Analyzing Safety Management Failure Paths in Coal Mines via the 24Model Accident Causation Framework and fsQCA

Abstract

This study investigated safety management performance in small- and medium-sized private coal mining enterprises (SMPCMEs) through an integrated application of the 24Model accident causation theory and fuzzy-set qualitative comparative analysis (fsQCA). Analyzing 40 sudden incidents (2013–2023), we examined six key factors—organizational, individual, and external dimensions—to identify nonlinear risk pathways. Results revealed four critical failure types—Internally Balanced (cultural–behavioral–environmental collapse), Safety Culture–Deficient (institutional hollowing), Cultural–External Environment (policy-implementation paradox), and External Environment–Integrated (technological-regulatory failure)—that collectively explained 83% of performance variance. Tailored strategies, including IoT-based real-time monitoring and AI-driven inspections, are proposed to transition from fragmented interventions to systemic governance. These findings provide actionable insights for enhancing safety resilience in high-risk mining sectors.

1. Introduction

Coal mine safety governance, as a critical enabler of China’s “Dual Carbon” goals of peaking carbon emissions by 2030 and achieving carbon neutrality by 2060, has improved significantly in recent decades. This is evidenced by a 68% reduction in the fatality rate per million tons of coal output, declining from 0.293 in 2013 to 0.094 in 2023 [1]. Yet, small- and medium-sized private coal mining enterprises (SMPCMEs) remain plagued by persistent safety challenges. The 2023 Xinjing Coal Mine collapse in Inner Mongolia, a disaster that resulted in 53 fatalities and revealed critical gaps in safety practices and regulatory oversight, highlights the pressing demand for more effective governance strategies. Despite ongoing safety reforms, SMPCMEs consistently report higher accident rates compared to state-owned mines. Safety deficiencies in SMPCMEs pose systemic challenges to sustainable development. These accidents seldom result from single factors; rather, they arise from the complex interactions of organizational weaknesses, outdated technologies, and insufficient regulatory oversight [2].

Previous studies often attribute safety management performance to isolated factors, such as leadership behavior or organizational culture [3,4,5]. However, these investigations predominantly use linear methods, including structural equation modeling and regression analysis, which simplify complex interactions among multidimensional elements [6,7]. A comprehensive review by Tubis et al. (2020) indicates that more than 80% of mining safety research adopts narrow methodologies, neglecting the concept that different factor combinations can lead to similar accident outcomes [8]. fsQCA overcomes this limitation by analyzing how combinations of conditions form sufficient pathways for outcomes. Sun et al. (2024) utilized fsQCA to examine the formation pathways of production safety accidents in China’s urban gas pipeline networks, exploring how interactions among multiple risk factors collectively influence accident dynamics [9].

Conventional models, such as the Human Factors Analysis and Classification System (HFACS) and the “Reason” model, focus on sequential cause–effect relationships [10]. While these frameworks identify individual contributors, they fail to capture how safety culture, management practices, and external pressures dynamically combine to compromise safety. For instance, Wu et al. (2008) demonstrated that leadership directly shapes safety climate but did not explore how economic fluctuations or weak enforcement alter this relationship [11]. High-Reliability Organization (HRO) theory, developed for large industries like aviation and nuclear power, emphasizes resilience through standardized protocols [12]. However, its application to SMPCMEs is limited by their resource scarcity and fragmented governance structures [13]. Liu et al. (2017) further showed that safety culture strategies effective in state-owned mines underperform in SMPCMEs, where profit-driven priorities frequently overshadow safety investments [14].

To address these limitations, this study pioneers an integrated analytical approach by combining 24Model accident causation framework [15] with fsQCA [16]. 24Model excels at identifying multi-level factors but struggles to capture complex interactions [15]; conversely, fsQCA lacks a structured theoretical framework for guiding variable selection [16]. 24Model provides fsQCA with a systematic, multi-level, and theory-driven framework for selecting and classifying variables, while fsQCA offers an empirical tool for 24Model to explore the nonlinear, configurational interactions and equifinal paths among the identified factors. This integration achieves complementary advantages by combining a “structured theoretical framework” with a “tool for analyzing complex interactions.” This synergy constitutes the key innovation of our study, distinguishing it from research using only the 24Model for case analysis or only fsQCA for data-driven exploration. 24Model, a tri-level framework encompassing organizational, individual, and external factors, identifies critical determinants of safety management performance in SMPCMEs. Building on this foundation, fsQCA is employed to analyze how these factors interact dynamically, providing actionable insights for systemic governance. The findings offer scientific guidance for policymakers and enterprise managers, facilitating a transition from “single-factor interventions” to “systemic governance” in coal mine safety management.

2. Materials and Methods

2.1. Case Selection

To ensure methodological rigor, stratified sampling was applied to select 40 sudden incidents involving SMPCMEs in China (2013–2023). These incidents were operationally defined as unplanned, abrupt events threatening personnel safety or causing substantial property damage, consistent with the Regulations on Reporting and Investigating Coal Mine Production Safety Accidents. The case selection adhered to the following criteria:

1. Incident Diversity. The cases comprised 13 gas-related accidents (32.5%), 7 water inrushes (17.5%), 15 roof collapses (37.5%), 3 fire and explosion accidents (7.5%), and 2 transportation accidents (5%) [17].

2. Geographical Coverage. The cases were distributed across principal coal-producing provinces in China (e.g., Shanxi, Inner Mongolia, Heilongjiang, Yunnan) to control for regional heterogeneity in safety management practices.

3. Severity Levels. Incidents were classified into four tiers (particularly severe/major/significant/general) based on fatalities and economic loss thresholds mandated by Chinese safety regulations.

This study employs 40 cases, aligning with fsQCA’s methodological requirement for small-to-medium-sized sample analysis. The core strength of fsQCA lies in its case-oriented approach, which reveals the complexity of condition combinations without relying on large-sample statistical significance. With a sample size (N = 40) within the methodologically recommended ideal range of 10–60 cases, it is sufficient to capture the multiple-path patterns of SMPCMEs accidents. Simultaneously, the design of the four-value fuzzy set (0, 0.33, 0.67, 1) strictly adheres to fsQCA methodological specifications and aligns logically with the classification levels of coal mine accidents.

The selected cases cover core coal-producing provinces, which constitute China’s primary coal heartland and experience a concentration of SMPCMEs, resulting in relatively rich and accessible accident data. Furthermore, the accident types within the cases encompass the main causes associated with typical geological risks. The dynamically coupled mechanisms reflected in the identified configurations can explain systemic failures in other high-risk management scenarios, such as construction and chemical industries. Thus, the findings provide a cross-contextual governance framework applicable to resource-constrained enterprises in other countries, demonstrating generalizability.

2.2. Theoretical Framework: 24Model

The 24Model served as a comprehensive theoretical framework to analyze the causation mechanisms of coal mine accidents through a tri-level classification system: organizational, individual, and external factors [18]. In this study, the 24Model was implemented with the following hierarchical structure.

Figure 1 incorporated factors at the organizational, individual, and external levels to account for the multidimensional nature of safety management. Each module explicitly labeled as the antecedent variables applied in this study. Safety culture (SC), safety management systems (SS), and organizational behaviors (OB) were derived from the organizational level, whereas employee safety competence (EQ) and work conditions (WC) originated from the individual level. External environmental influences (EI) were extracted from the external level. Safety culture played a crucial role in the establishment and implementation of the safety management system, while also enhancing employees’ safety competence. The safety management system, in turn, affected employee behavior, and the intensity of external supervision influenced the extent of safety management within enterprises. These factors interacted and influenced one another [19].

By mapping these factors onto the 24Model, we systematically analyzed their interactions, providing a comprehensive perspective on accident causation mechanisms and factors influencing safety management performance. Compared with existing models, such as the “Reason” model and HFACS, 24Model clearly defined each module, addressing the shortcomings encountered in the analysis of specific cases below [20].

2.3. Variable Definitions and Operationalization

2.3.1. Antecedent Variables

This study selected 40 coal mine sudden incidents as cases, constituting a moderate sample size. The condition variables were set with reference to 24Model, and a comprehensive examination of the actual circumstances of the cases was conducted, ultimately distilling 6 antecedent variables: safety culture, safety management system, organizational behavior, work conditions, employee safety quality, and external environmental impact. The details were as follows:

• Safety Culture (SC, Organizational Level) reflected an organization’s comprehension and implementation of safety values, encompassing managerial commitment, employee safety awareness, resource allocation for safety initiatives, and workplace safety climate. Organizations demonstrating characteristics such as ambiguous safety objectives, leadership neglect of safety protocols, or insufficient safety training investments were identified as exhibiting deficient safety culture [21].
• Safety Management System (SS, Organizational Level) served as a fundamental cause of accidents. Failures in implementing safety regulations and procedures, lax safety inspections and supervision, and the prevalence of unsafe behaviors such as violations of operating protocols occurred. Structural deficiencies in safety governance resulted in ambiguous departmental accountability, impeding operational safety protocols. Deficient subcontractor oversight further heightened accident risks due to inadequate safety compliance verification.
• Organizational Behavior (OB, Organizational Level) reflected the overall performance of SMPCMEs, including decision-making processes and adherence to safety protocols. For instance, some leaders made poor decisions in pursuit of production and economic benefits, or they issued illegal directives during production, forcing employees to engage in risky operations.
• Work Conditions (WC, Individual Level) denoted the prevailing working environment and contextual parameters. For example, certain coal mine working environments were harsh, with equipment that remained unmaintained and outdated over extended periods, rendering control measures for hazardous factors ineffective.
• Employee Safety Quality (EQ, Individual Level) gauged workers’ operational safety competencies and emergency response proficiency, encompassing three diagnostic indicators: unsuccessful safety certification attempts, deficiency in incident management experience, and recurrent violations of operational protocols. These manifestations collectively defined suboptimal safety qualification profiles.
• External Environmental Impact (EI, External Level) encompassed three key factors: regulatory enforcement, economic influences, and natural disaster risks. For instance, insufficient regulatory oversight of mining sectors by government agencies, reduced safety investments during economic downturns, and unpredictable safety disruptions caused by natural disasters were identified as critical manifestations of EI.

2.3.2. Outcome Variables

This study defined SMPCMEs’ safety management performance as the outcome variable. Safety management performance levels are systematically assessed by integrating accident severity classifications with multidimensional governance metrics, encompassing regulatory compliance, training effectiveness, and inspection rigor [22]. A four-tier scoring framework (Figure 2) is implemented:

• High Performance (1.00): Assigned to organizations with zero major accidents, rigorous regulatory enforcement, and demonstrably effective training protocols.
• Intermediate Level (0.67): Applicable to entities reporting minor accidents but maintaining prompt emergency response capabilities, despite localized operational vulnerabilities.
• Low Performance (0.33): Designated for systems experiencing major accidents with delayed remediation and identifiable management deficiencies.
• Critical Deficiency (0): Reserved for catastrophic failures involving systemic governance collapse.

These four levels corresponded to the assignment principles of the four-value set in fsQCA, allowing for distinct assignments to be made. This hierarchical valuation follows the dual principles of severity decrement and comprehensive weighting, with independent expert validation ensuring methodological rigor and impartiality. Furthermore, the outcome variable intuitively reflected the ability of SMPCMEs to suppress accidents in their daily safety management and production, serving as the primary basis for evaluating safety management performance [23].

2.3.3. Variable Assignment

As presented in

Table 1. Variable settings and assignments.

Category	Variable	Deficiency Indicators	Assignment Score
Antecedent Variables	Safety Culture (SC)	Unclear safety values -No annual safety goals established -Mine leadership fails to articulate safety-first principle in official documents	0 = No deficiency 0.33 = 1 deficiency 0.67 = 2 deficiencies 1 = ≥3 deficiencies
		Insufficient leadership attention -Shift leadership system by mine executives not implemented -Lack of follow-up and supervision on rectification of major hidden dangers
		Poor safety atmosphere -Prevalence of “Three Violations” -Absence of safety culture activities
		Low investment in safety initiatives -Insufficient extraction or misappropriation of safety funds -Aging or insufficient critical safety equipment
	Safety Management System (SS)	Non-implementation of safety regulations -Existence of unapproved operating procedures -Special operations personnel working without valid certification	0 = No deficiency 0.33 = 1 deficiency 0.67 = 2 deficiencies 1 = ≥3 deficiencies
		Disorganized safety management -Ambiguous division of safety responsibilities among departments -Missing safety meeting records
		Inadequate management of outsourced operations -Failure to rigorously vet the qualifications of outsourcing teams -Outsourced personnel not included in unified safety training and management
		Insufficient safety inspections and supervision -Safety inspections not conducted as per required frequency -Identified hazards not managed in a closed loop
	Organizational Behavior (OB)	Decision-making errors -Forcing production to continue despite known, unresolved major risks -Cutting necessary safety engineering time to pursue production targets	0 = No deficiency 0.33 = 1 deficiency 0.67 = 2 deficiencies 1 = ≥3 deficiencies
		Improper directives -Managers issuing operational directives that violate regulations -Forcing employees to work under unsafe conditions
		Unsafe operational practices -Prevalence of habitual violations on-site without correction -Failure to conduct pre-shift safety confirmations and risk identification
		Violated laws and regulations -Existence of mining beyond approved boundaries or layers -Concealing accidents or falsifying production safety reports
	Work Conditions (WC)	Inadequate safety conditions in the workplace -Inadequate or failed support strength in mining/heading faces -Roadways with severe deterioration/deformation	0 = No deficiency 0.33 = 1 deficiency 0.67 = 2 deficiencies 1 = ≥3 deficiencies
		Numerous equipment and facility hazards -Critical equipment operating with known defects -Electrical equipment lacks explosion-proofing
		Insufficient personnel and resource allocation -Understaffing in critical positions -Incomplete or expired emergency supplies
		Inadequate control measures for hazardous factors -Gas drainage fails to meet standards or monitoring system has frequent false alarms/misses -Water control measures not effectively implemented
	Employee Safety Quality (EQ)	Poor safety awareness -Employees unaware of job-specific risks and emergency response procedures -Multiple near-miss incidents occurred due to reckless operations	0 = No deficiency 0.33 = 1 deficiency 0.67 = 2 deficiencies 1 = ≥3 deficiencies
		Inadequate operational skills -New employees assigned to posts without effective training -Employees unfamiliar with equipment safe operating procedures
		Lacked sufficient emergency response capabilities -Employees unable to correctly use self-rescuers or fire extinguishers -Emergency drills are perfunctory
		Had ineffective safety training -Training content not relevant to job-specific risks -Training assessments are superficial or records falsified
	External Environmental Impact (EI)	Inadequate oversight from relevant departments -Only imposing fines for discovered major hazards without ordering production halt -Regulatory inspections are infrequent	0 = No deficiency 0.33 = 1 deficiency 0.67 = 2 deficiencies 1 = ≥3 deficiencies
		Government inaction -Ineffective crackdown on illegal production activities -Accident investigations downplay severity
		Experienced negative economic impacts -Significant cuts to safety investment budgets due to falling coal prices -Use of substandard equipment/materials to reduce costs
		Faced risks associated with natural disasters -Mine area located in geohazard-prone zone without effective mitigation -Failure to timely halt production and evacuate personnel upon extreme weather warnings
Outcome Variable	Safety Management Performance (SMP)	High Performance -Zero major accidents -Rigorous regulatory enforcement -Effective training protocols	1
		Intermediate Level -Minor accidents reported -Prompt emergency response	0.67
		Low Performance -Major accidents with delayed remediation -Identifiable management deficiencies	0.33
		Critical Deficiency -Catastrophic failures -Systemic governance collapse	0

. Following the identification of antecedent and outcome variables, a four-tier valuation framework was implemented. The scoring protocol adhered to two methodological principles: severity decrement and comprehensive weighting, as detailed in Table 1.

2.4. Analytical Method: fsQCA

2.4.1. Rationale for Method Selection

The fuzzy-set qualitative comparative analysis (fsQCA) method accounts for case heterogeneity and complexity [24]. It identifies multiple configurations associating with specific outcomes and detects functional equivalence across configurations [25]. fsQCA reveals conjunctural sufficiency between condition combinations and outcomes; when a specific combination of conditions exists, the outcome is highly probable. However, the same outcome may constitute through different paths, and no single condition is necessary for the outcome [26]. Variations in safety management performance stem from typical configurations formed by interdependent elements, exhibiting varying influence levels. Given this complexity, we employ fsQCA with four-value calibration for conditions, aligned with accident severity levels [27].

2.4.2. Analytical Procedure

• The analysis commenced with calculating consistency metrics for individual antecedent variables, where scores exceeding 0.9 were classified as necessary conditions.
• Configuration path construction followed, implementing threshold parameters of 0.8 (consistency), 1 (frequency), and 0.7 (PRI consistency). This process yielded parsimonious, intermediate, and complex solutions, with the intermediate solutions revealing six core configuration pathways through conditional combination extraction.
• Robustness validation was conducted through dual approaches: elevating the frequency threshold to 2 or modifying thresholds to 0.85 (consistency) and 0.75 (PRI consistency), thereby confirming solution stability and operational robustness.

2.4.3. Model Visualization

In light of this, this study developed a Coordinated Interaction Model for SMPCMEs’ safety management performance (as shown in Figure 3) to explore the intrinsic relationships among factors that influenced enterprise safety management performance [28,29].

Figure 3 illustrated dynamic interactions. The synergistic linkage model elucidates the dynamic interactions among multidimensional factors. At the organizational level, safety culture (SC) drives the establishment and refinement of safety management systems (SS) by shaping corporate values, while SS constrains organizational behaviors (OB) through institutional norms to ensure operational compliance. Organizational behaviors further influence the external environment (EI), such as regulatory interventions triggered by non-compliant practices. Concurrently, EI exerts feedback effects on SC and SS, forming a closed-loop regulatory mechanism. The fsQCA results show that EI and SS co-occur in multiple failure configurations (C3, C5, C6; see Figure 4). This empirically supports both the feedback effect of EI on SS depicted in Figure 3 and the dynamic relationship where the failure of SS jointly with the failure of EI leads to severe consequences (as seen in C5 and C6).

3. Results

3.1. Univariate Necessary Condition Analysis

To identify the necessary conditions for the outcome, consistency served as the primary indicator [30]:

$$\mathrm{Consistency}\left(X_i\le Y_i\right)={\displaystyle \frac{{\displaystyle \sum \left[\min \left(X_i,\left.Y_i\right)\right.\right]}}{{\displaystyle \sum \left(X_i\right)}}}$$

(1)

Here, X_i denoted the univariate variable, and Y_i represented the outcome variable.

A consistency value greater than 0.9 indicated that the variable was a necessary condition for the outcome [30]:

$$\mathrm{Converage}\left(X_i\le Y_i\right)={\displaystyle \frac{{\displaystyle \sum \left[\min \left(X_i,\left.Y_i\right)\right.\right]}}{{\displaystyle \sum \left(Y_i\right)}}}$$

(2)

However, if the coverage of X_i was below 0.8, it implied that X_i alone could not fully account for the outcome variable Y_i [30]. The results were presented in

Table 2. Results of the necessary condition analysis.

Condition Variable	Consistency	Coverage
SC	0.8037	0.7767
~SC	0.8216	0.7675
SS	0.9822	0.7071
~SS	0.5563	0.7766
OB	0.7501	0.8584
~OB	0.8753	0.7107
WC	0.7342	0.8383
~WC	0.8216	0.6682
EQ	0.7500	0.8087
~EQ	0.8753	0.7431
EI	0.7868	0.8627
~EI	0.8037	0.6736

.

Table 2 shows that the consistency of the safety management system exceeded 0.9, indicating that this factor was a necessary condition for the occurrence of sudden incidents in coal mines. However, since its coverage fell below 0.8, this factor alone could not sufficiently explain the incidents, requiring further configurational analysis.

3.2. Configurational Analysis

This study set the baseline consistency level at 0.8, the frequency threshold at 1, and the PRI consistency threshold at 0.70. The analysis produced complex, intermediate, and parsimonious solutions. The intermediate solution, with moderate complexity, did not permit the removal of necessary conditions [31]. Consequently, the configurational results for the safety management performance levels of SMPCMEs were derived from the intermediate solution, yielding six configurational paths, as presented in

Table 3. Configuration paths of safety management performance levels.

Configurations	Internally Balanced Type		Safety Culture Deficient Type	Cultural–External Environment Type	External Environment–Integrated Management Type
	C1	C2	C3	C4	C5	C6
SC	●	●		●		•
SS			●	⊗	●	●
OB	●	●	⊗		•
WC	●	●		•	⊗	⊗
EQ	•		⊗	•
EI		•	•	●	●	●
Original Coverage	0.3257	0.4263	0.4321	0.3363	0.3426	0.4317
Unique Coverage	0.0673	0.0178	0.0536	0.0165	0.0165	0.0466
Consistency	1
Overall Coverage	0.8253
Overall Consistency	1

Note: The presence (•) or absence () of core conditions are the primary drivers of failure. Marginal conditions play a secondary role. ●: Core condition present (critical factor exists); ⊗: Core condition absent (critical factor missing); •: Marginal condition present (supporting factor exists); : Marginal condition absent (supporting factor missing); Blank: Condition irrelevant to outcome.

.

As shown in Table 3, the consistency of each configuration and the overall consistency both exceeded 0.9, indicating that the combination of antecedent variables constituted a sufficient condition for the outcome variable [31]. The overall coverage was 0.8253, suggesting that approximately 83% of the safety management performance in SMPCMEs was explained. Based on the parsimonious and intermediate solutions, the combination paths were classified into four types: Internally Balanced Type, Safety Culture Deficient Type, Cultural–External Environment Type, and External Environment–Integrated Management Type:

• Corresponding to C1 and C2, Internally Balanced Type used the combination of “safety culture (SC),” “organizational behavior (OB),” and “work conditions (WC)” as core conditions. This path suggests that the combined absence of safety culture, non-compliant behaviors within the organization, and poor work conditions constitutes a sufficient configuration for low safety management performance. Regardless of whether deficiencies existed in the safety management system or employees’ safety awareness, or whether external environmental influences were present, the company’s safety management performance remained unsatisfactory. A representative case is the 2021 “11·10” roof collapse accident at Houzitian Coal Mine, Guizhou. The mine exhibited a deficient safety culture (SC●), characterized by ignored geological risks and absent safety training; organizational misconduct (OB●), including illegal command chains and critical staffing gaps; and poor working conditions (WC●), marked by structurally unstable roadways and obsolete support systems. Despite nominal safety protocols (SS/blank) and superficial regulatory oversight (EI/•), the co-occurrence of core conditions (SC●, OB●, WC●) was associated with the disaster, reflecting the Internally Balanced Type’s fatal logic of “cultural disorder—behavioral deviance—environmental collapse.
• Corresponding to C3, Safety Culture Deficiency Type utilized the combination of “safety management system (SS),” “organizational behavior (OB),” and “employee safety quality (EQ)” as core conditions, with external environmental influences serving as auxiliary conditions. This configuration indicated that, under the influence of external environments, the completeness of the safety management system played a dominant role in the safety management performance of enterprises. Regardless of whether poor work conditions existed, the safety management performance level of the company remained unsatisfactory. A representative case is the 2022 “10·15” water inrush accident at Xinglong Coal Mine, Yunnan. The mine exhibited formalized safety management systems (SS●) failing to control boundary-crossing risks, organizational misconduct (OB⊗) through illegal outsourcing, and deficient employee competence (EQ⊗) with untrained workers. Despite external oversight (EI●), the co-occurrence of core conditions (SS●, OB⊗, EQ⊗) aligned with the disaster scenario, aligning with the Safety Culture Deficiency Type’s logic of “institutional hollowing—behavioral anarchy—competence erosion.
• Corresponding to C4, Cultural–External Environment Type utilized the combination of “safety culture (SC),” “safety management system (SS),” and “external environmental influences (EI)” as core variables, with work conditions serving as supplementary conditions. This configuration indicated that, under the influence of poor work conditions, although the safety management system was lacking, safety culture and external environmental influences were core conditions affecting the subpar safety management performance of enterprises, accompanied by inadequate employee safety awareness. A representative case is the 2019 “11·25” coal and gas outburst accident at Sanjia Coal Mine, Guizhou. The mine exhibited formalized safety culture (SC●), marked by systemic data falsification and neglected risk awareness; compromised external oversight (EI●), with regulators imposing superficial penalties for violations; and collapsed safety management systems (SS⊗), lacking gas risk protocols. The co-occurrence of core conditions (SC●, EI●, SS⊗) was consistently observed in the disaster, reflecting the Cultural–External Environment Type’s pattern of “cultural alienation—regulatory failure—institutional collapse.
• Corresponding to C5 and C6, External Environment–Integrated Management Type utilized the combination of “safety management system (SS),” “work conditions (WC),” and “external environmental influences (EI)” as core variables. Regardless of the strength of safety culture or the degree of organizational behavioral compliance, SMPCMEs’ safety management performance remained at a low level. A representative case is the 2016 “10·29” gas explosion at Jingyou Coal Mine, Heilongjiang. The mine exhibited collapsed safety management systems (SS⊗), marked by structural deficiencies and protocol violations; systemic regulatory failure (EI●), with authorities neglecting enforcement and accountability; and hazardous working conditions (WC⊗), including critical ventilation and gas control failures. The combination of core conditions (SS⊗, EI●, WC⊗) aligned with the disaster scenario, characterizing the External Environment–Integrated Management Type’s profile of “institutional tokenism—regulatory vacuum—environmental collapse”.

3.3. Robustness Test

• The case frequency threshold was raised from one to two, while the original consistency and PRI consistency thresholds remained unchanged, and the configuration sufficiency analysis was re-conducted. The robustness test results indicated that the outcome remained stable.
• When the consistency threshold was increased from 0.80 to 0.85 and the PRI consistency from 0.70 to 0.75 (other parameters unchanged) [31], five out of six configurations remained stable, with only the C4 path showing a moderate decline due to sample heterogeneity, indicating overall robustness of the findings.

3.4. Configuration Path Correlation Analysis

To elucidate the synergistic interdependencies among core antecedent conditions, a co-occurrence heatmap (Figure 4) was constructed based on the six configuration pathways (C1–C6) delineated in Table 3.

This visualization quantifies pairwise condition interactions, with chromatic intensity reflecting the frequency of joint occurrence across configurations. Notably, external environmental impact (EI) demonstrated the most pronounced synergistic effects, exhibiting co-occurrence with safety culture (SC) in three configurations (C1, C2, C4) and with safety management systems (SS) in four configurations (C3, C5, C6). These empirical patterns corroborate the fsQCA findings, wherein EI persistently functioned as a regulatory modulator of internal governance efficacy through cross-configurational interplay.

4. Discussion

Compared to studies applying only the 24Model or only fsQCA, this study provides a more systematic mapping of failure paths. This study integrates 24Model accident causation theory and fsQCA to systematically identify multidimensional configurations associated with safety management performance in small and medium-sized private coal mining enterprises (SMPCMEs). Through configurational analysis of 40 sudden incidents in China (2013–2023), three core findings emerge:

• Multidimensional Failure Mechanisms. safety management performance deficiencies stem from synergistic interactions across organizational, individual, and external levels. Four critical configuration paths were identified: the Internally Balanced Type with “culture-behavior-environment” negative coupling; the Safety Culture–Deficient Type characterized by institutional hollowing and regulatory disconnection; the Cultural–External Environment Type reflecting policy-implementation paradox; and the External Environment–Integrated Management Type marked by technological lag and institutional failure.
• Tailored Governance Strategies for Configuration Types. For the Internally Balanced Type (C1/C2), safety culture revitalization programs should prioritize IoT-based monitoring networks to detect gas anomalies and roof displacement risks in real time, deploying Zheng et al.’s [32] system that integrates wireless sensors and Ethernet architectures, this directly intervenes in environmental risk transmission pathways through real-time dynamic monitoring and holistic personnel-equipment-environment data sharing. For Safety Culture–Deficient Type (C3), government regulators shall implement direct digital oversight to audit hazard rectification records, supported by anonymous reporting platforms for violation exposure, with mandatory management accountability agreements. Ningde City handled multiple industrial safety incidents through its Emergency Command and Rescue System, integrating enterprise-reported hazard source data and rectification reports. Regulatory authorities conducted audits via real-time data review to ensure timely issue resolution, enhancing overall safety culture. The Cultural–External Environment Type (C4) needs government-enterprise risk data hubs integrating regulatory databases and real-time operational metrics, which enable joint enforcement and emergency drills in high-risk zones. Chongqing utilized the data hub of the “Industrial Brain” platform to organize multiple simulated accident drills. Enterprises shared real-time operational indicators, and government command centers coordinated response actions. Deployed in high-risk chemical industrial parks and other zones, this initiative enhanced emergency response capabilities. The External Environment–Integrated Management Type (C5/C6) should adopt AI-powered robots for underground inspections and enhance infrastructure resilience through adaptive ventilation systems, implementing Tang’s [33] intelligent inspection system which automates hazard identification via multi-sensor integration and resolves data contention through directional queue buffering. Empirical results demonstrate over 8× improvement in inspection efficiency with 99.3% anomaly detection accuracy.
• This study identifies configurational pathways that arise from interactions among organizational, individual, and external factors. These pathways provide valuable insights for understanding safety failures extending beyond SMPCMEs.Similar systemic vulnerabilities exist in other high-risk domains involving geotechnically sensitive structures (e.g., tunnels, dams, heritage buildings) or complex industrial processes. Failures in such contexts, akin to those analyzed here, rarely stem from isolated causes; instead, they typically arise from synergistic interactions among management deficiencies, human factors, technological limitations, and external pressures (including environmental stresses and gaps in regulatory oversight) [34,35]. The integrated 24Model–fsQCA approach offers a transferable framework for diagnosing these multidimensional failure mechanisms across sectors. These sectors share similar complexity characteristics with SMPCMEs. To address limitations related to sample size and dynamic policy variables, future research should expand samples across different countries or high-risk industries, incorporate more accident cases, collect panel data, or design longitudinal studies. This will capture how policy changes and evolving safety management practices dynamically influence configurational paths. Such extensions will further validate the transferability of configurational approaches across high-risk industries.