Defining the Power and Energy Demands from Ships at Anchorage for Offshore Power Supply Solutions

Abstract

The maritime industry, following several conventions, regulations, and initiatives, is trying to adapt and limit its GHG and other local polluting emissions. A crucial aspect of these decarbonisation efforts is the provision of electric energy to vessels, either in port or stationed at anchorages. The latter prevents the option of receiving shore power to support their needs without operating their generators. Research and innovation efforts are attempting to fill this gap through several technology options. This study focuses on the systematic definition of the power and energy demands at anchorage to drive the design of such solutions, focusing on the materialisation of a modular and scalable power barge serving as an offshore power supply solution. Data for various ports and ship types were collected and analysed properly to extract significant insights. Results formulated baselines, per ship type and port, to be used for selecting power modules’ configurations and meeting these demands. This study, considering the lack of industry data regarding power demand, relies on existing studies, guidelines, and other literature to define power demand, which in turn introduces a great deal of uncertainty. Thus, a detailed statistical analysis was employed, together with probability modelling, in order to limit the uncertainty and provide a baseline for the power and energy demands to be verified by future studies capitalising on the accumulation of actual industry data.

1. Introduction

International shipping activities rely on fossil fuels [1], consuming approximately 300 million tons of fuel annually while representing more than 1000 million tons of carbon dioxide (CO₂). The world fleet is powered mainly by marine diesel engines for both propulsion and auxiliary loads fuelled by marine fuels. Heavy fuel oil (HFO) is the dominant fuel in international shipping, powering main ship engines. However, it has been significantly reduced by approximately 7% in the past few years, while the consumption of marine diesel oil (MDO), which is mainly used for powering auxiliary engines, has grown by approximately 6% [2]. This clearly indicates the dominance of residual fuels (e.g., HFO) but also the significant role of distillate marine diesel oils (e.g., MDOs) and yet justifies the continuously growing contribution of shipping to global anthropogenic emissions, as it is expected to increase from today’s 90% to as much as 130% of 2008 emissions by 2050, based on certain long-term economic and energy projections, in the “business as usual” scenario. However, in recent years, policies and legislation, mainly focusing on environmental sustainability, have pushed international shipping toward the process of decarbonisation. Regulatory bodies are pressing on the maritime world by adopting ambitious targets and by introducing several initiatives that will facilitate the transition to a sustainable maritime future. From the wider setting at the international level, with the Paris Agreement [3] triggering the initial International Maritime Organization (IMO) greenhouse gas (GHG) strategy [4], to the European Union’s (EU’s) more specific and recent ambitious regulations for renewable and low-carbon fuels in maritime transport [5], as well as the alternative fuels infrastructure [6], the regulatory framework has been enhanced, setting the scene for the industry. Thus, the industry is trying to adapt and limit its GHG and other local polluting emissions by introducing onboard energy efficiency technologies and measures while also exploring alternatives to conventional fossil fuels. These are the two main considerations for maritime decarbonisation, with solutions referring to either or both of them.

One important aspect of the decarbonisation efforts of the shipping industry is the provision of electric energy to vessels at port, which is already a mature technology. Shore-side Electricity (SSE), also known as cold ironing or Onshore Power Supply (OPS), refers to an electrical power supply system that replaces onboard generated power while a ship remains docked at the port. Thus, the vessel’s electricity generation-related emissions are eliminated locally and reduced globally, depending on the energy mix of the shore supply [7]. In ports, ship emissions are of increased concern, especially for sulphur oxides (SO_X), nitrous oxides (NO_X), and particulate matter (PM), for the local population. In this respect, SSE systems additionally aim to reduce the negative impact of ship emissions on the health of the population living in the port surroundings while also addressing noise and vibrations produced by auxiliary engines when vessels are docked, causing a nuisance and greatly affecting residents [8]. As for GHG emissions, it has been proven that emissions coming from ships’ activities close to a port area are responsible for approximately 5% of the total emissions deriving from navigation activities [9]; thus, SSE plays a part in wider shipping decarbonisation. Apart from cold ironing services, such systems also have the potential to support the electrification of the fleet—either in hybrid or full-electric formats—by charging onboard battery systems in the future [10], extending their decarbonisation capacities.

However, besides ports and the SSE systems to support vessels within, vessels are often stationed at anchorages, which prevents the option of receiving shore power to support their needs without operating their generators. On average and depending on ship size, ships spend up to 9% of their time waiting at anchorage emitting CO₂ while running auxiliary engines and boilers, which consume up to 10–15% of fuel. This anchorage time can be either due to traffic within ports and terminals, thus waiting to enter, or a variety of services or operations usually taking place at anchorage (e.g., bunkering, waste disposal, inspections, crew change, etc.) [11]. Research and innovation efforts are attempting to fill this gap through several options such as power barges and the combination of underwater cables and buoys. Each of them has its own challenges, but all of them are still lacking technology readiness at various levels as well as incentives to drive the needed investments.

To deal with this, extensive research and innovation efforts are required to explore the various alternatives for either of these technologies and assess their performance, both in environmental and financial terms. As part of these efforts, the ongoing research and innovation EU-funded project BlueBARGE is focusing on the design, development, and demonstration of a power barge solution, considering various power supply modules such as containerised batteries and fuel cells. The project has identified so far three main concept use cases. Supplying power to vessels when anchored is considered the main use case. Two secondary concept use cases are providing power to ships berthed in ports where SSE systems are not available due to limitations in the national grid or as an extension to an existing SSE system and to support the national electricity grid when a disruptive event causes a power outage or in the case of isolated inhabited areas (e.g., remote islands). In this context, and to carefully define the requirements for the design of such complex systems, it is necessary to determine the target power output and the required energy capacity to support the aforementioned concept use cases. The primary concern is of course to identify the needs of anchored ships, as the main use case, to also be applicable beyond the solution of power barges, which are a preferable option mainly due to their operational flexibility. Thus, due to the limited knowledge of the power needs at anchorages, this study aims to answer this question.

The following section provides the background, including a short literature review on available resources, as well as a definition of the methodology and data used for the study, while the next one presents in detail the study’s results, followed by a thorough discussion of the results and next steps of this work.

2. Materials and Methods

This section presents the background for the technology solutions under consideration after a thorough literature review, including the regulatory framework and similar systems and studies, followed by a detailed definition of the methodology and data that were used for the estimation and analysis of the power and energy demands.

2.1. Background

In more detail, regarding the regulatory framework for maritime decarbonisation, following the Paris Agreement, the International Maritime Organisation (IMO) established its greenhouse gas (GHG) strategy back in 2018 [4], updated in 2023 [12], while also developing several initiatives to limit carbon intensity. These include the already existing, updated, or newly introduced Energy Efficiency Design and Existing ship Indexes (EEDI, EEXI) addressing the technical efficiency of ships, the Carbon Intensity Indicator (CII) rating scheme addressing ships’ operational efficiency, and the enhanced Ship Energy Management Plan (SEEMP) addressing the management system for compliance with the aforementioned indicators [13]. In addition, the ambitious targets set by the EU are gradually introducing new policies and initiatives in terms of ship emissions. The EU’s “Fit for 55” package [14] puts together a set of measures and actions aiming at reducing net greenhouse gas emissions by at least 55% by 2030, facilitating the goal of climate neutrality by 2050, compared to 1990 levels. Important aspects of the “Fit for 55” are the extension of the EU Emission Trading System (ETS) [15] to the maritime industry, which puts a price on emissions, at least at an EU level, and FuelEU Maritime Regulation [5], starting this year, which aims to increase the demand for and consistent use of renewable and low-carbon fuels and reduce GHG emissions from the maritime sector while ensuring the smooth operation of maritime traffic and avoiding distortions in the internal market. Furthermore, the alternative fuels infrastructure Regulation (2023/1804) [6] was introduced mainly to ensure minimum infrastructure to support the required uptake of alternative fuels across all transport modes—and therefore ships—in all EU Member States and to ensure full interoperability of the infrastructure. Several proposals have also started to come forward at an international level on what type of regulations and measures should be applied on a worldwide scale to ensure that shipping achieves the strategy’s ambitions and continuous compliance with a set of rules and regulations following the EU’s example.

In technical terms, a variety of solutions work as enablers for meeting the aforementioned ambitions in maritime decarbonisation, including onboard efficiency measures and technology as well as alternatives to conventional fuels for either supporting ship propulsion or electricity needs. Relevant to this study’s focus, SSE systems and the solutions for offshore power supply are considered within the scope of both. On the one hand, energy efficiency may be greatly enhanced from the perspective of powering ships directly with electricity for the hotel loads, while on the other hand, when considering hybrid or full electric configurations in vessels, such solutions support the alternatives to conventional fossil fuels.

Both SSE and offshore supply solutions are out in the market. Of course, SSE infrastructures are not only continuously multiplying, covering more and more port facilities [16], but are also becoming mandatory for certain ship types, such as passenger and container ships, followed by the requirement from 2035 and onwards that vessels generally are no longer allowed to pollute when berthed at ports [5]. However, SSE infrastructure still seems not to have been developed enough to support the needs of power supply in ports. This may be due to a weak electricity grid; hence, there are many ports and harbours that require additional power and energy capacity to satisfy the needs of the vessels at ports. These excess power requirements of onshore energy providers can be achieved by additional energy supplied to the port electric grid or, ideally, by local energy provided by renewable sources. The expansions and upgrading of such infrastructures require a series of investments. In such an economic study, it is critical to ensure that this expansion is profitable in the long term, while considering obstacles related to short-term power fluctuations as well as the excess renewable energy that would not be stored or the low renewable energy production [17]. Currently, apart from challenges with respect to the proper dimensioning of an SSE system [18], the interest of this research is focused on its proper and more efficient implementation and manners to overcome challenges related to the economic costs and environmental benefits, emission reduction measures, and energy management aspects [19,20]. On the other hand, the focus of the current study, offshore supply solutions, has seen limited uptake so far due to issues such as the unavailability of technologies, technical and operational safety aspects, the lack of standardisation, as well as stakeholder resistance due to high uncertainty and a lack of financial incentives. As mentioned, such solutions, already explored to a certain extent, are either mobile floating units such as the power barge concept or combinations of underwater cabling and buoy solutions. The former option has already been seen through the concept of power ships, initially developed during World War II, usually defined as a special-purpose ship equipped with an onboard power plant to operate as a power generation resource [21]. Today’s concepts and solutions for power units cover just the energy demands of shore grids in occasions of a lack of infrastructure such as large power plants [22,23] or the energy needs of vessels [24,25,26]. The latter option, the combination of underwater cabling and buoys, is rather challenging from multiple aspects. However, there are efforts to establish such infrastructure [27].

Defining the vessels’ power and energy demand in either case has been quite a challenge for the industry. Regarding SSE infrastructure, the maturity and technological readiness are significantly high, with relevant standards and guidance already available [28,29], following several projects, including support actions and feasibility studies, such as EALING and ELECTRIPORT. On the other hand, efforts for offshore power supply for ships at anchorage are scarce. Of course, some studies have explored—among others—the loads at anchorage for ships [2] in order to estimate their consumption and generated emissions. Studies have also focused on the issue of long waiting times at anchorage, exploring other alternatives such as the just-in-time approach to resolve such issues [11,30]. Furthermore, combining these two perspectives, other studies, in order to calculate the effectiveness of a port call optimisation solution, have indirectly referred to consumption—and therefore energy demands—at anchorage [31,32]. From other perspectives, the issue at hand has been indirectly part of the comparative analysis by Łebkowski and Wnorowski of the energy consumption by conventional and anchor-based dynamic positioning of ships [33], while more focused studies have tried to explore consumption and GHG emissions at anchorage, specifically for bulk carriers in the Southern Gulf Islands [34], touching upon the scope of the current study regarding energy and power demands at anchorages. It is evident that there is no solid and holistic view on the matter, which justifies the importance of such a study.

2.2. Methodology

The methodology to approach the power and energy demand of vessels at anchorage, which is the main use case, was based on the analysis of port (anchorage) call data, including waiting time at anchorage for ships in various ports and the data for their power demands based on previous studies and assumptions. Thus, the main parts of the methodology are the definition of the data schema, the selection of ports for the analysis, the data collection and cleaning, the energy demand calculations, and finally the statistical analysis and modelling attempts. The final step serves as a way to present a more detailed view of the results while trying to set a literature-based baseline for the power and energy demands to be verified by future studies capitalising on the accumulation of actual industry data. The methodology is presented in detail in Figure 1.

The goal of the study is to approximate in as much detail as possible, based on available resources, the power and energy demands of anchored vessels, limiting the uncertainty around any claims while also setting the basis for future studies capitalising on actual industry data.

The primary data for the study, aside from the information derived from the 4th IMO GHG study and the EMSA SSE guidance as well as the load factors extracted from the literature, as illustrated in Figure 1, include port data, ship data, and port call data. The last is the most critical dataset, as it provides the waiting time at anchorage per port call (anchorage), specifically the duration of the vessel’s stay at anchorage. It should be noted that data are extracted solely for anchored vessels and not for vessels in port. In addition to this information, it is essential to have the IMO number of the vessel involved in the call, as it connects the analysis with the vessel’s characteristics, such as ship type, total auxiliary engine (AE) power, and port name, linking the analysis with the list of ports and their attributes, including location and categorisation within the Trans-European Transport Network (TEN-T) [35]—namely whether the port is a core port of the TEN-T network, a comprehensive port, or not part of the network. A detailed description of the data is provided in the next sub-section, while the data schema, which aids in understanding the data used for the present work, is illustrated in Figure 2.

The list of ports is the first step in starting the data collection, since port call data can only be collected based on identified ports. Thus, a representative list of ports was compiled with 37 ports, based on several criteria, supported by expert judgement to formulate the final list. These include the following:

• Focus on core ports of the TEN-T network: the final list includes 28 core ports, 4 comprehensive ports, and 5 ports not part of the TEN-T network, as the countries are not part of the EU.
• Ensure variation of EU or neighbouring counties (at least 20 counties): a total of 22 countries are represented, most of which are European (20), apart from Egypt and Turkey, which are considered EU adjacent.
• Ensure representation of sea basins, i.e., Atlantic, Mediterranean, Black Sea, North Sea, Baltic Sea: the final list includes 4 Atlantic ports, 18 Mediterranean ports, 1 Black Sea port, 9 North Sea ports, and 5 Baltic Sea ports.

On the other hand, for the power demand, which was also necessary to calculate the energy demand, actual data regarding the loading operation profile of the installed auxiliary system for ships could provide an optimal estimation. Thus, this study first attempted to accumulate data from shipping companies. However, these data were scarce and could not feed the entire analysis. Therefore, reference values for the power demand were obtained from the literature. These reference values mostly included specifications on the rated power output of auxiliary systems for ships, provided by both IMO and European Maritime Safety Agency (EMSA) studies, but also assumed loading factors for auxiliary engines for various ship types, as provided by multiple studies. More specifically, in the fourth IMO GHG study [2], there are detailed specifications for the power output of auxiliary systems at anchorage ${\overline{P}}_{d-IMO}$ (mean power demand), including ship type, size, and mode of operation. These were used in the study’s context to estimate emissions from auxiliary systems. The fourth IMO GHG study, apart from capitalising on the third IMO GHG study [36], used additional resources for establishing the auxiliary engine power output specifications (e.g., Starcrest’s vessel boarding program reports from 2012 to 2018, auxiliary engine and boiler fuel consumption data provided by ClassNK, auxiliary engine fuel consumption provided by continuous monitoring data at the time, and professional judgment from experts on the field [2]). In addition, EMSA [29], in its guidelines for SSE design and development, provides analytic data for power demand, mean ${\overline{P}}_{d-EMSA}$, and peak ${\check{P}}_{d-EMSA}$; these may be significantly different from the needs at anchorage but can be used as a reference for comparison. Finally, various publications [37,38,39,40,41,42,43] have tried to define load factors (LF) for the auxiliary engine systems of ships, which in combination with data regarding the total installed auxiliary power onboard (number of auxiliary engines $n_{AE}$ and AE power $P_{AE}$), can provide an approximation of the power demand $P_{d-LF}$ as their product. The latter is shown through Equation (1), which describes a basic calculation with respect to the LF approach, in order to be available for a more descriptive picture of the power demand of ships at anchorage, and to calculate, as a third approach, the energy demand of ships at anchorage.

$${\overline{P}}_{d-LF}=f_{LF}·n_{AE}·P_{AE}$$

(1)

where

$P_{d-LF}$ is the mean power demand based on the LF approach, in KW;

$f_{LF}$ is the load factor, per vessel type, based on literature references;

$n_{AE}$ is the number of the AEs onboard (excluding the emergency);

$P_{AE}$ is the rated power of the onboard AEs, in KW.

Furthermore, in order to calculate the energy demand per vessel per port (anchorage) call $E_{d-i}$ in KWh, the basic equation for the calculations is presented below as Equation (2), which includes the main two parameters, the mean power ${\overline{P}}_{d-i}$ in kW and time at anchorage $t$ in hours (h).

$$E_{d-i}={\overline{P}}_{d-i}·t$$

(2)

where

$E_{d-i}$ is the energy demand in kWh, based on the $i$ approach, namely the 4th IMO GHG study, EMSA SSE guidance, or LF approach;

${\overline{P}}_{d-i}$ is the mean power demand in kW, based on the $i$ approach;

$t$ is the time of stay at anchorage in hours (h).

The produced dataset from these calculations is then presented as aggregate results per ship type and port in order to provide a picture of the demand for all three approaches. Then, an analysis of the aggregate results concerning the more demanding vessels and ports provides more insights into possible use cases.

As no actual data from shipping companies could be used due to scarcity, this study, in addition to the above, employed methods for a more detailed view and elaboration on the results. These methods include exploring the probability distributions provided by the results, per ship type and size, and the attempts at probability modelling of the power and energy demand, which can be compared and verified by any future work capitalising on actual demand data. This study employs histograms to represent the distribution, complemented by the Kernel Density Estimate (KDE) curve, which provides a continuous smooth representation of the data distribution using the Gaussian function to aid in the visualisation of underlying trends that might be difficult to see with histograms alone. Furthermore, to model the probabilities, the following frequently used distributions were considered [44]: Normal Distribution (Gaussian), Exponential Distribution, Weibull minimum distribution, Gamma Distribution, and lognormal distribution. For each candidate distribution, the Maximum Likelihood Estimation (MLE) was used to estimate the distribution parameters. The best-fitting distribution was selected based on goodness-of-fit tests; this study employed the Kolmogorov–Smirnov (KS) test to evaluate the fit of each distribution [44]. The KS test statistic is defined in the following Equation (3):

$$D=\underset{x}{\sup }\left|F_{data}\left(x\right)-F_{model}\left(x\right)\right|$$

(3)

where

$F_{data}\left(x\right)$ is the empirical cumulative distribution function (ECDF) of the selected observed data,

$F_{model}\left(x\right)$ is the cumulative distribution function (CDF) of each of the fitted models from the above distribution models.

Based on the above selected method, the p-value from the KS test was used to assess the goodness of fit. A high p-value $(p<0.05)$ indicates that the observed data are not significantly different from the theoretical distribution, signifying a good fit, while a low p-value $(p>0.05)$ indicates that the observed data deviate significantly from the fitted distribution, signifying a poor fit [45]. Appropriate data handling, such as outlier detection and elimination, as well as data transformation (e.g., log transformation, normalisation) [45], were considered for appropriately plotting the data and modelling the distributions.

2.3. Data

In more detail, the data specifications, the collection process, and processing are presented, concluding with some key figures describing the final datasets. Apart from the data from the 4th IMO GHG study and EMSA SSE guidance, as well as the resources referring to the load factors approach, the datasets refer to the (i) port specifications, (ii) port call and time at anchorage, and (iii) ship specifications, as presented previously.

The port specifications data were the first to be defined, as they were used—for the selected list of ports—for accumulating port call data to be used for extracting the time at anchorage. These are presented in

Table 1. Port dataset specifications.

Attribute	Description	Type
Port name	The name of the port, as commonly referred to	String
Country	The country that the port belongs to	String
Latitude	The latitude of the port location	Geometry
Longitude	The longitude of the port location	Geometry
TEN-T	The categorisation of the port with respect to the TEN-T network	String

. Furthermore, the reference period for collecting port call data was the entire year 2023 in order to include seasonal variations while also covering a large period to collect a representative set of data. The specifications for the collected data are shown in

Table 2. Port call dataset specifications.

Attribute	Description	Type
IMO number	The 9-digit unique identification code for a vessel	Integer
Port name	The name of the port of call	String
Port zone	The specific zone/location of the port of call that the ship called	String
Arrival date	The date of arrival at the port of call	Date
Arrival time	The time of arrival at the port of call	Time
Hours in port	The time in hours that the ship remained at the port of call or at anchorage	Integer

. Consequently, based on the identified ships from the port call data, the ship specification data were extracted by including only the information useful for the analysis, as shown in the specifications of

Table 3. Ship dataset specifications.

Attribute	Description	Type
IMO number	The 9-digit unique identification code for a vessel	Integer
Ship name	The name of the ship, as given by its company	String
Ship type	The ship type/category, as defined by the relevant databases	String
Total ME power	The total power in kW (MCR) of the main engine	Float
AE power	The power in kW of the AE (not the emergency generator)	Float
Number of AEs	The number of auxiliary engines	Integer

.

Based on the aforementioned data specifications, a specific database was used for data exploration and extraction, considering all three datasets, driving also the definition of the data specifications to a large extent. The database was the Maritime Sea-web Online Ship Register [46]. The resulting datasets included 37 ports, a total of 302,538 records of port calls (anchorage calls), and a total of 23,157 ships.

The selected ports are shown in Figure 3, along with the average time at anchorage extracted from the port call dataset. The list was produced by industry experts considering a wide geographical coverage across Europe and adjacent regions as well as different port types in terms of traffic. The ports are visualised in Figure 4, along with metrics regarding port calls at their anchorages in the given timeframe and the average time at anchorage. Another useful visualisation is shown in Figure 5, which presents the most demanding ports considering anchorage calls and time at anchorage. A logarithmic scale was used to better present the wide range of ports and their corresponding characteristics.

The ship dataset was the most demanding in terms of processing since it needed additional effort to ensure consistency regarding ship types. The StatCode 5 coding system was utilised for a more standardised approach in the ship types’ naming and categorisation [47]. The fleet composition regarding ship types is shown in Figure 6.

Further, to provide results for the power and energy demands, as described in the methodology of the study, the below parameters were extracted or calculated (for the LF approach) for use in the calculation of energy demand:

• Mean power demand, per IMO GHG study, ${\overline{P}}_{d-IMO}$
• Mean power demand, per EMSA SSE guidance, ${\overline{P}}_{d-EMSA}$
• Peak power demand, per EMSA SSE guidance, ${\check{P}}_{d-EMSA}$
• Mean power demand, per load factors assumptions, ${\overline{P}}_{d-LF}$

Once again, it is noted that the numbers extracted from the EMSA SSE guidance do not correspond to anchorage power demands, as they refer to berthed vessels and are used as a reference for the design of SSE systems. On the other hand, the power demand calculated from the LF approach, even though this refers to hotel loads without specifying whether they may correspond to anchorage or berth loads, can be more accurate since the calculation involves actual vessel information regarding the installed AE power onboard. In contrast, the EMSA SSE guidance, and even the IMO GHG study approach, set specific values per ship type and size, which may lead to erroneous results. Given the above, as analysed below, the calculation based on the EMSA SSE guidance approach was used only as a reference, focusing on the numbers of the peak power demand as the only currently available. The calculations from the LF approach offer a more accurate view of the power demand, as they originate from actual ship information, while they serve as a reference for comparison with the energy demand results from using the IMO GHG study numbers.

During the analysis, the data were filtered appropriately to provide useful and reasonable results. The filtering included the exclusion of several ship types for which there was no clear information regarding their power demand, i.e., non-merchant ships, other dry cargo ships, other and non-ship structures, various missing values (e.g., if the total installed auxiliary power was missing), as well as cases of increased anchorage time. The last item is crucial, as it excludes cases where the anchorage time exceeded the threshold of 96 h (4 days). This was used to explore reasonable waiting at anchorage cases but not others (outliers), referring to abnormal situations in terms of operations (e.g., ship detained, ship under maintenance for long periods, ship waiting for commercial dispute reasons, etc.). Such filtering allowed for the accuracy of the results related to energy and power demand.

Upon a more detailed examination, based on the plot in Figure 5, the most prominent ports in terms of power and/or energy demand were identified as Istanbul, Rotterdam, Antwerp, and Marseille, indicating the importance of analysing the fleet calling at their anchorages. Consequently, Figure 7 offers an overview of the fleet composition for these ports. These ports were later utilised to present some results that necessitated a more detailed perspective. For this purpose, the main ship types were considered, which also represented the largest portion of the sample. These included cruise ships, Ro-Pax, container ships, bulk carriers, tankers, Ro-Ro, and general cargo ships. Oil tankers, chemical tankers, liquefied gas tankers, and other liquid tankers (as per Figure 6) were grouped together into a single category, adhering to the classification from StatCode 5 [47]. Offshore support vessels and fishing vessels were excluded from the visualisations due to their anchorage nature, which differs significantly from the rest of the commercial fleet.

3. Results

This section, based on the presented methodology and data, shows the results of the performed analysis regarding the power and energy demands. Both the power demand, in MW, and the calculated energy demand, in MWh, provide a clear picture of the anchorage demands for an offshore power supply solution such as a power barge. The subsections below refer to an overview of the results as aggregates, as well as results analysis per port and ship type, while also presenting a more detailed statistical analysis aiming for probability distribution modelling.

3.1. Aggregate Results per Ship Type and Port

In

Table 4. Power and energy demand results per ship type.

Ship Type	Total Port Calls (no.)	Average Time (h) at Anchorage	Average Power Demand (MW)			Average Energy Demand (MWh)
			IMO	EMSA	LF	IMO	EMSA	LF
Bulk carriers	35,694	16.63	0.28	0.61	0.42	4.68	10.14	6.78
Chemical tankers	50,996	13.4	0.4	7.98	1.35	6.19	115.35	20.79
Containers	23,231	14.07	1.06	2.45	1.2	14.28	33.1	15.39
Cruise	1378	7.63	5.15	7.07	7.71	53.28	64.23	77.54
Dredging vessels	1433	7.52	0.32	1.35	0.33	2.41	1.29	2.7
Fishing	7160	5.98	0.2	0.52	0.38	1.2	3.17	2.12
General cargo	47,972	16.33	0.17	1.51	0.17	2.7	24.58	2.69
Liquified gas tankers	9327	17.24	5.57	7.75	2.69	99.58	136.29	46.29
Offshore support vessels	17,448	4.86	0.32	1.03	1.31	1.56	5.2	22.78
Oil tankers	49,476	14.37	0.41	5.45	2.24	6.46	84.37	28.34
Refrigerated cargo	728	12.19	2.4	1.5	1.36	26.46	18.28	14.2
Ro-Pax	2744	6.86	0.35	2.19	1.1	3.15	17.42	9.89
Ro-Ro	4220	9.77	0.86	4.89	1.05	8.19	48.19	9.08

, the results per ship type are shown, both in terms of power and energy, including the statistics for the port calls and the time at anchorage. The results refer to all three approaches for calculating demand. As partly explained, the IMO GHG study approach is considered as the baseline approach, as it specifically refers to power output, namely the power demand, at anchorage. The EMSA SSE guidance approach was used only as a reference, focusing on the numbers of the peak power demand as the only currently available. The LF approach results were used for comparison purposes and to describe in more detail the power demand distribution, since it is based on actual ship information (i.e., the onboard installed AE power). In general, multiple differences among the approaches were observed, both in terms of actual numbers and descending order.

From the presented results, it is clear that the data corresponding to the EMSA SSE guidance approach are quite high. This is, of course, due to the different scopes, as the power output employed corresponds to berthed ships and not anchored, while these numbers are destined for the design of SSE systems, which may lead to over-dimensioning when it comes to offshore power supply solutions. Thus, as mentioned previously, these results were not considered for approximating the power and energy demand. However, as the other two approaches did not include peak power demands, the original numbers for power demand from the EMSA SSE guidance were needed as a reference or even target value. On the other hand, most of the results corresponding to the load factors approach were closer to the baseline results (IMO) than the EMSA approach. It is noted that, for the LF approach, there was a lack of data regarding the total installed auxiliary power for 33.6% (7788 from the total of 23,157 ships) of the examined cases. However, there was still enough to be used for comparison purposes and a more detailed view on the power demand. Their use for comparison can provide answers for the large deviations between the other two approaches. In more detail, for certain ship types, i.e., chemical tankers and oil tankers, the results according to the EMSA approach were way off the baseline estimation from the IMO GHG study approach, revealing significant differences. There, the load factors approach provided an estimation closer to the IMO approach, demonstrating that it probably is a more realistic one. Finally, by isolating the main ship types for the IMO approach, it is evident that cruise ships and container ships were the most demanding ones, aligning with expert judgement on the matter. This is depicted below in Figure 8, which summarises the relation between power and energy demand for the main ship types (and larger fleet size). Again, a logarithmic scale was used to provide a more comprehensive view.

In the same sense,

Table 5. Power and energy demand results per port.

Ship Type	Total Port Calls (no.)	Average Time (h) at Anchorage	Average Power Demand (MW)			Average Energy Demand (MWh)
			IMO	EMSA	LF	IMO	EMSA	LF
Algeciras	14,025	11.14	0.62	3.96	1.02	8.42	50.97	14.64
Antwerp	37,625	4.38	0.61	4.89	1.01	5.17	25.64	6.72
Barcelona	2129	16.83	0.82	4.4	1.19	13.17	92.5	20.12
Bilbao	2474	10.08	0.43	2.6	1.18	5.96	40.11	19.82
Bremerhaven	1610	1.33	0.38	2.72	0.38	0.49	3.16	0.42
Ceuta	1720	11.09	0.78	4.44	1.28	8.71	53.49	15.62
Constanza	5777	19.96	0.41	2.31	0.61	7.81	49.03	12.41
Copenhagen	317	15.19	0.24	2.88	0.37	3.4	44.84	6.39
Felixstowe	309	20.62	0.19	2.06	0.25	3.39	41.32	4.36
Gdansk	2808	18.97	0.82	3.03	0.9	17.32	61.09	17.95
Genoa	2250	11.82	0.95	3.78	1.34	9.51	51.96	18.06
Gibraltar	14,924	9.6	0.78	3.98	1.13	8.09	41.39	12.18
Gothenburg	7945	10.78	0.65	4.44	0.84	8.99	57.49	13.05
Hamburg	2598	22.59	1.06	4.27	1.62	26.92	104.59	39.33
Ibiza	61	11.02	0.69	2.94	0.82	3.1	32.39	10.4
Istanbul	72,630	17.32	0.34	2.67	0.65	5.8	47.14	11.69
Kiel	632	16.32	0.63	3.17	0.58	10.5	56.03	9.64
Klaipeda	585	20.43	0.66	3.31	0.71	12.25	73.45	15.43
Koper	1134	19.49	0.6	2.64	0.81	12.91	58.51	18.76
Las Palmas	5405	11.25	0.52	3.52	0.98	6.73	41.17	13.3
Le Havre	6197	17.4	0.7	4.14	1.15	14.9	87.92	26.29
Limassol	782	14.39	0.49	2.35	0.58	6.82	36.61	8.07
Lisbon	1462	11.51	0.52	2.53	0.64	6.6	31.19	8.1
London	2730	17.97	0.79	3.98	1.32	15.69	86.53	30.14
Marsaxlokk	13,289	14.32	0.63	4.6	1.22	10	77.75	20.89
Marseille	2459	25.78	1.18	5.96	1.22	35.91	173.41	33.87
Piraeus	8484	15.02	0.71	4.22	1.21	9.86	66.7	18.87
Port Said	12,989	14.12	0.99	4.45	1.58	14.37	64.28	22.58
Rotterdam	14,847	22.56	0.8	5.45	1.34	19.54	135.23	33.5
Sines	1403	17.28	1.17	5.1	1.79	22.53	109.65	32.47
Tallin	1576	10.17	0.67	4.7	1.38	6.94	54.15	16.95
Thira	2133	10.67	2.73	4.29	5.55	31.2	47.24	58.86
Trieste	931	20.03	0.64	4.35	1.36	13.3	94.76	31.29
Valencia	2047	11.96	0.95	3.24	1.32	9.41	25.29	12.34
Valletta	2927	6.72	0.42	2.99	1.1	2.73	39.28	8.12
Venice	1160	22.41	1.07	3.93	1.01	28.76	95.91	24.12
Wilhelmshaven	1950	3.92	0.31	1.23	0.16	1.39	4.75	0.8

illustrates the aggregate results regarding the considered ports for all three approaches, including statistics for the port calls and time at anchorage. In general, the approaches agreed on the set of most demanding ports, which are Marseille, Sines, Hamburg, Rotterdam, Venice, and Hamburg. Of course, there were differences in the descending order and actual numbers; thus, the comparison, based on the above reasoning, serves only for trend confirmation. In addition, the port of Thira (Santorini) in Greece was at the top of the list as the most demanding port for the IMO and load factors approaches, most likely due to the high demand for cruise ships at its anchorage. However, this was not observed with the EMSA approach, mainly due to the high demand allocated to other ship types, statistically ranking other ports above Thira.

The energy demand, as a product of time at anchorage and power demand, is equally affected by both parameters, while the distribution of the fleet among ship types is also crucial. In cases such as Thira, the result was influenced mainly by the power demand of cruise ships that are of increased number in its anchorage, while in cases such as Marseille, the long waiting times were the most influencing factor. Figure 9 provides a clear picture of the demand for ports, per the IMO approach, both in terms of power demand and energy demand. Once again, the plot is on a logarithmic scale to include the wide range of demands in the ports of scope. The figure shows the most demanding ports, including Thira.

Finally, in a holistic view of the results, Figure 10 and Figure 11 represent the overall distribution of power and energy demand, per the IMO GHG study approach, accompanied by the relevant percentage of total anchorage calls. From these figures, a designer could extract some target values for an offshore power supply solution. For instance, a 3 MW power output and a 35 MWh energy capacity solution would cover 90% to 95% of the demand. However, this cannot be easily considered as the correct way forward. A more detailed analysis of the ports and use cases to be supported would be more appropriate. In the next subsections, this is done for selected ports and ship types in order to show the variations among them.

3.2. Focused Results per Port and per Ship Type

Following the above reasoning, this subsection is dedicated to presenting the specific power and energy demand results for selected ports and ship types.

The selected ports are the ones that stand out in the previously shown Figure 5, namely, Rotterdam, Istanbul, Marseille, and Antwerp, with a variation in average anchorage time and number of port calls. Thus, Figure 12 presents the power demand distribution, while Figure 13 and Figure 14 depict the energy demand distribution for the aforementioned ports per the IMO GHG study approach.

For the power demand, it is evident that there is an increased concentration of up to 1 MW. However, the variations are large enough to affect the requirements for an offshore power supply solution if considering specific ship types in conjunction with the energy demand coverage. Indicatively, the aforementioned figures of 3 MW and 35 MWh seem to cover the demands in all ports except Marseille, mainly due to the composition of the fleet in its anchorage and the long waiting times. Furthermore, a more careful analysis regarding the ship types to be supported reveals additional concerns. This is showcased in the following subsection.

Following the analysis per port, the selected ship types are container ships, cruise ships, and Ro-Pax based on the regulatory trend of them being supported by shore electricity (at least for berths [5]). The first two are also the most demanding ships among the main ship types, as shown in Figure 8. Thus, Figure 15 and Figure 16 show the calculated energy demand distribution for each ship type per the IMO GHG study approach. The power demand distribution is not presented, as it is directly extracted from the IMO GHG study [2] for the selected ship types, which ranges from 0.91 to 1.95 MW for container ships, from 0.45 to 11.5 MW for cruise ships, and from 0.105 to 1.95 for Ro-Pax.

3.3. Statistical Analysis and Modelling

After presenting the aggregate results above, this subsection, based on the methodology discussed, provides a more detailed view of the results through comprehensive statistical analysis and probability distribution modelling. The analysis focuses on the identified container ships, as their larger sample size compared to the other two ship types previously explored offers greater analytical possibilities. Additionally, as demonstrated in Figure 16 regarding energy demand, modelling the distribution of cruise and Ro-Pax ships presents challenges due to the two peaks observed for cruise ships and the concentration of most Ro-Pax ships at lower energy values. The container fleet is divided into eight size bins, as outlined below, since demands vary significantly with ship size. For the energy demand analysis, the IMO approach was applied as the baseline method, while the LF approach was utilised solely for the power demand analysis, as it provides a detailed view of the distribution. The analysis followed the previously outlined methodology, including histograms for the distributions with various bin widths, a KDE curve to illustrate data trends, and fitting known probability distributions. However, on multiple occasions, additional techniques were required to appropriately plot the data, create the curves, and conduct the distribution fitting process. These techniques included normalisation for generating the KDE curve and log transformation of the data before applying the algorithm for distribution fitting. All calculations and plotting were carried out using Python 3.11.8 and necessary libraries such as pandas 1.5.3 for data processing, numpy 1.24.0 for numerical computations, matplotlib.pyplot 3.6.3 for plotting, seaborn 0.11.2 for enhancing plot aesthetics and scipy.stats 1.9.3 for distribution fitting.

The results presented below are intriguing both in terms of their original distribution and the attempts to fit distributions. For the former, Figure 17 illustrates the data histogram and the KDE curve of the power demand for container ships ranging from 1000 to 2999 TEU based on the LF approach, while Figure 18 displays the histograms and KDE curves of the energy demand for all size bins within the entire sample of container ships according to the IMO approach. In terms of power demand, it is evident that there is a trend indicating the data following a distribution close to normal, lognormal, or Weibull distributions, with a significant concentration of values from approximately 0.2 MW to 1.3 MW. The analysis of other size bins was not conducted, as the sample was limited compared to the one presented here. As explained, the LF approach provided a limited sample of results for all ship types due to a lack of data. Regarding energy demand, it is clear that the distributions were similar, with high concentrations near zero; this is a consequence of the limited duration of stay at anchorages, which in turn restricts demand and can be observed across various ship sizes. Naturally, with larger ship sizes, the histograms extend even further in terms of energy, ranging from approximately 42–43 MWh for containers from 0 to 999 TEU to as much as 180 MWh for containers exceeding 20,000 TEU. It is worth noting that the latter case represents extremely high demands, which may not be suitable for any offshore power supply solution.

Beyond the histograms and the KDE curves, the results of the distribution fitting attempts are presented. As discussed, the power demand distribution was close to several known probability distributions. After performing the respective calculations, the Weibull minimum distribution seemed to be the best fit, as presented in Figure 19, with the following parameters: shape (κ): 1.7748, location (t₀): 0.1347, scale (λ): 0.6746. However, the p-value test provided a value of 0.609%, which is significantly lower than the 5% threshold, indicating a good fit. Thus, this probability distribution fit cannot be considered valid, even if it is the best fit based on the MLE algorithm. For the energy demand, the distribution fitting of the container ships with a capacity of 1000 to 2999 (larger sample) is shown in Figure 20. In order to provide a valid distribution fitting, the data were log transformed. The best fit was a lognormal distribution with the following parameters: shape (σ): 0.0074, location (μ): −167.22, scale (exp(μ)): 169.16. In contrast with the previous distribution fitting, this one provided a p-value of 6.39%, which is higher than the 5% threshold, thus signalling a good fit.

4. Discussion

From the results above, the power and energy demands at anchorage are explored in detail using the three available approaches, as explained in the methodology of this study. The most acceptable values are the ones arising from the IMO GHG study approach when it comes to the calculated energy demand, as it is the only approach in which its reference values for power demand refer to the anchorage state. As already mentioned, the values derived from the reference numbers included in the EMSA SSE guidance are significantly higher, as they refer to berthed vessels, which typically have increased needs due to their primarily cargo operations. Furthermore, these reference numbers are supposed to be used for the design of SSE systems, and thus can be a bit excessive, probably due to the necessity of over-dimensioning SSE systems to cover power fluctuations at berth. However, EMSA’s peak demands are the only available source for the offshore power supply case since the other two approaches refer only to mean power outputs. As for the load factors approach, it generated the most detailed view of the mean power demand despite the smaller data sample (33.6% when compared to the other two), and it refers to hotel loads. Additionally, it is used for comparison purposes to provide answers regarding abnormal differences between the other two approaches (e.g., for tankers), apart from their systematically observed higher numbers for most ship types (e.g., cruise ships).

Based on the above discussion points and the presented results, a safe estimation that can be considered as a baseline for the design of offshore power supply solutions, targeting anchorage demands, would be to consider both the IMO and LF approaches for the mean power demand, the IMO approach for the energy demand, and the EMSA approach—as the only one currently available—for the peak power demand. Of course, if considered, caution is required so as not to result in over-dimensioning an offshore power supply solution. These are presented in

Table 6. Baseline values for the power and energy demand per ship type.

Ship Type	Mean Power Demand (MW)	Peak Power Demand (MW) 1	Mean Energy Demand (MWh)
Bulk carriers	0.35	0.7–2.8	4.68
Chemical tankers	1.35	9.0–20.0	6.19
Container ships	0.87	2.0–6.0	14.28
Cruise ships	6.43	4.5–20.0	53.28
General cargo	0.17	3.0–5.0	2.70
Oil tankers	1.33	6.0–10.0	6.46
Ro-Pax	0.73	4.0–6.5	3.15
Ro-Ro	0.96	-	8.19

¹ Values from the EMSA SSE guidance [29], which is the only available source for peak demands.

for the main ship types as well as in Figure 8. For the mean power demand, an average value is presented by combining the two preferred approaches.

Of course, the above baseline values represent the mean values for the entire fleet under consideration by ship type. Thus, they do not cover the most demanding vessels’ energy requirements, either due to power requirements or waiting time. To explore coverage over a specific fleet, whether for multiple or single types of ships, a more focused analysis is required. For instance, if considering a container ship, as per Figure 16 and Table 6, an assumed 3 MW–35 MWh solution would cover 100% of the mean power demand and approximately 90% of the fleet, but it would be lacking the peak demand as assumed by EMSA SSE guidance. In the case of cruise ships, the coverage, for such a solution, is much lower. Thus, in the design of an offshore power supply solution, such as a power barge, an iterative process would be necessary. A first design approach could use the aforementioned baseline values to produce several alternatives, ranging from lower to higher power outputs and energy capacities. Then, the assessment of these alternatives would require a more focused analysis of the fleet of scope to be supplied (e.g., at the port of interest) in order to understand the capabilities of the solution, including actual coverage. Ultimately, refining the initial design based on the target capabilities would yield an optimal design. Of course, other criteria need to be considered, such as financial, integration, and safety criteria, throughout the design process.

Further, a detailed statistical analysis and modelling of the probability distribution by considering specific ship types and sizes could be a more appropriate way of defining the power and energy demand requirements for the design of such a solution. However, from an academic perspective, this study lacks sufficient actual industry data from ships regarding their power demand, so it would be more accurate to use the probability distributions, which were proven to be a good fit, as in the presented case.

Future work in this scope would be to explore the demands for other use cases of such systems, such as the ones presented for the power barge solution. These include the provision of power to berthed ships, for ports that are not obligated to have SSE infrastructure, for the extension of the SSE infrastructure of large ports, or for the provision of power to the local grid in cases where a disruptive event causes a power outage or in the case of isolated inhabited areas (e.g., remote islands). Of course, as an extension of this study, an accumulation of enough industry data regarding power demand at anchorage from actual anchorage cases would allow for a more solid analysis of the presented results, either for validating them or refining them, as a complete picture of the demands from ship data would not be possible due to scarcity. More specifically, the way forward would be to compare the industry data for actual ship power demand and then energy demand, with the probability distributions proven to be a good fit. In this way, such probability distributions, if verified, could be widely used for the design of such solutions. Both aspects are explored within the ongoing EU-funded BlueBARGE project.