Systematic Cause Analysis of an Explosion Accident During the Packaging of Dangerous Goods

Abstract

Chemical plants inherently handle and operate with a wide range of hazardous materials, making them more prone to accidents compared to other industrial sectors. Consequently, safety management in chemical plants tends to be systematically organized based on elements of process safety management (PSM) systems. In June 2023, South Korea’s Ministry of Employment and Labor released the Serious Injury and Fatality (SIF) report, which summarized 4432 major accident cases that occurred over six years (2016–2021), including 1834 cases in manufacturing and related industries and 2574 cases in construction. The report provided an overview of these accidents, their causes, and measures to prevent their recurrence, with a focus on fatalities and severe injuries associated with critical losses across different industries. This study examined 16 accident cases that occurred at PSM-regulated facilities, which are managed on the basis of a systematic safety framework established by regulatory requirements. Among these, particular attention was paid to an explosion accident in the organic catalyst packaging process at a facility with no prior accident history and exhibiting unique accident characteristics. A systemic root cause analysis was conducted using the barrier-based systemic cause analysis technique (BSCAT) and the system theoretic accident model and process (STAMP-CAST) methodologies. The systemic analysis highlighted the critical importance of clearly identifying materials or factors that may inadvertently mix during the process design or mass production phases and evaluating whether such interactions could lead to accidents during the hazard assessment stage. Beyond incorporating the risk mitigation measures identified in the process design and procedural development phases without omissions, it is essential to periodically conduct “worker-centered risk assessments”. These assessments help evaluate the potential for accidents resulting from human errors, such as workers’ non-compliance with established procedures, which is a key aspect of preventing chemical accidents. Although this study did not include an evaluation of the impacts of high pressures or high temperatures on workers near chemical accident sites—hence, no specific recommendations regarding safe working distances are made—the findings are expected to contribute to the development of preventive measures for chemical accidents in smaller-scale plants where workers directly manage and operate processes.

1. Introduction

The advancement of science and technology, coupled with the expansion of chemical plants, has led to the production of a wide variety of chemical substances. These substances are now integral to the production of essential goods, driving a transformation of industrial structures. As a result, the role of chemicals in human life has significantly increased, leading to a growing dependence on these substances. Consequently, the global production of chemicals has exceeded 400 million tons annually and continues to rise each year [1,2,3]. Thus, the careless/unsafe production, use, storage, and transportation of chemical substances may result in accidents such as fires and explosions. In particular, fires and explosions can negatively affect human health, and in extreme cases, lead to large-scale fatal accidents [4,5]. Chemical substances with properties such as flammability, reactivity, and toxicity are often regarded as potential threats to human life and the natural environment, as evidenced by catastrophic chemical accidents like the 2015 Tianjin Port explosion and the 2020 Beirut Port explosion, which caused massive loss of life and property damage [6,7,8,9]. Chemical accidents, compared to general industrial accidents, often result in larger-scale accidents and pose significant risks to public safety. Consequently, various measures are implemented to maintain and operate the chemical industry safely, including the establishment and enforcement of safety protocols based on the hazards of chemical substances, as well as the development of systematic chemical accident prevention standards and the provision of incentives. However, most countries focus their safety policies on strengthening systems and regulations related to chemical safety management to prevent chemical accidents effectively [10,11,12].

Process safety management (PSM) is the most representative system for chemical safety management. It encompasses documented safe operating procedures, training programs for employees, including operators and management, measures to ensure the mechanical integrity of equipment, process hazard analyses, and overall process safety management practices. In the United States, to enhance PSM, additional systems are implemented under the Environmental Protection Agency (EPA), including the off-site consequence analysis (OCA) for chemical accident simulations and the risk management plan (RMP) program for hazard analysis and emergency response planning [13].

In South Korea, the 2012 hydrofluoric acid accident in Gumi highlighted that chemical accidents not only impact the internal environment and personnel within the plant but also cause severe contamination of the surrounding environment and significant harm to nearby residents [14]. As a result, a new chemical safety management law, similar to the U.S. off-site consequence analysis (OCA) and risk management plan (RMP) programs, was enacted and implemented. However, this legislation has faced some criticism for imposing excessive administrative regulations [15]. Despite the strengthening of various regulations and legal frameworks, chemical accidents continue to occur, primarily because most accidents within plants are closely related to human error. Consequently, accident investigation researchers recognize human error and accident models as the most critical factors for the stable operation of systems. As organizational activities have grown increasingly complex compared to the past, researchers emphasize that preventing accidents within organizations requires not only sequential analyses of the root causes from various perspectives but also comprehensive and integrated systemic analyses that consider social, technical, and organizational factors [16,17,18,19].

There are various systems thinking analyses and investigation techniques (Figure 1), and they are generally categorized into three generations based on their thinking model [20,21].

The first-generation accident survey model was based on linear cause–effect models, such as the well-known domino, FTA and SCAT models. In this model, accidents occur in a linear causal relationship with the root cause of the accident [22]. Accordingly, the model demonstrates that accidents can be prevented by eliminating only their root causes. However, as more accidents occurred later, it became necessary to expand the human and organizational factors and identify potential accident factors; thus, the second-generation accident investigation model was created in the 1990s.

The second-generation model is known as the Swiss cheese model of reason, which is based on an epidemiological model. In the first-generation accident investigation model, when the cause of the accident was identified as human error, no further accident analysis was conducted. However, the second-generation model included the behavioral factors of the accident victims and the management problems in the organization. The second-generation model was more effective than the first in investigating accidents caused by various and complex factors, but the second-generation model also assumed a linear causal relationship, which limited its ability to clearly analyze accidents caused by complex interactions [23].

The third-generation accident investigation model is a systemic investigation technique based on the systemic model, in which accidents are caused by emergent properties, where various factors in an organization interact in an unexpected and nonlinear manner [20,24,25,26]. Of these systematic accident investigation techniques, AcciMap was applied in the 1990s, and the system theoretic accident model and process (STAMP), CREAM, and functional resonance analysis method (FRAM) have been used for accident investigations and research since the 2000s [21].

The system theoretic accident model and process (STAMP) model is an analysis method released in 2004 to overcome the limitations of the sequential accident analysis of the first- and second-generation accident investigation models [27]. This model consists of a hierarchy of control structures for each stage, including system development and operation structures. The upper structure of the model determines the safety-related policies, internal standards, and procedures, and the lower structure includes procedures or controls that must be operated to realize the policies and procedures [16,17,18,19]. In the system theoretic accident model and process (STAMP) framework, safety is viewed not as a problem of preventing failures but as an issue of control. The constraints in this context may include environmental or financial conditions, rules, procedures, equipment, and technical designs. Causal analysis based on systems theory (CAST) is a structured technique for causality analysis based on systems theory, focusing on analyzing accident causation from a systemic perspective.

The STAMP itself is not an analytical method but rather a set of models or assumptions about how accidents occur. On the other hand, CAST is a causality analysis method grounded in systems theory that applies the principles of the STAMP [28]. The STAMP is a top-down analysis model, which has the advantage of enabling various analyses of causes related to accidents caused by complex factors by including the software, people, organization, procedures, and safety culture as the main factors for accident occurrence even in complex structures [29,30]. Due to these characteristics, the STAMP has been applied to analyze various types of accidents. In this study, the STAMP-CAST was also utilized as the accident investigation methodology [31].

The FRAM models the nonlinear interaction of system functions during the accident process based on the premise that accidents occur when normal variable factors in the system form unexpected combinations. Although it is a functional model that helps analysts understand the functionality of the entire system, it is also applied to safety models because such models can be used for accident analysis and risk assessment. The FRAM is based on the concepts and principles of Safety II and resilience engineering, a new safety paradigm [25,32]. The purpose of the functional resonance analysis method (FRAM) is to provide a concise and systematic description of normal work processes that occur under typical conditions. As a methodological framework, the FRAM is based on four principles: the equivalence of success and failure principle, the principle of approximate adjustments, the principle of emergent outcomes, and the principle of functional resonance in the relationships between functions. Using the FRAM allows for the representation of a system’s routine operational conditions without assuming any specific accident model. It also demonstrates the advantage of showing how the same type of accident can arise in different forms from various causes, highlighting the complexity and variability of accident causation [33].

The AcciMap model, introduced in 1997, is also a representative accident analysis framework. In this model, safety is regarded as an emergent property resulting from the interactions between different layers or levels of the system. Therefore, when an accident occurs, it is essential to examine whether there were issues in the interactions across the entire system [34]. In 1998, Hollnagel developed the cognitive reliability and error analysis method (CREAM), which proposed that factors such as the performance conditions, organizational suitability, working conditions, availability of work procedures and plans, work schedules, and appropriateness of training and experience interact with one another, influencing human error. These human errors can, in turn, be linked to accidents [35].

Various models can be used for accident investigation depending on their characteristics. Of the 216 accident investigation reports written in the past 20 years, 63 were conducted using accident investigation methods, including the STAMP, AcciMap, HFACS, FRAM, and FTA [36]. The STAMP is the most commonly used model for systemic accident investigations, as it is useful for analyzing accidents caused by complex systems and the relationships between stakeholders involved in the decision-making process. The FRAM is useful for job analysis, function assignment, accident analysis, hazard/risk assessment, and impact analysis of changes such as design alterations and improvements [37]. In South Korea, various systematic accident investigation studies have been conducted using the STAMP model. This technique is mainly used to determine organizational and systematic recurrence prevention measures by analyzing the lack of legal obligation measures, incorrect decisions and defects, and policy errors for each function of the government, original contractors, suppliers, production departments, workers, production processes, and facilities involved in accidents related to chemical handling in plants [38].

In this study, based on the technical analysis results of fire and explosion accidents that occurred during the packaging of organic catalysts in a chemical plant, we further analyzed the causes and expansion factors of accidents from a barrier perspective using the barrier-based systematic cause analysis technique (BSCAT), an accident investigation technique provided by DNV. The BSCAT is based on the systematic cause analysis technique (SCAT), an accident investigation technique introduced by Dr. Bird in 1985. This accident investigation technique focuses on identifying the basic/root cause of accidents and control areas for improvement actions based on the direct cause identified by the accident investigator [20]. The BSCAT analyzes accidents based on the barriers involved in preventing the spread of chemical accidents in factories with complexities among various causes and factors related to their occurrence, and the barrier is the concept of containment of threat factors [39].

The material involved in this accident investigation is an organic catalyst, classified under South Korea’s Hazardous Materials Act as a Class 3 hazardous material (spontaneously combustible substances and water-reactive substances). In South Korea, a total of 104 fires caused by this type of material occurred between 2013 and 2017 [40]. In Japan, there were over 12 reported accidents involving alkyl aluminum, a Class 3 hazardous material, between 2000 and 2016. Most of these accidents were related to the handling of the substance and occurred within manufacturing facilities due to internal leaks or during packaging, leading to external leaks [41]. Given the high hazard level of the materials handled and the clear recognition that accidents can be triggered by minor handling negligence, similar accidents continue to occur. In response to this issue, this study employed the barrier-based systemic cause analysis technique (BSCAT) as a tool for developing preventive measures and post-accident response strategies. Particular effort was made to specifically identify the barriers involved in mitigating and addressing the threat factors associated with the accident [39]. Additionally, along with technical analysis involving the recreation of the accident scenario and BSCAT analysis, the STAMP-CAST model was employed to systematically analyze various interrelationships among factors such as safety policies, internal standards, procedures, personnel, and organizational structure related to the organic catalyst explosion accident [42]. The objective of this study was to utilize multiple analytical models in an integrated manner to clearly identify areas for improvement in the relationships and functions among relevant departments during the stages of process research, process establishment, hazard assessment, and work procedure development. Based on these findings, this study aimed to comprehensively establish effective recurrence prevention measures, ensuring that accidents similar to or identical to this accident do not reoccur again.

2. Literature Review

A comprehensive accident analysis was conducted for an explosion accident in a chemical plant using the STAMP-CAST and BSCAT methods. While systematic analysis of the root causes of chemical accidents is critical, it is equally important to investigate the reasons behind the escalation of a minor event into a major accident. This necessitates a detailed analysis from the “perspective of Barrier which prevents the expansion of the accident”. To address this, two accident investigation methodologies were applied in this study. The BSCAT method, as illustrated in Figure 5, is a tool for analyzing the functionality and failure factors of the barrier designed to prevent accidents and contain their escalation [39]. In chemical plants, systems are designed to operate with both accident prevention and escalation mitigation mechanisms in place. As a result, minor events typically remain contained without progressing into major accidents. However, in cases where escalation mitigation systems are absent or fail, minor events can lead to catastrophic accidents. Thus, in addition to systemically analyzing the root causes of accidents, it is essential to clearly identify the factors that allow minor events to escalate into major accidents. This study utilized the STAMP-CAST method to analyze the root causes of the accident and the BSCAT method to investigate the factors that led to the escalation of a minor event into a major explosion.

2.1. STAMP Model

As discussed in Section 1, the STAMP serves as a model for accident analysis [31]. Leveson et al. (2002) [16] applied the STAMP to analyze accidental friendly fire injuries in the military, demonstrating that the model could address software, human decision-making, organizational, and social factors while identifying new types of hazard analyses [43]. In 2003, the STAMP was applied to analyze water pollution accidents, revealing its utility in providing information on organizational structures, engineering designs, manufacturing processes, and operational changes necessary for accident prevention [44]. Ferjencik (2011) adopted the STAMP to systematically analyze a major accident at Semtin’s explosive manufacturing plant. This analysis involved identifying external factors affecting safety management, such as the media, regulatory bodies, unions, parliament, and legislatures, followed by an investigation of additional root causes [45]. Altabbakh et al. (2014) validated the practicality and reliability of the STAMP in the oil and gas industry through case studies, concluding that the model surpasses traditional hazard analysis methods such as hazard and operability analysis (HAZOP) [46]. Thus, the STAMP has been validated as an effective tool applicable across various industries, including the military, public health, and chemical plants.

2.2. STPA Method

The system theoretic process analysis (STPA) is a hazard identification and analysis method based on the system theoretic accident model and process (STAMP), enabling improvements in system safety during development, operation, and maintenance [47,48].

The STPA has been widely applied to address safety issues in the aerospace industry. For example, Allison et al. (2017) utilized the STPA to analyze the rapid decompression event of an aircraft and identify the primary causes of the Helios 522 accident. This demonstrated the STPA’s ability to identify key safety factors related to individuals, organizations, and technologies [49]. In chemical plants, chemical leaks and explosions can result in significant harm to workers and the environment. Rodríguez et al. (2016) and Yousefi et al. (2019) applied the STPA to chemical plants, showing that the STPA offers a novel approach to hazard management in the chemical and natural gas industries, either as a supplement to or replacement for hazard and operability analysis (HAZOP) [50,51]. In the transportation sector, the STPA has also been widely applied. Read et al. (2019) used the STPA to establish Australia’s railway transport control system. Their research demonstrated the applicability of the STAMP model and related methods such as CAST for accident analysis and the STPA for hazard analysis, showcasing their versatile use in the railway industry [52].

2.3. CAST Method

The STAMP provides a theoretical framework for system safety, while its derived accident investigation method, causal analysis based on systems theory (CAST), is used to analyze and resolve cause-and-effect failures in non-dynamic systems that lead to unforeseen or hazardous conditions [53]. CAST offers a framework for reviewing entire processes related to an accident, identifying system flaws, and pinpointing control deficiencies. This allows investigators to clearly understand the root causes of an accident, derive the lessons learned, and propose necessary system improvements to prevent future accidents [51]. Puisa et al. (2019) applied CAST to analyze the Le Boreal cruise ship engine room fire, identifying potential flaws in maritime safety management systems and recommending comprehensive improvements to enhance the overall maritime safety management [54]. CAST has also been extensively applied in the petrochemical industry’s transportation and pipeline sectors. Gong et al. (2018) investigated the causes of the Donghuang oil transportation pipeline leakage and explosion in China and applied CAST for accident analysis in the long-distance oil pipeline transportation industry [55]. Similarly, Li et al. (2020) used CAST to analyze a gas explosion in an underground pipeline in Taiwan. Their findings confirmed that CAST is effective for identifying accident causes in pipeline systems and establishing reliable safety management systems [56].

3. Materials and Methods

3.1. Materials

A comprehensive analysis was conducted on a dataset comprising 4432 severe accidents identified over six years from 2016 to 2021. These data were extracted from the Serious Injury and Fatality (SIF) report that was publicly released by the Ministry of Employment and Labor in South Korea in June 2023. The dataset included 1834 cases related to manufacturing and associated industries and 2574 cases associated with the construction sector. The initial phase of the analysis involved categorizing the accidents associated with the handling of specific hazardous materials, namely explosive, water-reactive, spontaneously ignitable liquid/solid, and flammable liquid materials. Following this categorization, a subset of 62 cases in which the causative agent was identified as a flammable material was further delineated for a more detailed investigation. Among these 62 cases, 16 occurred within chemical and rubber product, active pharmaceutical ingredient, and petroleum product manufacturing plants. These industries fall under the jurisdiction of process safety management (PSM) workplaces, thus warranting an additional layer of scrutiny. An in-depth examination of these 16 cases was subsequently undertaken to gain insight into the summaries of the accidents and their underlying causes, the findings of which are presented in

Table 1. History of accidents involving hazardous materials in process safety management (PSM) plants in South Korea (2016–2021).

No.	Industries	Accident Date	Process Categorization	Disaster
1	Pharmaceutical raw materials, petroleum products manufacturing	January 2016	Fire in a cleaning process of a raw material concentrator	Death
2	Pharmaceutical raw materials, petroleum products manufacturing	January 2016	Fire in the process of feeding raw materials into a reactor	Death
3	Chemical and rubber products manufacturing	July 2016	Fire from a chemical leak in a reactor	Death
4	Chemical and rubber products manufacturing	October 2017	Dust explosion in a facility during pre-operational inspection	Death
5	Chemical and rubber products manufacturing	November 2018	Fire in the process of adding filler to the mixer	Death
6	Chemical and rubber products manufacturing	August 2018	Fire caused by oil vapor in the process of feeding raw materials into an agitator	Serious injury
7	Chemical and rubber products manufacturing	November 2019	Fire during a chemical splay operation	Serious injury
8	Chemical and rubber products manufacturing	December 2019	Explosion during a reactor internal cleaning	Death
9	Pharmaceutical raw materials, petroleum products manufacturing	May 2019	Fire during filtration of chemicals	Serious injury
10	Pharmaceutical raw materials, petroleum products manufacturing	February 2020	Explosion during a reactor internal cleaning	Serious injury
11	Chemical and rubber products manufacturing	March 2020	Explosion during welding on top of reactor	Death
12	Chemical and rubber products manufacturing	March 2020	Fire while cutting piping connected to a storage tank	Death
13	Pharmaceutical raw materials, petroleum products manufacturing	April 2020	Fire during powder feeding operation	Serious injury
14	Chemical and rubber products manufacturing	May 2020	Fire during flange bolt cutting operation	Serious injury
15	Chemical and rubber products manufacturing	May 2020	Fires and explosions in organocatalyst product packaging operations	Death
16	Chemical and rubber products manufacturing	June 2021	Fire during powdered chemical feed operation	Serious injury

. As shown in Table 1, most accidents involving hazardous materials occurred during raw material charging, cleaning processes, or equipment cutting operations. However, the fire/explosion accident that occurred in May 2020 within the catalyst process was distinct in nature, as it took place during the product packaging operation, unlike the other accidents. The 16 accidents resulted in 9 fatalities, substantial property damage, and financial losses, including production disruptions due to operational downtime. In particular, the Fires and Explosions in Organo-catalyst Product Packaging Operations accident of May 2020, analyzed in this study, caused severe economic losses. The plant, which had received an investment exceeding KRW 100 billion, remained non-operational for a year following the accident and was ultimately shut down permanently in 2023.

In May 2020, an accident occurred during the packaging of catalyst products at a PSM facility, and the analysis data from the Occupational Safety and Health Agency were documented by KOSHA-MIA-202014. A summary of the accident is presented in

Table 2. Accident overview.

Category	Contents
Accident overview	During the process of packing a finished catalyst product in a mobile container (1 m3), the pressure increased rapidly inside the packaging container, causing the internal material to be ejected through the safety valve into the packaging room.
Accident causing material	Organic catalysts to produce polyolefin materials used in the HDPE process
Packaging container

.

The Occupational Safety and Health Agency analyzed the accident, in which a worker died due to the explosion during the operation. The accident occurred as a sequential occurrence of two events (

Table 3. Estimated cause and result of the accident determined by the Korean Occupational Safety and Health Agency (KOSHA).

Sequence of Events	Accident Cause	Reason for Accident	Result
First Event	Improper connection between container and flange	Air enters between improperly connected flanges and reacts with the catalyst inside the container	Pressure builds up inside the vessel, causing the internal pressure to blow out through the process safety valve (PSV).
Second Event	Safety valve outlet not connected to a safety location	Catalyst dust is blowing into the room because the PSV outlet is not connected to a safety location	“Dust explosion” caused by dust released into the packing room

). The first event was as follows. During the packaging process, in a state where the catalyst transfer pipe and catalyst packaging container were not properly connected, external air flowed into the packaging container through the gap between the pipe and container. Subsequently, a reaction occurred between the catalyst inside the container and the air (Figure 2). Because of this reaction, high pressure formed inside the packaging container, which was relieved through the process safety valve (PSV), and catalyst dust burst into the packaging room, causing a dust explosion in the room.

To clearly analyze the cause of the accident and establish effective recurrence prevention measures, we applied the research methodology described in Section 2.2.

3.2. Methods

This study aimed to conduct an in-depth analysis of an explosion accident that occurred during the organic catalyst packaging process, a type of accident that had not been previously reported in academic or industrial contexts. This research began by identifying the fundamental characteristics of the accident-related process, the packaging containers involved, and the materials responsible for the accident through literature reviews and other sources. Subsequently, interviews with workers were conducted to catalog the factors that could have influenced the accident, followed by an evaluation of those factors. This study then proceeded by selecting the key contributing factors among the cataloged items and systematically analyzing the reasons the factors were not managed in advance and how they led to the accident. This approach was designed to identify the root causes and develop a comprehensive understanding of the accident from a systemic perspective.

3.2.1. Procedure

To dissect the intricacies of this fire/explosion accident, a dual-pronged approach, encompassing both technical and systemic analyses, was adopted. As shown in Figure 3, the framework for this analysis was meticulously followed by using accident data compiled by the Korea Occupational Safety and Health Agency (KOSHA), a premier national safety organization. Using this methodological approach, the complex interplay of factors that culminated in the accident was determined, providing a holistic understanding of the causative elements and potential preventive measures.

3.2.2. Systematic Accident/Accident Analysis

Accident investigation techniques have been developed in the order of technologies, human factors, and sociotechnical systems (Figure 1), and they are applied to accident investigations according to the characteristics of these techniques. Accident investigation methodologies approach the same accident from entirely different perspectives. Therefore, combining multiple accident investigation techniques, rather than relying on a single method, enables a more robust and reliable investigation [57,58].

The accident investigation methodologies are categorized, as shown in

Table 4. Characteristics of different accident investigation methods.

Method	Training Need	Levels of Analysis	Application Field
Root cause analysis	Specialist/novice	1–4	Across many industries especially quality management
Events and causal factors charting	Novice	1–4	Occupational accidents across all sectors
Barrier analysis	Novice	1–2
Fault tree analysis	Expert	1–2
Event tree analysis	Specialist	1–3
SCAT	Specialist	1–4	Occupational accidents across all sectors
MORT	Expert	2–4	Nuclear industry
MTO analysis	Specialist/expert	1–4	Nuclear, traffic, aviation industry
AEB method	Specialist	1–3	Nuclear industry
TRIPOD	Specialist	1–4	Oil industry
STAMP	Specialist	2–4	Occupational accidents across all sectors
AcciMap	Expert	1–6

Levels of analysis: 1. Work and technological system, 2. Staff level, 3. Management level, 4. Company level, 5. Regulators and associations level, 6. Government level.

, based on the complexity of their application, level of analysis, and target scope [59,60].

In this study, as outlined in Table 4, the STAMP method (Figure 4) and the BSCAT method (Figure 5), which are widely applicable across various industrial sectors and require systemic analysis by specialists, were applied. The interrelationships among various elements, such as accident-related safety policies, internal standards, procedures, personnel, and organizational structures, were analyzed based on the systemic components of the two methods.

Based on the accident scenarios derived from the technical analysis results, a root cause analysis of the accidents was conducted using the BSCAT method. Subsequently, the focus was placed on identifying the barriers that could have prevented progression to the next stage during the accident and analyzing the root causes of the barriers’ failure to operate.

4. Results

4.1. Technical Analysis Results

The details of the investigation are provided in the KOSHA-MIA-202014, conducted by the KOSHA, which postulates that the accident was caused by the unintended introduction of external substances (e.g., air or water) into the catalyst packaging container. Considering the manufacturing and facility attributes prevailing at the time of the accident, potential scenarios facilitating the ingress of material into the packaging container were meticulously analyzed and are presented in

Table 5. Possible scenarios that could lead to an accident.

Classification	Possible Scenario	Investigation Result
Packaging containers	Moisture residue due to poor container cleaning	Internal analysis of identically handled containers shows no moisture residue
	Moisture ingress due to improper container storage
Nitrogen inlet line	Moisture content in nitrogen	The issue did not occur with other products packaged under the same conditions
	Moisture in the nitrogen supply line
Catalyst filling line	Entry of air or moisture during the catalyst packaging line connection process	Nitrogen purging for 5 min after packaging line connection removes internal residue
	Entry of air or moisture during catalyst packaging operations	Need technical analysis
Vent line	Debris blocks the vent pipe releasing pressure from the packaging container	The pressure inside the packaging container (5 barg) is expelled through the vent pipe during the packing operation
	Oil inside the pressure gauge installed at the end of the container discharge line enters the packaging container	Even if oil leaks, it is structured in a way that prevents oil from entering inside the packaging container
Foreign matter present in the raw material	Foreign matter (such as moisture) in the raw material was mixed in during manufacturing	The issue did not occur with other products packaged under the same conditions

. This comprehensive analysis revealed that the entry of external substances into the packaging container was a circumstance unique to the catalyst packaging process. Following this revelation, a technical verification plan was formulated to evaluate the feasibility of external substances penetrating packaging containers. The failure to implement a safety location at the discharge section of the PSV is implicated as the cause of the second event (Table 3), which is a transgression of domestic safety regulations. Consequently, this aspect was omitted from the technical verification elements examined in this study. The primary objectives of the technical verification conducted in this study were two-fold. First, we aimed to determine whether external substances such as air or water could potentially infiltrate the container via minor leaks stemming from imperfect connections between the packaging container and the attached pipes. This scenario was evaluated during the procedure of infusing nitrogen at a pressure of 5 barg through pipes connected to a packaging container. Second, we sought to ascertain the reaction time between the substances that had entered and the catalyst chamber. This dual approach was key in unearthing the causative factors of the accident and propelling the study toward achieving a holistic understanding of the event.

4.1.1. Material Characteristics/Properties Analysis

According to KOSHA-MIA-202014 from the Occupational Safety and Health Agency in South Korea, the substance related to the accident is methylaluminoxane (MAO), which is an iso-Bu Me, branched, cyclic, and linear aluminoxane. This substance is classified as a Class 3 hazardous material according to the national criteria for dangerous materials classification. Its molecular weights and structural formulae are listed in

Table 6. Information on methylaluminoxane (MAO).

Classification	Contents	Classification	Contents
Chemical Name	Methylaluminoxane	Structural Formula
Molecular Weight	(58.02)m (100.10)n	CAS No.	146905-79-5

. Class 3 dangerous materials are spontaneously combustible and water-reactive chemicals that react upon contact with air or moisture.

In polymer-producing petrochemical processes, methylaluminoxane (MAO) is used as a co-catalyst that aids in the activity of metallocene, the primary catalyst used in polymer manufacturing, which, when combined, creates a supported catalyst (Figure 6).

4.1.2. Evaluation of Reactivity with Air

MAO is categorized as a Class 3 hazardous material with a propensity to react with air or water. To determine the thermodynamic nature of this reaction (exothermic or endothermic), a detailed analysis of the reaction heat was performed using computational chemistry tools (

Table 7. Measurement of the enthalpy change (∆E) in MAO’s reactions with air and water.

Reaction Case	Chemical Equation	∆E (kcal/mol)	Remark
MAO+O2	2(CH3AlO) + 3O2 → Al2O3 + 2CO + 3H2O	−55.2	O2 1 mol
MAO+H2O	2(CH3AlO) + H2O → Al2O3 + 2CH4	−30.5	H2O 1 mol
<Calculation Tool> ① Software Biovia Material Studio Dmol3 (Software Owner: LG Chem, Seoul, Republic of Korea) ② Method : Density Functional Theory (DFT), Becke Exchange Functional & Perdew-Wang 91 Correlation Functional ③ Basis Set: DND(Double-numerical + d Polarization Basis Set) ④ Calculation formula: ΔE°_rxn = [Sum of products for ΔH°_f] − [Sum of reactants for ΔH°_f]

).

For this analysis, the enthalpies of formation for both the reactants and the products involved in the chemical reaction were theoretically computed and juxtaposed for predictive insight. Table 7 shows that the enthalpies of the products, resulting from reaction of MAO with oxygen or water, are significantly more stable than those of the reactants.

This revelation conclusively confirms that the interaction between MAO and water or air is exothermic. Considering this thermodynamic characterization, the actual reactive behavior of MAO in conjunction with the metallocene-supported catalysts in the presence of air was observed. In a controlled environment within a fume hood, 100 g of the supported catalyst was transferred to a stainless-steel carrier. Adorned in flame-retardant apparel and equipped with essential safety gear, a researcher ensured thorough mixing of the supported catalyst and air by agitating the carrier. The ensuing reactive behavior between the supported catalyst and air was monitored at 1 s intervals.

As shown in Figure 7, upon exposure to air, the MAO+metallocene-supported catalyst exhibited a noticeable increase in temperature after 20 s, indicated by the color change. Subsequently, at the 30 s mark, smoke was observed, accompanied by ignition. Table 7 and Figure 7 corroborate that in the event of air infiltration during the catalyst packaging process, the exothermic reaction could potentially yield the reactants and generate carbon monoxide and water within the container. This comprehensive analysis underscores the importance of maintaining an airtight environment during catalyst packaging to pre-emptively mitigate the risk of hazardous reactions.

4.1.3. Assessment of Air Introduction During Packaging Process

Upon exposure to air, the MAO+metallocene-supported catalyst reaction initiated after 20 s. This finding highlights the need for stringent preventive measures against air entering the packaging container throughout the packaging process. The risk assessment report related to this accident, which was associated with the catalyst packaging operation, failed to adequately address the potential hazards stemming from air infiltration during the packaging stage. After the accident, a thorough site investigation revealed that the bolts connecting the piping to the packaging container were not secured according to the prescribed torque standards; therefore, we hypothesized that air entered through a gap between the packaging container and the piping. To validate this hypothesis, the packaging equipment was reassembled to replicate the conditions and status at the time of the accident, and an experiment was performed to evaluate the possibility of air entry through the identified gap. By leveraging the principles of Bernoulli’s equation, as shown in Figure 8, the experiment was conducted to investigate whether the pressure changes occurring around the gap could facilitate external air entry into the packaging container.

Two methodologies were considered: employing actual packaging equipment and applying simulation techniques using computational fluid dynamics (CFD). However, in this study, a deliberate gap was introduced at the connection point on top of the packaging container, using the same equipment used during the accident, as shown in Figure 9. The aim was to determine whether air can infiltrate through this specific area.

The catalyst manufacturing process concludes with the packaging phase, where the discharge line from the vessel containing the final product, after drying, is connected to the connection point (Figure 9). An inert gas line (N₂) is also connected to the discharge line during the packaging operation to mitigate the risk of a reaction between the catalyst and the oxygen within the packaging container. While the N₂ line maintains a pressure of 5 barg, the structure of the packaging container permits the release of N₂ through the upper vent line, maintaining the pressure of the packaging container below 0.5 barg.

For the experimental analysis of air infiltration, an artificial gap was created between the piping and the upper part of the packaging container. By intentionally under-tightening the four bolts on the flange and connecting the piping to the packaging container, a gap of approximately 1 mm was maintained, as shown in Figure 10. By maintaining the N₂ pressure at 5 barg, the product input valve on top of the packaging container was opened by approximately 30%. From the observations, we observed that smoke from the smoke stick entered the flange gap swiftly. Based on the experimental results, the pressure change inside the pipe was simulated using the Ansys fluent simulation software (version 11.0) (Figure 11). As shown in Figure 11, negative pressure was formed in a certain section around the gap between the flanges, and a flow rate of approximately 20 m/s was created in this section. The Ansys fluent simulation software was used to accurately reflect the flow and pressure formation process of the gas supplied in a high-pressure state within a pipe with a complex structure and to calculate numerous finely divided cells simultaneously. After 3D modeling the pipe, its volume was divided into 1,000,000 cells based on a volume of 0.003 m³ for accurate calculation, and each cell was calculated separately. As this did not consider the CFD analysis, the CFD calculations are not included.

This reconstruction test and analysis of the CFD experiment conclusively demonstrated that if the internal structure of the packaging container’s piping is similar to the conditions necessitated by Bernoulli’s equation, the external air has the potential to infiltrate through minute gaps. Therefore, it is important, especially when dealing with products that are sensitive to air, to ensure airtight connections and eliminate even the smallest gaps. Furthermore, if any small gaps arise between the piping and the packaging container, considering a straight piping design without changes in the inner diameter of the piping is crucial to prevent the infiltration of external air into the packaging container.

4.2. Systematic Analysis by Accident

The technical analysis of accidents that occurred within the PSM facility was conducted using the STAMP and BSCAT methods. Systematic recurrence prevention measures have been established based on the fundamental causes of accidents identified through these methods.

4.2.1. BSCAT Analysis

To prevent the recurrence of this accident, the design and management of manufacturing facilities from the research phase to the construction of production equipment in the plant must be performed with the utmost care to ensure that there is no air infiltration. Accordingly, using the BSCAT, an analysis was conducted sequentially from the causes of the improper connection between the packaging containers and the piping up to the barriers identified at the time of the accident (Figure 12).

The prevention controls identified using the BSCAT analysis included education on standard operating procedures (SOPs), supplementation of SOPs, the safety location of the PSV, and the “safety mgmt. system”. The accident could have been prevented if the prevention controls were properly executed before the workers performed the task. In addition, even if various prevention control items were not properly executed, the packaging process would have been organized as outside work when designing the process from the mitigation control perspective, or if the indoor fire protection facilities were properly configured, the accident would not have caused significant losses.

The barriers identified through the BSCAT analysis are listed in

Table 8. Status of different barriers that could have prevented the accident.

No.	Barrier	Status	Confidence	Challenge	Performance
1	Safety Mgmt. Sys.		Low	Contains regulations on procedures and work permits for safe process management	Not working properly
2	Safety valve discharge line connected to combustion/absorption/capture/recovery facilities		Missing	Dangerous substances discharged from safety valves must be treated by combustion, absorption, capture, and recovery	Not operational
3	Supplementation of SOP		Missing	In case the piping is not connected properly, the operator should reconfirm the piping connection on his own or have another operator reconfirm the piping connection Establish a procedure for waiting a certain amount of time after packaging to work on a product because adverse events may occur during packaging	Not operational
4	Education on standard of procedure (SOP)		Missing	Prevent similar accidents from occurring by training workers on work procedures	Not operational
5	Packaging process organized as an outside work		Missing	Operate catalyst packaging outside to ensure that catalyst dust does not reach explosive dust concentrations even if it is blown outside	Not operational
6	Properly configured fire protection facilities		Missing	Prevent secondary accidents such as fires and explosions by eliminating flammable materials near the workplace or establishing facilities indoors to prevent or respond to dust explosions	Not operational

. Predominantly, there was no connection between the safety valve discharge line and the combustion/absorption/capture/recovery facilities (no barrier was present), and the safety management system was an unreliable barrier. Furthermore, there were no additions to the work procedure documents (a barrier that did not exist) and education on work procedures (another nonexistent barrier). From the onset of the accident to the spread of the damage, the analyzed barriers were outdoor operations (barrier was not present) and fire safety facilities (barrier was not present).

The BCSAT analysis identified the following areas for improvement to prevent accident recurrence: process safety leadership, hazard assessment, risk management, emergency response, and risk monitoring (Figure 13).

4.2.2. STAMP-CAST Analysis

Through meticulous technical analysis and the application of the BSCAT, the causes of the accident, along with managerial lapses and deficiencies present at the time of the accident, were identified. The roots of these managerial shortcomings can be traced back to the initial phases of the chemical research and development process, extending to the pre-commissioning stage. During these critical periods, the identification and subsequent assessment of the risk factors associated with the materials, processes, and equipment were inadequate, resulting in potential hazards being overlooked. These risk factors necessitate vigilant management, either by designated personnel at each stage of the process or through comprehensive reviews conducted by pertinent departments. However, even in the moments leading to the accident, these overlooked risk factors remain undetected, culminating in an unfortunate event. Ensuring a proper connection between the packaging container and the piping is a seemingly straightforward and crucial process during product packaging operations; it is imperative that this is systematically evaluated during the risk assessment phase. The lack of a secure connection could allow air to infiltrate the packaging container, posing a significant risk of fire and explosion. Additionally, a comprehensive preoperative safety check should be an integral part of the procedures leading up to the commencement of work. By dissecting the identified factors through the lens of work stages and inter-organizational relationships, we created a visual representation (Figure 14). This approach not only clarifies the causative factors but also emphasizes the pivotal role of rigorous managerial oversight in preventing such accidents in the future.

5. Discussion

The analysis of the safety control structure model included the stages from early research and development (R&D), scale-up operations, and design and installation associated with new factory establishments, including the production team, safety and environment team, company management, Ministry of Employment and Labor, Occupational Safety and Health Agency, and research institutes. Using this model, communication strategies and necessary internal organizational improvements for the prevention of recurring similar accidents can be established (

Table 9. Management requirements for each department to prevent accidents during chemical processes.

Group	Safety Requirements	Improper Decision	Defect in Organization
MEL/KOSHA	(1) Enforcement of the Safety and Health Act (2) Thoroughly review a PSM document	-	-
R&D center	(1) Establishing safety standards and risk assessment criteria in R&D (2) Establish a safety training program for researchers (3) Establish procedures to identify the risks involved before transferring R&D products to production.	Underestimating the risk of chemicals and failing to consider the safeguards that should be in place during the scale-up phase	Enhance the ability to list risks identified in the research phase and clearly communicate them to production and design teams
Production team	(1) Establish procedures to identify key risks before transferring R&D products to production (2) Establish procedures for training workers on key points related to manufacturing new products	(1) Performing piping connections in unsafe locations and conditions without a worker-centric risk assessment (2) Lack of understanding of what happens when foreign objects are introduced during the packaging process	Lack of procedure regarding what needs to be reviewed when introducing new products and the ability to worker-centric risk assessments and manage accident cases
EHS team	(1) Clear cross-functional R&R by PSM element (2) Inventory legal requirements about PSM	Determined that indoor emissions from PSV lines are not in violation of the law	It is necessary to supplement the function that lists relevant laws and regulations when introducing new products and periodically check them during the construction process
D/G safety manager	Perform safety management tasks to meet legal requirements when handling hazardous materials	Determining the risk of the packaging step as low and not participating in the packaging process	Establish a system to ensure that safety managers are involved in all hazardous materials work
Process design team	(1) Establish procedures to incorporate risk factors identified in the research phase into the design. (2) Establish procedures to incorporate accident cases into design	(1) Failure to clearly review legislation during process design (2) Failure to account for human error in process design	Internal criteria for safe design guidelines (KOSHA guidelines) must be clarified

).

To prevent the recurrence of similar accidents through accident investigations, it is essential, as shown in Table 8 and Table 9, to incorporate the reaction requirements and hazards of substances identified during the research phase into the reactor design stage during chemical process design. This approach ensures that the process is designed to minimize the introduction of external materials into the system. Subsequently, it becomes the duty of the controller to proactively identify potential risk factors during the scale-up phase and throughout the establishment and operation of the actual production process, ensuring that these risk factors are considered in the risk assessment. In the permitting stage, it is crucial to compile a comprehensive checklist of items that require review in accordance with the regulations governing the establishment of new processes. A safety design, conceived from a foolproof perspective and informed by the KOSHA guidelines and other relevant technical materials, should be executed, considering the most unforeseeable circumstances.

Through this study, a newly identified finding is that when high-pressure piping is not properly connected, fluids or gases can enter the packaging container. If a pathway is formed for the introduced fluid or gas to escape from the container, even if the improperly connected high-pressure piping allows the rapid movement of the fluid or gas, small gaps in the piping can enable external air or gas to continuously flow into the packaging container. Given this reality, it is crucial to re-emphasize the importance of proper piping connections during the packaging of materials that react rapidly upon contact with oxygen.

As demonstrated in this study, when analyzing accidents where various factors are complexly interrelated, utilizing systemic analysis methods such as the BSCAT and STAMP-CAST techniques can be highly effective in clearly and specifically identifying the root causes of accidents. However, to reduce the casualties from similar accidents, further research should be conducted to measure the impact range of explosion accidents caused by such accidents and to establish additional safety management measures for workers within the affected area. This would enable the development of concrete strategies for both accident prevention and the minimization of damages resulting from accidents.

6. Conclusions

By investigating the technical intricacies of the causes of an accident during the catalyst packaging process, this study has successfully elucidated the immediate cause and further subjected it to a meticulous re-analysis employing systematic analytical methods. The accident analysis outcomes highlight the importance of recognizing the potential for substances to mix inadvertently during both the process design and stable operational stages, potentially resulting in an accident. It is of the utmost importance to distinctly identify these considerations during the risk assessment phase and establish, within the organization, a procedure to seamlessly incorporate risk mitigation measures for each identified risk factor into the design stage. Even if the respective risk factors are meticulously reflected in work procedures, it is crucial to adopt a “worker-centric risk assessment” approach to mitigate the risk of accidents owing to human error. When examining chemical accidents within a PSM framework, it is easy to mistakenly attribute them to a single causative factor when solely evaluating the outcomes. However, this study showed that a systematic analysis can reveal the complex web of underlying causes that lead to these accidents. This highlights that beyond conventional accident prevention measures, such as securing additional devices and functions within the manufacturing process, there is a need for proactive management and vigilance, underpinned by clearly defined roles and responsibilities (R&R) across the departments involved in process design, operation, safety environment, and R&D. The onus is on the production and safety departments within the factory to internalize these considerations to ensure their systematic application and operation in the field, bolstered by dual or even triple verification processes, extending beyond the R&D phase.

When the STAMP-CAST method and the BSCAT method were jointly applied for accident analysis, it was confirmed that these approaches not only enable the systematic identification of accidents’ root causes and the development of measures to mitigate or eliminate accident risks but also provide specific insights into the factors that allowed a minor event to escalate into a major accident. Additionally, they facilitate the formulation of strategies to prevent such escalation. However, the accident analyzed in this study involved a single worker who died due to an explosion while on duty. Since it was not possible to interview the worker most familiar with the situation at the time of the accident, there were limitations in precisely identifying the specific circumstances that led to the worker’s death. While this study primarily focused on analyzing the root causes of the accident and developing recurrence prevention measures, future research should include an analysis of the explosion pressure and the thermal energy of the explosion flame. This would allow for the establishment of standards for ensuring worker safety, such as safe distances and protective gear, in the event of similar accidents.

This study validated the applicability of a systematic analysis method integrating the BSCAT and STAMP for both direct and root cause analyses of chemical accidents. Moreover, it suggested that the adoption of such systematic techniques for process risk evaluation and preoperative inspections of newly established plants hold significant promise for reducing the risks associated with accidents.