Maritime transportation through the Turkish Straits (Istanbul and Çanakkale) connecting the Black Sea to the Mediterranean represents a critical global corridor with a high traffic density of approximately 48,000 commercial vessel transits annually. These waterways present unique operational challenges, including strong currents (Bosporus currents reaching approximately 2.1–3.1 m/s), variable weather conditions, particularly in the Sea of Marmara, confined navigation channels with limited maneuvering space, and complex vessel-traffic interactions among large commercial ships, tankers, passenger ferries, and small fishing vessels. Despite the continuous safety improvements and regulatory enhancements aligned with International Maritime Organization (IMO) conventions, maritime accidents persist globally. Recent data from the Allianz Global Corporate and Specialty (AGCS) Safety and Shipping Review 2024 documented 2,951 maritime casualties and incidents globally in 2023, representing a 3% reduction from 2022, yet underscoring the continuing challenge of maritime safety management (Inmarsat Maritime, 2024). Turkish waters experience similar patterns, with accidents resulting in loss of life, property damage, environmental pollution through oil spills and cargo releases, and disruption to global trade flows (Kamal & Çakır, 2022; Huang et al., 2023; Arıcan, 2024). The complex nature of maritime accidents involves multiple interacting factors across technical, operational, human, and environmental dimensions (Chauvin et al., 2013; Weng et al., 2019), requiring systematic analytical approaches to address the causal complexity, accommodate expert judgment uncertainty, and support multi-dimensional decision-making for effective safety intervention design (Uğurlu et al., 2018; Fan et al., 2020). Maritime safety research increasingly uses multi-criteria decision-making (MCDM) methodologies to address these operational challenges through integrated analytical frameworks. Recent developments show diverse methodological approaches. Collision avoidance technologies have been examined through integrated decision-making frameworks (Jansson et al., 2002). Tanker risk evaluations have used enhanced reliability analysis combining cognitive assessment with fuzzy approaches (Zhou et al., 2017), and network-based prevention strategies have incorporated Bayesian frameworks (Fan et al., 2022). Contemporary implementations address vessel docking safety (Li et al., 2024), ferry navigation risk (Pham and Hoang, 2025), and the assessment of grounding probability (Liu et al., 2024). Advances in maritime risk assessment include the integration of artificial intelligence (Wang et al., 2024), data-driven accident prediction models (Kim et al., 2023), and real-time decision support systems (Sur & Kim, 2024).

A comprehensive review of the literature on maritime accident analysis reveals persistent methodological limitations and geographic gaps. First, methodological fragmentation characterizes current research. Although individual analytical methods have demonstrated value, integrated approaches combining multiple perspectives are rare in maritime contexts. Zhou et al. (2024) developed data-driven Bayesian Network models for global maritime casualty analysis identifying risk influencing factors but did not integrate multi-criteria prioritization or performance evaluation. Sur and Kim (2024) applied the ordinal priority approach and Grey Relational Analysis to rank risky maritime accident types but did not reveal the causal structures or assess control effectiveness. Huang et al. (2023) analyzed intercontinental sea accidents in the Mediterranean and Black Seas using weighted association rule mining, focusing solely on characteristic patterns without a comprehensive decision support framework. Guo et al. (2025) analyzed Chinese maritime accidents using DEMATEL with fuzzy cognitive maps but did not prioritize accident types or evaluate performance gaps. Kuzu (2023) applied F-DEMATEL to Turkish anchor loss but did not extend it to comprehensive accident analysis or performance assessment. These single-method approaches answer only one question (causality OR priority OR performance) when maritime safety management requires an integrated understanding of all three simultaneously. Triple-method integration frameworks have been demonstrated in other domains. Abdullah et al. (2023) integrated fuzzy DEMATEL and fuzzy TOPSIS to evaluate Industry 4.0 technologies in manufacturing, showing how combined causal-performance analysis provides superior decision support. Sathyan et al. (2023) applied an integrated F-DEMATEL-F-AHP-F-TOPSIS in fashion supply chain management, showing the effectiveness of the framework in addressing complex multi-factor problems. Cross-sectoral MCDM developments include aircraft procurement (Mathew et al., 2023), sustainability evaluation (Işık & Aladağ, 2016), uncertainty modeling (Kahraman et al., 2018), and infrastructure assessment (Alhadidi & Alomari, 2024). Nevertheless, maritime accident analysis has yet to adopt such comprehensive integrated frameworks. Recent maritime applications combining multiple methods are limited. Port safety assessment (John et al., 2024), collision risk evaluation (Wu et al., 2017), and safety management system implementation (Aykuz & Celik, 2014) represent emerging efforts, but comprehensive causality priority performance integration specifically for accident analysis remains underexplored. Maritime accident analysis using proximity-to-ideal-solution techniques within integrated frameworks constitutes a significant methodological gap (Zhang et al., 2022). Second, inadequate uncertainty treatment persists across maritime accident studies. Most research uses crisp MCDM methods (Hwang & Yoon, 1981; Opricovic & Tzeng, 2004) requiring expert assessments expressed as precise numerical values. On the other hand, maritime experts naturally express judgments linguistically using terms such as ‘moderate influence, ‘high importance, or ‘weak performance’ rather than the exact numbers, reflecting genuine uncertainty from incomplete information, rare event frequencies, and complex causal chains. Imposing crisp numerical values on inherently vague expert assessments creates false precision, potentially leading to misleading conclusions. Fuzzy set theory provides mathematical frameworks for systematically representing and propagating uncertainty through analytical processes (Zadeh, 1965; Chang, 1996; Büyüközkan & Çifçi, 2012). Recent advances revealed the effectiveness of fuzzy logic in maritime contexts. Sur and Kim (2024) applied fuzzy evaluation for comprehensive maritime risk estimation, Fiskin (2023) combined fuzzy logic with MCDM for ship domain determination in high-density waterways, and Qu and Wang (2025) developed multi-source data-driven navigation safety frameworks integrating computational intelligence with fuzzy Bayesian networks. Fuzzy TOPSIS implementations address diverse decision contexts, with comprehensive analytical reviews documenting 184 applications revealing methodological evolution (Palczewski & Sałabun, 2019), while bibliometric studies confirm sustained research momentum (Ayan & Abacıoğlu, 2022). Methodological diversity includes rough set integration (Abdullah et al., 2023), machine learning hybridization (Nandi et al., 2024), best-worst methods (Haseli et al., 2024), Industry 5.0 applications (Anbarkhan, 2023), and randomized weighted fuzzy AHP for financial decision-making under uncertainty (Vasantha Lakshmi & Udaya Kumara, 2024). Despite these advances in fuzzy MCDM methods (Jiang et al., 2020; Gao, 2024), comprehensive maritime accident analysis integrating fuzzy approaches throughout the entire analytical pipeline—from initial expert assessment collection through final ranking generation—remains limited. Although some studies applied fuzzy methods to single analytical components, systematic uncertainty representation that ensures consistency and prevents loss of information during multi-stage analytical transitions represents an unmet need. Third, the lack of comprehensive multi-perspective analysis limits the effectiveness of decision support. Single-method studies identified either causal relationships among factors, priority rankings of accident types, or performance gaps in safety management, but they rarely integrated these complementary analytical perspectives within unified frameworks. Recent maritime safety research shows this fragmentation. Multi-source data-driven frameworks combining computational intelligence and Bayesian networks focus primarily on risk prediction without performance evaluation (Qu & Wang, 2025), whereas Grey Relational Analysis applications prioritize risky accident types without revealing causal structure (Sur & Kim, 2024). A Comprehensive understanding requires simultaneously addressing causality (what factors drive accidents?), priority (which accident types warrant the greatest attention?), and performance (where do current practices demonstrate critical deficiencies?). Isolated single-perspective analyses provide incomplete decision support for complex maritime safety challenges (Govindan & Chaudhuri, 2016; Stević et al., 2020), but recent integrated approaches in non-maritime domains show promise for adaptation (Zhou et al., 2024; Jansson et al., 2002). Fourth, geographic concentration and temporal focus characterize existing maritime accident research. Studies reveal a concentration in European waters, Chinese ports, and North American contexts, with limited focus on Turkish maritime operations despite unique operational characteristics. Recent global perspectives include Zhou et al. (2024), who analyzed worldwide casualty data across multiple regions, and Huang et al. (2023), who examined the accident characteristics in the Mediterranean and Black Seas accident, yet the distinctive features of the Turkish Straits remain understudied. Turkish waters present specific challenges: extremely high traffic density (48,000 annual transits through confined waterways), strong currents (Bosporus currents reaching 2.1–3.1 m/s), complex regulatory environment including stringent Turkish Environmental Code enforcement (44% fine increase from 2024 to 2025 adjusting for inflation and compliance requirements), European Union accession dynamics, evolving IMO convention implementation, and mixed fleet composition ranging from modern containerships to aging tankers and traditional fishing vessels. Recent Turkish maritime studies addressed specific aspects. Demirci & Gülmez (2021) analyzed Ro-Ro ship accidents using the Human Factors Analysis and Classification System (HFACS). Kamal and Çakır (2022) applied a data-driven Bayesian approach to the accident patterns in the Istanbul Strait, Demirci et al. (2022) examined human-error-related Ro-Ro ship accidents using a case-based HFACS analysis. Kuzu (2023) examined anchor loss incidents, and Arıcan (2024) analyzed accidents in Turkish territorial waters. Despite these studies, a comprehensive accident analysis integrating fuzzy MCDM methodology that simultaneously addresses multiple accident types and risk factors remains absent. Geographic diversity in maritime accident research strengthens the understanding of context-specific risk patterns and enables broader applicability of findings (İltar et al., 2025). This study addressed these identified gaps by developing an integrated F-DEMATEL-F-AHPF- TOPSIS methodological framework that combines causal relationship analysis, accident type prioritization, and contributing factor performance assessment within a unified analytical approach. The research pursued five specific objectives: (1) Methodological integration-Develop comprehensive triple-integration framework extending successful applications in manufacturing (Abdullah et al., 2023) and supply chain management (Sathyan et al., 2023) to a maritime safety context, enabling simultaneous causal analysis (F-DEMATEL revealing which factors drive accidents), hierarchical prioritization (F-AHP determining which accident types need priority intervention), and performance assessment (F-TOPSIS identifying where current controls are weakest); (2) Systematic uncertainty treatment-Apply fuzzy set theory consistently throughout entire analytical pipeline from expert assessment collection through final result generation, representing expert judgment vagueness via triangular fuzzy numbers (van Laarhoven & Pedrycz, 1983; Kahraman et al., 2015) rather than imposing false precision, advancing beyond single-component fuzzy implementations demonstrated in recent maritime applications (Sur & Kim, 2024; Fiskin, 2023); (3) Multi-perspective analysis-Provide comprehensive insights from causal, priority, and performance perspectives impossible to obtain through single-method approaches, advancing beyond isolated Bayesian network applications (Zhou et al., 2024; Qu & Wang, 2025) or singular Grey relational analysis (Sur & Kim, 2024); (4) Geographic specificity-Focus analytical attention on Turkish maritime domain through systematic selection and content analysis of 86 accident investigation reports from Turkish waters (2013–2024) combined with expert assessments from 15 maritime domain specialists with extensive Turkish operational experience, contributing to underrepresented geographic research area (İltar et al., 2025); (5) Actionable decision support-Deliver integrated framework producing three complementary outputs enabling evidence-based safety resource allocation, intervention sequencing, and performance improvement targeting based on empirical accident patterns and expert consensus rather than intuition or tradition. This study makes five distinctive contributions differentiating it from existing maritime safety research. The first is triple integration, which represents the first application of the integrated F-DEMATEL-F-AHP-F-TOPSIS framework to comprehensive maritime accident analysis, combining causality determination, hierarchical prioritization, and performance assessment within a unified methodology. Although previous maritime work examines Chinese accident patterns using partial integration (Guo et al., 2025) and Turkish anchor loss using single-method fuzzy DEMATEL (Kuzu, 2023), no study has integrated all three perspectives for multi-category accident analysis. The second is systematic uncertainty treatment, which is a consistent fuzzy approach throughout the entire analytical pipeline rather than an isolated application, ensuring consistent uncertainty representation and preventing information loss during analytical transitions, advancing beyond single-component fuzzy implementations. The third is the geographic contribution, which is the first comprehensive fuzzy MCDM accident analysis for Turkish maritime operations, addressing the underrepresented region with unique operational characteristics (48,000 annual Strait transits, stringent environmental enforcement with 44% increased penalties, mixed fleet composition, and confined high-current navigation). The f is the practical orientation, which is an actionable decision support framework directly applicable to safety resource allocation rather than a purely academic contribution, addressing the persistent research-practice gap where publications identify factors or rank risks but rarely provide integrated guidance for the following: (a) Which factors to 102 Emine Can et al. address first based on causal influence? (b) Which accident types warrant priority investment? (c) Where do current safety controls demonstrate critical deficiencies? Recent safety assessment frameworks focus on specific domains. Qu and Wang (2025) developed a multi-source navigation safety framework, and Tunçel et al. (2023) provided a probability-based risk quantification for specific accident types, but did not provide comprehensive resource allocation guidance across multiple accident categories and risk factors. The fifth is the empirical foundation. Eighty-six systematically selected accident reports combined with 15 expert assessments ensured that the findings were grounded in actual accident patterns rather than hypothetical scenarios, which is consistent with the recent emphasis on data-driven maritime safety analysis (Zhou et al., 2024; Wang et al., 2023). The remainder of this paper is organized as follows. Section 2 presents the materials and methods including the data collection procedures, risk factor identification, justification of fuzzy set theory, and detailed F-DEMATEL, F-AHP, and F-TOPSIS methodologies. Section 3 reports the results from causal analysis, prioritization, and performance assessment. Section 4 discusses the findings, strategic implications, and comparison with the existing literature. Section 5 concludes the paper by outlining the contributions, limitations, and future research directions.

Table 1 presents a categorized comparison with previous maritime safety research to demonstrate the uniqueness and contribution of this study. Although individual studies have applied various MCDM methods to maritime accident analysis, none have integrated causality determination (F-DEMATEL), hierarchical prioritization (F-AHP), and performance assessment (F-TOPSIS) within a unified framework for comprehensive maritime accident analysis.

The integrated F-DEMATEL–F-AHP–F-TOPSIS framework provides a multi-perspective analytical approach that significantly advances the theoretical understanding and practical implementation of maritime safety management. The causal analysis (F-DEMATEL) identified the structural risk and navigation as upstream drivers requiring preventive intervention, whereas Training/Experience and External Factors emerged as downstream vulnerabilities demanding protective reinforcement. This causal hierarchy establishes a sequenced, evidence-based intervention logic; addressing the root causes mitigates the cascading failures, while reinforcing the downstream vulnerabilities enhances resilience against upstream control failures and external stressors. The finding that Training and Experience function as an effect rather than a cause is particularly revealing. This suggests that the deficiencies in training represent symptoms of broader systemic weaknesses rather than isolated operational shortcomings. This interpretation aligns with recent human factors research showing that crew competency gaps often originate from inadequacies in organizational safety management systems rather than individual performance limitations (Qu & Wang, 2025; Wang et al., 2021).

The priority analysis (F-AHP) showed that fatal occupational accidents (40.7%) and fire/explosion incidents (39.8%) collectively accounted for 80.5% of the intervention priority, reflecting the strong empirical consistency between the expert-derived weights and observed frequencies (fatal occupational: 40.7%; fire/explosion: 39.5%). This national prioritization diverges from the global maritime trends, where collision and grounding dominate (Zhou et al., 2024; Huang et al., 2023). The distinction highlights the unique risk environment of Turkish waters—frequent fatal occupational accidents among small commercial vessels lacking safety infrastructure, and elevated fire risks among aging passenger and tanker fleets with outdated suppression systems. Accordingly, resource allocation strategies should follow a data-driven approach: 40% toward fatal occupational accident prevention (protective gear, working-at-height systems, and emergency response capability), 40% toward fire/explosion mitigation (suppression system modernization and firefighting training), and 20% toward other categories. This reallocation revealed the practical translation of MCDM findings into operational planning, addressing the longstanding research–practice gap where academic studies identified the risks but failed to guide actionable interventions (Tunçel et al., 2023; Arıcan, 2024).

The performance assessment (F-TOPSIS) indicated that Navigation represents the most critical performance gap (CC = 0.5462), suggesting that, despite the existing measures, such as pilotage and VTS support, navigational safety performance remains the furthest from the ideal benchmark under challenging operational conditions. The structural Risk exhibits the second-largest gap (CC = 0.6024), indicating continued exposure related to the vessel condition and equipment reliability. The psychosocial risk showed moderate gaps (CC = 0.6562), while eternal factors showed comparatively stronger performance (CC = 0.6986). In contrast, training and experience achieved the highest performance score (CC = 0.7586), suggesting that competency-related controls are relatively closer to the ideal benchmark than other dimensions.

The observed paradox—Structural Risk as the strongest causal factor among the better-performing areas—suggests that high investment in technical safety must be complemented by human and environmental risk mitigation. Effective upstream prevention must align with downstream protection, reinforcing the findings from recent research showing that holistic maritime safety depends on integrating technical, human, and organizational dimensions (Sur & Kim, 2024; Abdullah et al., 2023). This integrated framework uniquely exposes such interdependencies, validating its methodological strength. The framework also supports phased implementation: (1) Immediate (0–6 months)—simulation-based emergency training and competency enhancement; (2) Short-term (6–12 months)—environmental resilience measures including weather routing and visibility management; (3) Medium-term (12–24 months)—psychosocial support and fatigue management programs; (4) Ongoing—preventive maintenance for sustained structural integrity; and (5) Continuous—navigation system and pilotage upgrades. This sequential prioritization ensures efficient resource use while achieving systemic improvement, consistent with the recent implementation guidance (Qu and Wang, 2025; Inmarsat Maritime, 2024).

Context-specific strategies are essential for Turkish waters, which feature confined high-density traffic, strong Bosporus currents (2.1–3.1 m/s), frequent fog in the Sea of Marmara, and high vessel aging rates. Recent national initiatives—VTS expansion, pilotage enforcement, and environmental compliance intensification (44% penalty increase from 2024–2025)—illustrate progress, but integrated frameworks addressing all identified risk factors are limited (Arıcan, 2024; Demirci & Gülmez, 2021).

A comparison with existing studies confirmed the alignment and advancement. Guo et al. (2025) reported causal mapping via F-DEMATEL in China but did not conduct performance analysis. Kuzu (2023) used F-DEMATEL to assess Turkish anchor loss, and Sathyan et al. (2023) applied the integrated fuzzy framework in supply chain contexts. This study extended these applications by incorporating five accident categories, integrating performance assessment (F-TOPSIS), and grounding results in Turkish empirical data, thereby contributing geographic specificity and methodological breadth. Similarly, Zhou et al. (2024) and Sur and Kim. (2024) conducted global-scale analyses, while the present research added regional granularity and multi-dimensional insight. Collectively, the framework addressed all four research gaps:

Recent MCDM innovations, such as entropy weighting (Delgado & Romero, 2016), spherical fuzzy sets (Gündoğdu & Kahraman, 2018), and grey relational analysis (Wei, 2011), offer valuable complementary techniques used across sustainability, energy, and port studies (Akram et al., 2022; Li et al., 2024; Gao,2024). Nevertheless, the F-DEMATEL–F-AHP–F-TOPSIS framework remains particularly suitable for maritime accident analysis because of its balance of methodological rigor and operational feasibility. Its strengths include the following: (1) integrated causal-priority insights; (2) transparency of expert-driven evaluation; (3) comprehensive yet computationally efficient fuzzy treatment; and (4) robust validation (CR = 0.086; ρ > 0.85). These results provide credible, defensible recommendations for maritime safety management and resource allocation (Vasantha Lakshmi & Udaya Kumara, 2024).

Fuzzy number representation throughout the analytical pipeline effectively captured the expert uncertainty—a critical feature given limited statistical data and the high stakes of maritime safety decisions. For example, the Fatal Occupational Accident fuzzy weights (0.348, 0.407, and 0.465) accurately represent the consensus variability (±6%), while the Training/Experience (CC range [0.71, 0.81]) confirms the ranking distinction under uncertainty. This explicit representation enhances transparency and decision defensibility, enabling policymakers to balance confidence and caution when allocating safety resources (Sur & Kim, 2024; Wang et al., 2021). 108 Emine Can et al.

This study developed an integrated F-DEMATEL–F-AHP–F-TOPSIS framework to conduct a comprehensive analysis of maritime accidents in Turkish waters, addressing the critical methodological and geographic gaps in maritime safety research. The analysis combined 86 accident investigation reports (2013–2024) with 15 expert assessments, providing a robust empirical and experiential foundation. The causal analysis (F-DEMATEL) identified the Structural Risk and Navigation as the primary upstream factors requiring preventive intervention, while the Training/Experience and External Factors emerged as downstream vulnerabilities demanding enhanced protective measures. The priority assessment (F-AHP) identified the Fatal Occupational Accidents (40.7%) and Fire/Explosion incidents (39.8%) as the dominant safety priorities with strong empirical consistency between the expert weights and observed frequencies. The performance evaluation (F-TOPSIS) revealed Navigation as the most critical control deficiency (CC = 0.5462), highlighting the need for reinforcement of the urgent navigational system, while Training and Experience showed the strongest relative performance (CC = 0.7586).The framework contributes to the field by (1) introducing the first integrated causality–priority–performance assessment for maritime accidents, (2) implementing systematic fuzzy set theory across the analytical pipeline to preserve expert uncertainty, (3) providing multi-perspective analysis that reveals interdependencies among risk factors, (4) offering a geographically grounded contribution focused on the underrepresented Turkish maritime context, and (5) delivering actionable decision support, enabling evidence-based safety resource allocation following a 40%–40%–20% distribution and a phased implementation strategy.

Although the research exhibited strong methodological validity, certain limitations remain. The geographic concentration on Turkish waters may constrain global generalization, and the 11-year temporal window may not fully capture the evolving risk dynamics. The 15-expert panel, though diverse, represents primarily national perspectives, and the five-factor model prioritizes analytical tractability over fine-grained differentiation.

Future studies should extend the framework to other confined and high-density waterways, conduct longitudinal and cross-national comparative analyses, and explore hierarchical or hybrid extensions that integrate entropy-based or spherical fuzzy weighting for enhanced uncertainty modeling.

Overall, the integrated F-DEMATEL–F-AHP–F-TOPSIS framework provides a replicable, data-driven, and decision-oriented methodology applicable to diverse maritime contexts worldwide. These findings confirm that sustainable maritime safety improvement requires simultaneous attention to causal origins, risk priorities, and performance deficiencies—a multidimensional insight unattainable using single-method approaches.