An Integrated Multi-Criteria Decision Support Model for Sustainable Ship Queuing Policy Application via Vessel Traffic Service (VTS)

Abstract

The International Maritime Organization (IMO) persistently improves policies to mitigate greenhouse gas (GHG) emissions from maritime operations, emphasizing the significance of operational measures. Simultaneously, heightened recognition of collaborative efforts within the maritime sector has increased the applicability of arrival policies like Just-In-Time Arrival (JITA), aimed at curtailing unnecessary anchorage time and emissions affecting adjacent communities in port vicinities. Nevertheless, ongoing initiatives advocate adopting JITA over the prevailing First Come, First Served (FCFS) policy, which is perceived as inefficient and, in the meantime, fair in the shipping industry. This research introduces an integrated decision support model to facilitate the implementation of a sustainable ship queuing policy by the VTS. The model addresses critical concerns, including the priorities of relevant authorities, the duration of nautical services for incoming vessels, and carbon dioxide (CO₂) emissions attributable to anchorage waiting times. The decision support framework presented integrates the Fuzzy Analytical Hierarchy Process (FAHP) and PROMETHEE II methodologies; the study’s outcomes suggest that the model significantly reduces ships’ unnecessary CO₂ emissions during anchorage waiting periods compared to the FCFS policy, with reduction rates ranging from 32.8% to 45% based on case analysis. Moreover, the proposed model ensures fairness by treating competing arriving ships equitably according to predefined criteria.

1. Introduction

The shipping industry is held accountable for 2.89% of worldwide anthropogenic greenhouse gas (GHG) emissions [1]. In response, the International Maritime Organization (IMO) remains committed to combating climate change through proactive measures to minimize GHG emissions from global shipping, safeguard the environment, and foster our planet’s secure and sustainable future. Recently, the updated IMO Strategy for Reducing GHG Emissions from Ships, adopted on 7 July 2023, highlighted the necessity of improving the optimization and planning of logistical chains and ports, among other potential mid-term measures for reducing GHG emissions [2]. With the ongoing rise in demand for maritime transportation, the industry is compelled to apply optimal operational measures to mitigate GHG emissions effectively.

In operational strategies, the Virtual Arrival (VA) policy is employed within tanker shipping, wherein vessels are asked to decrease their sailing speed to synchronize their arrival at a port within a predefined timeframe, aligning with the port’s readiness to accommodate the ship. Additionally, following the acknowledgment of the advantageous outcomes stemming from collaboration between the port and shipping domains, the Just-in-Time Arrival (JITA) initiative, introduced by the IMO-Global Industry Alliance (GIA), advocates for vessels to optimize their speed to reach the pilot boarding place (PBP) at the requested time of arrival (RTA), contingent upon the assurance of berth availability, fairway clearance, and the provision of nautical services by the Port Authority (PA) [3]. Moreover, the researchers [4,5,6,7,8,9] concurred that both JITA and VA represent promising sustainable measures, as they effectively decrease fuel consumption while maintaining transportation service quality, efficient traffic management, and minimizing in-transit inventory costs for cargo. Although both VA and JITA share a mutual goal of alleviating congestion and minimizing emissions that impact neighboring port communities, JITA emphasizes the readiness of subsequent infrastructure and the preparedness of nautical service providers within the port to diminish the waiting time at anchorage.

On average, 15% of emissions emanate from ships during port calls [5]. While dry and wet bulk vessels typically spend approximately 9% of their annual duration at ports, impacting the port city, container ships and liquid natural gas carriers consume only 4.5% and 4%, respectively [3]. The provided data underscore the potential for further enhancement within the dry bulk shipping sector, which is predominantly characterized by vessels engaged in the tramp trade, unlike the container sector. Nevertheless, a notable impediment to the adoption of sustainable practices in the dry bulk industry is the entrenched FCFS policy. Zhang et al. [10] argue that while FCFS may not represent the most efficient scheduling policy, its merit lies in equitably allocating port resources to vessels based on their actual arrival times. However, Senss et al. [11] emphasize the necessity for a departure from the prevailing FCFS queuing policy in the dry bulk sector towards a more fair, impartial, and dependable berthing prioritization scheme. Merkel et al. [12] suggested that the widespread adoption of sustainable practices in dry bulk shipping could be facilitated by the implementation of a queuing policy by the Port Authority (PA).

Many researchers [6,8,12,13,14] stated that the success of sustainable measures hinges on the decision-making process of relevant authorities through already established VTS systems, effectively coordinating by aligning the interests of stakeholders and their varying priorities, planning the arrival time rationally, and considering carriers’ contractual obligations. To support the JITA concept, the local PA should incorporate a compatible queuing policy into their corporate strategy and promulgate that to the dry shipping sector participants. In the meantime, the management of the JITA policy itself and a compatible queuing policy present minimal challenges for the Port Authority (PA), primarily due to the essential role of the Traffic Organization Service (TOS) fulfilled by the VTS, which commonly oversees the regulation of maritime traffic within the PA’s jurisdiction during port calls.

Furthermore, contemporary research mainly sheds light on adjusting ships’ speed in transit or during approaching port waters to reduce GHG emissions, optimizing operational processes to reduce waiting time, and scheduling vessel traffic, considering the temporary limitations of operational processes. The existing literature’s methodologies for assessing fuel savings and estimating emissions encompass simulation models and mathematical modeling. These approaches entail modeling fuel consumption and CO₂ emissions and analyzing outcomes using descriptive statistics. Established models are also employed to calculate fuel consumption and emissions, while genetic algorithms are utilized for specific problem definitions [4,7,8,12]. A combination of FAHP and the Expert System is another methodological approach to managing traffic flow in confined waters. These methodologies predominantly rely on a quantitative research approach utilizing AIS data or data acquired from the PA. However, a smaller number of research studies combine the priorities of the PA and supporting organizations: the JITA initiative requirement on nautical service preparedness, the role and capacity of the VTS system, and alternative queuing policies to FCFS using multi-criteria decision-making models directly compatible with the existing VTS system.

On the other hand, it is crucial to determine the complete ranking of arriving ships to ensure the establishment of a fairer queuing system to support the principles of JITA and consider the capacity of nautical services for subsequent operations to eliminate unnecessary CO₂ emissions. The significance of a complete ranking of arriving ships is inherent in the fair competition of arriving ships under the preliminarily determined decision criteria by the PA and subordinate organizations responsible for managing marine traffic in the maritime domain.

After identifying research gaps and delineating challenges, this study aims to propose an integrated decision support model tailored for the VTS system to execute a fair queuing policy for arriving ships compatible with the JITA initiative. In order to accomplish our aim, we have identified the following research objectives:

• To identify the hierarchical structure of decision criteria that represents the priorities of the PA and sets the groundwork for a fair queuing policy for Iskenderun Bay, Turkey.
• To examine current vessel traffic management and TOS practices overseen by the VTS system and elucidate the information/data flow accessible for model integration.
• To propose an integrated multi-criteria decision support model comprising the Fuzzy Analytical Hierarchy Process (FAHP) and PROMETHEE II applicable by the VTS, which determine priority weights of criteria and provide a complete ranking of arriving ships for the management of a queuing policy compatible with the JITA policy.
• To analyze the reduction in CO₂ emissions during the anchorage period of arriving ships within the port limits by implementing the proposed model that considers subsequent nautical service duration.

This study focuses on ship queuing policies in port domains. Firstly, we introduce a novel queuing policy that integrates the priorities of the PA with existing VTS systems, enhancing operational efficiency and safety. Secondly, the fairness of this policy is ensured through a comprehensive ranking of arriving ships, determined by decision criteria and weights established by the institutions responsible for regulating vessel traffic. Thirdly, by incorporating the FAHP and PROMETHEE II methodologies, our integrated decision support model incorporates nautical service duration considerations, further refining the queuing policy. Through a rigorous case study application, our model demonstrates its efficacy in reducing CO₂ emissions from waiting ships in anchorage areas, offering tangible environmental benefits. The detailed framework provided alongside the proposed model facilitates its adaptation to other port domains, contingent upon a thorough preliminary analysis of marine traffic dynamics, data availability, and VTS capabilities. This comprehensive approach addresses immediate challenges in port management and lays a foundation for sustainable maritime operations on a broader scale.

The subsequent sections of this paper are structured as follows: Section 2 presents an extensive review of the relevant literature. Section 3 offers an examination of vessel traffic management in the Gulf of Iskenderun and outlines the methodology employed in this research. Section 4 elucidates the findings of the study. Section 5 encompasses the discussion of these findings. Section 6 provides concluding remarks and delineates potential directions for future research.

2. Literature Review

Through various approaches, stakeholders in the shipping industry will persist in responding to the demand for IMO GHG strategies. A scholarly categorization has outlined the principal measures as technological advancements, fleet-related operational enhancements, market-based mechanisms, management strategies, and decision support models [14]. Management measures like the JITA and the VA scheme are targeted methods for mitigating shipping activities’ environmental footprint, thereby aiding in achieving the IMO’s GHG reduction objectives. Regarding the dry bulk sector, sustainable management measures and policies might face resistance due to perceived impracticality or a lack of alignment with fundamental safety, operational, and trade requisites [11]. To capture the essence of the literature, the initial focus will be on understanding the JITA policy, the VA policy, the FCFS policy, and their latest implementations and developments. Subsequently, attention will be directed towards vessel traffic optimization, the VTS, and its current and prospective role.

The JITA policy primarily focuses on reducing idling and waiting times in port anchorage areas, aiming to achieve cost savings, minimize fuel consumption of auxiliary engines and boilers, and reduce greenhouse gas (GHG) emissions [3]. Research conducted by Winnes et al. [15] utilized AIS data and analyzed CO₂ emissions across various operational modes within the Port of Gothenburg in 2010. The study revealed varying emission rates across different modes, with “at berth” being the highest emitter at 53%, followed by “in the fairway” at 23%, “at anchor” at 10%, “in the port basin” at 9%, and “maneuvering” at 5%. The distribution of emissions across operational phases highlights opportunities for improvement, particularly for the “in the fairway” and “at anchor” modes, through the adoption of the JITA policy.

On the other hand, reducing time spent at anchor not only decreases emissions from ships but also mitigates harmful air pollutants in the port city, such as sulfur oxides, nitrogen oxides, and other pollutants within the scope of the International Convention for the Prevention of Pollution from Ships Annex VI. In addition to reducing congestion and traffic, the JITA policy also prevents hull fouling caused by long anchorage wait times in tropical waters, reduces the risk of accidents, and improves navigational safety and security in the maritime domain [3,11]. Despite the compelling evidence highlighting the benefits of the JITA policy for reducing GHG emissions and air pollution, the dry bulk shipping sector continues to prefer the widely adopted FCFS queuing policy. This preference aligns with the unique characteristics of tramp businesses [7,8,9,11].

In general, the FCFS practice is available where the dry bulk terminals have no contract with the carriers [3]; therefore, the FCFS queuing practice tends to result in extended waiting periods for incoming ships at anchorage [16]. This policy incentivizes vessels to rush their arrival at the port, consequently leading to prolonged waits for berthing opportunities [12,16]. One driving factor behind this behavior may be the potential for higher earnings from demurrage rates compared to daily charter rates or the desire to avoid risks associated with charter party cancellations [17]. Consequently, such a policy contributes to increased fuel consumption and exacerbates the emission of harmful air pollutants and GHGs in port areas [16,17,18]. While contractual barriers have historically been a primary concern for the dry bulk shipping industry, recent efforts to develop charter party clauses have been successful. Incorporating such clauses in voyage charter parties necessitates owners sharing vessel arrival information with relevant third parties to facilitate preplanning. However, while owners’ consent is essential, it should not be unreasonably withheld, provided it remains within the safe operational limits of the vessel in question [19]. In general, it is imperative to prioritize operational measures to diminish GHG emissions; nevertheless, adjusting prevailing policies must be undertaken carefully, considering the present business landscape, for a smooth transition to new approaches.

Additionally, the VA policy represents an operational procedure introduced by industry associations within the tanker industry, sharing many resemblances with the Just-In-Time Arrival (JITA) concept, thus warranting further clarification. Its primary objective is to mitigate potential delays at the destination port by adjusting a vessel’s speed during transit to ensure it arrives at the RTA [7,8]. Integrating a provision into the charter party agreement is essential, given that the VA policy encompasses the entire operational phase of a ship’s voyage. Consequently, fuel savings and additional time utilized are rigorously assessed at the voyage’s conclusion, leveraging all pertinent data, including weather conditions, wave patterns, speed projections, and other technical factors, to determine demurrage payable at a previously agreed-upon rate [20].

Implementing the VA policy contributes to GHG emission savings in transporting wet bulk cargo’s transit and arrival phases. Alvarez et al. [21] developed a simulation model integrating berth allocation, land-side equipment assignment, and speed optimization. Their findings indicate that adopting the VA policy could reduce fuel consumption by approximately 6% compared to traditional FCFS policies. In a study by Jia et al. [7], an assessment was made of a fleet of 483 very large crude carriers to evaluate global emissions and fuel consumption reductions from 2013 to 2015. The average sea passage speed during voyages was reduced by accounting for idle time at the destination port using Automatic Identification System (AIS) data. The study revealed that a 25% reduction in waiting time could lead to 7.26% fuel savings, while a complete elimination of waiting time could result in 19% fuel savings. Andersson et al. [4] estimated the impact of the VA policy by comparing the estimated anchorage times of ships in Baltic Sea countries using AIS data. The research findings indicated that implementing speed reductions of 10%, 25%, or 50% for durations of 1, 4, 12, or 24 h yields significant benefits. Approximately two-thirds of these benefits are attributed to emission costs, with the remaining one-third attributed to fuel costs. Merkel et al. [12] investigated the waiting time period at anchorage. They employed speed-linked elasticities to ascertain the speed/fuel correlation across diverse speeds and assessed the prospective fuel economizations stemming from the VA policy. Their findings indicated that reducing speed 4–12 h before planned arrival could result in fuel savings ranging from around 1.7% to 4.7%. Shao et al. [8] introduced a model incorporating vessel scheduling with the VA policy, drawing from real-world data from the Ningbo-Zhoushan port. They applied the non-dominated sorting genetic algorithm II, emphasizing safety, efficiency, and equity in port management. The results demonstrated a 31.4% reduction in total waiting time compared to FCFS policies. Furthermore, upon implementation of the VA policy, carbon emissions were reduced by 23.39%. To achieve these results, researchers selected vessel length, draft, speed, estimated time of arrival (ETA), and similar variables for their models. Based on the reviewed studies, Alvarez et al. [21] observed a decrease in overall fuel consumption and reduced CO₂ emissions. However, they did not provide a specific numerical value for the latter. Conversely, other researchers [4,7,8,12] utilized various methods, such as emission coefficients, emission factors, or constant figures used on specific fuel consumption, to estimate CO₂ emission reductions. While each approach has merits, the U.S. Environmental Protection Agency (EPA) [22] proposed another methodology for calculating CO₂ emission reductions, considering the average power rating of auxiliary engines and boilers used during anchorage by different ships.

Previous research has heavily focused on optimizing ship scheduling through various methods with the usual priorities and mainly aims to reduce waiting time. For instance, Zhang et al. [23] introduced a vessel traffic optimization and scheduling model for compound waterways using a multi-objective genetic algorithm to minimize waiting and waterway occupancy times. Compared to the FCFS policy, their results suggested improved scheduling efficiency, enhanced navigation safety, and reduced workload pressure on the Vessel Traffic Service Operator (VTSO). Similarly, Zhang, Zheng, and Wang [24] developed a model and algorithm for scheduling vessels through two-way tidal channels, aiming to decrease waiting time and enhance operational efficiency. Their model yielded a waiting time of 57.29 h, notably lower than 74.17, 70.87, 67.36, and 61.77 h resulting from the FCFS policy, corresponding to large draft vessel first, random scheduling, and manual scheduling methods, respectively. Both models share similarities regarding selected variables such as breadth, speed, draft, underkeel clearance, and estimated time of arrival (ETA) of ships, all quantitative data directly utilized by the proposed models. Moreover, Liang et al. [25] introduced an alternate methodology merging the FAHP and Expert System to formulate optimal traffic directives for individual vessels navigating the regulated water passages of the Yangtze River. Their framework prioritized vessel sequences through a hierarchical arrangement encompassing static and dynamic attributes, including vessel category, cargo type, navigational status, vessel trajectory, and voyage duration. Experimental results demonstrated that the proposed model reduced waiting time by an average of approximately 22 min compared to the existing system. In other words, the proposed model underlined another fact that such rare models relying on the judgment of domain experts may additionally contribute to maintaining navigation safety, efficiency, and emissions reductions from ships due to decreased waiting time in port areas.

The VTS systems operate in maritime regions where increased risks to navigation safety, the need for optimizing vessel traffic, and safeguarding environmental integrity against pollution events are identified [26]. The recent revision of the VTS Guidelines by the IMO has mandated the implementation of TOS as a requisite service type for all existing VTS systems. Essentially, contemporary VTS systems are expected to fulfill a spectrum of duties, including monitoring standard information dissemination, preplanning vessel movements, executing the JITA principle, and allocating space between vessels within their operational jurisdiction [26,27,28]. Furthermore, the forthcoming role of VTS systems was underscored at the 14th International Association of Marine Aids to Navigation and Lighthouse Authorities (IALA) Symposium on VTS, convened in Rotterdam in 2021. Participants from esteemed organizations like the International Hydrographic Organization (IHO) and IALA delineated the evolution of the VTS system toward a digitalized service model, akin to an aviation-style traffic control system and a core element in E-Navigation and Port Collaborative Decision Making (P-CDM) concepts [27,29,30].

In summary, established methods for operational measures exist to manage GHG emissions from shipping activities in port areas, aiming to mitigate adverse environmental and community impacts. While the VA policy governs ship transit from berth to berth, the recently introduced JITA policy is additionally concerned with optimizing traffic management during the operational decision-making phase of arriving vessels in port areas. Existing research predominantly focuses on landside dynamics [21], speed management in transit [7], speed reduction at the approaching phase of sea passage [4,8,12], reducing waiting time by optimizing the scheduling of vessels [23,24] to contribute to reducing GHG emissions or improving safe and efficient management of vessel traffic flow by using static and dynamic data regarding the ships [8,23,24,25], and mainly considering binding constraints in port waterways [23,24]. Due to the fairness feature, the FCFS policy still challenges transitioning to the JITA policy, especially in the dry bulk sector. However, there are partial attempts to develop a fair queuing policy that simultaneously balances the PAs’ and their supporting public entities’ priorities for arriving ships and focuses on nautical services and the anchorage period of vessels, which is underlined by the JITA initiative. In addition, the realization of the queuing policy can only be achieved by the participation of the VTS system, rendering TOS in port jurisdiction. Therefore, developing a queuing policy is mainly subject to the existing capabilities and practical availability of relevant information/data in the VTS system. Developing a decision support model that enables the implementation of alternative sustainable queuing policies via the VTS system is likely to make moving away from the inefficient and environmentally unfriendly FCFS policy possible in the dry bulk shipping sector. This study aims to bridge the stated research gap by proposing a methodology and demonstrating its practical application through a case study in the Gulf of Iskenderun, Turkey.

3. Material and Method

3.1. Review of Vessel Traffic Management in the Gulf of Iskenderun

Apart from providing Turkish Straits Vessel Traffic Services, the Vessel Traffic Management System project has rendered vehicle traffic services in Izmit, Izmir, and Mersin in Turkey. The Directorate General of Coastal Safety controls 24 unmanned traffic surveillance stations that are connected to the VTS centers mentioned above [31].

The Mersin VTS Centre oversees three distinct sub-sectors: the Mediterranean sector, the Mersin sector, and the Iskenderun sector. Within the region, eight unmanned traffic surveillance stations are strategically positioned to monitor designated areas and maintain communication with vessels under their jurisdiction as necessary. Leveraging radio link technology facilitates the seamless transmission of gathered data and information from these surveillance stations to the Mersin VTS Centre, thereby supporting the facilitation of organizing ship traffic for vessels entering and departing the Gulf of Iskenderun. The Gulf of Iskenderun encompasses 30 port facilities under the purview of the two regional Port Authorities. According to Ministry of Transportation and Infrastructure of Turkiye statistics, the annual average number of ship visits surpasses 5000, with 5987 ships recorded in 2022 [32].

During vessel traffic management, the established VTS system receives specific reports from approaching vessels. The details provided in the Sailing Plan (SP) 1 report, which is submitted 24 h prior to vessels entering the Mediterranean Sector, are logged into the VTS database. The principal aim of this report is to establish an initial repository of vessel data. Subsequently, the information received through the SP 1 report undergoes validation against data from the local Port Management Information System (PMIS) integrated within the Maritime Single Window (MSW), and any necessary corrections or updates are made accordingly. Upon entry into the Mediterranean sector, masters of arriving vessels submit a report known as SP 2. As vessels approach the Iskenderun Sector and pass the port limit, the master submits a final report known as the Position Report.

On the other hand, the Pilotage and Tug Organization, which provides nautical service, is not directly involved in the entry of ships into the maritime area. The PMIS is used for information exchange that must be carried out prior to the organization’s provision of nautical service to arriving ships.

Regarding traffic management in Iskenderun Bay, the PA exercises direct oversight over both the VTS system and the Pilotage and Tug Organization, which can collectively support establishing a sustainable queuing policy. However, it is crucial to underline the circumstances that could disrupt the regular traffic flow, leading to a shift in the hierarchical control between central and local authorities. For instance, meteorological events are monitored unilaterally by local authorities and organizations, as depicted in Figure 1, and findings are subsequently reported to the local Port Authority. Following an evaluation of the gathered data concerning adverse weather conditions that pose significant risks to maritime operations’ safety, the local Port Authority, empowered with the authority to enforce regulations issued by the Ministry of Transport governing port facilities within its maritime jurisdiction, promptly suspends all berthing and unberthing operations without delay by promulgating a written order. Moreover, emergency situations such as collisions and consequential environmental pollution could lead to a transfer of control from local to central authorities, contingent upon the severity of the impact. Following an assessment of the evolving circumstances, the determination of whether the emergency responses are managed locally or centrally is subject to the consent of the central authority.

3.2. Methodology

Defining a fairer queuing policy compatible with the JITA approach that enables the management of marine traffic with lower GHG production from the ships is not a straightforward task. As Näslund [33] stated, logistical challenges encountered in real-world scenarios are complex and disorderly, making qualitative inquiries valuable in addressing these issues. Given this complexity, employing diverse methodologies and maintaining a modest approach is necessary [34]. Moreover, the prevailing body of research concerning ports and logistics has predominantly relied on quantitative approaches, prompting calls for qualitative and interpretative studies to enhance comprehension of empirical phenomena [35]. Therefore, to overcome the complexity and gather details regarding marine traffic management and the priorities of respective stakeholders, the exploratory sequential mixed methods research approach was adopted for the study. The selected research approach is a strategy for the design and implementation of research [36] and an alternative method that can be chosen based on the research problem or field [37].

Field observations and semi-structured interviews, among the qualitative research methods, were applied to collect primary data. The Iskenderun Regional Harbor Master Office, Mersin Vessel Traffic Management System, and Isdemir Pilotage and Towage Organization headquarters were visited for observation purposes to collect data regarding operational practices and established data collection and sharing systems. The Isdemir port facility and routinely visiting ships are the additional public and private sites selected in the Bay of Iskenderun. Additional primary data were obtained from domain experts via semi-structured interviews, as presented in

Table 1. List of maritime domain experts.

Expert Code	Maritime Sector	Current Position	Education	Experience
E1	Maritime transportation	Master mariner	BSc degree	17
E2	Port management	Port management executive	BSc degree	20
E3	Port management	Port operations manager	BSc degree	15
E4	VTS center	Chief VTSO	BSc degree	25
E5	Maritime transportation	Master mariner	BSc degree	20
E6	VTS center	VTSO	BSc degree	18
E7	VTS center	VTSO	BSc degree	16
E8	Maritime transportation	Master mariner	BSc degree	24
E9	VTS center	VTSO	BSc degree	11
E10	Pilotage and towage organization	Maritime pilot	BSc degree	20
E11	Pilotage and towage organization	Maritime pilot	BSc degree	19
E12	Harbor master office	Regional harbor master	BSc degree	25

. The selection of the domain experts was facilitated through a non-probability judgmental sampling technique, whereby samples were chosen based on specific characteristics, as it is impractical to include all individuals within the statistical population [38]. Especially in maritime transportation and logistics research, a preference exists for participants employed within private or public institutions within the industry [39]. Furthermore, with the active involvement of the designated domain experts, a series of meetings were also scheduled to elicit perspectives on the accessibility and reliability of existing data for enacting a queuing policy with specified criteria and the viability of the proposed methodology in the Gulf of Iskenderun.

On the other hand, the framework of the proposed decision support model, illustrated in Figure 2, comprises a FAHP for prioritizing criteria weights and PROMETHEE II (Preference Ranking Organization Methods for Enrichment Evaluation) for achieving a complete ranking of incoming vessels. Determining the weight is crucial in most multi-criteria methods [40]. The PROMETHEE II method is contingent upon the premise that all decision-makers possess the acumen to judiciously assess the criteria [41]. To eliminate the limitation mentioned above, an integrated decision support tool has been proposed to take advantage of the synergy and produce fair, complete rankings of arriving ships.

Process steps are as follows:

• Identify ship queuing criteria and organize them hierarchically.
• Utilize the FAHP to compute the priority weight for each criterion in the hierarchy.
• Determine ship-wise nautical service duration to facilitate the RTA at the PBP, estimate waiting idle/anchorage periods, and organize the berthing maneuvers of arriving ships.
• Employ the PROMETHEE II complete ranking decision-making tool to rank incoming ships.
• Analyze the result via the Geometrical Analysis for Interactive Aid (GAIA) plane and walking weights method.
• Develop a hierarchical structure encompassing the goal, criteria, and decision criteria at pertinent stages, as depicted in Figure 3. In addition to the available secondary data in the literature, the contribution of the domain experts enabled the segmentation of the queuing policy criteria into four categories: safety, environmental, business, and efficiency.

An additional group discussion was conducted to select experts from the sample group who should evaluate the importance of the defined criteria. To obtain results consistent with actual practices, the experts’ involvement with enforcement and management authority to manage vessel traffic in the maritime domain reached a consensus. Then, a chief VTSO (E4), maritime pilot (E10), and harbor master (E12) performed a pairwise comparison of each criterion by completing a questionnaire using the evaluation scale illustrated in

Table 2. Membership function of the linguistic scale.

Fuzzy Number	Linguistic Scale	Membership	Inverse
${\tilde{A}}_1$	Equally important	(1, 1, 1)	(1, 1, 1)
${\tilde{A}}_2$	Moderately important	(1, 3, 5)	(1/5, 1/3, 1/1)
${\tilde{A}}_3$	More important	(3, 5, 7)	(1/7, 1/5, 1/3)
${\tilde{A}}_4$	Strongly important	(5, 7, 9)	(1/9, 1/7, 1/5)
${\tilde{A}}_5$	Extremely important	(7, 9, 9)	(1/9, 1/9, 1/7)

.

3.2.1. Determining the Weights of Ship Queuing Policy Criteria via FAHP

The Analytical Hierarchy Process (AHP) concerns the information gathered from domain experts, yet more is needed to reflect uncertainties and fuzziness. Given the fuzzy nature of the comparison process, decision-makers typically articulate interval assessments more frequently than precise values. This is because decision-makers are unable to explicitly articulate their preferences [42]. The FAHP enabled experts to make comparisons using fuzzy ratios rather than crisp values [43].

Initially, van Laarhoven and Pedrycz [44] conducted a pioneering study on the FAHP method, contrasting fuzzy ratios using triangle fuzzy numbers (TFN). While Buckley adopted trapezoidal fuzzy numbers to represent comparison ratios’ priorities, Chang employed triangular fuzzy membership values for pairwise comparisons and introduced a novel FAHP approach called “extent synthesis analysis” [45]. Weck et al. [46] introduced a methodology for evaluating diverse production cycle alternatives, incorporating fuzzy logic mathematics into the traditional AHP framework. Lee et al. [47] revisited the fundamental principles of the AHP, introducing comparison intervals and proposing a methodology reliant on stochastic optimization to ensure global consistency and accommodate the fuzzy nature of comparisons. Duru et al. [48] improved upon Chang’s method by integrating coefficients derived from the Centric Consistency Index (CCI) for individual participants. The FAHP exhibits versatility, allowing seamless integration with diverse techniques, especially in decision-making models where determining criteria weights is crucial.

Anojkumar et al. [49] emphasized defining a problem and creating a hierarchical structure with at least three levels: goal, criteria, and decision criteria, when using FAHP methods. After constituting the hierarchical structure, decision criteria are evaluated by pairwise comparisons of selected domain experts from Table 1.

Furthermore, triangular fuzzy numbers (TFNs) are utilized in operations delineated within Equation (1):

$${\mu }_{\check{A}}\left(x\right)=\left\{\begin{array}{c}0, x<l,\hfill \\ \frac{\left(x-l\right)}{m-l}, l\le x<m,\hfill \\ 1, x=m,\hfill \\ \frac{\left(u-x\right)}{\left(u-m\right)}, m<x\le u,\hfill \\ 0, u<x, \hfill \end{array}\right.$$

(1)

Wherein “l” and “u” denote the lower and upper bounds, respectively, of a fuzzy number denoted as Ã, and “m” signifies the midpoint. A TFN is expressed as à = (l, m, u) in accordance with Zadeh’s definition [50].

The next step in determining criteria weights involves preparing a questionnaire for the selected domain experts to compare each criterion pairwise. Nonetheless, the evaluation scale employed in the FAHP method diverges from the numerical scale. Instead, it utilizes fuzzy linguistic variables corresponding to Saaty’s AHP nine-point fundamental scale [51]. This comparison involves linguistic terms, with their corresponding fuzzy numbers delineated in Table 2.

After obtaining the individual fuzzy judgment matrix of domain experts established on the qualitative evaluation scale demonstrated in Table 2, a consensus matrix is acquired by the aggregated TFN of each matrix by using the max–min arithmetic mean operation shown in Equation (2) [52].

$$u_{ij}=\underset{t=\mathrm{1,2},\dots ,q}{\max }\left(u_{ij}^{\left(t\right)}\right), m_{ij}=\frac{1}{q}{\sum }_{t=1}^q{m}_{ij}^{\left(t\right)}, l_{ij}=\underset{t=\mathrm{1,2},\dots ,q}{\min }(l_{ij}^{\left(t\right)})$$

(2)

Before initiating operations, the consistency of the matrix comprising subjective judgments from decision-makers needs assessment. Thus, ensuring individual consistency of the pairwise matrices can be achieved through the utilization of the CCI, as suggested by Duru et al. [48]. The algorithm for the CCI lets $A=\left(a_{Lij}{{,w}_{Mij},w}_{Uij}\right)nxn$ be a fuzzy judgment matrix, while $w=\left[\left(w_{L1}{{,w}_{M1},w}_{U1}\right),\dots ,\left(w_{L1}{{,w}_{M1},w}_{U1}\right)\right]T$ represents the priority vectors obtained from matrix $A$ using the row geometric mean method. The CCI is computed by Equation (3):

$$CCI\left(A\right)=\frac{2}{\left(n-1\right)\left(n-2\right)}\sum _{i<j}(\mathrm{l}\mathrm{o}\mathrm{g}(\frac{a_{Lij}+a_{Mij}+a_{Uij}}{3})-\log (\frac{w_{Li}+w_{Mi}+w_{Ui}}{3})+\log (\frac{w_{Lj}+w_{Mj}+w_{Uj}}{3}))2$$

(3)

When CCI(A) = 0, the $A$ is fully consistent. Once CCI($A$) = 0.31 for n = 3, CCI($A$) = 0.35 for n = 4, and CCI($A$) = 0.37 for n > 4, threshold values are obtained, which means that $A$ is sufficiently consistent.

According to Chang’s [45] extent synthesis FAHP, there are two sets, namely $X=\{x1,x2,x3,\dots xn\}$, which is an object set, and $G=\{g1,g2,g3,\dots gn\}$, which represents the goal set. Extent analysis is conducted individually for each goal on each object. Consequently, it is feasible to procure $m$ extent analysis values for each object, denoted by the following symbols:

$$M_{gi}^1,M_{gi}^2{,\dots ,M}_{gi}^m, i=\mathrm{1,2},\dots ,n,$$

(4)

where $M_{gi}^j(j=\mathrm{1,2},\dots m)$ are TFNs.

Chang’s extent analysis is delineated through the following steps:

Step 1: Define the value of fuzzy synthetic extent for the $i$th object.

$$S_i=\sum _{j=i}^m{M}_{gi}^j\otimes {\left[\sum _{i=1}^n{\sum }_{j=1}^m{M}_{gi}^j\right]}^{-1}$$

(5)

To calculate $\sum _{j=i}^m{M}_{gi}^j$, perform the fuzzy addition operation on the $m$ extent analysis values for a specific matrix, as follows:

$$\sum _{j=1}^m{M}_{gi}^j=\left(\sum _{J=1}^m{l}_j,\sum _{J=1}^m{m}_j,\sum _{J=1}^m{u}_j,\right)$$

(6)

To obtain ${\left[\sum _{i=1}^n{\sum }_{j=1}^m{M}_{gi}^j\right]}^{-1}$, perform the fuzzy addition operation of $M_{gi}^j\left(j=\mathrm{1,2},\dots ,m\right)$ values, as follows:

$$\sum _{i=1}^{n}a\sum _{j=1}^m{M}_{gi}^j=\left(\sum _{J=1}^m{l}_j,\sum _{J=1}^m{m}_j,\sum _{J=1}^m{u}_j,\right)$$

(7)

The computation of the inverse of the vector in Equation (7) follows:

$${\left[\sum _{i=1}^{n}a\sum _{j=1}^m{M}_{gi}^j\right]}^{-1}=\left(\frac{1}{\sum _{i=1}^n{u}_j},\frac{1}{\sum _{i=1}^n{m}_j},\frac{1}{\sum _{i=1}^n{l}_j}\right)$$

(8)

Step 2: The degree of possibility of $M_2=\left(l_2{,m}_2{,u}_2\right)\ge M_1=\left(l_1{,m}_1{,u}_1\right)$ is defined as:

$$V\left(M_2\ge M_1\right)=\underset{y\ge x}{\sup }\left[\mathrm{m}\mathrm{i}\mathrm{n}({\mu }_{M1}\left(x\right),{\mu }_{M2}\left(y\right))\right]$$

(9)

The expression of this can be as follows:

$$V\left(M_2\ge M_1\right)=hgt\left(M_1\cap M_2\right)={\mu }_{M_2}\left(d\right)=\left\{\begin{array}{c}1 ,m_2\ge m_1 \hfill \\ 0 ,l_1\ge u_2 \hfill \\ \frac{l_1-u_2}{\left(m_2-u_2\right)-(m_1-l_1)} ,otherwise\hfill \end{array}\right.$$

(10)

Step 3: The degree of possibility for a convex fuzzy number to exceed $k$ convex fuzzy numbers $M_i=(\mathrm{1,2},\dots ,k)$ can be defined as:

$$\begin{gathered} V\left(M\ge M_1,M_2,M_k\right)=V\left[\left(M\ge M_1\right)ve\left(M\ge M_2)ve\dots ve\left(M\ge M_k\right)\right)\right] \\ =minV\left(M\ge M_i\right),i=\mathrm{1,2},\dots ,k \end{gathered}$$

(11)

Assuming that $d^{\text{'}}\left(A_i\right)=minV(S_i\ge S_k)$ for $k=\mathrm{1,2},\dots ,n,k\ne i$. Then the weight vector is:

$$W^{\text{'}}={\left(d^{\text{'}}\left(A_1\right),d^{\text{'}}\left(A_2\right),\dots ,d^{\text{'}}\left(A_n\right)\right)}^T$$

(12)

where $A_i\left(i=\mathrm{1,2},\dots ,n\right)$ are $n$ elements.

Step 4: The weight vectors are normalized.

$$W={\left(d\left(A_1\right),d\left(A_2\right),\dots ,d\left(A_n\right)\right)}^T$$

(13)

where $W$ is a non-fuzzy number.

3.2.2. Production of Ship-Wise Nautical Service Duration Table

Before the complete ranking of ships is obtained, the crucial step involves determining the rendering duration of nautical services for the queued ships to produce the RTA at the PBP. Correlation analysis and simple linear regression are employed to create a table using the raw dataset.

Correlation analysis endeavors to ascertain both the extent and the functional configuration of the relationship between two or more variables. When one variable exhibits a parallel increase or decrease alongside another, the relationship between them is characterized as linear [53]. The correlation coefficient, symbolized as $r$, is calculated utilizing the $n$ pairs of values. This process facilitates a more decisive appraisal of our confidence level regarding the adherence of the data to a normal distribution. Obtaining a p-value less than 0.05, which means the null hypothesis is true in the normality test, concludes that the data do not follow a normal distribution [54].

As an alternative approach under the stated situation, Spearman’s rank correlation coefficient, a nonparametric (distribution-free) rank statistic, is posited as a metric for gauging the magnitude of the relationship between two variables. It indicates a monotonic association, particularly valuable in instances where the data distribution renders Pearson’s correlation coefficient unsuitable or prone to misinterpretation [55].

As expressed in Equation (14), Spearman’s correlation coefficient involves computing the discrepancy between the two ranks, represented as $D$. Subsequently, these disparities are squared, resulting in the summation $\sum D^2$ [56].

$$R=1-\frac{6\sum D^2}{N^3-N}$$

(14)

Within the simple linear regression method, $Y$ signifies the dependent variable, whereas $X_1$ denotes the independent variable. The parameters $B_0$ and $B_1$ symbolize this variable’s unknown parameters, while ${\epsilon }_i$ is utilized to denote the random error terms. The simple linear regression model is articulated in Equation (15) [57].

$$Y=B_0+B_1{X}_{1i}+{\epsilon }_i i=\mathrm{1,2},\dots ,n$$

(15)

The acquired data were statistically analyzed using Minitab software (21.4.3) [58], and the resulting outcomes are presented in the subsequent section.

3.2.3. Determining the Complete Ranking of Arriving Ships via PROMETHEE II

Brans et al. [59] devised the PROMETHEE techniques, notably PROMETHEE I for partial ranking and PROMETHEE II for comprehensive ranking of alternatives. The outranking methods are suitable for ranking problems and useful in conceptual development and application for multi-criteria analyses [60], offering both flexibility and convenience to the user [61]. Additional benefits of the PROMETHEE methods include their successful application to real-world planning issues while simultaneously maintaining a level of simplicity [62]. The method was further extended by Brans and Vincke [63] and Brans and Mareschal [64]. Several versions were developed, encompassing the PROMETHEE III for interval-based ranking, the PROMETHEE IV for comprehensive and partial ranking of alternatives in scenarios where the pool of feasible solutions is continuous, the PROMETHEE V tailored for segmentation dilemmas, and the PROMETHEE VI devised to emulate the cognitive processes of the human brain.

Various software packages are accessible for conducting the analyses above, providing decision-makers with visual aids to scrutinize the circumstances. For instance, software such as Procalm and Decision Lab enables users to generate partial and comprehensive rankings. Additionally, these applications offer modules for conducting sensitivity analyses, allowing users to evaluate the impact on alternative outcomes by swiftly adjusting criterion weights [65]. Furthermore, Visual PROMETHEE, developed by VP Solutions under the guidance of Professor Bertrand Mareschal, serves as an integrated software solution designed specifically for implementing the PROMETHEE and GAIA methodologies. For example, within Visual PROMETHEE, a decision scenario can be delineated by a collection of actions, a set of criteria, and a criteria group. Decision-makers express their preferences by specifying whether each criterion should be minimized or maximized and assigning a preference function to each criterion. Furthermore, the statistics section offers essential metrics such as minimum, maximum, mean, and standard deviation [66]. A notable benefit of the GAIA plane is its capacity to visually present outcomes [40] and depict alternatives as points and criteria as vectors. Furthermore, the decision axis delineated by the thick red vector signifies the most favorable alternatives for the decision-maker. Alternatives near the criteria vectors signify those that should be prioritized primarily for the criterion under examination. Vectors aligned in the same direction pertain to criteria sharing analogous properties. Conversely, criterion bars pointing in divergent directions denote conflicting criteria [67].

The steps of the PROMETHEE II method [40] are defined as follows:

Step 1: Determining the deviation through pairwise comparison.

When examining a collection of alternatives within a multi-attribute decision framework, let $g_j\left(a\right)$ denote the assessed value of alternative $\left(a\right)$ regarding the criterion under consideration, $\left(g_j\right)$. To establish a specific preference function $P_j\left(a, b\right)$, quantifying the discrepancy between the evaluations of two alternatives $\left(a and b\right)$ on criterion $\left(g_j\right)$, a function must be devised to translate this difference into a preference degree spanning from 0 to 1. This preference function is characterized as a non-decreasing function of the observed $d_j$ between the evaluations of the alternatives on the given criterion ($g_j\left(a\right)-g_j\left(b\right))$, as illustrated in Equations (16) and (17).

$$d_j\left(a,b\right)=g_j\left(a\right)-g_j\left(b\right)$$

(16)

Step 2: Application of the preference function.

$$P_j\left(a,b\right)=F_j\left[d_j\left(a,b\right)\right]j=1,\dots ,k$$

(17)

The function $P_j\left(a, b\right)$ represents the preference of alternative $a$ over alternative $b$ for each criterion, as expressed by the function $d_j\left(a, b\right)$. Six preference functions have been developed: Gaussian, Linear, Level, V-Shape, U-Shape, and Usual. The selection of these functions is based on the preference or threshold of indifference [59].

Step 3: Calculating the global preference index.

$$\pi \left(a,b\right)=\sum _{i=1}^k{P}_i(a,b)w_j$$

(18)

where $P_j(a,b)$ is the preference function and $\sum _{i=1}^k{w}_j$ are the weights of the criteria sum. Preference index $\pi \left(a,b\right)$ expresses the intensity of the decision maker’s preference for alternative $a$ over $b$ while simultaneously considering all criteria.

Step 4: Calculating outranking flows.

Each alternative $a$ is compared to $(n-1)$ other alternatives in $A$. The positive outranking $\left({\Phi }^{+}\right)$ and negative outranking $\left({\Phi }^{-}\right)$ flows are determined by using Equations (16) and (17):

$$\left({\Phi }^{+}\right)=\frac{1}{n-1}\sum _{x\in A}\pi \left(a,x\right)$$

(19)

$$\left({\Phi }^{-}\right)=\frac{1}{n-1}\sum _{x\in A}\pi \left(x,a\right)$$

(20)

Step 5: Obtain PROMETHEE I partial ranking ${(P}^I,I^I,R^I)$ from the positive outranking $\left({\Phi }^{+}\right)$ and negative outranking $\left({\Phi }^{-}\right)$ flows. It is important to note that both flows may not induce the same rankings, and their intersection represents PROMETHEE I.

$$\left\{\begin{array}{c}a{P}^{I}b if \left\{\begin{array}{c}{\Phi }^{+}\left(a\right)>{\Phi }^{+}\left(b\right) and {\Phi }^{-}\left(a\right)<{\Phi }^{-}\left(b\right), or \\ {\Phi }^{+}\left(a\right)>{\Phi }^{+}\left(b\right) and {\Phi }^{-}\left(a\right)={\Phi }^{-}\left(b\right), or \\ {\Phi }^{+}\left(a\right)={\Phi }^{+}\left(b\right) and {\Phi }^{-}\left(a\right)<{\Phi }^{-}\left(b\right); \end{array}\right.\hfill \\ a{I}^{I}b if {\Phi }^{+}\left(a\right)={\Phi }^{+}\left(b\right) and {\Phi }^{-}\left(a\right)={\Phi }^{-}\left(b\right); \hfill \\ a{R}^{I}b if \left\{\begin{array}{c}{\Phi }^{+}\left(a\right)>{\Phi }^{+}\left(b\right) and {\Phi }^{-}\left(a\right)>{\Phi }^{-}\left(b\right), or\\ {\Phi }^{+}\left(a\right)<{\Phi }^{+}\left(b\right) and {\Phi }^{-}\left(a\right)<{\Phi }^{-}\left(b\right); \end{array}\right. \hfill \end{array}\right.$$

(21)

where $P^I,I^I,R^I$ respectively represent preference, indifference, and incomparability.

Step 6: The PROMETHEE II provides the option of a complete ranking ${(P}^{II},I^{II})$ in case the decision-maker requires it. Subsequently, the net outranking flow can be taken into consideration.

$$\Phi \left(\mathrm{a}\right)={\Phi }^{+}\left(a\right)-{\Phi }^{-}\left(a\right)$$

(22)

4. Results

4.1. Operational Process and Information/Data Management in the Local VTS System

The Mersin VTS system also undertakes TOS responsibilities for vessels entering the Iskenderun sector, utilizing a robust watch system for VTSOs. This system facilitates shifts organized into three teams of VTSOs, meticulously designed to accommodate operational hours and requisite rest periods. All VTSOs possess significant maritime experience gained through previous seafaring roles. The allocation of operator consoles is determined by the number of sectors, providing VTSOs with the capability to efficiently manage diverse data and information streams originating from multiple sources, including MSW, PMIS, AIS, RADAR, Long-Range Identification and Tracking (LRIT), and Direction Finder (DF). Apart from the system’s collective management of the data received from different sensors and equipment, the availability of an additional computer system integrated into VTSO consoles enables the management and processing of information regarding the proposed decision support model.

Furthermore, Figure 4 delineates the comprehensive process of collecting and disseminating information/data via equipment, communication channels, and information-sharing systems. Typically, MSW and its subsystem PMIS are provided with information/data by authorized shipping agency personnel, and the VTSO overseeing the Iskenderun sector enters further updates upon receiving crucial information via the SP reports or very high frequency (VHF) radio communication with the ships.

To summarize, the VTS system encompasses various tools dedicated to data collection, information exchange, and a centralized database housing records pertinent to ships and operational details. Within the VTS station, the VTSO console facilitates access to all data/information and control over sensors/software. Nevertheless, the current system solely operates on an FCFS basis for admitting ships into the maritime zone. Thus, there arises a requirement for bespoke or readily available software capable of establishing a complete ranking of incoming vessels according to predefined weighted decision criteria. As a result, the compatible software screenshot illustrated in Figure A1 can effectively interface the model with the actual operational processes of the VTS system.

4.1.1. Resources for the Information/Data Requested for the Model Input

The management of the model is subject to the timely collection of the requested datasets regarding arriving vessels in the maritime domain. Subsequent discussions aimed to elucidate the rationale behind selecting these criteria, leading to a comprehensive exposition of eight decision criteria. As communication and data sharing appear complicated and consist of both dynamic and static ships’ data, domain experts especially underlined that fair queuing policy criteria should use as many static ship data as possible; otherwise, it is challenging to justify the established queuing system as ensuring equitable opportunities for arriving ships.

For instance, the safety criteria, namely the SRPV, CDET, and CDEF decision criteria, are accessible through the PMIS, where the relevant scores appear and support the evaluation of whether or not a port state control inspection is needed. The available data in PMIS are also accessible through the European Maritime Safety Agency database called THETIS or EQUASIS, which is open to the public/private sector. As fairness is the main apprehension in the shipping industry, the quality of the data sources represents the objectivity level of the PA’s assessment of competing ships. Therefore, utilizing information sourced from dependable agency databases facilitates the development of an acceptable, comprehensive ranking system for sustainable queuing policies within the maritime sector.

On the other hand, the ATPL criterion is based on the regularly updated ETA of arriving ships and considers the remaining time in minutes for the final report tendering position. The stated data are also available through the PMIS, and defined conversions can be made before ranking the arriving ships. The ATPL criterion is significant since the FCFS policy uses it as a reference for rendering nautical service and determining the berthing turn of ships. As long as the derived data based on incoming ships’ ETA are accessible to all relevant stakeholders in the maritime sector, their reliability is deemed significant.

The DEMR criterion values are computed based on recent demurrage rates provided by the port management, as presented in

Table 3. Corresponding demurrage rate with DWT tonnage.

Ship DWT Tonnage	Daily Demurrage Rate ($)
Ship < 5000	5500
5000–9999	7400
10,000–29,999	9500
30,000–54,999	15,500
55,000–79,999	19,500
80,000–189,000	29,000

. These values are converted to the ratio by dividing the prevailing demurrage cost by the corresponding ship’s deadweight capacity, ensuring equitable treatment regardless of whether vessels arrive fully laden or in ballast condition. Nevertheless, it is essential to note that Table 3 undergoes periodic revisions as the demurrage rate of the respective tonnage oscillates in harmony with the freight market.

The MMSP and LOAS stand for operational criteria, especially as stated by the maritime pilots in the expert group. The assertion posited the necessity of MMSP in facilitating nautical services, mainly due to the potential occurrence of unforeseen circumstances. It was argued that guiding faster ships is comparatively easier than guiding slower vessels in such situations. Even minor operational strategy adjustments could delay subsequent service times for queued ships. The LOAS was another concern regarding queuing arriving ships; maritime pilots advised that giving berthing priority to the largest ships would automatically affect the efficiency level of nautical services. The reason for berthing a smaller ship between the largest ships is that it takes less maneuvering time than vice versa.

Finally, it is paramount for the effective deployment of our proposed decision support model that discerning the CO₂ emission output of ships be elaborated. Therefore, the following subsection has been included for submitting further details on how the COEA data are complemented to facilitate the queuing policy.

4.1.2. Establishing Ship CO₂ Emission Production Quantity

The recommended COEA criterion value was not available in established systems reachable by the VTS; therefore, a relatively easy-to-apply CO₂ emission estimation methodology was used as proposed by the EPA [22]. The document provided by the EPA contains tables presenting average auxiliary engine and boiler loads (kW) at anchorage for specific ship types and subtypes and default auxiliary and boiler CO₂ emission factors (g/kW-hr). The subsequent formula is furnished for the computation of the hourly CO₂ emission levels pertaining to the distinct vessel categories.

$${EP}_{{CO}_2}=\left(AP\times DR\times AEF+VB\times DR\times BEF\right)\times C$$

(23)

where:

${EP}_{{CO}_2}$ = CO₂ emission production (tons)

$AP$ = Total auxiliary power (kW)

$DR$ = Anticipated time at anchorage for the vessel (hours)

$AEF$ = Auxiliary engine CO₂ emission factor (g/kWh)

$VB$ = Total boiler power

$BEF$ = Boiler CO₂ emission factor (g/kWh)

$C$ = Conversion factor from grams to tons (10⁻⁶ tons/g)

The implementation of procedures outlined in the EPA’s methodology facilitated the generation of

Table 4. Ships’ CO₂ emission production per hour (created using the EPA methodology [22]).

Ship Type	Subtype	Auxiliary (kW)	Boiler (kW)	Auxiliary Engine Emission Factor (g/kW-hr)	Boiler Emission Factor (g/kW-hr)	CO2 Emission Production at Anchorage (tons/h)
Bulk Carrier	Handysize	190	50	695.702	961.8	0.180
	Handymax	260	100			0.277
	Panamax	420	200			0.485
	Capesize	420	200			0.485
General Cargo	Up to 10,000 DWT	60	0			0.042

. This table illustrates the emitted CO₂ levels by particular ship types during their waiting period at anchorage.

4.2. Application of the FAHP Method to Determine Priority Weights of Decision Criteria

The CCI values of each matrix were obtained by applying Equation (2). It was found that the matrix produced by the Chief VTSO was CCI = 0.34, the matrix produced by a maritime pilot was CCI = 0.28, and the matrix produced by the Regional Harbor Master was CCI = 0.25, as illustrated in

Table A1. Individual pairwise comparison fuzzy judgment matrix.

DM1	SRPV	CDET	CDEF	COEA	ATPL	DEMR	MMSP	LOAS
SRPV	(1, 1, 1)	(1, 1, 1)	(1, 1, 1)	(3, 5, 7)	(1, 3, 5)	(3, 5, 7)	(3, 5, 7)	(1, 3, 5)
CDET	(1, 1, 1)	(1, 1, 1)	(3, 5, 7)	(0.2, 0.33, 1)	(0.2, 0.33, 1)	(0.14, 0.2, 0,33)	(0.2, 0.33, 1)	(1, 3, 5)
CDEF	(1, 1, 1)	(0.14, 0.2, 0.33)	(1, 1, 1)	(0.2, 0.33, 1)	(0.2, 0.33, 1)	(0.2, 0.33, 1)	(0.2, 0.33, 1)	(0.2, 0.33, 1)
COEA	(0.14, 0.2, 0.33)	(1, 3, 5)	(1, 3, 5)	(1, 1, 1)	(3, 5, 7)	(0.2, 0.33, 1)	(1, 3, 5)	(1, 3.66, 7)
ATPL	(0.2, 0.33, 1)	(1, 3, 5)	(1, 3, 5)	(0.14, 0.2, 0.33)	(1, 1, 1)	(3, 5, 7)	(3, 5, 7)	(3, 5, 7)
DEMR	(0.14, 0.2, 0.33)	(3, 5, 7)	(1, 3, 5)	(1, 3, 5)	(0.11, 0.14, 0.2)	(1, 1, 1)	(3, 5, 7)	(3, 5, 7)
MMSP	(0.14, 0.2, 0.33)	(1, 3, 5)	(1, 3, 5)	(0.2, 0.33, 1)	(0.11, 0.14, 0.2)	(0.14, 0.2, 0.33)	(1, 1, 1)	(1, 1, 1)
LOAS	(0.2, 0.33, 1)	(0.2, 0.33, 1)	(1, 3, 5)	(0.14, 0.2, 1)	(0.11, 0.14, 0.2)	(0.14, 0.2, 0.33)	(1, 1, 1)	(1, 1, 1)
CCI = 0.34
DM2	SRPV	CDET	CDEF	COEA	ATPL	DEMR	MMSP	LOAS
SRPV	(1, 1, 1)	(1, 1, 1)	(1, 1, 1)	(3, 5, 7)	(1, 3, 5)	(1, 3, 5)	(3, 5, 7)	(1, 3, 5)
CDET	(1, 1, 1)	(1, 1, 1)	(1, 1, 1)	(0.2, 0.33, 1)	(0.2, 0.33, 1)	(0.2, 0.33, 1)	(0.2, 0.33, 1)	(1, 3, 5)
CDEF	(1, 1, 1)	(1, 1, 1)	(1, 1, 1)	(0.2, 0.33, 1)	(0.2, 0.33, 1)	(0.2, 0.33, 1)	(0.2, 0.33, 1)	(0.2, 0.33, 1)
COEA	(0.14, 0.2, 0.33)	(1, 3, 5)	(1, 3, 5)	(1, 1, 1)	(3, 5, 7)	(0.2, 0.33, 1)	(1, 3, 5)	(1, 3, 5)
ATPL	(0.2, 0.33, 1)	(1, 3, 5)	(1, 3, 5)	(0.14, 0.2, 0.33)	(1, 1, 1)	(3, 5, 7)	(3, 5, 7)	(3, 5, 7)
DEMR	(0.2, 0.33, 1)	(1, 3, 5)	(1, 3, 5)	(1, 3, 5)	(0.11, 0.14, 0.2)	(1, 1, 1)	(3, 5, 7)	(3, 5, 7)
MMSP	(0.14, 0.2, 0.33)	(1, 3, 5)	(1, 3, 5)	(0.2, 0.33, 1)	(0.11, 0.14, 0.2)	(0.14, 0.2, 0.33)	(1, 1, 1)	(1, 1, 1)
LOAS	(0.2, 0.33, 1)	(0.2, 0.33, 1)	(1, 3, 5)	(0.2, 0.33, 1)	(0.11, 0.14, 0.2)	(0.14, 0.2, 0.33)	(1, 1, 1)	(1, 1, 1)
CCI = 0.28
DM3	SRPV	CDET	CDEF	COEA	ATPL	DEMR	MMSP	LOAS
SRPV	(1, 1, 1)	(1, 1, 1)	(1, 1, 1)	(0.2, 0.33, 1)	(1, 3, 5)	(1, 3, 5)	(3, 5, 7)	(1, 3, 5)
CDET	(1, 1, 1)	(1, 1, 1)	(1, 1, 1)	(0.2, 0.33, 1)	(0.2, 0.33, 1)	(0.2, 0.33, 1)	(0.2, 0.33, 1)	(1, 3, 5)
CDEF	(1, 1, 1)	(1, 1, 1)	(1, 1, 1)	(0.2, 0.33, 1)	(0.2, 0.33, 1)	(0.2, 0.33, 1)	(0.2, 0.33, 1)	(0.2, 0.33, 1)
COEA	(1, 3, 5)	(1, 3, 5)	(1, 3, 5)	(1, 1, 1)	(3, 5, 7)	(0.2, 0.33, 1)	(1, 3, 5)	(1, 3, 5)
ATPL	(0.2, 0.33, 1)	(1, 3, 5)	(1, 3, 5)	(0.14, 0.2, 0.33)	(1, 1, 1)	(3, 5, 7)	(3, 5, 7)	(3, 5, 7)
DEMR	(0.2, 0.33, 1)	(1, 3, 5)	(1, 3, 5)	(1, 3, 5)	(0.11, 0.14, 0.2)	(1, 1, 1)	(3, 5, 7)	(3, 5, 7)
MMSP	(0.14, 0.2, 0.33)	(1, 3, 5)	(1, 3, 5)	(0.2, 0.33, 1)	(0.11, 0.14, 0.2)	(0.14, 0.2, 0.33)	(1, 1, 1)	(1, 1, 1)
LOAS	(0.2, 0.33, 1)	(0.2, 0.33, 1)	(1, 3, 5)	(0.2, 0.33, 1)	(0.11, 0.14, 0.2)	(0.14, 0.2, 0.33)	(1, 1, 1)	(1, 1, 1)
CCI = 0.25

. As each individual judgment matrix met the consistency criterion, the aggregation of individual fuzzy judgment matrices was made using the max–min arithmetic mean operation shown in Equation (1). Then, the aggregated fuzzy judgment matrix CCI = 0.29 was calculated as less than the threshold value of 0.37, as depicted in

Table 5. The aggregated fuzzy judgment matrix for criteria.

	SRPV	CDET	CDEF	COEA	ATPL	DEMR	MMSP	LOAS
SRPV	(1, 1, 1)	(1, 1, 1)	(1, 1, 1)	(0.2, 3.44, 7)	(1, 3, 5)	(1, 3.66, 7)	(3, 5, 7)	(1, 3, 5)
CDET	(1, 1, 1)	(1, 1, 1)	(1, 2.33, 7)	(0.2, 0.33, 1)	(0.2, 0.33, 1)	(0.14, 0.28, 1)	(0.2, 0.33, 1)	(1, 3, 5)
CDEF	(1, 1, 1)	(0.14, 0.73, 1)	(1, 1, 1)	(0.2, 0.33, 1)	(0.2, 0.33, 1)	(0.2, 0.33, 1)	(0.2, 0.33, 1)	(0.2, 0.33, 1)
COEA	(0.14, 1.13, 5)	(1, 3, 5)	(1, 3, 5)	(1, 1, 1)	(3, 5, 7)	(0.2, 0.33, 1)	(1, 3, 5)	(1, 3.66, 7)
ATPL	(0.2, 0.33, 1)	(1, 3, 5)	(1, 3, 5)	(0.14, 0.2, 0.33)	(1, 1, 1)	(0.2, 5, 7)	(3, 5, 7)	(3, 5, 7)
DEMR	(0.14, 0.28, 1)	(1, 3.66, 7)	(1, 3, 5)	(1, 3, 5)	(0.11, 0.14, 0.2)	(1, 1, 1)	(3, 5, 7)	(3, 5, 7)
MMSP	(0.14, 0.2, 0.33)	(1, 3, 5)	(1, 3, 5)	(0.2, 0.33, 1)	(0.11, 0.14, 0.2)	(0.14, 0.2, 0.33)	(1, 1, 1)	(1, 1, 1)
LOAS	(0.2, 0.33, 1)	(0.2, 0.33, 1)	(1, 3, 5)	(0.14, 0.28, 1)	(0.11, 0.14, 0.2)	(0.14, 0.2, 0.33)	(1, 1, 1)	(1, 1, 1)
CCI = 0.29

.

The computation of the fuzzy synthetic extent value for the pertinent objects has been conducted utilizing Equation (4).

$$\begin{gathered} S_{SRPV}=\left(\mathrm{0.05,0.19,0.63}\right), S_{CDET}=\left(0.03,0.08,0.34\right), S_{CDEF}=\left(0.02,0.04,0.15\right), \\ {S}_{COEA}=\left(0.04,0.18,0.67\right), S_{ATPL}=\left(0.05,0.20,0.62\right), S_{DEMR}=\left(0.05,0.19,0.63\right), \\ {S}_{MMSP}=\left(0.02,0.08,0.26\right), S_{LOAS}=\left(0.02,0.16,0.20\right) \end{gathered}$$

Upon obtaining the aforementioned results, the degree of possibility values were computed utilizing Equation (9), as presented in

Table A2. The degree of possibility values.

V(SSRPV≥SCDET) = 1	V(SSRPV≥SCDEF) = 1	V(SSRPV≥SCOEA) = 1	V(SSRPV≥SATPL) = 0.98	V(SSRPV≥SDEMR) = 1	V(SSRPV≥SMMSP) = 1	V(SSRPV≥SLOAS) = 1
V(SCDET≥SSRPV) = 0.73	V(SCDET≥SCDEF) = 0.78	V(SCDET≥SCOEA) = 0.75	V(SCDET≥SATPL) = 0.75	V(SCDET≥SDEMR) = 0.73	V(SCDET≥SMMSP) = 1	V(SCDET≥SLOAS) = 0.8
V(SCDEF≥SSRPV) = 0.40	V(SCDEF≥SCDET) = 0.75	V(SCDEF≥SCOEA) = 0.44	V(SCDEF≥SATPL) = 0.38	V(SCDEF≥SDEMR) = 0.76	V(SCDEF≥SMMSP) = 0.76	V(SCDEF≥SLOAS) = 0.52
V(SCOEA≥SSRPV) = 0.98	V(SCOEA≥SCDET) = 1	V(SCOEA≥SCDEF) = 1	V(SCOEA≥SATPL) = 0.97	V(SCOEA≥SDEMR) = 0.98	V(SCOEA≥SMMSP) = 1	V(SCOEA≥SLOAS) = 1
V(SATPL≥SSRPV) = 1	V(SATPL≥SCDET) = 1	V(SATPL≥SCDEF) = 1	V(SATPL≥SCOEA) = 1	V(SATPL≥SDEMR) = 1	V(SATPL≥SMMSP) = 1	V(SATPL≥SLOAS) = 1
V(SDEMR≥SSRPV) = 1	V(SDEMR≥SCDET) = 1	V(SDEMR≥SCDEF) = 1	V(SDEMR≥SCOEA) = 1	V(SDEMR≥SATPL) = 0.98	V(SDEMR≥SDEMR) = 1	V(SDEMR≥SLOAS) = 1
V(SMMSP≥S_SRPV) = 0.66	V(SMMSP≥SCDET) = 1	V(SMMSP≥SCDEF) = 1	V(SMMSP≥SCOEA) = 0.69	V(SMMSP≥SATPL) = 0.66	V(SMMSP≥SDEMR) = 0.66	V(SMMSP≥SLOAS) = 0.75
V(SLOAS≥SSRPV) = 0.83	V(SLOAS≥SCDET) = 1	V(SLOAS≥SCDEF) = 1	V(SLOAS≥SCOEA) = 0.89	V(SLOAS≥SATPL) = 0.79	V(SLOAS≥SDEMR) = 0.83	V(SLOAS≥SMMSP) = 1

. Subsequently, the priority weights were determined using Equation (10).

$$\begin{gathered} d\text{'}\left(S_{SRPV})=\mathrm{m}\mathrm{i}\mathrm{n}(1,1,1,0.98,1,1,1\right)=0.98 \\ d\text{'}\left(S_{CDET})=\mathrm{m}\mathrm{i}\mathrm{n}(0.73,0.78,0.75,0.75,0.73,1,0.8\right)=0.73 \\ d\text{'}\left(S_{CDEF})=\mathrm{m}\mathrm{i}\mathrm{n}(0.40,0.75,0.44,0.38,0.76,0.76,0.52\right)=0.38 \\ d\text{'}\left(S_{COEA})=\mathrm{m}\mathrm{i}\mathrm{n}(1,1,1,1,1,1,1\right)=1 \\ d\text{'}\left(S_{ATPL})=\mathrm{m}\mathrm{i}\mathrm{n}(0.98,1,1,0.97,0.98,1,1\right)=0.97 \\ d\text{'}\left(S_{DEMR})=\mathrm{m}\mathrm{i}\mathrm{n}(1,1,1,1,0.98,1,1\right)=0.98 \\ d\text{'}\left(S_{MMSP})=\mathrm{m}\mathrm{i}\mathrm{n}(0.66,1,1,0.69,0.66,0.66,0.75\right)=0.66 \\ d\text{'}\left(S_{LOAS})=\mathrm{m}\mathrm{i}\mathrm{n}(0.83,1,1,0.89,0.79,0.83,1\right)=0.98 \end{gathered}$$

Priority weights form vectors in accordance with Equation (11):

$$W\text{'}=(0.98;0.73;0.38;1;0.97;0.98;0.66;0.98)$$

After normalization of the values, the priority weights assigned to each criterion were computed as per Equation (12).

$$\begin{gathered} W_{SRPV}=0.15\hspace{1em}{W}_{CDET}=0.11\hspace{1em}{W}_{CDEF}=0.06\hspace{1em}{W}_{COEA}=0.15 \\ {W}_{ATPL}=0.14\hspace{1em}{W}_{DEMR}=0.15\hspace{1em}{W}_{MMSP}=0.1\hspace{1em}{W}_{LOAS}=0.14 \end{gathered}$$

4.3. Ship-Wise Nautical Service Duration

Apart from the criteria specifications, the individual nautical service durations for ships were essential in ascertaining the cumulative waiting time of incoming vessels at anchorages, awaiting their turn. To seek a solution for this matter, the dataset obtained from Isdemir Pilotage and Towage Organization contains the tugboat service durations for 323 vessels in the year 2022, statistically analyzed with respect to gross tonnage and service duration.

Following the application of a normality test to the available data, the obtained p-value was less than 0.05, indicating that the data do not follow a normal distribution. Consequently, Spearman’s rank correlation coefficient method was employed for correlation analysis, resulting in a correlation coefficient of 0.796. This coefficient serves to quantify the strength of the relationship between two variables. The correlation between maneuver duration and gross tonnage has been depicted in Figure 5. Due to the presence of a statistically significant correlation between gross tonnage and maneuvering duration, a simple linear regression analysis was conducted. Subsequently, a model was developed using Minitab software to predict maneuvering times based on gross tonnage values ranging from 10,000 to 90,000 [58].

A linear regression analysis was deemed appropriate, with gross tonnage considered the independent variable and maneuvering time as the dependent variable. This analysis aimed to derive a formula ($y=a+bx)$, expressing the relationship between the input and output variables, thereby obtaining a formula for maneuvering duration in terms of gross tonnage. The R-squared value of 69.31% indicates that the output can be effectively modeled using the provided input(s). The analysis showed that the R-squared and predicted R-squared values were 69.21% and 68.77%, respectively. These figures suggest statistically significant modeling of maneuvering time based on gross tonnage input by using constants a and b in the regression equation $(Maneuver duration=\mathrm{39,941}+0.000576\ast gross tonnage)$. Furthermore, the p-value of the statistical indicators, being less than 0.05, further underscores the robust relationship between the input and output variables. The model was used to predict maneuvering times with the Minitab software for gross tonnage values between 10,000 and 90,000. The derived results from the model presented in

Table 6. Nautical service duration with respect to ship gross tonnage [58].

Gross Tonnage	Maneuver Duration (min)
Ship < 9999	45
10,000–19,999	51
20,000–29,999	57
30,000–39,999	62
40,000–49,999	68
50,000–59,999	74
60,000–69,999	80
70,000–79,999	86
80,000–90,000	91

were validated by experts in the field, prompting our decision to incorporate these results into the case study undertaken in this research.

4.4. Application of PROMETHEE II in a Case Study

The FCFS policy, frequently employed within the dry bulk shipping domain, aligns with prevailing contractual norms in the industry. Nonetheless, advocating for an extensive decision-making framework to support an alternative queuing strategy that prioritizes incoming vessels based on predetermined criteria presents a novel standpoint. This approach advocates for mitigating vessel congestion at port boundaries, potentially reducing CO₂ emissions. In order to measure the feasibility of this proposed methodology, a case study scenario is constructed to assess the outcomes of the model and ascertain its potential effectiveness against the FCFS policy concerning the reduction in CO₂ emissions. The case study delineates procedures such as prioritizing incoming vessels, rendering nautical service in accordance with the obtained complete ranking of the ships, and evaluating waiting times and CO₂ emissions generated within the anchorage zone.

As depicted in Figure 6, the ship traffic flow in this case study is based on actual operational data sourced from the Mersin VTS organization and the Isdemir dry cargo port facility in Iskenderun Bay. Incoming ships rush to pass the port limit for tendering notice of readiness and acquiring priority for nautical service to maneuver for berthing the vacant berths in the dry cargo port. The distance from the port’s official boundary to the PBP spans 35 nautical miles and has neither draft restriction nor tidal dept change; in other words, depths in the fairway and tide range do not pose a risk to navigational safety. In the defined case study, there are no constraints concerning berth availability; however, the scarcity of nautical services due to a limited number of tugboats necessitates incoming vessels to await their turn within the designated anchorage area while preceding vessels receive the required services. While ships navigate within the fairway and approach the PBP position, transit time may be extended because of changing main engine loads at different phases. This case study assumes that ships can maneuver at high speeds in anchorage areas, eliminating the impact of anchorage maneuvering duration on waiting time at anchorage.

The FCFS policy grants precedence to the initial vessel crossing the port limit. On the other hand, the proposed decision support model for queuing policy assigns priority to the vessel with the highest rank, regardless of its arrival time at the PBP. As depicted in

Table 7. Determination of the waiting time of each arriving ship based on FCFS.

Ship Code	Gross Ton	MMSP	ETA to Port Limit	Transit From Port Limit to PBP or Anchorage	Arrival Time to PBP or Anchorage	Waiting Time for Berthing Turn (h)	Berthing Maneuver Commencement Time (RTA to PBP)	Maneuver Duration	Berthing Completion Time
Ship 1	2390	10	01:50	03:30	05:20	00:00	05:20	00:45	06:05
Ship 2	22,662	13	01:55	02:41	04:36	01:29	06:05	00:57	07:02
Ship 3	5164	11	02:05	03:11	05:16	01:46	07:02	00:45	07:47
Ship 4	26,058	13	02:15	02:41	04:56	02:51	07:47	00:57	08:44
Ship 5	41,655	14	02:15	02:30	04:45	03:59	08:44	01:08	09:52
Ship 6	92,758	15	02:20	02:20	04:40	05:12	09:52	01:31	11:23
Ship 7	6478	12	02:30	02:55	05:25	05:58	11:23	00:45	12:08
Ship 8	17,027	14	02:40	02:30	05:10	06:58	12:08	00:51	12:59
Ship 9	6201	13	02:55	02:41	05:36	07:23	12:59	00:45	13:44
Ship 10	2265	11	03:10	03:11	06:21	07:23	13:44	00:45	14:29

and Figure 6, pertinent data and time stamps are available for computing to produce results under the FCFS policy, including the actual ETA of ships to the port limit. The transit duration within the fairway, from the port limit to the anchorage area or PBP location, is derived using available data regarding ship speed measured in knots and fairway distance in nautical miles. The only directly applied dataset pertains to maneuver duration, contingent upon the gross tonnage of ships, as delineated in Table 6 (Section 4.1.2). The VTSO announces the time stamp of the RTA at the PBP in advance, ensuring the readiness of the vessel in line for access to nautical service, as illustrated in Table 7.

The outcomes of the FCFS policy under defined conditions illustrated in Figure 7 reveal a cumulative waiting time of 42.98 h with a corresponding CO₂ emission of 7.71 tons, according to Table 4, due to waiting turns for nautical service at the anchorage area.

Before executing the PROMETHEE II method for complete ranking,

Table 8. Arriving ships’ particulars and respective criteria scores.

Ship Code	Ship Type	Dwt	Grt	SRPV Ratio	CDET Ratio	CDEF Ratio	COEA (Ton/h)	ATPL (m)	DEMR ($/Ton)	MMSP Knots	LOAS (m)
				Min	Min	Min	Min	Min	Min	Max	Max
Ship 1	General Cargo	4325	2390	6	2.41	1.9	0.042	110	1.27	10	89
Ship 2	Bulk Carrier	37,196	22,662	3	3.58	2.9	0.18	115	0.42	13	186
Ship 3	General Cargo	7431	5164	3	3	0.18	0.042	125	0.99	11	119
Ship 4	Bulk Carrier	45,719	26,058	3	6.3	0.38	0.277	135	0.34	13	185
Ship 5	Bulk Carrier	75,725	41,655	4	9.1	1.9	0.485	135	0.26	14	225
Ship 6	Bulk Carrier	181,396	92,758	3	2.4	0.15	0.485	140	0.16	15	292
Ship 7	General Cargo	9641	6478	3	3.2	0.72	0.042	150	0.77	12	116
Ship 8	Bulk Carrier	28,208	17,027	4	2.6	1.4	0.18	160	0.34	14	169
Ship 9	General Cargo	8314	6201	3	2.41	1.9	0.042	175	0.88	13	120
Ship 10	General Cargo	3241	2265	3	3.58	0.42	0.042	190	1.7	11	98

displays the input particulars for ten incoming ships, featuring the ships’ characteristics and corresponding criterion scores. In order to prevent commercial conflicts, ship names have been substituted with ship codes. The input details include whether each criterion is to be minimized (Min) or maximized (Max), along with the unit for each criterion. All criteria except LOAS and MMSP are non-beneficial; hence, lower values are desirable. All data entries to the Visual PROMETHEE software (1.4.0.0) are presented in Figure A1. By utilizing the data provided in Table 8, employing linear preference functions for each criterion, and integrating weights derived from the FAHP method, a complete ranking is established via the PROMETHEE II approach using Visual PROMETHEE software, as illustrated in

Table 9. The PROMETHEE II complete ranking.

Ship Code	Phi+	Phi-	Phi	Rank
Ship 6	0.3597	0.0778	0.2819	1
Ship 2	0.2969	0.1008	0.1961	2
Ship 5	0.2708	0.1899	0.0809	3
Ship 4	0.2565	0.1867	0.0698	4
Ship 3	0.2192	0.1756	0.0436	5
Ship 8	0.2038	0.1744	0.0294	6
Ship 9	0.1408	0.2310	−0.0902	7
Ship 7	0.1403	0.2356	−0.0953	8
Ship 1	0.1686	0.3874	−0.2188	9
Ship 10	0.0729	0.3705	−0.2976	10

.

Once the ship ranking for berthing priority is established and nautical services commence upon the arrival of the vessel having first rank at the PBP, the total waiting time for all ships is calculated to be 40.32 h, as illustrated in Figure 8. This waiting period results in a corresponding CO₂ emission of 4.19 tons due to ships awaiting their turn at the anchorage area. Moreover, the dissemination of information regarding the arrival of ships at designated geographic coordinates, along with the RTA at the PBP data outlined in

Table 10. Port operational process time stamps, including the RTA at the PBP.

Ship Code	Gross Ton	MMSP (Knots)	ETA to Port Limit	Transit from Port Limit to PBP or Anchorage	Arrival Time to PBP or Anchorage	Waiting Time for Berthing Turn	Berthing Maneuver Commencement Time (RTA to PBP)	Maneuver Duration (m)	Berthing Time
Ship 6	92,758	15	02:20	02:20	04:40	00:00	04:40	01:31	06:11
Ship 2	22,662	13	01:55	02:41	04:36	01:35	06:11	00:57	07:08
Ship 5	41,655	14	02:15	02:30	04:45	02:23	07:08	01:08	08:16
Ship 4	26,058	13	02:15	02:41	04:56	03:20	08:16	00:57	09:13
Ship 3	5164	11	02:05	03:11	05:16	03:57	09:13	00:45	09:58
Ship 8	17,027	14	02:40	02:30	05:10	04:48	09:58	00:51	10:49
Ship 9	6201	13	02:55	02:41	05:36	05:13	10:49	00:45	11:34
Ship 7	6478	12	02:30	02:55	05:25	06:09	11:34	00:45	12:19
Ship 1	2390	10	01:50	03:30	05:20	06:59	12:19	00:45	13:04
Ship 10	2265	11	03:10	03:11	06:21	06:43	13:04	00:45	13:49

, which includes time stamps of port operational procedures, not only facilitates JITA planning for incoming vessels but also empowers ship captains to adjust ship speed to mitigate fuel consumption proactively.

4.5. Analysis of the Results via GAIA Plane and Walking Weights

The two-dimensional GAIA plane, characterized by a quality level of 79.3% for visualization purposes, is depicted in Figure 9, facilitating an analysis of the most influential criteria for ranking incoming ships. Decision-makers have accorded relatively higher priority to the ATPL, LOAS, MMSP, SRPV, and DEMR criteria. Consequently, Ship 2, Ship 3, Ship 4, Ship 5, and Ship 6 have secured positions in the upper echelons of the list. While Ship 2 closely aligns with the ATPL criterion, Ship 6, holding the foremost rank, also exhibits considerable proximity to multiple criteria. However, it is noted that DEMR and COEA exert relatively less influence on decision-makers compared to previously stated criteria.

The Visual PROMETHEE software provides a unique feature in the form of sensitivity analysis, facilitated through the “walking weights” option. This functionality allows for the adjustment of significant weights to observe the resulting changes in the PROMETHEE II ranking. Figure 10, presented herein, depicts two distinct scenarios characterized by varying distributions of priority weights. In the first scenario, the priority weights of ATPL, DEMR, and COEA were increased by 20%, 18%, and 20%, respectively, to examine arriving ships’ ranking and corresponding CO₂ emissions. Similarly, the priority weights of SRPV, CDET, CDEF, and COEA were raised by 15% equally in the second scenario to obtain an alternative ship ranking for further examination. The upper bar chart presents the PROMETHEE II complete ranking for incoming ships, while the lower bar chart showcases the redistributed priority weights to criteria groups.

In the first scenario, waiting for incoming ships at the anchorage totals 40.22 h, coinciding with an emission output of 5.17 tons of CO₂. Conversely, in the second scenario, where safety considerations are integrated with environmental factors and proportionally distributed across other criteria by 10%, the aggregated waiting time diminishes to 39.93 h, concurrent with a calculated CO₂ emission of 5.35 tons. Despite the marginal variance between the two scenarios, the outcomes derived are superior to those yielded by the FCFS policy, as appears in

Table 11. Queuing arriving vessels under different policies and scenarios.

FCFS			Proposed Model			Walking Weight Scenario 1			Walking Weight Scenario 2
Ship Ranking	Waiting Duration(h)	CO2 Emission (Tons)	Ship Ranking	Waiting Duration(h)	CO2 Emission (Tons)	Ship Ranking	Waiting Duration(h)	CO2 Emission (Tons)	Ship Ranking	Waiting Duration(h)	CO2 Emission (Tons)
Ship 1	0.00	0.00	Ship 6	0.00	0.00	Ship 6	0.00	0.00	Ship 6	0.00	0.00
Ship 2	1.48	0.27	Ship 2	1.58	0.29	Ship 2	1.58	0.28	Ship 2	1.58	0.29
Ship 3	1.77	0.07	Ship 5	2.38	1.16	Ship 3	1.86	0.08	Ship 3	1.86	0.08
Ship 4	2.85	0.79	Ship 4	2.20	0.61	Ship 4	2.95	0.82	Ship 8	2.71	0.49
Ship 5	3.98	1.93	Ship 3	3.95	0.17	Ship 5	4.08	1.98	Ship 4	3.80	1.05
Ship 6	5.20	2.52	Ship 8	5.13	0.92	Ship 1	4.63	0.19	Ship 5	4.93	2.39
Ship 7	5.97	0.25	Ship 9	5.22	0.22	Ship 8	5.55	1.00	Ship 7	5.40	0.23
Ship 8	6.97	1.25	Ship 7	6.15	0.26	Ship 7	6.15	0.26	Ship 9	5.96	0.25
Ship 9	7.38	0.31	Ship 1	6.98	0.29	Ship 9	6.71	0.28	Ship 10	5.96	0.25
Ship 10	7.38	0.31	Ship 10	6.72	0.28	Ship 10	6.71	0.28	Ship 1	7.73	0.32
Total	42.98	7.71	Total	40.32	4.19	Total	40.22	5.17	Total	39.93	5.35

.

The employment of the two-dimensional GAIA plane and the walking weights function within Visual PROMETHEE has yielded invaluable insights into decision-making processes and the responsiveness of rankings to alterations in criteria weights. The findings underscore the significance of factors such as ATPL, LOAS, MMSP, SRPV, and DEMR in determining ship rankings while highlighting the relatively lower influence of COEA. Moreover, the sensitivity analyses demonstrate the potential implications of adjusting criteria weights on waiting times and emissions, emphasizing the need for careful consideration in decision-making processes. While the FCFS policy yields the most unfavorable outcomes regarding waiting duration at anchorage and CO₂ emissions, the proposed decision support model for the queuing policy and alternative scenarios surpassed it. Although the model did not yield the shortest waiting time, it achieved the lowest cumulative CO₂ emissions of 4.19 tons. Conversely, the alternative scenarios exhibited reduced waiting time at anchorage but resulted in higher CO₂ emissions than the proposed decision support model outputs.

5. Discussion

The results imply that stakeholders place considerable emphasis on safety, business, efficiency, and environmental factors when assessing the scheduling of incoming ships. Significantly, the decision criteria established by experts exhibit similarity to dynamic and static data employed in prior research on vessel traffic scheduling and optimization [23,24,25]. However, safety criteria are accorded the most significant priority weight at 32%, reflecting their salience and relevance to marine operations, as emphasized by scholars [3,11,25]. Subsequently, business and efficiency criteria are allocated 29% and 25%, respectively, while environmental considerations receive a weight of 15%. The proposed framework enables experts in the field to assign priority weights, considering the diverse dynamics within the port vicinity, primarily relying on their intuition rather than strict constraints. The GAIA model furnished significant insights into the decision-making process concerning CO₂ emissions, with the COEA abbreviation indicating that decision-makers assign these decisions relatively lower precedence. Nevertheless, prioritizing efficiency alongside business and safety criteria suggests that the PA and affiliated organizations prioritize the optimal utilization of infrastructure and resources to enhance port productivity while considering safety and environmental concerns concurrently. Consequently, the proposed model assesses competing arriving ships under predetermined criteria, which is the essential feature expected from an alternative fair queuing policy [11].

According to the findings delineated in the study, the FCFS policy incurs a cumulative waiting duration of 42.98 h at anchorage, resulting in the emission of 7.71 tons of CO₂. Conversely, establishing a queuing policy using the proposed decision support model that determines the lineup for nautical services led to a reduction in the cumulative waiting time for all incoming vessels by 2.66 h, coupled with a decrease in cumulative CO₂ emissions produced at the anchorage area by 3.52 tons. Despite the operational phase at anchorage constituting a relatively minor portion of a ship’s port call, as indicated in the research conducted by Winnes et al. [15], considerable CO₂ emissions were saved from ten ships awaiting their turn for maritime services. The case study focuses on approximately the final 12-h operational period for incoming vessels; however, the reductions in CO₂ emissions based on various scenarios range between 32.8% and 45%, highlighting the potential to mitigate greenhouse gas emissions by minimizing fuel consumption [3]. As a result, this study emphasizes the significance and effectiveness of operational tactics and strategies in attaining positive outcomes related to GHG emissions reduction within a restricted timeframe, even in the absence of speed adjustments for incoming vessels.

The future role of VTS systems consists of applying the JITA policy and setting allocation between vessels [26,27,28], acting as aviation-style traffic control [27,29,30]. The compatibility of the queuing system with the JITA policy is only a matter of disseminating the maneuver commencement time available in Table 7 as the RTA at the PBP with the arriving dry cargo vessels in advance, in accordance with the existing operational measures to reduce GHG emissions [3,7,8]. If the VTSO instructs incoming vessels to adjust their ETA accordingly, the resulting CO₂ emissions of 4.19 tons would be further diminished. This adjustment would ensure complete compliance with the JITA policy and yield fuel savings during the approach phase. It is evident that the VTS system is the only established system that can provide TOS for arriving vessels to the port jurisdiction; hence, the VTS system should form the cornerstone of operational measures to ensure the sustainable management of vessel arrivals. Nevertheless, the stipulated ETA within the regulated maritime domain must be legally binding for respective parties to start the laytime clock for vessels bound for their designated destination; failure to do so would pose challenges in mitigating the tendency of dry cargo vessels to hasten toward port boundaries. This advancement will facilitate the dry bulk shipping sector’s voluntary engagement in adhering to the JITA initiative, contingent upon the port of arrival offering a framework aligned with the sector’s contractual prerequisites.

The implementation of this complete ship ranking system by using the proposed decision support model is also notably straightforward, leading to a reduction in workload pressure on VTSOs [23]. The model’s features are also suitable for generating ship rankings based on a pre-assigned set of weights for criteria set by relevant authorities. For instance, if specific weather conditions significantly impact navigation safety, VTSOs utilize weight distributions allocated under the safety-intensive mode. Similarly, authorities can prioritize environmental, efficiency, and economic-intensive criteria in advance for the perusal of the VTSO in charge. However, considering different priorities due to geographic difficulties and port governance approaches in various jurisdictions may raise the need for preliminary adjustments to implement the proposed queuing policy. On the other hand, the model relies heavily on the data obtained from the different resources; thus, the quality of the data, such as whether they are real-time or not, directly affects the obtained results. In addition, a smooth flow of real-time data would further relieve the VTSO workload; otherwise, an additional officer may be imperative for data handling and organizing vessel traffic under the requirements of sustainable operational measures.

6. Conclusions

This paper proposes an integrated decision support model tailored explicitly for the VTS system that effectively manages ship berthing prioritization through a sustainable queuing policy for the dry cargo shipping sector. The model incorporates the FAHP and PROMETHEE II decision-making techniques, which are crucial in ensuring the model’s compatibility and effectiveness. Furthermore, considering the priorities of the respective parties with control over port marine traffic and providing a complete ranking of arriving ships that is agreeable to stakeholders strengthens the fairness feature of the model. On the other hand, this ranking system sets the groundwork for eradicating hasty attempts to cross port boundaries, as nautical services for arriving ships are predefined in accordance with the published time stamps and the ranking of arriving ships. The application of a case study in the Gulf of Iskenderun enabled testing of the model outcomes, and the results demonstrated the model’s effectiveness and compatibility with the JITA initiative. The main conclusion remarks of this study can be summarized as follows:

• The Port Authority (PA) and affiliated public entities are queuing arriving ships while prioritizing safety criteria, which take precedence, followed by business interests, operational efficiency, and environmental impacts.
• Augmenting the VTS system with the proposed model enhances the ability to render TOS under the PA’s equitable queuing policy, which aligns with JITA requirements.
• The implementation of the proposed model reduces CO₂ emissions based on various scenarios, ranging between 32.8% and 45% in comparison with FCFS.
• The model can be applied to other ports after the application procedures underlined in the integrated decision support model framework.

Future research endeavors could extend beyond the dry bulk shipping sector to explore the integration of various sectors and terminals within the maritime domain and address real-world challenges comprehensively. Furthermore, there is a movement towards integrating contemporary concepts such as P-CDM, which presents novel challenges in engaging a broader spectrum of stakeholders to establish priorities. Additionally, improving data flow in the model has the potential to expedite processes by eliminating manual data entry tasks. Therefore, efforts to integrate the model with E-Navigation may broaden the scope of criteria and establish a predefined set of criteria for alternative conditions. In order to enhance transparency and fairness in the maritime domain, it is suggested that complete ship ranking tables, including the RTA at the PBP data as depicted in Table 10, be made accessible on a dedicated website. This initiative will also enhance the credibility of the proposed model and introduce transparency in the decision-making process regarding the queuing policy applicable in the port areas.