire Drills
	Floor of Fire Occurrence
	Time of Low Alertness
	Firefighter Rescue
	Response of Dispatcher

.

Resident Characteristics. Numerous studies on fatal fires have highlighted that the personal characteristics of building residents significantly impact their likelihood of survival during a fire. These characteristics include age, gender, income, socioeconomic status, occupation, education, physical health, state of consciousness, and the use of substances such as alcohol or drugs. These factors critically influence an individual’s decision-making capabilities in the event of a fire, thereby affecting their survival prospects. Survival during a fire often hinges on the residents’ physical control, response, and cognitive abilities to navigate through smoke and flames. Poor judgment during a fire can lead to inappropriate escape actions, which might result in fatalities, even in relatively minor fires.

Building-Related Factors. The impact of a building’s structural design on fire safety is significant. Research indicates that various aspects, such as the building’s structure, interior materials, compartmentalization, the installation and functionality of fire safety equipment, and even the everyday items placed by residents, can influence the behavior and spread of smoke and flames during a fire. These factors collectively affect the residents’ ability to escape and their chances of survival. High-rise buildings pose unique risks during fires when compared to traditional one- or two-family homes and low-rise apartment buildings. The fire at the Cheng Chung Cheng Building exemplifies the challenges associated with high-rise fires. With urban populations growing and the push towards vertical development to maximize living space, high-rise residential buildings have become more common. However, the structural features of these buildings also introduce significant safety concerns. As buildings increase in height, critical elements such as the coverage and functionality of fire safety equipment, the design of smoke ventilation systems, emergency evacuation routes, and the configuration of internal spaces become pivotal to managing fire incidents and facilitating safe evacuations. Since the 21st century, several severe fires in high-rise buildings have resulted in significant casualties, underscoring the heightened risks associated with such structures. Notable examples include the 2010 Jing’an Apartment Fire in Shanghai, China, which resulted in 58 fatalities [50]; the 2017 Grenfell Tower Fire in the UK, claiming 71 lives [33,34]; the 2017 Plasco Building Fire in Tehran, Iran, which killed 30 people [51]; and the 2021 fire at the Cheng Chung Cheng Building in Taiwan, which led to 46 deaths [1]. These incidents highlight that the risk factors in high-rise fires are more complex and numerous, involving variables such as building height, the effectiveness of fire safety equipment, the design of smoke ventilation systems, and the structural integrity of internal compartments.

Fire Situation. Fire situation factors encompass a broader scope than resident characteristics and building-related elements. While characteristic factors pertain mainly to residents’ intrinsic traits and social backgrounds, situational factors focus on the behavioral responses of residents during fire scenarios. These can be divided into two phases: before and after the fire ignites. Before a fire, influential factors include the regularity of fire drills and the day-to-day maintenance and management of the building. These elements are crucial for effective fire prevention and initial response capabilities. During and after a fire, pertinent factors encompass the time and location of the fire, its cause, the actions taken by firefighters, the responses of emergency dispatchers, and the residents’ evacuation behaviors and escape strategies. These aspects directly impact the rate at which a fire spreads and the survival rates of the individuals involved. Thus, situational factors primarily affect the interactive relationship between people and buildings, the dynamic progression of a fire, and the impact of external environmental elements. Factors that do not relate directly to the physical attributes of the building or the inherent characteristics of the residents are categorized as situational factors for analysis.

3. Results and Discussion

3.1. Systems Thinking Model

The analysis of the Cheng Chung Cheng Building fire, drawing on survey reports and relevant literature, identifies key factors categorized into three interrelated aspects: residents’ characteristics, building features, and situational elements. Understanding these factors through a systems thinking perspective gives a more comprehensive grasp of their inherent interactions. In addressing fire safety, two pivotal issues are central: the likelihood of a fire occurring and the potential for residents to escape or survive such an event. This study adopts Marchant’s evacuation theory concepts of Available Safe Egress Time (ASET) and Required Safe Egress Time (RSET) [52].

ASET is defined as when a fire starts to when conditions become life-threatening, marked by the emergence of smoke, heat, and toxic gases. This timeframe is influenced by the psychological effects of the fire on residents, the building’s fire load, compartmentalization settings, and the presence of active fire suppression systems. Extending ASET provides residents with additional time to respond to a fire.

RSET represents the duration necessary for a safe evacuation once a fire is detected or an alarm is activated. This includes the time required to detect the fire and initiate a response. Factors affecting RSET include residents’ familiarity with fire protocols, physical and mental states, the accessibility of escape routes, building compartmentation, and the effectiveness of fire alarm systems. The shorter the RSET, the higher the chances of survival, underscoring the importance of early detection and rapid response capabilities.

The critical determinant for survival in a fire is ensuring that ASET exceeds RSET. The greater this margin, the higher the survival probability. In constructing a proto systems thinking model, “Cheng Chung Cheng Building fire”, this study selects “fire incidence rate”, “ASET”, and “RSET” as the critical stock variables to examine the impact of each key factor (see Figure 5). The three key factors are further derived and discussed in establishing this basic model as subsystems. The resident characteristics subsystem focuses on the residents’ reactions and decision-making abilities; the building-related subsystem covers the influence of building structure and fire safety equipment; and the fire situation subsystem includes the timing of the fire situation, the location of the fire, the rescue actions of the firefighters and the residents, and dynamic factors such as response actions.

3.2. Residents’ Characteristics Subsystem

The resident characteristics subsystem of the Cheng Chung Cheng Building fire analysis focuses on residents’ ability to assess conditions during a fire. This capability is crucially linked to the ASET and RSET, significantly influenced by residents’ mental and physiological states. An optimal mental and physiological state can extend ASET, reduce RSET, and increase the likelihood of survival in a fire. The details are shown in Figure 6.

The operational dynamics within this subsystem are driven by two reinforcing causal loops and one balancing causal loop. These include the R1 physiological condition loop and the R2 mental state loop, which are reinforcing; and the B1 fire incidence rate loop, which is balancing. These loops are affected by personal characteristics such as gender, age, income, occupation, and education level. Collectively, these factors determine the socioeconomic status of the residents, which, in turn, impacts their psychological and physiological states, influencing both the likelihood of fire occurrence and their decision-making capabilities during a fire. Furthermore, when residents successfully escape a fire, this experience can inform and enhance their fire prevention knowledge, thus feeding back to the loop.

Empirical studies consistently show that residents from disadvantaged backgrounds—particularly those who are elderly, with low education levels, or under the influence of alcohol or drugs—are more likely to become fire victims [39,40,41,47]. For instance, Marshall et al. found that residents with impaired judgment due to cognitive decline or intoxication had significantly reduced survival chances during fires [39]. Our model reflects this evidence by illustrating how vulnerable groups face dual threats: reduced responsiveness and an increased likelihood of triggering fire hazards (e.g., hoarding, unsafe practices), aligning with cascading vulnerability.

Regarding mental states and physiological conditions, the survey report indicates that many living in the Cheng Chung Cheng Building were from disadvantaged backgrounds with low socioeconomic status. According to Maslow’s hierarchy of needs [53], individuals who struggle to meet their basic survival needs are less likely to prioritize higher-level security needs. This connection between socioeconomic status and the ability to respond to emergencies like fires suggests that socioeconomic standing and the fulfillment of basic needs are critical factors in the casualties of such incidents. Additionally, the influence of substance abuse, such as alcohol and drugs, on residents’ physical and mental states cannot be overlooked. Intoxication can severely impair judgment and responsiveness in emergencies, significantly increasing the risk of fatalities.

However, the investigation report only noted the generally low-income and poor economic conditions of the residents, and the floor of victims was found without delving into the personal characteristics of the victims, their exact locations at the time of death, or their daily living conditions before the fire (see

Table 3. Distribution of casualties and use of floors.

Floor	Death Toll	Male	Female	Actual Use
12F				Vacant
11F	7	3	4	Suites
10F	9	6	3	Suites
9F	6	4	2	Suites
8F	10	8	2	Suites
7F	11	9	2	Suites
6F	2	0	2	Vacant
5F				Vacant
4F				Vacant
3F				Vacant
2F				Vacant
1F	1	1	0	Mixed-use (store, living area, storage)
Total death toll	46	31	15	−

). A more thorough examination of these individual characteristics and empirical data would provide a fuller understanding of the human factors involved in fire incidents and could inform more targeted policy adjustments in the future.

3.3. Building-Related Subsystem

In the building subsystem, the primary focus is controlling the spread of flames and smoke during a fire. A rapid spread can significantly shorten the ASET and extend the RSET, negatively impacting residents’ chances of survival. This subsystem emphasizes the importance of “Compartmentation, B2”, “Ventilation systems, B3”, “Fire safety equipment, B4”, and “Quantity of interior items, B5”, as depicted in Figure 7. (The notation of <grey words> in the figure is used to clarify the spatial relationships among variables within the system. The <grey words> indicate existing variables that are part of the same system but are shown separately from new variables to enhance the clarity of the diagram.)

Compartmentation is critical in extending the ASET and providing residents with sufficient time to respond during a fire. Its effectiveness determines whether fire and smoke can be confined long enough for safe evacuation. A well-compartmentalized building slows the spread of heat and smoke between units and floors, thus preserving egress routes and enhancing survivability. The effectiveness of compartmentation is shaped by several interacting factors, including the number of residential units, the complexity of internal decorations and furnishings, the building’s age, property value, structural integrity, and the stringency of legal regulations. These factors collectively form a balancing loop (B2): in newer or higher-value buildings, residential density tends to be lower, and fire loads are typically more manageable, thereby supporting more effective compartmentation. However, as buildings age and property values decline, particularly in economically disadvantaged neighborhoods, the number of residents often increases. This leads to overcrowding, more combustible contents, and irregular or obstructed layouts, which reduce the functional integrity of compartment boundaries. These physical and social changes accelerate the spread of fire and smoke and shorten the ASET, reinforcing the need for code revisions and regulatory adaptations. Empirical evidence supports the critical influence of compartmentation on fire survivability. Ronchi and Nilsson demonstrated that ineffective compartmentation in high-rise structures causes smoke to quickly spread into stairwells and corridors, rendering them impassable within minutes [13]. Rein et al. further described how vertical shafts and shared risers in old residential buildings become conduits for smoke when not adequately sealed [12]. These findings directly align with the observed failures in the Cheng Chung Cheng Building, constructed in 1981, before the enforcement of modern compartmentation regulations.

Unlike the Grenfell Tower, the Cheng Chung Cheng Building lacked exterior wall cladding, and the fire primarily propagated vertically through smoke. The official investigation report highlighted that, during the fire, the failure of fire doors between living spaces and emergency staircases allowed smoke from a ground-floor fire to rise through the three stairwells, penetrating the living spaces between the 7th and 11th floors and causing significant casualties, as described in

Table 4. The condition of compartmentation on each floor.

Floor	Compartmentation of the Stairwells
12F	Incomplete (Stairwell A: ○; Stairwell B: ○; Stairwell C: ●)
11F	Incomplete (Stairwell A: ●; Stairwell B: ●; Stairwell C: ●)
10F	Incomplete (Stairwell A: ⊙; Stairwell B: ⊙; Stairwell C: ⊙)
9F	Incomplete (Stairwell A: ⊙; Stairwell B: ⊙; Stairwell C: ⊙)
8F	Incomplete (Stairwell A: ⊙; Stairwell B: ●; Stairwell C: ●)
7F	Incomplete (Stairwell A: ●; Stairwell B: ●; Stairwell C: ●)
6F	Incomplete (Stairwell A: ○; Stairwell B: ⊙; Stairwell C: ○)
5F	Incomplete (Stairwell A: ⊙; Stairwell B: ⊙; Stairwell C: ⊙)
4F	Incomplete (Stairwell A: ⊙; Stairwell B: ●; Stairwell C: ●)
3F	Incomplete (Stairwell A: ⊙; Stairwell B: ○; Stairwell C: ○)
2F	Incomplete (Stairwell A: ●; Stairwell B: ●; Stairwell C: ⊙)
1F	Incomplete (Stairwell A: ●; Stairwell B: ●; Stairwell C: ●)

N.B. There are three types of compartment incompletion. ○: Stairwell fire door closed, fire door cannot open (welded), or a solid wall replaces the existing fire door ●: No stairwell fire door (including part of the fire door being removed); ⊙: hindrance (unable to complete the functionality of fire doors), such as pile-up of items, door panels unstable (or fallen), an improperly installed additional latch hindered the fire door from closing properly.

and Figure 8. Additionally, the absence of independent compartmentation for ducts, due to the building being constructed in 1981 before such regulations were mandated, allowed smoke from lower floors to rise through the ducts and pervade the building. Furthermore, the building’s age of 40 years, coupled with economic decline in the surrounding area leading to decreased property values, led to an increase in socioeconomically disadvantaged residents. These changes have complicated the management of living spaces, decorations, and contents, adversely affected the maintenance of communal facilities, and compromised the integrity of the building’s compartmentation.

In summary, the Cheng Chung Cheng Building illustrates how compartmentation is a structural feature and a systemic vulnerability when degraded by aging infrastructure, weak enforcement, and socioeconomic decline. The observed failure to contain smoke and protect egress paths exemplifies how cumulative structural and demographic stressors can weaken the balancing loop B2.

Ventilation systems are pivotal in managing fire and smoke behavior in high-rise buildings. Their design and capacity directly influence the spread rate of smoke and its expulsion during emergencies. In buildings with significant vertical height, thermal buoyancy effects, called the “chimney effect”, cause hot smoke to rise rapidly through stairwells, atriums, and shaft spaces, often compromising evacuation routes before residents can respond. The interplay between the building’s height, spatial layout, and fire code requirements constitutes a balancing loop (B3). The taller and more complex a building, the more sophisticated its ventilation and smoke extraction systems must be to counterbalance this risk. When such systems are lacking or underperforming, the natural upward flow of smoke is left unchecked, accelerating the degradation of tenable conditions and shortening the ASET.

The Cheng Chung Cheng Building, constructed before mandatory smoke control provisions were implemented, did not have a dedicated fire ventilation or mechanical smoke exhaust system. Its L-shaped configuration and internal atrium design—with long, narrow corridors flanked by units—created a semi-enclosed environment particularly vulnerable to smoke accumulation. As illustrated in Figure 8, smoke rapidly ascended through the atrium and external window wells when the fire ignited on the first floor, reaching the 7th to 11th floors in minutes. The building’s structure and inadequate smoke extraction allowed smoke to infiltrate all three emergency stairwells. This infiltration was especially consequential. The stairwells, which could have functioned as temporary safe refuges or vertical evacuation paths, were rendered unusable. Empirical studies, such as Rein et al., confirm that older buildings without dedicated pressure differential systems or automatic smoke vents are highly susceptible to stairwell smoke penetration [12]. Ronchi and Nilsson further emphasized that once stairwells are lost to smoke, not only are the occupants trapped, but firefighter access is also severely compromised [13]. Because firefighters could not use internal stairwells, all rescue operations had to be carried out externally using aerial ladders—an inherently slower and more hazardous approach. This delayed intervention prevented targeted rescues on the most affected floors.

The absence of a dedicated fire ventilation system disrupted both passive egress routes and active firefighting access. In the context of our system model, B3 represents a critical balancing mechanism whereby increasing building height and complexity are counteracted by increasingly robust ventilation designs. This balance is essential to maintain tenable conditions for evacuation and fire control. Without a functional ventilation system, this compensatory mechanism was disabled, allowing smoke to ascend rapidly and uncontrollably through vertical shafts and stairwells. Consequently, the building’s vulnerabilities were fully exposed, rendering evacuation routes impassable and severely hindering rescue operations.

Fire safety equipment does not prevent fire ignition, but it is critical in extending ASET and reducing RSET once a fire begins. This equipment includes fire alarms, sprinklers, extinguishers, and emergency lighting systems, and functions within a balancing loop (B5). When functioning correctly, these systems serve as an early warning and initial response that moderates fire growth and enhances occupant response time. Automatic fire alarm systems detect heat, smoke, or flames during the early stages of combustion and immediately alert building occupants. This can significantly reduce RSET by prompting faster evacuation. Concurrently, sprinkler systems suppress or contain the fire, limiting its spatial growth and prolonging the tenability of egress routes, thus extending ASET. Studies show that the early activation of alarms and sprinklers significantly reduces casualty risk in residential fires [48,49]. However, the effectiveness of these systems is contingent on proper installation, regular maintenance, and accessibility for inspection, all of which were lacking in the Cheng Chung Cheng Building. The structure’s fragmented ownership, comprising hundreds of individual unit holders, and the absence of a centralized resident management committee resulted in no systematic upkeep of communal fire protection systems. Consequently, the fire alarm system failed to activate during the incident.

This failure had critical consequences. Residents on upper floors, who could have had a crucial head start in evacuating, remained unaware of the fire until smoke and heat reached their units. The delay in recognition and response sharply reduced survivability, especially for those with impaired mobility or no direct line of sight to stairwell entrances. Furthermore, without a functioning alarm system, the RSET was significantly extended beyond safe margins, illustrating a failure in loop B5. The lack of management organization also impeded the fire department’s ability to conduct pre-incident inspections or equipment checks. Reports noted that inspectors were sometimes denied access due to unclear property rights or absent representatives. These operational blind spots highlight that even the most robust technical systems require active institutional support to remain functional over time. This organizational breakdown demonstrates that no matter how well designed, technical solutions cannot operate independently of social and administrative infrastructure. In buildings like Cheng Chung Cheng, where fragmented ownership, resident turnover, and an absent management committee intersect, these social deficits become just as dangerous as missing equipment. This outcome underscores that fire safety in aging and socially complex buildings is not solely a matter of physical design, but a systemic challenge embedded in the broader organizational and social context.

Regarding fire incidence rates, the internal items of a building interact with the ASET to form a balancing loop. As the building’s occupancy increases, so does the quantity of personal belongings, which can significantly impact fire safety. This is particularly evident among residents with poor mental health, who may engage in hoarding behavior, leading to an accumulation of excessive items. Such clutter provides additional fuel for fires, accelerating their spread, decreasing ASET, and affecting the RSET. Excessive items can obstruct passageways, slowing evacuation and complicating rescue operations during emergencies. Therefore, this accumulation of belongings poses a dual threat by rapidly fueling fires and hindering timely escape and rescue efforts.

3.4. Fire Situation

The fire situation subsystem in fire safety analysis examines residents’ preparatory behaviors before a fire and their responsive actions after ignition, exploring how these behaviors impact outcomes.

Pre-fire behavior: In the Cheng Chung Cheng Building fire, the residents’ pre-fire behaviors critically influenced the system’s vulnerability to a significant incident. According to the administrative investigation report [2], the building lacked a management committee and had no records of organized fire drills before the fire, nor any visible evacuation instructions within common areas. The absence of systematic fire prevention activities weakened residents’ awareness and preparedness, particularly in recognizing fire hazards and executing timely evacuation. This deficiency in pre-incident fire management formed a balancing causal loop “B6”, where reduced fire incidents over time led to decreased safety vigilance among residents and building managers, as shown in Figure 9. This decline in vigilance contributed to the absence of routine drills and safety education, further lowering residents’ readiness and reinforcing the system’s fragility [20,23]. However, a particularly alarming observation from the official report is that the Cheng Chung Cheng Building had experienced over ten smaller-scale fire incidents in the years leading up to the 2021 disaster [2]. These included electrical fires, smoke incidents in commercial units, and unattended stove flare-ups. Despite this warning signal, the building lacked a functional management organization to coordinate follow-up actions or improve preventive infrastructure. This absence of institutional response mechanisms meant that even when fire events occurred, they failed to activate the B6 loop effectively. There was no actual increase in perceived risk, safety education, or structural investments in fire preparedness. As a result, the potential for pre-incident learning and risk mitigation was lost. This led to a sustained state of vulnerability in which residents, despite recurring fire events, remained unprepared and unfamiliar with evacuation protocols. During the 2021 fire, many victims were found on floors 7F to 11F (see Table 3), indicating that they either failed to perceive the imminent threat or could not act on it effectively. Occurring at 2:54 A.M., a time associated with low cognitive alertness, the combination of poor preparedness and delayed response critically extended the RSET, leading to tragic outcomes. The system dynamic interpretation is thus twofold: not only was the B6 safety vigilance loop weak, but it also failed to engage despite multiple trigger events, underscoring the depth of systemic inertia in this high-risk, under-managed environment.

Post-fire behavior: The analysis of post-fire behavior considers initial fire details such as cause, location, and timing, which influence the fire’s development and spread. For instance, fires during early morning hours often catch residents off-guard, delaying response times and hindering evacuation efforts. Additionally, external responders, including firefighters and emergency dispatchers, are crucial. The official report on the Cheng Chung Cheng fire focuses mainly on the decisions made by fire commanders, with less emphasis placed on the actions of the first responding firefighters or the detailed processes of emergency dispatchers, which contrasts with the thorough documentation found in the Grenfell Tower investigation.

In Taiwan, emergency dispatchers instruct callers on immediate actions during fire reports, such as alerting others or attempting to escape. A “stay-put” strategy may be advised when heavy smoke or entrapment occurs. However, this strategy’s suitability often depends on the specific conditions of the fire, suggesting that it should not be a one-size-fits-all recommendation [14]. The Cheng Chung Cheng report lacks a thorough discussion on the appropriateness of the stay-put strategy and the critical interaction between dispatchers and on-site rescuers. This type of interaction, involving information exchange among callers, dispatch personnel, and rescuers, is critical to the effectiveness of rescue operations. While this process was well documented in the Grenfell Tower report, it was notably missing from the Cheng Chung Cheng investigation report. A comprehensive record of this interaction would enable a deeper analysis of the information needed during a fire, which could help refine dispatch strategies, rescue operations, and fire prevention education policies.

3.5. Combined System

Integrating the subsystems of situational factors and resident characteristics and buildings uncovers significant interrelations and mutual influences among these elements. This synthesis demonstrates how complex system operations often involve multiple interconnected factors whose interactions can precipitate significant outcomes [54]. These dynamics introduce new feedback loops and increase the complexity of existing ones.

Investments in fire safety, including the judgment capabilities within the resident characteristics subsystem, compartmentation, fire safety equipment, and ventilation systems, are organized into three balancing loops: B7, B8, and B9. The quantity of interior items connected with the mental state into a balancing loop, B10, as shown in Figure 10. These loops are further influenced by socioeconomic status and education levels. The integrated system reveals that enhancing overall fire safety is contingent upon comprehensive fire safety investments. This encompasses regulatory requirements for fire safety equipment, the formation of building management committees, and factors such as residents’ characteristics and educational levels, all of which significantly influence survival probabilities during a fire. Additional factors include the timing of the fire, its cause, and external or situational elements such as firefighter response.

The Cheng Chung Cheng Building fire was triggered by a resident’s negligent disposal of unextinguished agarwood into a trash can. The act ignited 59 motorcycles parked nearby, producing vast amounts of dense smoke that spread rapidly through unprotected stairwells, the internal atrium, and ducts, quickly engulfing the building. The malfunctioning fire safety equipment failed to alert residents promptly, significantly delaying initial response times. Furthermore, the absence of regular fire drills left residents unprepared, impairing their ability to respond effectively and evacuate swiftly, leading to numerous casualties. This study underscores the complex interdependency of various factors, including residents’ behaviors, building design, and the effectiveness of safety equipment and support systems during emergencies. The compounded effects of these elements culminated in a large-scale disaster, emphasizing the profound impact that the interactions among different subsystems have on fire incidents.

It is, therefore, important to revisit the initial discussions, evaluate the issues and causes identified in official investigation reports, and assess whether they truly address the core of the problems and whether the remedies proposed are adequate or merely address symptoms.

Cause of the fire: The official investigation into the Cheng Chung Cheng Building fire attributed the cause to a tenant’s careless disposal of unextinguished agarwood on the first floor. This individual was subsequently sentenced to life imprisonment on 25 January 2024. While this outcome suggests a degree of closure, it merely skims the surface of the disaster. The investigation did not delve into critical aspects such as the tenant’s awareness of the risks associated with their actions or their life experiences and educational background that might inform their understanding of such dangers. This oversight indicates a reliance on punitive measures rather than a comprehensive approach to learning and mitigating the root causes of the disaster. Merely focusing on arrest and sentencing fails to educate the public about the underlying issues or to enhance fire safety education and social policy. This approach reflects a significant gap in systemic analysis, highlighting a missed opportunity to derive meaningful insights for developing fire safety education and behavioral interventions from the disaster. A more thorough investigation could have provided valuable lessons on preventing similar tragedies by addressing the symptoms and the systemic factors contributing to such events.

Issues in establishing a building management committee: The causal loop related to investments in fire safety underscores the importance of a robust building management committee to oversee and manage communal affairs effectively. The investigation into the Cheng Chung Cheng Building fire highlighted the absence of such a committee as a contributing factor to the disaster, prompting legislative changes on 11 May 2022 [55]. These amendments mandate the establishment of management committees in all apartment buildings. However, the effectiveness of this legislation remains uncertain. It primarily emphasizes “enforcement” and “penalties”, lacking supportive policies that could enhance its impact. The systemic analysis conducted in this study reveals that the formation of a management committee and the efficacy of fire safety measures are intricately linked to factors such as the residents’ socioeconomic status, educational background, occupation, and income. In the case of the Cheng Chung Cheng Building, most residents were tenants, not property owners, which likely diminished the owners’ commitment to proactive fire safety management. Implementing legislative amendments without addressing the underlying systemic issues will not resolve the core problems. A comprehensive approach is necessary to enhance fire safety and ensure the effective operation of building management committees. This should include educational initiatives to raise awareness about fire safety, incentive mechanisms to encourage active participation from all stakeholders, and policy support to empower management committees.

Compartmentation and fire safety equipment deficiencies: The investigation report into the Cheng Chung Cheng Building fire thoroughly addresses the compartmentation and fire safety equipment issues. It identifies the fact that the building’s structural and safety problems can be traced back to the 1981 building regulations, applicable when it was initially constructed primarily for commercial use. Floors 7 to 11 were later converted for residential use, but these modifications did not meet the legal standards required to trigger oversight by building management authorities. Consequently, the compartmentation and facility design of the converted areas failed to comply with residential standards. The report also notes that damaged safety doors and compartmentation failures significantly influenced the spread of the fire and compromised the residents’ escape routes. Additionally, in mixed-use buildings, ensuring the maintenance of fire safety equipment in unused spaces presents significant challenges. In response, the government revised the fire safety regulations on 11 May 2022 to mandate that maintenance obligations for fire safety equipment can be waived only if an entire building is out of use [56]. Although the report effectively outlines the compartmentation issues and the lapses in building management and fire department enforcement, it stops short of suggesting concrete measures for improvement. Instead, it recommends punitive actions against personnel who fail to enforce regulations effectively, which addresses only the problem’s superficial aspects. This focus on penalizing enforcement failures does not tackle the root issues, such as the applicability of outdated regulations or the need for policy enhancements. Moreover, while the policy revisions address the maintenance of fire safety equipment, they overlook the broader implications of building compartmentation evolution and its impact on residential safety over time.

This study applied systems thinking to examine the Cheng Chung Cheng Building fire and identified three interrelated subsystems: resident characteristics, building factors, and situational conditions. Among the key findings, one crucial contribution is the demonstration that systemic fragility can persist and accumulate even in prior warning events, such as more than ten minor fire incidents, if institutional mechanisms for learning and adaptation are absent. Compared to existing literature, this study’s findings are consistent with past quantitative research emphasizing the role of socioeconomic status in shaping residents’ vulnerability during fire events. For example, studies by Brennan [4], Runyan et al. [41], and Warda et al. [44] have shown that residents with lower income, education, and health status face significantly higher risks of fire-related fatalities. Our case study reaffirms this relationship, as most victims in the Cheng Chung Cheng fire came from economically disadvantaged backgrounds, lacking the physical and cognitive capacity for timely evacuation. However, this study also reveals distinctive differences from international cases, especially in the role of informal institutional arrangements. Unlike the Grenfell Tower case in the UK, where failures were predominantly attributed to policy oversight and regulatory capture [33,34], the Cheng Chung Cheng Building case illustrates how the absence of formal building management organizations in most of Taiwan’s old condominium governance structure plays a critical role in system failure. The decentralized ownership, weak enforcement of the Condominium Administration Act, and lack of coordinated fire management created conditions where even repeated minor fires failed to stimulate corrective action. This phenomenon reflects a structural gap not widely discussed in fire safety literature, other studies, or cases. Moreover, while the concept of “learning from incidents” is emphasized in safety literature (e.g., Reason, [20]; Dekker, [54]), this study demonstrates that such learning is not automatic. Without organized feedback loops or mandated communal governance, the systemic loop, such as B6 (Safety Vigilance), may be disrupted or never triggered, resulting in repeated failure accumulation.

3.6. Recommendations

Based on the systems thinking model integrating resident characteristics, building-related factors, and fire situations, and combined with the results of the official investigation report, we propose multi-level policy recommendations to reduce fire-related casualties in high-rise buildings. These recommendations target the root causes identified through causal loops and empirical evidence, aiming to extend the ASET, shorten the RSET, and mitigate fire occurrence risks (see Figure 11).

• Resident-Level Interventions:
  • Targeted risk communication campaigns: Develop customized fire prevention education for socioeconomically vulnerable groups (e.g., people who are elderly, on a low-income, or use substances) using accessible media formats (e.g., community radio, social media, local dialects, and home visits).
  • Regular health and mental state screenings in aging or public housing communities to identify and support residents with limited judgment capacity or physical mobility, particularly those living alone.
  • Community-based hoarding intervention teams, coordinated through public health and housing departments, to reduce excessive fire loads in residential units.
• Building-Level Interventions:
  • Mandate periodic audits of compartmentation, smoke extraction systems, and escape route signage in mixed-use and aging high-rise buildings. The results should be made publicly accessible to promote transparency and accountability.
  • Subsidize fire safety retrofits (e.g., fire doors, automatic smoke vents, pressurized stairwells, alarm systems) through a government-led incentive scheme targeted at low-income or collectively owned buildings.
  • Incentivize the establishment of active management committees under the Condominium Administration Act, with legal responsibilities for fire safety maintenance, emergency drills, and equipment upkeep.
• Emergency Response and Coordination:
  • Integrate stay-or-evacuate decision trees into 119 command center systems using building-specific data (e.g., sprinkler status, fire load level), supported by AI-based decision support tools.
  • Formalize dispatcher–resident interaction protocols, ensuring early and accurate guidance. Lessons from cases like the Grenfell Tower fire should inform protocols.
  • Conduct inter-agency scenario drills, including dispatchers, firefighters, and community leaders, simulating high-casualty fire incidents in dense residential environments.
• Cross-System Monitoring and Data Transparency:
  • Establish a centralized high-rise fire risk registry, integrating building characteristics, incident history, inspection records, and resident vulnerability indices. This can guide resource allocation and emergency preparedness.
  • Require post-incident data-sharing protocols among fire departments, urban planning authorities, and academic institutions to enable feedback-driven policy adjustment and modeling to refine recommendations that aim to address the direct symptoms of fire-related issues and intervene at systemic issue points identified through our model. This way, more resilient fire safety ecosystem can be developed by bridging individual behavior, infrastructure quality, and institutional responses.

4. Conclusions

This study implemented a systems thinking model to examine the intricate relationships between fire occurrences, resident characteristics, building factors, and the dynamics between humans and their situations during fires. This approach helped deconstruct the complex factors influencing fire systems and unveiled numerous systemic issues. These issues encompass the needs of vulnerable groups, the influence of resident characteristics, the variability inherent in high-rise building fires, and the high uncertainty of external factors. Analyzing these interconnected elements aimed to penetrate deeper into the core of disaster-related problems and develop more substantive solutions.

Following disasters, public responses often veer towards “witch hunts” and “shallow thinking”, where immediate coping strategies are proposed and idealistic future blueprints are envisioned. Such reactions typically address only the symptoms of issues without tackling their fundamental causes. While each disaster is undoubtedly tragic, it also presents opportunities for societal learning, problem-solving, and progress. Building a more comprehensive fire safety framework requires legislative amendments, enhanced enforcement by government agencies, and robust participation from the broader community. The effective resolution of these issues demands collaborative efforts from all stakeholders to prevent future major fire disasters.