Ship emissions reduction via slow steaming without disrupting the logistical supply chain: A case study of the port of Felixstowe

ABSTRACT In this era of global supply chains and just-in-time logistics, the speed with which shipping companies can deliver goods to their customers is an economic competitive advantage. The growth of international trade means more ships and more voyages which will contribute to the increasing global emissions inventory. Nonetheless, developments in ship environmental performance have not matched the increase in shipping activities. Several environmental protection and emissions reduction measures have been identified but are yet to be implemented globally due to technological gaps and their capital-intensiveness. This paper focused on addressing the challenges of applying “slow steaming” as a measure of cutting emissions. It demonstrated that by applying analytical models, the optimal berthing capacity can be defined and that by implementing the model, ships can employ slow steaming to reduce emissions. The paper also proposed a collaboration between port and ship owners/companies where real-time information is exchanged to facilitate access and swift cargo operations at ports so that ships can employ speed reduction techniques during a sail.


Background
The International Maritime Organization (IMO) has set for 2030 an agenda for sustainable shipping.Critical among issues of concern for achieving sustainable shipping are Ballast Water Management (BWM), measures to reduce Greenhouse Gas (GHG) emissions, efforts to reduce marine litter, protect the polar regions, and reduce the sulphur content of ship fuels.According to a UNCTAD report released in early 2019, total world ship fleet had a 2.6% capacity growth year on year.The report further stated that global seaborne trade will continue to rise at a compound annual growth rate of 3.4% from 2019 to 2024 (UNCTAD, 2019).This requires additional ship tonnage to meet the increasing demand and to replace ageing fleet.The operationalisation of a large fleet size and its growing characteristics is that shipping's contributions to global emissions will continue to increase and thus leaves its sustainability efforts in further jeopardy.
The International Maritime Organization's (IMO) fourth greenhouse gas (GHG) study painted a concerning picture of future emissions trends.According to their projections, GHG emissions were expected to climb from approximately 90% of 2008 levels in 2018 to a staggering 90-130% increase by 2050 (IMO, 2020).This alarming trajectory is graphically illustrated in Figure 1.These projections stand in stark contrast to the ambitious IMO GHG emissions reduction strategy established in 2018, which aimed to slash GHG emissions by a minimum of 50% by 2050 compared to the 2008 baseline.Such a divergence between goals and reality has raised concerns and ignited a sense of urgency among stakeholders and regulators within the maritime industry.A previous study conducted by Cames et al. (2015) offered an even bleaker outlook, estimating that if no significant technological advancements or policy interventions were implemented, CO2 emissions from shipping could account for a troubling 17% of global CO2 emissions by 2050.These disconcerting findings have further heightened the sense of urgency within the industry, prompting a call for immediate solutions.
In response to this mounting crisis, researchers and experts have identified two broad spectra of measures to mitigate emissions: technical and operational.As Psaraftis and Kontovas (2009) noted, one significant operational approach that has gained traction in the pursuit of emissions reduction is the concept of "slow steaming."This strategy, as argued by Lindstad et al. in Lindstad et al. (2011), involves deliberately reducing vessel speeds, leading to a decrease in fuel consumption.Since fuel consumption is directly correlated with emissions, this reduction in speed translates into a tangible reduction in emissions -a promising step toward addressing the looming environmental challenges in the maritime industry.
Maritime infrastructures ashore struggling to keep up with rising volumes of international trade lead to port congestion.Port congestion became rife during the early 2000s as the global container demand was growing at 10-15% per year (Lind et al., 2018).It remains that port congestion is an issue of great concern given that it causes delays, inefficiencies, and increasing GHG emissions at port cities.A 2017 study of European ports revealed that ships spent more than 71,000 hours waiting to access berths for cargo operations, which in monetary terms could amass to an industry value loss of about EUR 100 million per year (Lind et al., 2018).This certainly doesn't bode for an industry such as the maritime industry where time is most important and capital resources are required for more productive ventures.It also means that the industry still has a long way to go in terms of achieving overall sustainability.
Slow steaming ultimately increases the shipping time from port to port, thereby having ships tend to arrive at ports later than they would if they sailed at fast speed.Additionally, many ports around the globe operate by the principle of first-in-first-out (FIFO), based on which cargo operations on ships are executed.This motivates ships and ship operators to hurry to ports to be in a queue for the earliest possible access but end up spending material times waiting, sometimes running into days.Coupled with waiting times, slow steaming has the potential of causing massive logistical supply chain issues.From the supply chain point of view, port congestion and the extra time spent at sea makes slow steaming as an emissions reduction measure quite counterproductive.

Problem statement
The practice of slow steaming has emerged as a costeffective and environmentally friendly method for reducing emissions in the maritime industry (Andersson, 2008;Cariou, 2011;Lee et al., 2015;Lindstad et al., 2012Lindstad et al., , 2013;;Psaraftis & Kontovas, 2010).Nevertheless, a critical challenge arises from the operational constraints imposed on ships that are obligated to adhere to strict time and location schedules to meet customer demands.The primary issue associated with the implementation of slow steaming pertains to port congestion and its impact on ship schedules.
Ships, particularly liner vessels, are obligated to adhere to predetermined route schedules within specified time frames to fulfil customer expectations.The integration of slow steaming into these schedules presents a potential disruption to the logistical supply chain that ships operate within.This challenge is further compounded by the contemporary issues of port congestion, which can lead to unexpected delays at ports.As global port volumes continue to rise, many ports are struggling with capacity constraints, resulting in extended waiting times for vessels.When coupled with the practice of slow steaming, these delays have far-reaching consequences, including an overall increase in shipping duration, escalated inventory costs, and customer dissatisfaction.
Hence, the main problem addressed in this academic paper is the intricate interplay between slow steaming as an emissions reduction strategy and the logistical challenges posed by port congestion and stringent ship schedules.This problem necessitates an analysis of strategies to mitigate the negative impact of slow steaming on shipping operations, including, revising scheduling mechanisms, and exploring innovative solutions to harmonize environmental goals with customer satisfaction and economic efficiency in the maritime industry.

Research aim and objectives
The aim of this paper is to model the ship port interactions such that, ship emissions can be reduced without negatively impacting the logistical supply chain.The specific objectives are; • To identify ship idle times during port calls.
• To define the optimal berthing capacity and arrival rate that effectively eliminates ship idle times during port calls.

Scope and organization
The scope of the research is limited to finding a pragmatic solution to the increasing emissions from maritime transportation, by addressing the challenges of slow steaming as an operational measure.The rest of this study is organized as follows.The literature review is discussed in Section 2. The methodology and model are described in Section 3. The mathematical model for solving the queue problem is introduced in section 4. The results are discussed in Section 5. Finally, conclusions, limitations, and future work are outlined in the last section.

Composition of GHG emissions from shipping
Greenhouse gas (GHG) emissions encompass a diverse array of gases that exert varying influences on the climate, including carbon dioxide (CO2), methane (CH4), nitrous oxide (N2O), and hydrofluorocarbons (HFCs).The combustion of hydrocarbon-based fuels inevitably yields CO2 emissions.Concurrently, the carbon content of these fuels also gives rise to byproducts such as soot, carbon monoxide, and hydrocarbon compounds, which, in comparison to the carbon contributing to CO2, are considered negligible (Styhre & Winnes, 2019).Consequently, it is conventionally assumed that all carbon constituents in fuels ultimately manifest as CO2 emissions.Presently, CO2 is recognized as the most environmentally detrimental GHG arising from shipping activities, garnering the lion's share of efforts aimed at emissions control.Nonetheless, the maritime industry also harbours concerns regarding other GHGs, notably methane, hydrofluorocarbons (GL Reynolds Environmental Sustainability Consultants, 2019), and nitrous oxide.While these gases possess emission factors that are modest and often deemed inconsequential, accounting for less than 2% of the global warming potential within port emission inventories attributable to ship emissions (Styhre & Winnes, 2019), it is imperative to underscore the significance of methane emissions.
Methane emissions, stemming from the incomplete combustion of methane-rich fuels, commonly referred to as methane slip, warrant particular attention due to their potent heat-trapping capabilities.Over a 20-year time horizon, methane has been reported to possess a heat-trapping potential approximately 86 times greater than an equivalent quantity of CO2 (ICCT, 2020).Consequently, although methane emissions from shipping constitute a relatively minor fraction of GHG emissions, they are characterized by a rapidly expanding influence.
Of concern, the fourth International Maritime Organization (IMO) GHG study has unearthed a disconcerting surge in methane emissions, recording a staggering 150% increase between 2012 and 2018 (IMO, 2020).This unprecedented upswing in methane emissions, as elucidated by the International Council on Clean Transportation (ICCT, 2020), is primarily attributed to the proliferation of LNG-fueled vessels.

Ship emissions
Emissions originating from maritime transportation are a complex concern encompassing two principal settings: emissions on the high seas and emissions within port areas.It is noteworthy that the predominant share of emissions occurs during high-speed voyages at sea, given the substantial fuel consumption associated with this phase of maritime operations (Styhre & Winnes, 2019).Nevertheless, ship emissions within port vicinities present a distinctive and pressing issue.Notably, ship emissions rank as the single largest source of emissions within port environments, surpassing those generated by the ports' internal operations by a factor of 10 (Habibi & Rehmatulla, 2009).The classification of emissions occurring during the ship-port interface phase includes emissions accrued during a ship's waiting period for berthing, the actual berthing process, and departure.The severity of this issue is exacerbated by extended Turnaround Times (TATs) and further by waiting times.On a global scale, ship CO2 emissions within port areas collectively amount to approximately 18 million tonnes, a figure projected to rise fourfold by the year 2050 (Alamoush et al., 2020).
Research by Moon and Woo (2014) has demonstrated the direct correlation between reduced ship TATs and a corresponding decline in CO2 emissions, highlighting a potential reduction of up to 37%.Conversely, an annual increase of 30% in TATs, as evidenced by Alamoush et al. (2020), precipitates a corresponding surge of 30.7% in CO2 emissions.Notably, Styhre et al. (2017) conducted a comprehensive study across four ports situated on four different continents, revealing that ships were responsible for emissions ranging from 8% to an alarming 88% of the total GHG emissions.Container ships, in particular, emerged as the most prominent emitters of GHGs, an observation substantiated by Psaraftis and Kontovas (2010) and Styhre and Winnes (2019).Furthermore, these vessels, owing to the escalating scale of container trade, find themselves disproportionately affected by longer TATs, thereby exerting additional pressure on existing terminal infrastructure.Qi and Song (2012) undertook a study to design optimal vessel schedules that reduce both fuel costs and emissions.Their approach considered uncertain port times at each port of call and liner schedule frequency requirements.Using simulation-based stochastic approximation methods for cases with 100% service level constraints, the research identified the most challenging segment of shipping routes concerning achieving a perfect service level and minimizing emissions as the shortest leg.Furthermore, the study demonstrated the convexity and continuous differentiability of the objective function, elucidating that the optimal schedule could be obtained through the solution of a nonlinear optimization problem.Bouman et al. (2017) conducted an extensive literature review to identify measures with high potential for CO2 reduction in the maritime sector.The findings emphasized that no single measure alone can achieve substantial sector-wide reductions.Instead, they concluded that significant emissions reductions exceeding 75% are attainable by rapidly adopting and combining multiple measures by 2050.Nonetheless, they acknowledged that certain individual measures, such as the utilization of biofuels or speed optimization, can yield significant reduction potentials if systematically and robustly implemented.Lee et al. (2015) proposed a comprehensive model to quantify the interplay between shipping time, bunker cost, and delivery reliability.Using industry data to characterize port times and vessel fuel consumption statistically, the study underscored the potential benefits of slow steaming with speed flexibility.This approach was found to reduce fuel consumption, emissions, and mitigate the impact of random port time variations.However, it should be noted that this study primarily considered still water conditions, with some exceptions.

Empirical review
Psaraftis and Kontovas (2010) investigated the consequences of speed reduction on ship CO2 emissions and other ship attributes.The study found that a one-knot reduction in speed led to a 25-minute increase in voyage time while concurrently reducing emissions and fuel costs by 10%.However, Psaraftis and Kontovas (2010) highlighted the potential drawbacks of speed reductions, including extended transit times and revenue loss, necessitating the addition of ships to maintain demand.
These studies collectively contribute valuable insights into the multifaceted challenges of maritime emissions reduction and optimization strategies.

Methodology
The research was guided by mixed methods of research.We reviewed past relevant works to establish the premises upon which we base our optimization model.We also employed queuing model to determine the optimal berthing capacity at a port of call.

Model development
Determining the berthing capacity per period is key to determining the optimal speed.The model considers the time stamps of arrivals and berthing times of ships at the port.Basic parameters used for the calculation of adequate indices of the queuing system function are the intensity of arrival flow (λ) and intensity of servicing

Input data
Secondary data from ship tracking databases were the main source of data.Marine Traffic proved a very useful database for collecting data to determine ship idle time.Additionally, AIS data from "Fleetmon" database was used to calculate the optimal berthing capacity.The data span a 1-month period from 28/09/2019 to 1/11/2019, with two hundred and thirty-seven (237) port calls of container ships.
The case considered for the dissertation is the port of Felixstowe.The choice of this port is because it is Britain's biggest and busiest container port, welcoming approximately 3000 ships per year and handling over 4 million TEUs, it also connects to about 700 ports worldwide.The port is also a state-of-the-art technology port with many berths and relatively large number of ship calls.

Ship-port idle time
As shown in Figure 2, the median time of ship's waiting per week is illustrated.Captioned as "time at anchorage," the chart depicts the median ship idle times of port calls at the port of Felixstowe per week.
Table 1 is a tabular representation and processing of data from Figure 2. The Manipulation of the data reveals the total and average of ships' waiting time at the port within the period.The summation of the weekly median idle times of container ship calls at the port collectively amount to 165 hours -an average of about 18 hours in total per day at the port.The data sourced were not presented per ship, i.e., per ship call.Therefore, the average waiting time per each ship port call could not be determined with these data.However, a prior comprehensive study conducted of Europe's' Tier 1 and Tier 2 ports 2017 revealed that 38,425 container ships that called at such ports collectively spent 71,202 hours waiting to enter the ports -an average of 1.85 hours per port call (Lind et al., 2018).

Berthing as a queuing system
The berthing process represents a queuing system with the following structure: customers are ships forming a waiting line at anchorage (depending on the situation) in order to be served (berthed) in the berthing area and after the service (cargo operations) has been completed after a certain amount of time, they exit the system.According to the characteristics of the intensity of ships' arrival flow and the service time mentioned in section 3.2, non-uniformity in the use of the berthing capacity can be deducted, if the number of ships arriving into the berthing area is greater than the number that the existing capacity can serve during a time causing waiting lines; and in reverse, berthing area capacity is greater than the ships' arrivals, but then berthing capacities are not fully utilised.Figure 3 below presents average port call data of the port for the period 28/09/2019 -1/11/2019 secured from "Fleetmon" database.Table 2 is a processed form of the data See appendix-A for full datasheet.
To determine the optimal berthing capacity, there needs to be a definition of the optimal berthing capacity, taking into consideration all the factors influencing the operations of the berthing area.The optimal berthing capacity is that which provides a satisfactory level of service to the ships, along with good economic and environmental benefits.For the selected system of incoming ships into the berthing area, the arrival rate λ (lambda) represents the average number of ships arriving into the berthing area in a unit of time (hours, day, etc.) under observation; service rate μ (mule) also represents the average number of ships served within a unit of time under observation.
The intensity of servicing is the reciprocal value of the average time required for the ship to berth and the length of service time (μ = 1/t).Service time t is expressed in the  number of units of time required for the operations of one ship at berth.This data is used to express the throughput of the port.The relation between the arrival rate and the intensity of the flow of the servicing of the ships at berths is the degree of load of the berth ρ=λ/μ, i.e., the coefficient of utilization of the berthing area; ρs =λ/Sμ Where S represents the number of berths in the terminal or port.
If all the berths are occupied, an arriving ship waits in line until served.The berthing area is defined as a multiple-server queuing system with finite capacity, and an infinite queuing line.This is because the length of the queue line can extend as far into the high seas.
For a berthing system, the following conclusions can be drawn; • The berthing area has a fixed capacity, • Considering that there are many berths of different characteristics in the berthing area, we can talk about a multi-server queuing system, • Arrivals of ships into berthing area are distributed according to a Poisson distribution, • Service time is also distributed according to an exponential distribution, • Servicing of ships is done according to the FIFO method (first-come-first-served), • There is unlimited space for ships waiting to enter the berthing area, so therefore there is a waiting line with infinite length.
The berthing area capacity is expressed in the number of berth spaces available for ships.The capacity includes both statistical and dynamic capacities in the berthing area.The statistical capacity is expressed in the number of ships which can be berthed at the same time, whereas the dynamic capacity refers to the total number of ships counted in a unit of time.There are 13 container berths at the port of Felixstowe as shown in Table 1.This is the statistical berthing area capacity, which means that 13 ships of varying sizes can dock at the berthing area at a given time all other factors held constant; S = 13.Therefore, the observed servicing process is classified as a queuing problem with a definite number of ships that can berth at a time, M/M/13.Any ship calling the port after this, i.e., the 14 th ship, will be required to be probably the first in the waiting line because there would be no space at the berthing area.The dynamic capacity is calculated by considering the number of ships entering the berthing area in a unit of time, the average berthing time, and the total working time of the berthing area.Therefore, the number of required berthing spaces can be calculated using the formula below.

ΣZM = λ × t0/T
where: ΣZM -average number of occupied berthing spaces during the day, λ -average number of ships arriving during the day, t -average duration of berthing for one ship (hours), T -total daily work time for berthing (hours).Therefore, the dynamic capacity, i.e., the average number of occupied berthing spaces within a day = λ × t0/T = 6.97 × 19.2/24 = 5.576, approximately 6 berthing spaces per day.
The ships' arrival rate: using the average number of ships calling at the port daily in October 2019; λ = 237/ 34 = 6.97 or 7 ships per day rounded up (with 24 hours working time).
Service rate: μ is obtained from the calculation of the average servicing time, i.e., cargo operations time; 4558/237 = 19.2 from the data of the month October 2019, hence 19.2 hours per ship.
From the datasheet, it is evident that more ships call at the port during peak hours, but the dissertation will stick to the average across the month to make the model simple.Table 3 above lists all container berths at the port of Felixstowe operational at the time of preparing to write this paper.Total berths are 13.Based on the characteristics of the berthing system, we arrive at ρ = λ/μ*13 = 7/19.2× 13 = 35/1248 or 2.8%.This gives a waiting time of: expected waiting time in queue, W = Lq/λ, where Lq, length of queue = 0 given that the model entails immediate access to berth for ships arriving at the berthing area.Therefore, the expected waiting time W is zero.
The question arises whether the existence of waiting lines at the port is due to insufficient number of berthing spaces or the high intensity of inter-arrival times.The reality, however, is that the port cannot provide enough berths to serve each arriving ship on time with the current FIFO system.This is not only financially impossible but also uneconomical for many reasons, for example, peak and slow periods as well as non-uniformity of ship calls.Therefore, the berthing system represented as a queuing problem simulates an ideal situation where the queues are eliminated.It demonstrates that with accurate parameters and appropriate mathematical models, the optimal berthing capacity can be defined.And that by applying the laid-out model, the waiting times can be eliminated by ensuring that ships that call at the port at any time arrive at such a time that their arrival intensity ≈ berthing area capacity.At zero waiting time (Lq and W = 0), ships will have direct access to the berthing area upon arrival at port.
Figure 4 shows the relationship between the variables.The effective implementation of this model necessitates a departure from the prevailing berth allocation system, FIFO approach.It is imperative for ports to embrace an alternative policy for berth allocation grounded in a prearrival booking system.Such a transition is anticipated to discourage vessels from fast steaming when heading towards ports without assured berthing access, consequently leading to a reduction in waiting times.Notably, the utilization of pre-booked port slots is observed to substantially diminish the waiting duration for vessels compared to the conventional first-come-first-served system, as affirmed by Kontovas and Psaraftis (2011).
Correspondingly, Steenken et al. (2004) have advanced a comparable viewpoint, emphasizing the capacity of ports to mitigate Turnaround Times (TATs) for container ships through the strategic optimization of operations via the Berth Allocation Problem (BAP).BAP involves the proactive planning of ships' berthing times and allocation of quay space in advance.By adopting this proactive approach, vessels would be alleviated from the need to accelerate their approach to ports and endure extended waiting periods before securing access to berths for cargo operations.Consequently, such a shift is anticipated to incentivize vessels to adopt slower speeds.Consequently,

Conclusion
Ship port calls have large oscillations during the hour, day, month, and year.This makes it difficult to predetermine the number of ships arriving into or being served per period.However, for the purposes of planning, it is useful to determine whether there is a certain regularity in the arrival and service rate of the ships in the berthing area.
Given that ship calls and length of time of their servicing can be taken as probabilistic variables and the empirical distribution of these variables then manipulated with adequate theoretical distributions, it becomes possible to apply analytical approaches.Hence, the use of formulas set out in queue modelling to eliminate idle time during port call.To achieve the second objective, the dissertation applied queue modelling technique to historical data on berthing at the port of Felixstowe to define the optimal berthing capacity in order to eliminate the queues and idle time.While the queuing model appeared successful (eliminating the queues) in the case examined, it is worth noting that many other mathematical models and tools can equally achieve the same result and even provide better clarity, for example simulation and BAP.The dissertation also argued for a switch from the current FIFO principle in berthing arrangements to a pre-booking system where ships register their place with the next port of call so that berth space can be allocated prior to arrival.The optimal berthing capacity along with the pre-booking principle will allow for ships to have direct access to berths at each port call when implemented together in a robust manner.
Importantly, this multifaceted approach holds the potential to offer economic incentives to shipowners, as the need for reduced fuel consumption aligns with their financial interests.Furthermore, this approach presents environmental advantages by contributing to a reduction in emissions, thereby promoting eco-friendly practices within the maritime sector.In essence, this paradigm shift in berth allocation policies stands to benefit shipowners economically while concurrently advancing environmentally responsible practices within the industry.

Recommendations
To reach the desirable outcome of this dissertation, the following recommendations can be implemented: • The global maritime community needs to be informed about the collective benefit of collaborating to ensuring sustainability in shipping.
• The IMO should adopt the strategy as a shortterm GHG measure as part of its agenda to decarbonise the shipping industry.• The IMO should speed up developing policy and regulations based on this measure by 2023 and hold member countries to implement it.• Tax incentives should be made available for actors that take initial steps to implement the collaborative framework, so that others will be motivated to follow.

Limitations
The biggest limitation encountered during the research was the inability to secure data directly from companies.
Secondly, the dissertation intended to use the same set of data to determine the idle time and find the optimal berthing capacity using the queue modelling.This could not happen because the two data sets secured both lacked elements necessary to satisfy both objectives two and three.
Finally, the data secured for identifying the idle time were not comprehensive.It was already in a processed form, and only the median idle times per week were plotted on the graph that was used.This prevented the dissertation finding the average ship idle time per ship call.

Future work
A study to be conducted based on defining the optimal berthing capacity along keeping an inventory of emissions for the period under study.This will enable the study to compare the savings on emissions by the measure to current situation without the measure.Whether the measure is best for reducing emissions depend on a study that includes emissions inventory.
Application of digital twins to the model, representing the real scenario with a virtual model simultaneously and a seamless transmission of data.
A further area of interest lies with the application of search-based heuristics.These could be used to optimise a defined set of parameters that define the design or operation of a port to minimise time at berth, and idle time.
(μ) which in practical examples are determined based on data obtained by statistical observation or evaluation depending on the objective and task of the research.The M/M/ S queuing model employed in this paper follows a Poisson distribution of arrival (ƛ) and service rate (µ) with multiple servers.Service time of each of the servers (S) is exponentially distributed with parameter.Additionally, service times are independent and independent of the arrival times.Here: Utilization rate, ρ = λ/μ* S Waiting time in the queue, W = Lq λ Length of queue, Wq = p μ 1À p ð Þ

Figure 2 .
Figure 2. Ships' weekly average waiting time at the port of Felixstowe.Source: Marine Traffic (2020)

Figure 3 .
Figure 3. Graphical representation of ship hours per day at the berthing area of port of Felixstowe for the period October 2019.Source: Author' construction

Figure 4 .
Figure 4. Relationship between the optimal berthing capacity and emissions reduction.

Table 1 .
Weekly averages of ship waiting time at port of Felixstowe from week 29 to 37 of 2020.

Table 2 .
Number of ships arrived into the port of Felixstowe berthing area in October 2019.

Table 3 .
Container berth capacities at the port of Felixstowe.