With the growing complexity of sociotechnical systems in maritime transportation, system-theoretic approaches such as STPA, STAMP, and CAST have emerged as effective tools through their focus on control problems. To strengthen both theoretical foundations and practical implementation in the industry, it is necessary to review the development of systems theory within this domain, addressing current limitations and outlining future research directions. This article presents a comprehensive review of system-theory applications in maritime transportation, encompassing modelling processes, application areas, and associated challenges. A bibliometric analysis of 76 papers published between 2015 and 2024 was conducted to identify thematic clusters, keywords, and research trends. The review examined how systems theory has been applied, the data sources employed to generate models, and how these methods are integrated with conventional risk assessment approaches. Findings indicate that STPA is the most widely adopted technique in the maritime sector for systems-based safety analysis. More than half of the reviewed studies applied systems theory to risk analysis for autonomous vessels, demonstrating researchers’ confidence in STPA’s capacity to address the challenges of highly complex systems such as autonomous ships. At the same time, there is a growing interest in combining STPA with other methodologies to produce quantifiable results for safety assessments. Consequently, an increasing proportion of studies are evolving towards integration with computational methods, algorithms, and Quantitative Risk Assessment (QRA). Overall, this paper provides a critical overview of system-theory applications in the maritime domain, highlighting limitations while offering recommendations to guide future research and practice.

As maritime operations become increasingly dynamic and safety-critical, supporting effective learning and decision-making remains a central challenge in maritime education. Within such training contexts, feedback plays a central role in helping seafarers reflect on their weaknesses. Building on this, this study examines seafarers' perceptions of artificial intelligence (AI)-based feedback in maritime training by comparing three feedback formats following a COLREGs assessment: traditional feedback, AI-generated global performance summaries, and AI-generated personalised feedback at the question level. Forty-nine seafarers evaluated each feedback format across six instructional dimensions: correctness, sufficiency, usefulness, clarity, adaptiveness, and motivational impact.Statistical analysis confirmed that both AI-based feedback approaches significantly outperformed the traditional method by providing more relevant guidance, better addressing individual learning needs, and enhancing the overall instructional value. Participants expressed a clear preference for AI-generated feedback tailored to individual questions, which they described as more motivating, clearer, more informative, and more learner-centred. These findings highlight the potential of AI to deliver scalable, faster, and more cost-efficient feedback to support engagement and learning in maritime training.

While human factors are recognised as the primary cause of maritime accidents, the discipline's overarching structural and methodological landscape remains unmapped. Following the PRISMA 2020 guidelines, this systematic review evaluated 1,516 maritime human factors studies published through 2026. To manage this scale, screening and pre-classification across a nine-field ontology were conducted using a large language model (LLM), standardised by explicit criteria and validated by expert review. The findings show a relatively emerging and highly concentrated field, with over 80% of the literature published since 2016. However, its structural makeup shows a significant methodological lag: while computational and predictive methods are advancing, theoretical frameworks have not kept pace. More than half of the corpus (55.6%) discusses the “human element” without adopting a dedicated human-centred approach, a shortcoming that is exacerbated when machine learning methods are employed. Moreover, although accident investigation reports form the primary empirical evidence base in the examined dataset (22.1%), only 1.1% of studies utilise scalable text-mining techniques. As a result, the field depends heavily on manual, unscalable extraction of human factors data. Research attention also misaligns with empirical realities of causality, leaving critical factors such as fatigue and communication severely underinvestigated. Synthesising these mismatches, the review identifies six fundamental, evidence-based gaps, spanning methodological scalability, theoretical framing, and empirical validity, that collectively isolate retrospective accident analysis from prospective risk prediction. Bridging these gaps is essential to systematically integrate historical casualty data with advanced risk modelling.

The maritime domain has relied on advanced technology that has been developing progressively, which increases the risk associated with new challenging threats regarding cybersecurity. Many researchers and technical reports mainly focus on the technical measures to prevent cyber-attacks; however, the human factor is still the crucial reason for cyber-attacks, similar to maritime accidents. This study aims to quantitatively assess human factors' role in cybersecurity by evaluating the seafarers' perception of cybersecurity risks and best practices in the maritime domain. The structured questionnaire was designed to measure the seafarers' cybersecurity awareness, perceptions, and knowledge, as well as their understanding of cybersecurity rules and protocols and their ability to identify and respond to potential cyber-attacks. The collected data was analysed using statistical methods to identify the relation between human factors and cybersecurity domain. The finding reveals that an organization's cybersecurity policies and guidelines influence an individual's security-related behaviours. Additionally, cybersecurity perception, knowledge, awareness, and behaviour of seafarers are positively associated. The study's results would have significant implications for maritime organizations, shipping companies, and training and education centres, which would be needed to develop effective policies, strategies, and tailored training programmes to address the specific needs of seafarers.

Cyber-attacks can compromise the reliability of onboard digital systems, navigation, sensing, control and decision support. An undetected attack may mislead bridge teams and increase the risk of collisions and groundings. Building maritime cyber resilience is therefore essential to the safety of navigation. In this study, 61 certificated seafarers participated in an experimental full-mission bridge simulator study replicating an Istanbul Strait transit. Participants were allocated to a test group or a control group and were exposed to a timed GNSS (Global Navigation Satellite) System spoofing event in a single continuous run. A multimodal dataset was acquired using physiological measures, including EEG and eye-tracking, together with objective and subjective instruments. Relative to the routine briefing, the cyber-situational awareness briefing produced statistically significant improvements across modalities. The test group exhibited higher EEG-derived vigilance (p = 0.027) and lower stress and workload than the control group. Participants' self-reports indicated higher anomaly awareness, more frequent noticing of unusual device behaviour and greater suspicion (p = 0.006; p = 0.004). Eye-tracking metrics revealed more independent cross-verification, with reduced reliance on the primary electronic display. Objective performance, summarised by the composite performance index, was greater in the test group (p < 0.001), and the proportion of safe passages was higher. This study investigated whether a situational awareness intervention favourably shifts human factors, trust calibration, and performance under cyber-attacks, specifically GNSS spoofing. To our knowledge, this research provides the first empirical evidence from seafarers exposed to cyberattacks in a simulator environment and offers new insights into maritime human factors.

Traditional wooden cargo vessels, or Pelayaran Rakyat (Pelra), remain vital to Indonesia's inter-island logistics but face persistent operational and financial constraints. This study assesses policy-performance misalignment in Pelra by combining policy cycle analysis with stakeholder-based the Simple Multi-Attribute Rating Technique (SMART) Key Performance Indicators (KPI) ranking. Stakeholder perspectives were obtained from 70 semi-structured interviews across seven key maritime groups. A multi-criteria decision-making (MCDM) approach using the SMART was applied to evaluate five KPI domains: Operational Safety (OS), Accessibility to Finance (AF), Integration into National Logistics (NI), Infrastructure Support (IS), and Environmental Compliance (EC). The findings reveal pronounced perception asymmetries among stakeholders and identify severe structural gaps, particularly in AF and EC, which hinder Pelra's participation in national maritime programs. Despite strong policy intent, fragmented enforcement, regulatory distance, and limited stakeholder engagement remain key barriers. Ultimately, this study proposes an evidence-based, participatory policy framework to support adaptive reform, improve access to financing and environmental compliance, and strengthen the logistical role of traditional shipping in Indonesia.

Maritime operations involve multinational crews operating in safety-critical and time-constrained environments. Differences in professional background, training pathways, and prior learning experiences increase the need for training approaches that can accommodate learner heterogeneity while remaining operationally efficient. Learner profiling plays a central role in enabling such adaptive training, but lengthy assessment instruments can increase profiling time and reduce response quality, limiting their practical applicability in maritime contexts. In this study, we introduce a data-driven method to develop a 20-item short form of the Felder-Silverman Index of Learning Styles (ILS) tailored for maritime trainees. The responses of 155 experienced seafarers were analysed using the full 44-item ILS questionnaire, and the most descriptive five questions within each dimension were identified using ANOVA F statistics. The resulting 20-item form was evaluated on an independent test set. Across all four dimensions, the short form achieved balanced accuracies of 0.84-0.89 and F1 scores of 0.83-0.96, with the Input dimension performing best (accuracy = 0.89, F1 = 0.96). The proposed short form substantially reduces respondent burden while preserving predictive performance, enabling efficient learner profiling and supporting the design of adaptive maritime training.

Near-miss incident reports provide valuable but underutilised insights for improving maritime safety. However, existing research emphasises failures, with limited focus on positive safety practices. This study addresses this gap by examining connections between incidents and preventive measures to identify what works and what doesn't. To overcome the limitations of traditional topic modelling approaches, this study applies BERTopic, a transformer-based method. Using this method, we analysed 4360 near-miss reports spanning five years from a ferry operator to identify recurring risk patterns and the preventative measures linked to them. GPT-assisted labelling improved interpretability, with outputs validated through structured multi-expert review. Over 75 distinct incident topics were identified across operational, environmental, and human factors domains, aligning with existing categories and uncovering previously unclassified risk themes. Time-series analysis highlighted seasonal variations, with passenger-related risks peaking in summer and infrastructure failures more common in winter. Prevention measures were modelled separately, producing 87 topics, with association rule mining identifying clear linkages between incident types and mitigations; for example, navigation-related incidents were frequently linked to improved bridge communication. Overall, combining unsupervised topic modelling with LLM-assisted labelling produces interpretable and operationally meaningful outputs. This approach offers a scalable framework for prioritising preventative measures using data-driven evidence.

Human error continues to account for most maritime accidents, highlighting persistent challenges in safety management and organisational learning. Although post-incident investigations generate rich narrative data, extracting meaningful insights is hindered by manual coding, subjective interpretation, and inconsistent application of human-factors taxonomies. Despite increasing interest in applying NLP to safety analysis, no existing approach provides a taxonomy-ready, domain-adapted framework capable of reliably classifying fine-grained human and organisational factors in maritime accident narratives. This study presents a natural language processing (NLP) and machine learning (ML) framework to automate the identification of human and organisational contributory factors in maritime accident reporting using an adapted SHIELD taxonomy. The approach analyses individual causal and contributory text segments extracted from accident investigation narratives, which are classified into human and organisational factor codes. Central to the approach is SeaSafeBERT, a domain-adapted BERT model fine-tuned on an international corpus of maritime investigations and combined with TF-IDF lexical features to support structured, multi-level classification. To address sparse taxonomy classes, we developed SHIELD-NLP, a consolidated coding scheme preserving SHIELD's intent but ensuring sufficient class representation across a dataset of 330 maritime accident investigations and enabling classification into 53 SHIELD-NLP codes. The system achieved strong performances for many operationally important SHIELD-NLP codes, with several reaching F1 scores above 0.80. Association rule mining further revealed consistent interaction patterns between perceptual, cognitive, and organisational factors across incident types. Together, these contributions represent, to our knowledge, one of the first domain-specific maritime safety language models and integrated pipelines for automated SHIELD-based analysis, demonstrating the potential of AI-enabled methods to accelerate incident learning, improve code consistency, and strengthen proactive safety management in maritime operations.

Abstract The use of alternative fuels such as methane, methanol, and ammonia is quintessential for the shipping decarbonization. However, to minimize their carbon footprint and achieve net-zero targets, these fuels’ production must involve energy from renewable sources. This study aims to perform a comparative thermodynamic assessment of methane, methanol, and ammonia production pathways using renewable energy, determining the efficiencies of the renewable electric power conversion to chemical fuel power. The methodological approach defines the fuels production process boundaries and considers both electrical and thermal power demands to identify the dominant sources of irreversibility. A thermodynamic approach is applied considering the e-fuel production of 1,000 kg, along with the inflows of energy, compounds, and chemical reactions for each e-fuel. The required input parameters for the involved chemical conversion processes are identified by thorough literature review. The results demonstrate that the second-law efficiency for e-methane, e-methanol, and e-ammonia production is 29.2%, 32.7%. and 30.9%, respectively. Most irreversibilities occur due to the electrolysis required to produce the hydrogen feedstock. The derived results are used to discuss the e-fuels’ suitability for shipping, supporting system optimization, and evidence-based decision-making for maritime decarbonization strategies, while informing stakeholders and policy makers, hence providing, to the best of our knowledge, a first-of-its-kind input to policymakers pertaining to e-fuels production insights.

The maritime industry is highly concerned about cybersecurity due to the growing utilisation of digital technologies for navigation and other operations on-board ships. The connectivity of the systems creates vulnerabilities that make crucial ship functions prone to cyber-attacks, such as manipulating Global Positioning System (GPS) signals through spoofing. These threats provide substantial risks, particularly while navigating congested waterways. Identifying spoofing attacks requires consistent alertness and sophisticated monitoring from navigation professionals already dealing with demanding workloads. This research comprehensively analyses human reliability, specifically in detecting GPS spoofing on Electronic Chart Displays and Information Systems (ECDIS) in congested waterways. In order to accomplish this, the Human Error Assessment and Reduction Technique (HEART), as a robust tool, is expanded using the Evidential Reasoning (ER) approach. The HEART provides a comprehensive tool for calculating human error probability, whereas the ER incorporates raters' assessments in decision-making. The research findings indicate that human reliability for detecting GPS spoofing on ECDIS in congested waterways is found to be 6.74E-01. In addition to having practical implications for marine cybersecurity, the suggested technique demonstrates how a thorough understanding of human errors can be obtained. This allows for systematic quantification of the probabilities of human errors associated with identifying spoofing to enhance operational reliability.

In this study, the dominant process parameters in the air-jet continuous galvanizing line on coating thickness were estimated by computational fluid dynamics and machine learning approaches. First, 128 different cases consisting of different levels of process parameters were created with the Taguchi method. Then, numerical analyses were performed for each case, calculating the maximum pressure gradient and maximum shear stress values on the strip, which were then used in the analytical model developed based on one-dimensional lubrication theory to obtain coating thickness values. Lastly, artificial intelligence techniques based on different machine learning algorithms such as K-Nearest Neighbors, linear regression, random forest and Adaboost, the relative effects of the process parameters influencing the coating thickness were compared through the feature importance values. It was observed that the dominant process parameters differ in low and high jet pressure cases. Accordingly, in the case of low jet pressure, air jet pressure, nozzle slot opening and velocity of the steel strip stand out as the dominant parameters, while in the case of high jet pressure, the most effective parameters influencing the coating thickness are air jet pressure and nozzle slot opening. In addition to this, the effect of the distance between the nozzle and the zinc pot influencing the coating thickness can also be neglected in both low and high pressure cases. Moreover, it was also noticed that the effects of nozzle angle and the distance between the nozzle and the steel strip influencing the coating thickness increase with increasing jet pressure.

The fisheries industry faces increasing sustainability challenges from environmental, economic, and social perspectives, which directly affect fishing vessels as its primary infrastructure. This study conducted a systematic literature review following the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) 2020 guidelines to evaluate technological innovations that improve the sustainability of fishing vessels. Comprehensive searches were performed in Scopus, Web of Science, ScienceDirect, and IEEE Xplore, covering the period 2020–2024. The searches identified 756 articles, of which 105 met the predefined eligibility criteria after screening titles, abstracts, and full texts. Each innovation was categorised and analysed based on its functional vessel domain, contribution to environmental, economic, and social sustainability, maturity level using the Technology Readiness Levels (TRLs) framework, and relevance to Circular Economy (CE) principles. The results indicate that most innovations focus on environmental sustainability, particularly on emission reduction and energy efficiency. Social sustainability remains under-addressed, especially in terms of labour conditions and gender equality. CE principles are present in some initiatives but are not yet fully integrated into vessel design or operation. Most innovations are at medium TRL stages, with adoption limited by financial, infrastructural, and institutional barriers, especially in small-scale fisheries. Future research should address these gaps by enhancing CE integration and promoting a more balanced attention across all three sustainability dimensions.

The concept of maritime circularity has gained increasing attention to address challenges arising from the net-zero targets of the maritime industry. The circular economy provides potential solutions to address these challenges through reuse, remanufacturing, and recycling practices. However, the industry faces complex challenges, including inefficient reverse supply chains, a lack of awareness about circular economy principles, standardisation issues, and the need for digital infrastructure to provide vital information in the sector. These challenges prevent the implementation of circularity practices, as access to crucial data throughout the vessel’s life cycle is obstructed. This novel research aims to create a robust first-of-its-kind database solution specifically designed to support the industry’s shift towards circularity. The database will facilitate fast and transparent information flow between the stakeholders, providing foundations for asset tracking and a robust reverse supply chain. A case study was conducted to show that a database could help extract higher financial value from end-of-life ships by over 80%. The ageing fleet increases the urgency of utilising such a database, which could be a pivotal strategy for a sustainable and circular industry. This digital solution offers significant benefits to all industry stakeholders and allows holistic resource management, influencing maritime operations’ sustainability, resilience, and profitability.

A feasibility study was conducted on the energy and peak power demand of ships for utilising the Onshore Power Supply (OPS) and transitioning to using alternative fuels. The port of Plymouth was adopted as a case study. Four types of ships, Ro-Pax, Tanker, Bulk Carrier and General Cargo, were in operation at the port. A representative vessel was selected for each ship type to simulate the average ship's cargo capacity and engine power. One year of real port operations, including material handling equipment and trucks, were simulated. The peak power and annual energy demand for the OPS system were calculated to be 5.95 MW and 7.1 GWh, respectively. Implementing an OPS system saved 83.6% of total CO2. Fuel volumes were calculated for conventional and alternative fuels, the volume of liquid hydrogen was around 3.5 times that of the conventional fuel, whereas methanol required less mass and volume than ammonia and hydrogen.

This study introduces an innovative deep-learning approach for fuel demand estimation in maritime transportation, leveraging a novel convolutional neural network, bidirectional, and long short-term memory attention as a deep learning model. The input variables studied include vessel characteristics, weather conditions, sea states, the number of ships entering the port, and navigation specifics. This study focused on the ports of Jazan in Saudi Arabia and Fujairah in the United Arab Emirates, analyzing daily and monthly data to capture fuel consumption patterns. The proposed model significantly improves prediction accuracy compared with traditional methods, effectively accounting for the complex, nonlinear interactions influencing fuel demand. The results showed that the proposed model has a mean square error of 0.0199 for the daily scale, which is a significantly higher accuracy than the other models. The model could play an important role in port management with a potential reduction in fuel consumption, enhancing port efficiency and minimizing environmental impacts, such as preserving seawater quality. This advancement supports sustainable development in maritime operations, offering a robust tool for operational cost reduction and regulatory compliance.

The maritime sector predominantly relies on subjective evaluations of seafarers' skills and experience in conventional recruiting procedures. Nevertheless, subjective evaluation methods are highly susceptible to biases and inconsistencies. This study proposes a novel recruitment process within the maritime industry by merging psychological tests and machine learning methods in the recruitment process. Using psychological tests such as MMPI-I as features in machine learning methods for the recruitment process represents a step-change approach within the industry to promote more objective assessments of maritime professionals during recruitment processes and identify suitable candidates based on data from same-rank maritime professionals. This new methodology contributes innovatively to traditional maritime sector recruitment methods and potentially addresses a significant gap in the existing literature. The proposed methodology aims to predict future values by analysing existing data sets. Data were collected from 183 volunteer cadets from different backgrounds using an application form and the MMPI-I Personality Inventory. The dataset was classified using several machine learning algorithms, and their performance metrics were compared. The top five classification algorithms (Decision tree, PNN, random forest, gradient boost trees, and naive Bayes) with the best performance were evaluated, and the most accurate classification performance was achieved with the GBT algorithm. The results show that the highest values are for Gradient Boosted Trees (86%) and Random Forest (80%). The GBT algorithm has scored higher values for other metrics as well. The findings of this study indicate that it is possible to train algorithms that can adequately forecast the credentials and appropriateness of marine recruits. Through more development, these data-driven solutions could enhance existing subjective recruitment practices. This approach could improve the objectivity and prediction accuracy of marine recruiting processes, hence facilitating the selection of highly qualified individuals.

Anchoring is one of the frequent processes undertaken by the crew on-board ship. Improper anchoring has significant consequences, such as anchor dredging, which may pose serious risks to the ship crew, marine environment and commodities. This paper aims to present a robust approach combining bow-tie approach with SLIM (Success Likelihood Index Method) under fuzzy sets theory. Since the fuzzy bow-tie represents a comprehensive quantitative risk assessment (QRA) method, the SLIM predicts the probability of a human error occurring during a specific task's completion. In this context, anchor dredging risk can be calculated for a specific case. This paper's novelty lies in consideration of human error's contribution to risk for ship anchor dredging and combining quantitative and qualitative analysis in maritime risk assessment. The finding of the paper shows that the occurrence probability of anchor dredging risk for cargo ships is 7.31E-02. Besides its robust theoretical background, cargo ship owners, superintendents, safety inspectors, and health, safety, environment and quality (HSEQ) managers can utilise the paper's findings to minimise potential risk and enhance operational safety during anchoring operations.