This paper presents a large-scale, data-driven case study using real operational and meteorological data to optimize maritime vessel routing by jointly considering fuel consumption, voyage time, and safety risks under diverse environmental conditions. A multi-objective optimization framework based on a weighted-sum approach was applied to systematically capture trade-offs among objectives. In practice, selecting appropriate objective weightings, particularly for safety, is challenging, as planners often lack quantitative guidance on their operational implications. This study addresses that gap by evaluating a broad range of weighting scenarios and identifying when safety criteria have minor, moderate, or major impacts. Five major safety issues, namely dynamic stability, bow slamming, green water on deck, and newly developed parametric rolling and surf-riding/broaching-to risk functions, were incorporated through combined critical and non-critical penalties. The case study covered multiple global routes over an entire year and across different vessel sizes and loading conditions. Results show that moderate safety weighting reduces total risk by 30–35% with only a 1–2% increase in voyage cost, while further safety prioritization yields diminishing economic returns. The framework also provides a transferable methodology for systematic parameter selection, supporting transparent and practical multi-objective decision-making in maritime routing.
Platform Supply Vessels (PSVs), which provide essential logistical support to offshore oil and gas installations, are a significant source of CO2 emissions. The emergence of hybrid-electric propulsion technologies, combined with offshore charging infrastructure, offers new opportunities to mitigate these emissions. This paper studies the Hybrid-Electric Platform Supply Vessel Routing Problem with Speed Optimization (HEPSVRP-SO). It considers routing, sailing speed, charging, and power management decisions for a fleet of PSVs with the objective of minimizing operational costs while meeting given emission reduction targets. We formulate the problem as a mixed-integer programming (MIP) model, which is computationally tractable for a commercial MIP solver only for small-scale instances. To address larger instances, we develop a three-stage Path-Flow-Based (PFB) heuristic that (i) generates a set of promising candidate routes, (ii) optimizes sailing speeds and charging decisions along each individual candidate route, and (iii) assigns routes to vessels by solving a path-flow model. Computational experiments using real-world data from the Norwegian continental shelf confirm the heuristic’s effectiveness. The results further indicate that significant emission reductions can be attained at moderate additional cost, and that optimizing sailing speeds is crucial for minimizing these costs.
Prior research on the lead/lag relationship between freight rates and macroeconomic indicators has focused heavily on the Baltic Dry Index (BDI) and global or U.S. macroeconomic indicators. This study aims to extend this research by examining the lead/lag relationship between freight rates and local macroeconomic indicators in 12 countries encompassing the G7 and original BRICS countries. In addition to examining dry bulk freight rates measured by the BDI, this study includes tanker, container charter, and container route rates. Short-term lead/lag relationships are examined through vector autoregressive models, with long-term relationships examined through autoregressive distributed lag models. Effect size relationships are analyzed using impulse response functions. The results show that freight rates have strong predictive ability over industrial production and import/export volume, but are not as effective at predicting inflation and stock prices. In turn, macroeconomic indicators show strong predictive ability over dry bulk and tanker freight rates but little predictive ability for container freight rates.
This paper evaluates multi-mode multi-skill personnel scheduling in Roll-on/Roll-off (RoRo) terminals, focusing on the loading of vehicles onto a RoRo ship by internal and external personnel. In this personnel-intensive loading, vehicles in parking areas of the RoRo terminal are grouped into prioritized batches and driven by teams through a ramp onto the decks of the RoRo ship. The hierarchical objectives are to load the batches so that the highest possible sum of priorities is achieved and the lowest possible number of external teams is used. We formalize the problem by identifying the essential components of decision making and by formulating a Constraint Programming (CP) model. Furthermore, we quantify the problem by comparing the objective values of different multi-skill allocations, a two-stage optimization, and a human planner with those of the CP model. Specifically, we evaluate the resilience of multi-skill allocations of internal teams to changes in those of external teams. Moreover, a two-stage optimization improves the resilience of decision making by increasing the operational flexibility. Our findings on manual personnel scheduling by a human planner indicate that decision support systems improve the loading. This not only contributes to the theoretical understanding of personnel scheduling in RoRo terminals, but also offers managerial implications for RoRo terminal managers to achieve improvements and resilience in their real-life personnel scheduling.
The objective of this research is to assess the interaction of maritime decarbonization policy with Capesize iron ore freight rates and optimal speeds in order for shipowners to maximize Time Charter Equivalent (TCE) earnings in the spot market. Error Correction Models (ECMs) are developed for estimating the equilibrium relationship of freight rates with bunker prices and a proxy for China’s steel margins, for two major trade routes, namely Australia-China and Brazil-China. Freight rates are forecast in scenario analysis and optimal speeds are modeled before and after the International Maritime Organization’s (IMO) Greenhouse Gas Fuel Intensity (GFI) targets and the accompanying Greenhouse Gas (GHG) pricing mechanism are assumed to be implemented. It is found that when the GHG pricing is internalized into bunker prices in 2030, the rise in freight rates may be able to absorb the GFI compliance costs at varying rates across the speed-fuel consumption curves of different conventional Capesize engine specifications. As a result, the incentive to adjust optimal speeds downwards is mitigated, particularly for eco Capesize without scrubber. Instead, when shipowners fully absorb the GFI compliance costs (i.e. GFI costs are treated as external to bunker prices), downward adjustment of optimal speeds may be incentivized more. In addition, the speed that optimizes the Carbon Intensity Indicator (CII) rating is calculated and compared with the optimal speed, in order to establish the trade-off with optimal earnings. Optimizing the CII rating using speed only is found to require a higher deviation from optimal speeds in 2030 compared to 2026.
This paper evaluates the efficacy and efficiency of fleet decarbonization pathways. We compare three different regulatory approaches, including the recent International Maritime Organization (IMO) Net-Zero Framework. To compare each pathway, we define a Strategic Tramp Fleet Renewal and Retrofit Problem with emission reduction integrated with tactical and operations decisions. The strategic decisions are the power system and fuel selection for the fleet over the planning horizon. The tactical and operations decisions include the fleet deployment and the transportation demand satisfaction. We analyze a case study using data from a tramp shipping operator, where diesel, LNG, ammonia, LPG, and methanol are the available fuels. The experimental results confirms that Bio-methanol and E-ammonia are most relevant alternative fuels to achieve consistent emissions reductions. We conclude by studying the marginal contribution of the most relevant fuels, discussing the different trajectories’ cost efficiencies, and observing that the IMO Net-Zero Framework is very expensive for shipping operators.
Quay cranes (QCs) play a vital role in ship-to-shore operations, enabling the seamless transfer of cargo between sea and land. However, increasing trade volumes require faster and more cost-effective container handling, exerting significant pressure on QCs and leading to greater wear on critical components such as wires, hoists, and rope clamps. While operations research has explored maintenance scheduling to improve terminal performance, comparatively little work has examined how machine learning can exploit the growing volume of QC monitoring and operational data to predict breakdowns before they occur. This study contributes to this area by integrating terminal operations data, QC monitoring logs, and meteorological observations into a unified analytical framework. We employ explainable artificial intelligence (XAI), using both global and local SHapley Additive exPlanations (SHAP) to identify the operational and environmental factors most strongly associated with QC failures and to illustrate concrete, instance-level examples of how specific conditions contribute towards breakdowns. In parallel, we develop a robust machine learning pipeline built around nested cross-validation to assess the predictive capability of multiple classifiers for forecasting QC breakdowns. Our XAI analysis reveals that breakdown risk is closely linked to QC working time, the distribution of moves across simultaneously operating QCs, hoist overload and trolley alignment warnings, and adverse weather conditions. Among the evaluated models, LightGBM achieved the highest predictive accuracy, reaching up to 83% in identifying breakdown-prone scenarios. These findings demonstrate the feasibility and value of data-driven predictive maintenance for QCs, providing insights that support safer, more reliable, and more efficient terminal operations.
The maritime industry, particularly deep-sea operations, is facing increasing challenges related to crew shortages, fatigue, and compliance with rest and work-hour regulations. One proposed solution is the concept of a periodically unmanned bridge (B0), in which the bridge may remain unattended under certain operational conditions (e.g. open sea, clear weather, no technical defects). However, safe implementation of B0 requires addressing human, technological and regulatory challenges.This exploratory study presents one of the first empirical evaluations of B0 operations using a full-mission bridge simulator. Eight experienced navigators were interviewed after completing four realistic operational scenarios under two conditions: continuous bridge manning and B0. Post-scenario interviews were triangulated with data on navigator performance, workload ratings, and Situational Awareness (SA) measures.Participants highlighted opportunities and concerns related to B0, particularly regarding trust in automation, failure detection and the need for rapid re-engagement following alarms. They also noted that the concept may require new forms of training, information presentation, and decision-support systems. Results for performance, workload, and situational awareness were broadly similar between the two bridge types. Although tentative due to the limited sample size, the findings indicated a potential advantage for anticipatory situational awareness (projection) in the B0 condition.Overall, the findings suggest that B0 may be feasible within constrained operational envelopes, if alarm strategies, training, and human-automation interaction are carefully designed. The study contributes early empirical evidence to inform the development of decision-support requirements and regulatory adaptations in ongoing discussions regarding the potential implementation of periodically unmanned bridges.
To achieve the emission reduction targets set by the maritime industry, large investments must be made in renewing and retrofitting the existing fleets in various shipping segments. This paper studies the strategic supply vessel fleet renewal and retrofit problem (SVFRRP) for guiding the transition towards a greener fleet of platform supply vessels (PSVs) used for offshore oil and gas logistics. Since the future costs of the different relevant fuel types are uncertain, we propose a new two-stage integer stochastic programming model for the SVFRRP. In each time period of the planning horizon, the model includes strategic decisions to meet the specified emission requirements. Existing vessels might be retrofitted to a new power system, or scrapped, while new vessels can be acquired and introduced to the fleet. Furthermore, the fleet deployment is considered, controlling the use of each vessel in order to fulfill the required cargo demands at each offshore installation serviced by the PSV fleet. We perform an extensive assessment of available fuel technologies for PSVs and use this in computational analyses for two real case studies from the Norwegian oil and gas industry. We compare the optimal fleet renewal strategies and corresponding costs from different emission reduction trajectories, and the results from the analyses provide valuable insights to the impact of various emission reduction policies to the optimal fleet renewal strategies.
Maintaining offshore wind farms (OWFs) is essential to ensure stable power production, but poses significant logistical and operational challenges due to the harsh marine environment, making it complicated to transport technicians and equipment.This study introduces the problem of finding the route and assignment of technicians minimizing the total time required to complete all scheduled maintenance tasks at OWFs under three concurrent assumptions: (1) the compatibility of technicians’ skill-set and the maintenance task to be performed, (2) technicians’ routing in the form of drop-off and pick-up, and (3) synchronized two-echelon system, composed of accommodation vessel (AV) and crew transfer vessels (CTVs). Together, these assumptions create a realistic and operationally meaningful foundation for the problem. Skill-task compatibility ensures that only appropriately qualified technicians are assigned to each maintenance job, reflecting real workforce constraints. Modeling technicians’ movements as coordinated drop-off and pick-up routes cuts idle time of the vessels, as service times at turbines are typically much longer than the short travel times between them. Finally, the synchronized two-echelon system addresses the long commute distance from shore, enhancing operational efficiency.The problem is modeled as a mixed integer linear program (MILP) model that accounts for various types of vessels and allows technicians to continue working independently after being dropped off. Finding feasible solutions to this problem is challenging, and solving it to optimality is extremely computationally complex. Thus, an adaptive matheuristic is designed to find high-quality solutions efficiently. Preliminary experiments on test instances based on Norwegian OWFs demonstrate that the proposed method yields robust and near-optimal solutions. We have also compared this method to a genetic algorithm that is adjusted to solve this specific problem and observe that, in most cases, the adaptive matheuristic reached better solutions with higher robustness. The impact of technician availability is also analyzed, showing that reducing crew size can significantly affect total operation time.
In today’s uncertain and rapidly evolving global landscape, maritime container terminals are increasingly affected by operational disruptions such as yard congestion, unbalanced resources utilization and delays in container handling. Truck Appointment Systems (TAS) have emerged as a key strategy to regulate truck arrivals and smooth peak demand, yet their effectiveness remains limited by insufficient integration with yard-side dynamics. Appointment allocation is typically designed without accounting for the spatial and temporal variability of yard operations. Designing a robust TAS requires the definition of realistic and representative operational scenarios. This paper proposes an enhanced DBSCAN-based clustering framework designed to identify recurring operational scenarios in container terminals. This is performed by jointly analysing spatiotemporal truck arrival patterns and container handling behaviours. The algorithm extends traditional density-based clustering by incorporating multi-dimensional distance metrics that capture spatial proximity, temporal alignment and operational similarity between container movements. The application of the proposed approach to a real-world case study from an Italian container terminal demonstrates its ability to extract five recurrent operational scenarios, covering more than 90% of container movements, with a noise ratio of 9.7% and a High-Density Score (HDS) of 0.4253, indicating a good balance between cluster cohesion, coverage, and operational interpretability. The identified scenarios were further analysed and qualitatively validated through interactions with terminal planners and operational managers. Overall, the resulting clusters provide actionable insights to support robust TAS design, enabling the development of data-driven decision support tools that explicitly account for operational variability and enhance the resilience of terminal planning and management.
The Northern Sea Route (NSR) has gained prominence as an international maritime corridor due to the retreat of Arctic sea ice, offering significant distance and time savings, particularly for shipping between Asia and Europe compared to conventional maritime routes. However, navigating the NSR involves considerable risks due to severe ice conditions along its navigation paths, making efficient icebreaking services critical for safe and reliable operations. The demand for icebreakers fluctuates spatially and temporally, depending on sea ice conditions and the ice class of cargo vessels. Additionally, icebreaking costs constitute a substantial portion of the overall voyage costs along the NSR. Strategically prepositioning icebreakers at optimum locations can help reduce these costs, enabling cargo vessels to request services more efficiently and minimize response times. This study focuses on the preparation stage of icebreaking services and introduces a weighted-demand response model to determine the optimum prepositioning of icebreakers before serving cargo vessels. The model considers expected vessel movements, navigation paths, and prevailing ice conditions. Eight Russian seaports are evaluated as potential prepositioning locations, and six ice classes—IC, IB, IA, IA-super, PC6, and PC5 are considered. The findings reveal that the optimum prepositioning locations and their priorities vary monthly in response to changing ice conditions, the composition of ice-class vessels, and their navigation directions. Moreover, Pevek Port consistently emerged as the highest-priority prepositioning location in most months. This study highlights the operational and policy implications of optimizing icebreaking services to reduce operating costs and improve the competitiveness of the NSR as an international maritime corridor.
Accurate and real-time ship trajectory prediction is a premise for high-stake tasks such as risk reduction, route planning, energy saving, etc., and becomes more feasible based on the processing of AIS data with sophisticated algorithms, so as to ensure high-standard navigation by providing efficient trajectory-based maritime traffic management. In contrast to current prevailing research striving to improve short-term prediction accuracy, this paper focuses on whereabouts estimation in order to improve longer-term predictions for vessels. Taking the meaningful whereabouts as implicit destinations, the novel Destination-Guided Trajectory Prediction (DGTP) model is proposed, which employs a cascaded Seq2Seq architecture with BiGRU to simultaneously predict both vessel destination and trajectory. Trajectory Alignment Loss (TAL) is also introduced to encourage precise matching between the predicted and true trajectories in optimizing the DGTP model. Experiments conducted on a large volume of AIS data demonstrate that both destination prediction and TAL loss can independently improve trajectory prediction performances. Moreover, the synergistic combination of destination prediction and TAL within the DGTP model leads to substantial accuracy enhancements, demonstrating the promising results in long-term prediction.
Traditional route recommendation systems optimize navigation paths using environmental variables such as weather and sea conditions, but often fail to account for real-world factors encountered by mariners. To address this gap, this study proposes a knowledge transfer Q-learning (KT-QL) algorithm, a reinforcement learning method built upon the Q-learning framework. The proposed KT-QL algorithm integrates expert trajectory knowledge derived from Automatic Identification System data into the Q-learning process, enabling the agent to combine trial-and-error exploration with data-driven guidance. Experimental results show that KT-QL reduces Hausdorff distances by approximately 39 % compared with conventional reinforcement learning and traditional search methods, and enhances fuel consumption prediction accuracy by approximately 2 %. These findings highlight the potential of KT-QL to enhance maritime operational efficiency, safety, and environmental sustainability.
Carbon emission reduction has been the focus of the International Maritime Organization (IMO), and restrictive mandates are considered by the Marine Environment Protection Committee (MEPC). The new guidelines consider carbon dioxide (CO2) emissions based on the propulsion system efficiency, distance, and dead weight, which are called the carbon intensity indicator (CII). In this research, this factor was calculated based on the large available data from a chemical tanker ship to analyze the ship rating using artificial intelligence techniques. The available data, consisting of global positioning system (GPS) location, wind speed and direction, draft and trim, engine power and speed, and vessel speed, are used for the CII prediction by the artificial neural network (ANN) modeling. Two types of ANN are considered for modeling: multilayer feedforward with two hidden layers, called deep neural networks (DNN), and generalized regression neural networks (GRNN). The attained, required, and referenced CII are calculated, and the system rating is determined and compared with the predicted CII. The best performance of the DNN is achieved with 15 neurons in the first and second hidden layers. The performance of the two types of ANN is robust and close to each other. However, the GRNN has slightly better predictive efficiency, considering the faster convergence and setup configuration complexity. The GRNN model shows a mean absolute error of 0.0928 with an unacceptable prediction ratio of 0.06 % and a coefficient of determination R2 = 0.998, which can capture the CII metric values and trend in transient mode robustly.
Efficient and timely vessel arrival planning is crucial for smooth operations in maritime transportation networks, ensuring optimal resource utilization and minimizing operational costs. When proforma schedules are disturbed by arrival deviations of vessels, waiting time and unnecessary fuel consumption become problems that shipping lines are faced with. Using a simulation model of a single-berth terminal, we test speed selection strategies for vessels that aim to minimize fuel, sailing, and waiting costs under varying availability of information. In different scenarios, we find optimality gaps ranging from 0.1% to 19.6% and show that knowing and communicating service end time to the vessel calling next could be valuable to integrated shipping lines and terminals.
This paper studies a maritime inventory routing problem (MIRP) faced by fish feed suppliers responsible for distributing different types of fish feed from one or several production facilities to a number of fish farms located at sea with a given heterogeneous fleet of specialized vessels. The feed supplier needs to maintain sufficient inventory levels at the farms at all times while minimizing the distribution costs. We propose a discrete-time mixed-integer programming (MIP) model for the fish feed MIRP. Since a commercial MIP-solver can only solve small problem instances, we also propose a matheuristic for solving real-life instances. The matheuristic employs a memetic algorithm, a metaheuristic combining a genetic algorithm with local search to decide how to route the vessels, coupled with a linear program for assigning quantities along the vessel routes. We perform a computational study on a number of realistic test instances generated using data from one of Norway’s largest fish feed suppliers. We show that the matheuristic produces reasonable solutions where the commercial MIP-solver fails, and as such can provide valuable decision support.
The adoption of Maritime Autonomous Surface Ships (MASS) in commercial shipping presents significant challenges despite rapid technological advancements. This study explores the barriers to the commercial adoption of MASS. Through a systematic literature review, 60 barriers were identified and categorized into four themes: (1) human factors, (2) data and risk management, (3) technology and connectivity, and (4) operations and policy. To reveal the most critical barriers, the importance-improvement (A-B) analysis was conducted utilizing data collected from maritime stakeholders. The analysis revealed that the most critical barriers include the trustworthiness of autonomous technology, managing loss of autonomous control system, vulnerabilities to cyberattacks, and the complexities of regulatory compliance in system development and deployment. Future resources and investments should be directed towards addressing the most critical barriers identified in this study for ensuring the successful integration of MASS in commercial shipping.
This paper develops a composite Port Resilience Index (PRI) to address the specific vulnerabilities and operational challenges of Greek ports in respect to climate-related hazards. Based on stakeholder engagement from Living Labs in three key ports (Chios, Volos, and Heraklion), the study identifies and quantifies the impacts of climate-related hazards using a structured Multi-Criteria Decision Analysis (MCDA) framework. Specifically, the Analytic Hierarchy Process (AHP) is used to elicit expert judgments and prioritize resilience criteria across five impact areas: Infrastructure, Operational and Supply Chain, Digital, Socioeconomic and Environmental, and Governance and Compliance Resilience. Nineteen indicators, spanning physical infrastructure, operational reliability, digital readiness, and socioeconomic factors, are evaluated to construct a composite PRI, enabling a transparent and stakeholder-informed benchmarking process. The results reveal significant variation in resilience levels, with Volos exhibiting the highest PRI (0.643) and Chios the lowest (0.217), thereby highlighting port-specific adaptation needs. Conducting a sensitivity analysis we validated the robustness of the PRI construction methodology across various weighting scenarios. The key contributions of this study are: (i) the development of a replicable, data-driven PRI model; (ii) the integration of local stakeholder input via Living Labs; and (iii) the innovative application of AHP to climate resilience planning in the port industry. Moreover, while focused on Greek ports, the framework offers a replicable model that can be adapted to other regions facing similar climate challenges. Ultimately, the PRI serves as both a diagnostic and strategic tool to guide policy, investment, and disaster preparedness in ports
The relocation of containers is essential at port terminals to increase operational efficiency during container retrieval from the yard. When a container must be retrieved, any container placed on top of it must be moved to another stack, delaying the retrieval process. The container premarshalling problem (CPMP) aims to tackle this issue by finding a sequence of minimal container relocations to achieve a bay arrangement where no container needs to be moved during retrieval. The classical formulation of this problem assumes that all premarshalling relocations occur within the bay being arranged. However, this study demonstrates that practical applications of premarshalling can benefit from more efficient use of available resources. We introduce a novel problem variant that allows the use of an auxiliary bay as additional space for relocating containers during the arrangement process. We present constraint programming solution methods for this variant that reveal a significant reduction in premarshalling relocations when an auxiliary bay is used. The results demonstrate that bays where high occupancy rates prevent premarshalling can be successfully arranged with an auxiliary bay. Additionally, we propose two alternative formulations allowing different rates of relocations between bays, offering adaptability to varying port terminal requirements.
