Optimal Alternative Fuel Selection for Dual-Fuel Ships Under FuelEU Maritime Regulations: Environmental and Economic Assessment

Abstract

To address greenhouse gas (GHG) emissions from the maritime sector, the European Union (EU) has introduced the FuelEU Maritime regulation to incentivize ships to adopt diversified compliance pathways and energy solutions. This study aims to determine the optimal alternative fuel configurations for dual-fuel ships of different types under environmental, economic, and regulatory constraints. An integrated environmental and cost assessment model from a well-to-wake (WtW) perspective to systematically evaluate the environmental benefits and economic feasibility of fossil-based, bio-based, and renewable electricity-based alternative fuels applied in dual-fuel ships. By incorporating the PROMETHEE II method within a multi-criteria decision analysis (MCDA) framework, together with the CRITIC objective weighting method, the study enables a robust ranking of alternative fuel configurations across three key dimensions: environmental performance, cost feasibility, and regulatory compliance. The results indicate that, regardless of ship type, the very low sulfur fuel oil (VLSFO) + marine gas oil (MGO) and VLSFO + methanol (MEOH) combinations fail to meet the GHG intensity targets for 2025–2050. Only the VLSFO + electrolytic liquid hydrogen (E-LH₂) and VLSFO + electrolytic ammonia (E-NH₃) configurations are compliant. Although e-fuels incur the highest annual costs, the EU compliance penalty associated with fossil fuels increases exponentially. In contrast, e-fuels retain long-term cost advantages, ultimately driving a sector-wide transition toward e-fuel-dominated energy structures by 2050. Their superior environmental performance and regulatory compatibility emerge as the core drivers of the maritime energy transition.

1. Introduction

Greenhouse gas (GHG) emissions from the maritime sector account for nearly 3% of global total emissions, making it a critical issue in the context of international climate governance [1,2]. In response, the International Maritime Organization (IMO) released the “2023 IMO Strategy on Reduction of GHG Emissions from Ships”, aiming to regulate GHG emissions from shipping through an increasingly complex and dynamically evolving regulatory framework [3,4]. To align with the IMO’s strategic objectives, regional bodies such as the European Union (EU) have introduced complementary regulatory initiatives. The EU has legally committed to achieving climate neutrality by 2050 and, in July 2021, launched the “Fit for 55” legislative package to support this target [5,6]. As part of this package, the maritime sector has been incorporated into the EU Emissions Trading System (ETS) starting in 2024. Furthermore, the FuelEU Maritime regulation is scheduled to take effect on 1 January 2025, with the aim of accelerating the decarbonization of maritime transport [4].

The FuelEU Maritime regulation mandates that all commercial ships with a gross tonnage above 5000—regardless of flag—engaged in the transport of passengers or cargo must progressively reduce their GHG intensity when calling at EU ports. The regulation sets a clearly defined reduction trajectory: a 2% decrease starting in 2025, reaching an 80% reduction target by 2050. FuelEU Maritime adopts a technology-neutral and performance-based compliance framework [7]. It utilizes a well-to-wake (WtW) life-cycle approach to account for emissions of carbon dioxide (CO₂), methane (CH₄), and nitrous oxide (N₂O), thereby enabling ships to comply through diverse energy system configurations and operational profiles. The scope of application varies by voyage type: for intra-EU voyages between ports of EU Member States, the regulation applies to the total energy used throughout the voyage; for voyages involving outermost EU regions or between EU and non-EU ports, it applies to 50% of the energy used during the journey [8].

With the implementation of the FuelEU Maritime regulation, ships are under dual pressure to improve energy efficiency and reduce emissions, positioning the adoption of alternative fuels as a key strategy to meet compliance challenges. Complementing alternative fuels is the deployment of innovative marine propulsion systems, among which dual-fuel engines have gained particular attention for their enhanced flexibility and adaptability to a variety of fuel types [9]. In the context of marine alternative fuels, liquefied natural gas (LNG) exhibits lower carbon intensity compared to conventional fossil fuels and is regarded as a promising short-term option. However, fossil-based LNG is unlikely to meet the stringent requirements of deep decarbonization in maritime transport [10,11]. In contrast, bio-LNG derived from biomass offers renewable and environmentally friendly advantages, making it a greener alternative [10]. Methanol is another widely recognized low-carbon fuel, notable for its ease of storage and transportation, and can be produced from both natural gas and biomass sources [11]. Non-fossil-based methanol is considered a strong medium- to long-term candidate for maritime decarbonization [12]. Hydrotreated vegetable oil (HVO), currently the most compatible direct substitute for conventional fuels, remains costly and is still under development [11]. Hydrogen-based e-fuels, such as electrolytic liquid hydrogen (E-LH₂) and ammonia (E-NH₃) from renewable electricity, are regarded as long-term solutions due to their zero-carbon characteristics [13]. Nevertheless, their large-scale application in the shipping industry faces complex technological, logistical, and infrastructural challenges. The production of e-fuels is heavily reliant on high-cost water electrolysis technologies. For liquid hydrogen, large-scale storage and transport are hindered by high energy consumption for liquefaction and boil-off losses. In the case of liquid ammonia, the development of efficient catalysts and the minimization of energy losses during cracking remain critical obstacles [14,15]. High infrastructure costs and the need for large-scale deployment are shared challenges for both E-LH₂ and E-NH₃ [14,16]. Moreover, one of the most significant barriers to widespread adoption is the high production cost of e-fuels, which is substantially greater than that of conventional fuels [11].

A substantial body of research has emerged evaluating the environmental and economic implications of marine alternative fuels. Wang et al. conducted a comprehensive analysis of the properties of LNG, hydrogen, methanol, ammonia, and biofuels, and developed a life-cycle assessment (LCA) framework for marine applications. They reviewed the existing LCA literature with a focus on environmental impacts and economic viability [10]. Biu et al. proposed a life-cycle cost (LCC) framework to evaluate the full LCA performance of innovative marine dual-fuel engines. Their findings indicate that, under baseline scenarios, dual-fuel engines outperform conventional diesel engines in terms of cost-effectiveness [17]. Lee et al. further conducted an LCA of nine alternative fuel production pathways and integrated carbon pricing analysis to assess their economic performance. The results identified biomass-based Fischer–Tropsch diesel (FT-diesel), e-methanol, and E-NH₃ as the most environmentally favorable options [11]. In a comparative study of 22 alternative fuel pathways, Law et al. highlighted that biofuels offer a balanced trade-off in terms of cost, availability, and technological readiness level (TRL), although challenges remain in mitigating non-GHG pollutants. In contrast, hydrogen and ammonia demonstrated the highest energy consumption and economic cost burdens [18]. These studies have laid a solid foundation for the environmental and economic assessment of alternative marine fuels, establishing relatively systematic LCA and LCC frameworks. However, most of the existing literature treats fuels as the primary object of analysis, lacking an in-depth exploration of ship–fuel compatibility and long-term policy adaptability.

Several scholars have assessed the competitiveness of marine fuels within the context of the European Union’s regulatory framework. Solakivi et al. evaluated the cost implications and long-term feasibility of low-carbon and carbon-neutral marine fuels under the EU’s “Fit for 55” legislative package. Their findings suggest that the most cost-effective regulatory compliance pathway involves a gradual transition from fossil fuels to biofuels, ultimately shifting toward e-fuels [19]. Flodén et al. focused on the emission reduction potential of different ship types within the EU ETS, analyzing associated mitigation measures, modal shifts, and cost variations [20]. Vierth et al. estimated the impact of various EU ETS expansion scenarios and revisions to the Energy Taxation Directive (ETD) on fuel costs for cargo ships calling at Swedish ports, simulating how increased fuel costs would affect transport and logistics costs, port choice, transport modes, and ship selection [2]. To explore the potential decarbonization pathways for various ship types in Sweden’s maritime sector, Trosvik and Brynolf developed a scenario-based modeling tool to assess the effects of EU ETS, FuelEU Maritime, and related policy instruments [21]. Within the context of EU climate policy, numerous studies have conducted scenario-based simulations focusing on carbon pricing mechanisms and regulations such as FuelEU Maritime, providing quantitative support for evaluating policy impacts. Nevertheless, these analyses tend to emphasize fuel pathways or macro-level trends in the maritime sector, while overlooking the heterogeneity in responses across ship types and technical configurations. Moreover, they often fall short in identifying actionable, ship-specific optimal configurations.

Beyond the single-dimensional assessment of environmental and economic feasibility, scholars have increasingly advocated for the incorporation of multi-criteria decision analysis (MCDA) into fuel pathway evaluation, emphasizing the need for integrated consideration of environmental performance, economic viability, and technological maturity [22]. Ren and Liang developed a comprehensive sustainability evaluation framework for marine alternative fuels, proposing a fuzzy group MCDA model to assess methanol, LNG, and hydrogen as candidate fuels [23]. Soltani Motlagh et al. applied a hybrid AHP–TOPSIS approach to analyze shipowners’ and stakeholders’ preferences regarding various emission reduction options [24]. Moshiul et al. introduced TOPSIS as a decision-support tool to holistically evaluate the performance of alternative fuels across environmental, technological, and economic dimensions [25]. Strantzali et al. proposed an MCDA framework based on the PROMETHEE II ranking method, incorporating economic, technological, environmental, and social criteria to facilitate comprehensive evaluation and decision-making for marine alternative fuel adoption [26]. The above-mentioned studies have incorporated MCDA methods into the evaluation of marine fuels, achieving notable progress in balancing environmental, economic, and technological dimensions. However, most existing methods rely heavily on subjective expert weighting and lack a systematic integration and quantitative analysis of the regulatory compliance dimension, which limits their applicability in policy-oriented evaluations.

Despite the growing body of research on alternative marine fuels, existing studies have yet to establish a systematic evaluation framework aligned with the tightening regulatory landscape of the EU. In particular, the current literature lacks comprehensive LCA that account for both the environmental impacts and economic performance of alternative fuels across diverse ship types. Moreover, few studies provide a robust basis for identifying optimal fuel-ship configurations to support practical decision-making. Although works such as [2,19] have examined cost implications under EU carbon policy scenarios, their analyses are limited to single-criterion assessments and omit MCDA approaches. Similarly, while [20,21] consider multiple ship categories, they do not adopt a full WtW life-cycle assessment, thereby limiting their ability to evaluate the compatibility between fuel options and ship operational profiles.

Addressing these gaps, this study makes two key contributions: (1) It develops an integrated WtW life-cycle assessment framework that incorporates both environmental and economic dimensions, specifically tailored to evaluate dual-fuel propulsion systems under the constraints of the FuelEU Maritime regulation. Unlike previous studies that often assess fuel properties in isolation from ship-specific characteristics, this research systematically analyzes the life-cycle GHG and cost feasibility of fossil-based, bio-based, and renewable electricity-based fuels across various ship types. This approach enhances the practical relevance and contextual applicability of the evaluation results; (2) Building upon detailed environmental and economic modeling, this study introduces an integrated MCDA approach combining the CRITIC objective weighting method and the PROMETHEE II ranking method. This robust evaluation framework accounts for environmental performance, cost feasibility, and regulatory compliance, thereby improving the model’s adaptability to tightening carbon intensity targets under EU regulations and strengthening its support for evidence-based decision-making in maritime energy transitions.

2. Method and Data

2.1. Method

This study focuses on dual-fuel ships equipped with alternative fuels. First, an environmental impact assessment model and an annual total cost estimation model are developed from a WtW perspective. Second, based on the annual GHG intensity limits stipulated in the FuelEU Maritime regulation, the compliance surplus and potential penalty costs of each fuel option are calculated. Third, three core quantitative indicators—WtW GHG intensity, annual total cost, and compliance surplus—are selected. The CRITIC objective weighting method is employed to determine the relative importance of each criterion, and the PROMETHEE II MCDA method is used to assess the adaptability of various fuel alternatives across different ship types, ultimately identifying the optimal fuel choice. Figure 1 illustrates the analytical framework of this study, which aims to provide a comprehensive WtW assessment of alternative fuels—encompassing environmental impact, annual economic performance, and regulatory penalty costs—while supporting fuel selection decisions tailored to specific ship types.

2.1.1. Calculation of Fuel Consumption for Dual-Fuel Ships

This study comprehensively considers the operational status of ships under two distinct conditions: at-sea navigation and port berthing. Fuel consumption is estimated based on specific fuel consumption rates. During navigation, the main engine load factor is estimated using the propeller law, as adopted in [27], which expresses the load as the cube of the ratio between average operational speed and design speed, thereby reflecting the load characteristics of different ship types under real-world conditions. Given the weak correlation between auxiliary engine load and ship speed, a fixed load factor of 50% is uniformly applied, in accordance with [27,28]. The fuel usage mode for both main and auxiliary engines also follow the approach in [27]. Specifically, dual-fuel ships are assumed to operate with 95% alternative fuel and 5% very low sulfur fuel oil (VLSFO) as ignition fuel for the main engine, while the auxiliary engines are assumed to use alternative fuels exclusively throughout the voyage, with marine gas oil (MGO) included in the alternative fuel category. Notably, ref. [27,28] also provide a substantial number of model input parameters, which serve as key data sources supporting the computational setup of this study.

Auxiliary engine fuel consumption is initially calculated based on the use of MGO as the reference fuel. When other alternative fuels are employed, the equivalent energy consumption is determined through conversion based on their respective lower calorific values (LCV), ensuring energy-equivalent substitution. According to reference [27], the hourly fuel consumption of the main engine ($F_{ME}$) and the auxiliary engine ($F_{AE}$) is calculated as follows:

$$F_{ME}={\left({\displaystyle \frac{v_{ave}}{v_{design}}}\right)}^3\times P_{ME}^{design}\times SFO{C}_{ME}\times {10}^{-6}$$

(1)

$$F_{AE}=L{F}_{AE}\times P_{AE}^{design}\times SFO{C}_{AE}\times {10}^{-6}$$

(2)

where $v_{ave}$ is the average sailing speed of the ship (kn); $v_{design}$ is the ship’s design speed (kn); $P_{ME}^{design}$ is the design power of the main engine (kW); $SFO{C}_{ME}$ is the specific fuel consumption of the main engine, set at 206 g/kWh [28]; $L{F}_{AE}$ is the auxiliary engine load factor, set to be 0.5 [27,28]; $P_{AE}^{design}$ is the design power of the auxiliary engine (kW); and $SFO{C}_{AE}$ is the specific fuel consumption of the auxiliary engine, set at 221 g/kWh [28]. The annual consumption of alternative fuel for the main engine ($F_{ME}^{ALT}$), the annual consumption of VLSFO for pilot fuel ($F_{ME}^{VLSFO}$), and the annual consumption of alternative fuel for the auxiliary engine ($F_{AE}^{ALT}$) are calculated as follows:

$$F_{ME}^{ALT}={\displaystyle \frac{95\%\times F_{ME}\times LC{V}^{VLSFO}}{LC{V}^{ALT}}}\times T^{sail}$$

(3)

$$F_{ME}^{VLSFO}=5\%\times F_{ME}\times T^{sail}$$

(4)

$$F_{AE}^{ALT}={\displaystyle \frac{F_{AE}\times LC{V}^{MGO}}{LC{V}^{ALT}}}\times \left(T^{berth}+T^{sail}\right)$$

(5)

where $LC{V}^{VLSFO}$ is the lower calorific value of VLSFO (MJ/kg), $LC{V}^{ALT}$ is the lower calorific value of the alternative fuel (MJ/kg), and $LC{V}^{MGO}$ is the lower calorific value of MGO (MJ/kg). $T^{sail}$ and $T^{berth}$ represent the ship’s annual sailing time and berthing time, respectively, measured in hours. The annual total consumption of alternative fuel ($F_{total}^{ALT}$) and the annual total consumption of VLSFO ($F_{total}^{VLSFO}$) are given by

$$F_{total}^{ALT}=F_{ME}^{ALT}+F_{AE}^{ALT}$$

(6)

$$F_{total}^{VLSFO}=F_{ME}^{VLSFO}$$

(7)

2.1.2. Well-to-Wake GHG Intensity Calculation Model

This study calculates the annual average GHG emission intensity of dual-fuel ships based on the full life-cycle fuel pathway, encompassing both the well-to-tank (WtT) and tank-to-wake (TtW) stages. The assessment includes three major GHGs: CO₂, CH₄, and N₂O. Excluding the use of shore power during port stays, this study adaptively modifies the relevant formulas provided in ANNEX I of the FuelEU Maritime regulation, as referenced in [29], to better reflect the operational conditions of dual-fuel ships. Specifically, the calculation of TtW emission intensity (gCO₂eq/gFuel) consists of the following two components. Combustion emissions: These refer to the GHG emissions resulting from the combustion of alternative fuels in both main and auxiliary engines, as well as the VLSFO ignition fuel used in the main engine. These emissions are converted into carbon dioxide equivalents (CO₂eq). Unburned emissions: These include methane and nitrous oxide emissions arising from the use of alternative fuels in main and auxiliary engines due to incomplete combustion or leakage. These emissions are also converted into CO₂eq. The corresponding formulas, along with definitions of each variable and computational details, follow the methodology described in [29].

$$C{O}_{2eqTt{W}_{ALT,ME}}=\left(C_{fC{O}_2,ALT,ME}\times GW{P}_{C{O}_2}\right)+\left(C_{fC{H}_4,ALT,ME}\times GW{P}_{C{H}_4}\right)+\left(C_{f{N}_{2}O,ALT,ME}\times GW{P}_{N_{2}O}\right)$$

(8)

$$C{O}_{2eqTt{W}_{ALT,AE}}=\left(C_{fC{O}_2,ALT,AE}\times GW{P}_{C{O}_2}\right)+\left(C_{fC{H}_4,ALT,AE}\times GW{P}_{C{H}_4}\right)+\left(C_{f{N}_{2}O,ALT,AE}\times GW{P}_{N_{2}O}\right)$$

(9)

$$C{O}_{2eqTt{W}_{VLSFO,ME}}=\left(C_{fC{O}_2,VLSFO,ME}\times GW{P}_{C{O}_2}\right)+\left(C_{fC{H}_4,VLSFO,ME}\times GW{P}_{C{H}_4}\right)+\left(C_{f{N}_{2}O,VLSFO,ME}\times GW{P}_{N_{2}O}\right)$$

(10)

$$C{O}_{2eqTt{W}_{slip,ALT,ME}}=\left(C_{sfC{O}_2,ALT,ME}\times GW{P}_{C{O}_2}\right)+\left(C_{sfC{H}_4,ALT,ME}\times GW{P}_{C{H}_4}\right)+\left(C_{sf{N}_{2}O,ALT,ME}\times GW{P}_{N_{2}O}\right)$$

(11)

$$C{O}_{2eqTt{W}_{slip,ALT,AE}}=\left(C_{sfC{O}_2,ALT,AE}\times GW{P}_{C{O}_2}\right)+\left(C_{sfC{H}_4,ALT,AE}\times GW{P}_{C{H}_4}\right)+\left(C_{sf{N}_{2}O,ALT,AE}\times GW{P}_{N_{2}O}\right)$$

(12)

where $C_{fC{O}_2,ALT,ME}$, $C_{fC{H}_4,ALT,ME}$, $C_{f{N}_{2}O,ALT,ME}$, $C_{fC{O}_2,ALT,AE}$, $C_{fC{H}_4,ALT,AE}$, $C_{f{N}_{2}O,ALT,AE}$, $C_{fC{O}_2,VLSFO,ME}$, $C_{fC{H}_4,VLSFO,ME}$, and $C_{f{N}_{2}O,VLSFO,ME}$ denote the emission factors (gGHG/gFuel) of CO₂, CH₄, and N₂O released during the combustion of alternative fuels in the main engine and auxiliary engine and of VLSFO in the main engine, respectively. Similarly $C_{sfC{O}_2,ALT,ME}$, $C_{sfC{H}_4,ALT,ME}$, $C_{sf{N}_{2}O,ALT,ME}$, $C_{sfC{O}_2,ALT,AE}$, $C_{sfC{H}_4,ALT,AE}$, and $C_{sf{N}_{2}O,ALT,AE}$ related terms represent the emission factors (gGHG/gFuel) of unburned methane and nitrous oxide leakage associated with alternative fuels in the main and auxiliary engines, with $C_{sfC{O}_2,ALT,ME}=C_{sfC{O}_2,ALT,AE}=C_{sf{N}_{2}O,,ALT,ME}=C_{sf{N}_{2}O,ALT,AE}=0$, $C_{sfC{H}_4,ALT,ME}=C_{sfC{H}_4,ALT,AE}=1$. No unburned emissions occur during the combustion of VLSFO in the main engine.

The 100-year global warming potential (GWP) values for CO₂, CH₄, and N₂O are denoted by $GW{P}_{C{O}_2}$, $GW{P}_{C{H}_4}$, and $GW{P}_{N_{2}O}$, and are set at 1, 28, and 265, respectively [30].

Equations (13)–(15) are used to calculate the annual average GHG emission intensity (gCO₂eq/MJ) for dual-fuel ships. This calculation encompasses both the upstream GHG emission intensity—arising from the fuel’s extraction, production, transportation, and bunkering processes—and the direct GHG emissions resulting from fuel combustion in ship engines and potential leakage. The final value of $GHGI{E}_{actual}$ is obtained by summing these two components.

$$WTT={\displaystyle \frac{F_{total}^{ALT}\times C{O}_{2eqWt{T}_{ALT}}\times LC{V}^{ALT}+F_{total}^{VLSFO}\times C{O}_{2eqWt{T}_{VLSFO}}\times LC{V}^{VLSFO}}{F_{total}^{ALT}\times LC{V}^{ALT}+F_{total}^{VLSFO}\times LC{V}^{VLSFO}}}$$

(13)

$$\begin{array}{cc}\hfill TTW& \\ & F_{ME}^{ALT}\times \left[\left(1-\frac{1}{100}{c}_{slipME}\right)\times C{O}_{2eqTt{W}_{ALT,ME}}+\frac{1}{100}{c}_{slipME}\times C{O}_{2eqTt{W}_{slip,ALT,ME}}\right]\times {10}^3\hfill \\ & +F_{ME}^{VLSFO}\times C{O}_{2eqTt{W}_{VLSFO,ME}}\times {10}^3\hfill \\ \hfill =& {\displaystyle \frac{+F_{AE}^{ALT}\times \left[\left(1-\frac{1}{100}{c}_{slipAE}\right)\times C{O}_{2eqTt{W}_{ALT,AE}}+\frac{1}{100}{c}_{slipAE}\times C{O}_{2eqTt{W}_{slip,ALT,AE}}\right]\times {10}^3}{F_{total}^{ALT}\times LC{V}^{ALT}\times RW{D}_{ALT}+F_{total}^{VLSFO}\times LC{V}^{VLSFO}\times RW{D}_{VLSFO}}}\hfill \end{array}$$

(14)

$$GHGI{E}_{actual}=WTT+TTW$$

(15)

where $c_{slipME}$, $c_{slipAE}$ represent the fuel leakage rates for the main engine and auxiliary engine, respectively. The parameter $RWD$ is a key adjustment factor defined in the FuelEU Maritime regulation, introduced to incentivize the use of renewable energy fuels. For fuels derived from non-biological origins, a multiplier of 2 may be applied to the GHG Intensity calculation during the period from 1 January 2025 to 31 December 2033. Otherwise, $RWD$ is set to 1 [29].

2.1.3. Annual Cost Calculation Model

The annual cost $C_{total}$ primarily consists of four components: annualized capital investment cost $C_{investment}$, fixed operating cost of the ship $C_{OPEX-fixed}$, fuel-related variable operating cost $C_{OPEX-varible}$, and the opportunity cost associated with cargo space loss $C_{lost}$. The calculation is expressed in Equation (16) as follows:

$$C_{total}=C_{investment}+C_{OPEX-fixed}+C_{OPEX-varible}+C_{lost}$$

(16)

$$C_{investment}={\displaystyle \frac{r{\left(1+r\right)}^{life}}{(1+r{)}^{life}-1}}\times CAPEX$$

(17)

Equation (17) represents the capital recovery factor formula [31], which is used to annualize the one-time CAPEX over the ship’s operational lifetime. Here, $life$ denotes the ship lifetime, which is set to be 30 years in this study [32]; $r$ is the discount rate, set at 5% according to reference [27]; and $CAPEX$ represents the capital investment cost of the ship [31].

$$C_{OPEX-fixed}=\left(1+\delta \right)\times {\displaystyle \frac{CAPEX}{life}}$$

(18)

$$C_{OPEX-varible}=F_{total}^{ALT}\times f_{ALT}+F_{total}^{VLSFO}\times f_{VLSFO}$$

(19)

The fixed operating cost of a ship varies significantly with ship size. Equation (18) presents the calculation method for fixed operating costs [27], where $\delta$ denotes the ratio of fixed operating cost to average annual shipbuilding cost. This ratio is ship-size dependent and exhibits a strong correlation with ship scale [27,33]. Equation (19) defines the annual variable operating cost—specifically, the annual fuel cost. In this equation, $f_{ALT}$ represents the price of the alternative fuel, and $f_{VLSFO}$ denotes the price of VLSFO.

$$C_{lost}=TEU\times \rho \times c_{TEU}^{lost}\times \phi$$

(20)

$$C_{lost}=DWT\times \rho \times c_{dwt}^{lost}\times \left(T^{berth}+T^{sail}\right)$$

(21)

When alternative fuels are adopted, the lower energy density relative to conventional fuels may require fuel storage tank volumes that are 1.8 to 4.5 times larger to maintain equivalent voyage ranges, thereby encroaching on available cargo space [27]. The cost of lost cargo capacity is calculated using differentiated approaches based on ship type. Equation (20) applies to containerships, while Equation (21) is used for other ship types. In these equations, $\rho$ represents the cargo space loss rate; $TEU$ is the nominal container capacity of the containership (in twenty-foot equivalent units); $c_{TEU}^{lost}$ is the unit cost per lost TEU; $\phi$ is the annual voyage frequency for containerships; $DWT$ is the deadweight tonnage of non-container ships; and $c_{dwt}^{lost}$ is the unit cost per lost deadweight tonnage.

2.1.4. FuelEU Penalty Calculation Model

ANNEX IV of the FuelEU Maritime regulation provides the methodology for calculating the EU compliance balance and associated penalty costs, as expressed in Equations (22) and (23) [29]:

$$\begin{array}{cc}\hfill Compliance balance& \\ \hfill =& \left(GHGI{E}_{target}-GHGI{E}_{actual}\right)\hfill \\ \hfill \times & \left(F_{total}^{ALT}\times 10^3\times LH{V}^{ALT}+F_{total}^{VLSFO}\times {10}^3\times LH{V}^{VLSFO}\right)\hfill \end{array}$$

(22)

$$\begin{array}{c}FuelEU Penalty\hfill \\ =\left\{\begin{array}{c}{\displaystyle \frac{\left|Compliance Balance\right|}{GHGI{E}_{actual}\times 41000}}\times 2400,GHGI{E}_{target}<GHGI{E}_{actual}\\ 0,GHGI{E}_{actual}\le GHGI{E}_{target}\hfill \end{array}\right.\hfill \end{array}$$

(23)

where $GHGI{E}_{target}$ denotes the regulatory GHG intensity limit (gCO₂eq/MJ) for energy used onboard the ship; $GHGI{E}_{actual}$ is the calculated annual average GHG intensity of energy used onboard during the relevant reporting period; and $Compliance balance$ is the compliance balance, calculated as the absolute value of the difference between $GHGI{E}_{target}$ and $GHGI{E}_{actual}$, expressed in grams of CO₂ equivalent (gCO₂eq). $FuelEU Penalty$ refers to the monetary penalty imposed under the FuelEU Maritime regulation, expressed in euros (EUR). The penalty is calculated based on a reference cost of EUR 2400 per ton of VLSFO-equivalent emissions, assuming 1 ton of VLSFO corresponds to 41,000 MJ of energy.

2.1.5. PROMETHEE II Method

MCDA is commonly used to support the selection of optimal compromise solutions based on multiple, often conflicting, criteria. PROMETHEE is a preference ranking method designed to rank and select a finite set of alternatives under such conflicting criteria [34]. In this study, the PROMETHEE II method is applied to a set of fuel configuration alternatives composed of VLSFO, MGO, and seven emerging alternative fuels (including LNG, methanol, ammonia, etc.), with the aim of identifying the optimal substitute fuel option. To objectively determine the weights of the evaluation criteria, the CRITIC method is employed. This method posits that the decision-making information entropy of a criterion increases with its degree of dispersion (standard deviation) and decreases with its correlation (covariance) with other criteria [35]. Moreover, the PROMETHEE II method requires the specification of preference thresholds for each criterion, reflecting the decision-maker’s sensitivity to differences among alternatives. To avoid the uncertainty introduced by subjective threshold assignment, this study adopts an automated approach: the 75th percentile of the normalized difference distribution for each criterion is extracted as its preference threshold. This enhances the reasonableness and adaptability of the preference function in distinguishing between alternatives. To evaluate the impact of changes in preference thresholds and criterion weights on the ranking results, a sensitivity analysis is further conducted in Section 3.4. This includes both single-criterion perturbation tests and Monte Carlo simulations of weight variations, thereby verifying the robustness of the decision outcomes under different parameter settings.

Step 1: The decision matrix is normalized using Equations (24) and (25), corresponding to beneficial (profit-type) and non-beneficial (cost-type) criteria, respectively. Here, $m$ denotes the number of criteria, $n$ represents the number of alternatives, and $x_{ij}$ is the performance value of the $i$ alternative with respect to the $j$ criterion.

$$x_{ij}^{\prime }={\displaystyle \frac{x_{ij}-\mathit{min}{x}_{ij}}{\mathit{max}{x}_{ij}-\mathit{min}{x}_{ij}}}$$

(24)

$$x_{ij}^{\prime }={\displaystyle \frac{\mathit{max}{x}_{ij}-x_{ij}}{\mathit{max}{x}_{ij}-\mathit{min}{x}_{ij}}}$$

(25)

Step 2: The CRITIC method is employed to objectively determine the weights of the criteria. This method evaluates the objective importance of each criterion by considering both the contrast intensity (i.e., variability) of the criterion and its conflict (i.e., correlation) with other criteria [35]. The detailed procedure is defined by Equations (26)–(29), which involve computing the standard deviation of each criterion, followed by the assessment of conflict intensity, information content, and final objective weights.

$${\sigma }_j=\sqrt{{\displaystyle \frac{{\sum }_{i=1}^n{\left(x_{ij}^{\prime }-{\overline{x}}_j\right)}^2}{n-1}}}$$

(26)

$$f_j=\sum _{i=1}^m\left(1-\left|r_{ij}\right|\right)$$

(27)

$$C_j={\sigma }_j\times f_j$$

(28)

$${\omega }_j={\displaystyle \frac{C_j}{{\sum }_{j=1}^m{C}_j}}$$

(29)

where $x_{ij}^{\prime }$ is the normalized value of the $i$ alternative with respect to the $j$ criterion; ${\sigma }_j$ is the sample standard deviation of criterion $j$, representing the contrast intensity; ${\overline{x}}_j$ is the mean value of criterion $j$; $f_j$ denotes the conflict measure of criterion $j$ with all other criteria; $r_{ij}$ is the correlation coefficient between criterion $i$ and $j$ criterion, measured using the Pearson correlation coefficient; $C_j$ represents the amount of information conveyed by criterion $j$; and ${\omega }_j$ is the final normalized weight of criterion $j$.

Step 3: A pairwise comparison is performed among all alternatives based on their criterion values, to determine the evaluative differences between each alternative and all others [34].

$$d_j\left(i,k\right)=g_j\left(i\right)-g_j\left(k\right)$$

(30)

where $d_j\left(i,k\right)$ represents the evaluation difference between alternative $i$ and alternative $k$ with respect to criterion $j$, and $g_j\left(·\right)$ denotes the preference function for criterion $j$, applied to the normalized evaluation difference.

Step 4: The preference function is calculated. Brans and Vincke [36] proposed six types of preference functions. In this study, two types are adopted: the Usual Criterion and the V-shape with Indifference Criterion, as defined in Equations (31) and (32).

$$F_j\left(i,k\right)=\left\{\begin{array}{c}0,d_j\left(i,k\right)\le 0\\ 1,d_j\left(i,k\right)>0\end{array}\right.$$

(31)

$$F_j\left(i,k\right)=\left\{\begin{array}{c}0,d_j\left(i,k\right)\le q\hfill \\ {\displaystyle \frac{d_j\left(i,k\right)-q}{p}},q<d_j\left(i,k\right)\le q+p\hfill \\ 1,d_j\left(i,k\right)>q+p\hfill \end{array}\right.$$

(32)

where $q\ge 0$ is the indifference threshold—below which no preference is expressed—set to 0 in this study, and $p>0$ is the linear preference threshold, defining the interval $\left(q,q+p\right)$ within which the preference strength increases linearly. In this study, $p$ is defined as the third quartile of the normalized evaluative differences.

Step 5: Compute the aggregated global preference index [34].

$$\pi \left(i,k\right)=\sum _{j=1}^m{\omega }_j{F}_j\left(i,k\right)$$

(33)

where $\pi \left(i,k\right)$ denotes the aggregated preference index of alternative $i$ over alternative $k$.

Step 6: Calculate the positive outranking flow and the negative outranking flow [34]. The positive outranking flow ${\varphi }^{+}\left(i\right)$ reflects the degree to which alternative $i$ outranks other alternatives, while the negative outranking flow ${\varphi }^{-}\left(i\right)$ reflects the degree to which alternative $i$ is outranked by the others.

$${\varphi }^{-}\left(i\right)={\displaystyle \frac{1}{n-1}}\sum \pi \left(x,i\right)$$

(34)

$${\varphi }^{+}\left(i\right)={\displaystyle \frac{1}{n-1}}\sum \pi \left(i,x\right)$$

(35)

Step 7: Calculate the net outranking flow $\varphi \left(i\right)$ [34], which is used to rank the alternatives. A higher net outranking flow value indicates a stronger overall performance of the alternative.

$$\varphi \left(i\right)={\varphi }^{+}\left(i\right)-{\varphi }^{-}\left(i\right),\varphi \left(i\right)\in \left[-\mathrm{1,1}\right]$$

(36)

2.2. Data

2.2.1. Case Study Ships

In accordance with the FuelEU Maritime regulation, this study selects five representative ship categories—tankers, bulk carriers, containerships, general cargo ships, and ferries—covering a total of 17 sub-types of ships with gross tonnage exceeding 5000. Based on actual operational data, the sailing and berthing durations within EU waters were extracted in line with the regulatory scope, serving as inputs for evaluating the environmental and economic performance of ships under the FuelEU Maritime framework. The main parameters are summarized in

Table 1. Main parameters of the selected ships.

Ship Category	Ship Subtype	DWT/ TEU	Main Engine Power (kW)	Auxiliary Engine Power (kW)	Design Speed (kn)	Average Speed (kn)	Sailing Time Within EU Waters (h)	Berthing Time at EU Ports (h)	Newbuilding Price (Million EUR)
Tanker	UL/VLCC	296,174	22,200	4620	14.8	11.4	4913	529	115.92
	Suezmax	147,630	18,200	4400	14.0	11.8	3963	2487	80.96
	Aframax	108,553	14,070	2310	15.3	9.9	1630	858	68.54
	Panamax	71,573	22,400	4200	16.0	9.8	5453	3410	54.74
	Handysize	37,993	9960	1776	14.2	11.2	4110	3662	45.08
	Small tanker	6870	2400	945	10.0	8	4272	4585	16.56
Bulker	Capesize	181,709	18,660	2400	11.5	10.8	579	486	68.08
	Panamax	81,290	9930	2400	14.3	11.1	2461	1237	34.04
	Handymax	82,094	9710	2400	14.0	10.9	984	1247	34.04
	Handysize	21,353	9720	2400	17.5	10	3092	3942	19.32
Containership	Post-Panamax	18,270	59,360	15,000	19.0	15.3	2649	773	196.42
	Neo-Panamax	14,074	72,240	13,790	24.2	15.1	3921	698	140.76
	Intermediate	6750	57,099	9000	25.5	15.8	2769.5	1242	81.88
General Cargo	General Cargo I	9500	4000	1030	13.0	10.7	3312	1491	17.00
	General Cargo II	10,952	3360	1170	12.0	10.5	2031	1605	17.94
Ferry	Ro-Pax	12,662	33,600	5850	23.0	17.5	4759.12	3473.9	60.68
	Pass/Car	6133	31,200	7840	21.0	19	6567.71	1803	45.25

Note: (a) General Cargo I refers to ships with a deadweight tonnage (DWT) between 5000 and 9999; General Cargo II refers to those with a DWT greater than 10,000. (b) Data sources: https://www.clarksons.net.cn (accessed on 5 March 2025) and https://www.myvessel.cn (accessed on 5 March 2025).

.

2.2.2. Fuel Data

This study adopts a WtW perspective to analyze three categories of energy carriers: conventional fossil fuels, biomass-based fuels, and renewable electricity-based fuels. A dual-fuel internal combustion engine is selected as the propulsion system, with the combination of VLSFO and MGO serving as the reference configuration. When MGO is replaced by alternative fuels, both the newbuilding price adjustment factor (reflecting changes in construction cost) and the associated deadweight loss during operation must be considered. Detailed fuel parameters are provided in

Table 2. Fuel-related parameters.

Fuel Type	LCV (MJ/kg)	WtT Emission Factor (gCO2eq/MJ)	TtW Emission Factor			Newbuilding Price Adjustment Factor	Cargo Deadweight Loss Rate for This Fuel (%)
			(CO2 gCO2eq/g Fuel)	CH4 (gCH4eq/g Fuel)	N2O (gN2Oeq/g Fuel)
VLSFO	41 [37]	13.2 [38]	3.114 [39]	0.00005 [39]	0.00018 [39]	1	–
MGO	42.7 [38]	14.4 [38]	3.206 [39]	0.00005 [39]	0.00018 [39]	1	–
LNG	49.1 [38]	18.5 [40]	2.750 [39]	0 [39]	0.00011 [39]	1.23 [41]	1 [27]
MEOH	19.9 [38]	31.3 [40]	1.375 [38]	0 [38]	0.00018 [38]	1.1 [42]	2 [27]
Bio-LNG	50 [19]	−25 [38]	2.750 [38]	0 [38]	0.00011 [38]	1.23 [41]	1 [27]
Bio-MEOH	20 [38]	−55.4 [38]	1.375 [38]	0 [38]	0.00018 [38]	1.1 [42]	2 [27]
HVO	44 [19]	−33.4 [38]	3.115 [38]	0.00005 [38]	0.00018 [38]	1 [43]	–
E-LH2	120 [44]	0 [40]	0 [39]	0 [38]	0.00018 [38]	1.58 [45]	5 [27]
E-NH3	18.6 [38]	0 [40]	0 [40]	0 [38]	0.00018 [38]	1.25 [45]	3 [27]

Note: (a) The WtT emission factor for VLSFO is based on the corresponding value for LFO. (b) “–” indicates no cargo deadweight loss.

. The loss of cargo deadweight results in cargo space opportunity costs, which vary depending on ship type and voyage length; the unit loss costs are shown in

Table 3. Unit cargo loss cost by ship type.

Voyage Length	Short	Medium	Long
Tanker (€/dwt/day)	0.59 [46]	0.33 [46]	0.12 [46]
Bulker (€/dwt/day)	0.30 [46]	0.21 [46]	0.08 [46]
Ferry (€/m3 of space loss)	6 [47]	8 [47]	10 [47]
General cargo (€/dwt/day)	0.1 [47]	–	–
Containership (€/TEU/trip)	600 [47]	900 [47]	1100 [47]

Note: “–” indicates that the unit cargo loss cost for this ship type is not affected by voyage length.

.

Based on reference [7], this study establishes annual GHG intensity target values for the period 2025–2050. Fuel price trajectories are determined with reference to the trends outlined in [19,21], incorporating differentiated annual adjustment strategies. Fossil fuels and HVO follow a progressive price increase path, while biofuels such as Bio-LNG and Bio-MEOH adopt a gradual price reduction scheme. Renewable electricity-based fuels are modeled using a stepwise cost-decline trajectory toward 2050. Specific values are presented in

Table 4. Annual GHG intensity targets and fuel prices.

Year	GHG Intensity Target	Fuel Price (EUR/t)
		VLSFO	MGO	LNG	MEOH	Bio-LNG	Bio-MEOH	HVO	E-LH2	E-NH3
2025	89.34	540.14	731.5	543.60 [4]	460 [48]	1383.90 [4]	1193 [49]	1000 [50]	5059.54 [51]	1072.52 [51]
2030	85.69	514.46	847	665.03	575	1231	1152	1031.12	5299.2	872.71
2035	77.94	692.58	885	715.22	582	1138.5	1147.5	1177.86	4857.6	804.26
2040	62.90	870.7	923	752.87	589	1046	1143	1324.6	4416	735.82
2045	34.64	1015.30	935.5	790.51	585.5	980	1125	1467.97	3974.4	663
2050	18.23	1159.89	948	828.15	582	914	1107	1611.33	3532.8	590.36

. The 2025 baseline prices for VLSFO and MGO are estimated based on historical average port prices from 2020 to 2024 obtained from the Clarkson database. The USD-to-EUR exchange rate is converted using the fixed value as of 24 March 2025.

3. Results

3.1. Environmental Assessment of Alternative Fuels

According to the FuelEU Maritime regulation [29], renewable fuels of non-biological origin are granted time-dependent reward factors. As a result, the evaluation of WtW GHG intensity is divided into two periods: 2025–2030 and 2035–2050. Figure 2, Figure 3, Figure 4, Figure 5 and Figure 6 present the WtW GHG intensity results for alternative fuels across five ship types for the period 2025–2030, with horizontal lines indicating the annual GHG intensity targets from 2025 to 2050. It is evident that, among the alternative fuel options, the VLSFO + MEOH combination exhibits the highest WtW GHG intensity, even exceeding that of the baseline VLSFO + MGO. This is primarily due to the higher WtT intensity of methanol compared to MGO, which offsets the TtW advantages. Both VLSFO + MGO and VLSFO + MEOH fail to meet the GHG intensity targets for 2025 and beyond. While VLSFO + LNG offers moderate improvement, it is projected to become non-compliant by 2035. Although the WtT GHG intensity of biofuels such as VLSFO + HVO and VLSFO + BIO-LNG offers notable advantages, their relatively high TtW emissions are expected to pose compliance challenges beyond 2040. By comparison, the VLSFO + Bio-MEOH configuration demonstrates a well-balanced performance across both WtT and TtW stages—second only to e-fuels—resulting in a significantly lower overall WtW GHG intensity than other biofuel options. This finding aligns with the conclusions of related studies [11,12], confirming the viability of Bio-MEOH as a medium- to long-term decarbonization pathway in maritime transport. E-fuel options such as VLSFO + E-LH₂ and VLSFO + E-NH₃, characterized by their ultra-low emissions profiles, can meet the long-term GHG intensity reduction targets set for the 2025–2050 period. Among them, E-LH₂ is considered the most environmentally favorable option and is prioritized by policymakers for promotion—an assessment that is consistent with the perspective presented in [52].

Figure A1, Figure A2, Figure A3, Figure A4 and Figure A5 illustrate the WtW GHG intensity results for alternative fuels across five ship types during the 2035–2050 period. As the reward factor only applies to E-LH₂ and E-NH₃, a noticeable increase—approximately twofold—in their WtW GHG intensity can be observed. Nevertheless, both fuel combinations still meet the 2050 GHG intensity targets.

The comparison across different ship types also reveals pronounced structural differences in WtW GHG intensity, which stem from the combined effects of ship design parameters, operational conditions, and emission profiles. For instance, the highest value appears in the VLSFO + MEOH scenario for the Handysize bulker, reaching 102.56 gCO₂eq/MJ, while the lowest is also found in the same ship type under the VLSFO + E-LH₂ configuration, at just 1.18 gCO₂eq/MJ. Moreover, the VLSFO + Bio-MEOH combination enables compliance with 2025–2050 GHG targets for several ship types, including Aframax, Panamax, and Small tankers, as well as Handymax and Handysize bulkers. However, this compliance is not observed across all ship types. This variation highlights the need for policy considerations that tailor alternative fuel configurations to the specific characteristics of different ship types. Regardless of ship category, VLSFO + MEOH consistently exhibits the highest WtW GHG intensity, while VLSFO + E-LH₂ remains the lowest. The overall compliance trends of different fuels remain largely consistent across ship types, which is fundamentally determined by the intrinsic properties of each fuel.

3.2. Economic and Compliance Assessment of Alternative Fuels

Figure 7, Figure 8, Figure 9, Figure 10 and Figure 11 present the annual cost and EU compliance penalty cost for five ship types. From these figures, it is evident that, between 2025 and 2050, annual cost trends vary significantly across different ship–fuel combinations. According to reference [53], fuel costs typically account for approximately 50% of a ship’s operational costs (excluding capital expenditure and cargo handling), making them the primary driver of annual cost variation. For all ship types, the cost of fossil fuels increases gradually over time, leading to a corresponding rise in annual costs, albeit at a relatively moderate rate. Among the evaluated configurations, VLSFO + LNG consistently demonstrates strong performance across most ship types. From the shipowner’s economic perspective, it frequently ranks among the lowest annual cost options. LNG also receives the highest overall ranking, with evaluation results broadly consistent with the analysis presented in [52]. While fossil fuels show moderate cost growth, alternative fuels exhibit more dynamic trends in both pricing and cost competitiveness over time. In particular, the price of e-fuels declines over time, resulting in decreasing annual costs. However, despite this trend, fossil fuels still maintain significantly lower annual costs compared to e-fuels. Notably, the VLSFO + E-LH₂ combination remains the most expensive option in ship types such as Aframax, Capesize, and Handymax throughout the 2025–2050 period. Among biofuels, HVO demonstrates relatively strong economic feasibility before 2035, with annual costs slightly higher than those of MGO and LNG in most ship types, followed by a steady increase thereafter. Conversely, the annual cost of Bio-LNG decreases steadily after 2035. Bio-MEOH, however, shows poor economic performance, with annual costs surpassing those of e-fuels across ships such as RoPax and Pass/Car throughout the entire analysis period.

In terms of compliance assessment, consistent with the environmental evaluation, the VLSFO + MEOH combination exhibits the highest WtW GHG intensity, leading to high EU compliance penalties as early as 2025. For fossil fuel blends such as VLSFO + MGO and VLSFO + LNG, significant increases in compliance penalties begin around 2030 and 2035, respectively, reflecting a tightening regulatory stance on high-emission fuels. Among biofuels, VLSFO + Bio-LNG begins to incur notable penalties in certain ship types by 2050. In contrast, e-fuels—owing to their low WtW GHG intensity—demonstrate a clear advantage in compliance cost avoidance across the entire 2025–2050 horizon. The varying responses of different ship types to fuel-related economic and compliance factors highlight the need for differentiated, ship-specific, and phased policy strategies in future regulatory design.

3.3. Optimal Selection of Alternative Fuel Options

This study applies the PROMETHEE II multi-criteria decision-making method to rank alternative fuel options for 17 ship subtypes. Three quantitative indicators are selected to represent the dimensions of environmental performance, cost, and regulatory compliance: WtW GHG intensity, annual cost, and compliance surplus. The CRITIC method is used to determine the objective weights of each criterion, based on both contrast intensity and inter-criteria conflict.

Table 5. Optimal alternative fuel choices for ships from 2025 to 2050.

Ship/Year	2025	2030	2035	2040	2045	2050
UL/VLCC	VLSFO + HVO	VLSFO + HVO	VLSFO + HVO	VLSFO + HVO	VLSFO + BIO-LNG	VLSFO + E-NH3
Suezmax	VLSFO + HVO	VLSFO + HVO	VLSFO + HVO	VLSFO + BIO-LNG	VLSFO + BIO-LNG	VLSFO + E-NH3
Aframax	VLSFO + HVO	VLSFO + HVO	VLSFO + HVO	VLSFO + HVO	VLSFO + HVO	VLSFO + HVO
Panamax	VLSFO + HVO	VLSFO + HVO	VLSFO + HVO	VLSFO + BIO-LNG	VLSFO + BIO-LNG	VLSFO + E-LH2
Handysize	VLSFO + HVO	VLSFO + HVO	VLSFO + HVO	VLSFO + HVO	VLSFO + BIO-LNG	VLSFO + E-NH3
Small tanker	VLSFO + HVO	VLSFO + HVO	VLSFO + HVO	VLSFO + HVO	VLSFO + BIO-LNG	VLSFO + E-NH3
Capesize	VLSFO + HVO	VLSFO + HVO	VLSFO + HVO	VLSFO + HVO	VLSFO + HVO	VLSFO + E-NH3
Panamax	VLSFO + HVO	VLSFO + HVO	VLSFO + HVO	VLSFO + HVO	VLSFO + BIO-LNG	VLSFO + E-NH3
Handymax	VLSFO + HVO	VLSFO + HVO	VLSFO + HVO	VLSFO + HVO	VLSFO + HVO	VLSFO + E-NH3
Handysize	VLSFO + HVO	VLSFO + HVO	VLSFO + HVO	VLSFO + BIO-LNG	VLSFO + BIO-LNG	VLSFO + E-NH3
Post-Panamax	VLSFO + HVO	VLSFO + HVO	VLSFO + HVO	VLSFO + HVO	VLSFO + BIO-LNG	VLSFO + E-NH3
Neo-Panamax	VLSFO + HVO	VLSFO + HVO	VLSFO + HVO	VLSFO + BIO-LNG	VLSFO + BIO-LNG	VLSFO + E-NH3
Intermediate	VLSFO + HVO	VLSFO + HVO	VLSFO + HVO	VLSFO + HVO	VLSFO + BIO-LNG	VLSFO + BIO-LNG
General Cargo I	VLSFO + HVO	VLSFO + HVO	VLSFO + HVO	VLSFO + HVO	VLSFO + BIO-LNG	VLSFO + E-NH3
General Cargo II	VLSFO + HVO	VLSFO + HVO	VLSFO + HVO	VLSFO + HVO	VLSFO + E-NH3	VLSFO + E-NH3
Ro-Pax	VLSFO + HVO	VLSFO + HVO	VLSFO + HVO	VLSFO + E-LH2	VLSFO + E-LH2	VLSFO + E-LH2
Pass/Car	VLSFO + E-LH2	VLSFO + HVO	VLSFO + E-LH2	VLSFO + E-LH2	VLSFO + E-LH2	VLSFO + E-LH2

presents the optimal alternative fuel options identified for different ship types over the period 2025–2050. Results show that the rankings remain relatively stable across all ship types between 2025 and 2035, while greater diversity emerges from 2040 to 2050. In 2025 and 2035, all ship types—except for Pass/Car—select VLSFO + HVO as the optimal fuel option, primarily due to its relatively low WtW GHG intensity and moderate annual cost, making it the dominant choice. From 2040 onward, ships such as Suezmax and Panamax begin to shift toward VLSFO + Bio-LNG as the preferred alternative, while Ro-Pax and Pass/Car ships transition to e-fuels. By 2050, most ship types ultimately select e-fuels as their optimal transition fuel. This overall trend is consistent with the alternative fuel transition pathway outlined in [11]. However, due to the incorporation of ship-type heterogeneity in this study, the specific results exhibit a certain degree of divergence, reflecting the nuanced differences in fuel performance and feasibility across ship types. Notably, VLSFO + MGO and VLSFO + MEOH are not selected as the optimal option for any ship type, underscoring the overall advantages of biofuels and e-fuels in terms of environmental, economic, and compliance performance.

3.4. Sensitivity Analysis

In the PROMETHEE II method, this study adopts the V-shape with indifference criterion preference function for the two indicators—annual cost and compliance surplus. The preference threshold $p$ is one of the key parameters influencing the ranking outcomes. To assess the impact of variations in p on the stability of rankings, 20 evenly spaced values were set within the range [0.1, 1] for each indicator. Sensitivity curves were then constructed using Kendall’s Tau correlation coefficient to capture the ranking consistency under these perturbations.

The results indicate that the ranking structure is highly sensitive to the preference threshold: in certain intervals, even small variations in $p$ can significantly alter the ranking outcomes. However, within specific sub-ranges, the Kendall’s Tau values stabilize and approach 1, suggesting that parameter values within these intervals yield greater consistency and robustness in ranking, thus providing quantitative support for informed parameter selection.

Furthermore, recognizing the potential uncertainty associated with the CRITIC-derived criterion weights in practical applications, this study incorporates weight settings as a second core dimension of the sensitivity analysis. A Monte Carlo simulation with 10% perturbation amplitude and 1000 iterations was conducted to evaluate the ranking robustness of each alternative under varying weight configurations. The standard deviation of net flow values is employed as the primary metric for evaluating sensitivity to weight fluctuations, with lower values indicating reduced susceptibility to weight changes and enhanced robustness of the ranking outcomes.

Using the UL/VLCC ship type as an illustrative case, Figure 12, Figure 13, Figure 14, Figure 15, Figure 16 and Figure 17 present the sensitivity analysis results for different years between 2025 and 2050. For the annual cost indicator, the recommended $p$ values are primarily distributed in the 0.38–0.81 range, within which Kendall’s Tau values remain consistently close to 1, indicating high ranking stability. Similarly, for the compliance surplus indicator, the recommended $p$ values fall mainly within the 0.29–0.76 interval, also demonstrating strong robustness.

In the weight perturbation analysis, the VLSFO + MEOH configuration exhibits persistently high standard deviations in net outranking flow across the entire time period, reflecting strong sensitivity to weight variations and an inability to achieve stable rankings—consistent with the findings in Table 5, where this option never emerges as the top-ranked solution. In contrast, VLSFO + HVO shows the lowest standard deviation values during the 2025–2040 period, indicating robust and stable ranking outcomes. Although VLSFO + Bio-LNG and VLSFO + E-NH₃ do not always exhibit the minimum standard deviation, their relatively low variability and consistently favorable performance across multiple ship types suggest strong adaptability and potential advantages across key evaluation dimensions.

4. Conclusions and Discussion

This study addresses the critical question of how to determine the optimal alternative fuel configurations for different ship types under the constraints of the FuelEU Maritime regulation. It proposes a multi-criteria evaluation framework that incorporates life-cycle thinking and aligns with regulatory requirements. Through a systematic assessment, the study identifies the preferred fuel pathways for 17 representative subtypes across five major categories of ships from 2025 to 2050. The findings provide scientific support for shipping companies, shipbuilders, and policymakers in planning newbuilds, retrofitting existing ships, and developing fuel infrastructure. In particular, the results offer quantitative guidance for practical decision-making in the maritime energy transition, especially in the context of tightening carbon intensity targets and growing fuel diversification. The key findings are summarized as follows:

(1) Fuel type is the decisive factor that influences WtW GHG intensity and regulatory compliance, while ship-type differences have a relatively limited impact. E-fuels (VLSFO + E-LH₂/E-NH₃), due to their ultra-low life-cycle emissions, are the only solutions that enable full compliance with the 2025–2050 GHG reduction targets across all ship types. In contrast, conventional fuels (VLSFO + MGO/MEOH) and biofuels (HVO/BIO-LNG) fail to achieve long-term compliance due to excessive emissions in either the WtT or TtW stages;

(2) Although fossil fuels exhibit relatively low short-term annual costs, their compliance penalty costs increase sharply starting around 2030, ultimately eroding their long-term cost advantage. E-fuels (E-LH₂/E-NH₃), despite high initial costs, gain competitiveness by 2050 due to declining fuel prices and persistently low compliance penalties. Biofuels, by contrast, display transitional characteristics: they are more economically viable than e-fuels in the near to medium term but offer a limited window of environmental compliance. This suggests that biofuels can serve as an interim solution to balance economic and emission reduction objectives before a full-scale transition to e-fuel systems;

(3) In the short term, biofuels—particularly HVO—emerge as the most favorable options; however, by 2050, e-fuels (E-LH₂/E-NH₃) become the optimal choices for most ship types due to their superior environmental compliance. Life-cycle carbon performance and regulatory compatibility are the decisive factors shaping the competitiveness of alternative fuels. Conventional fuel options such as VLSFO + MGO and VLSFO + MEOH are systematically phased out, as they fail to strike a balance between economic feasibility and emissions reduction.

While this study develops a foundational assessment framework across environmental, economic, and regulatory dimensions, future research may expand in the following directions to enhance decision-support capabilities:

(1) Integrate technological maturity and social acceptability into the evaluation framework to establish a more comprehensive decision-making model that captures technical, economic, and societal dimensions;

(2) Develop dynamic, time-sequenced models informed by market supply–demand dynamics and policy evolution, employing Monte Carlo simulations or machine learning to forecast alternative fuel price fluctuations and to enhance the credibility of cost-effectiveness assessments.

(3) Move beyond the assumption of fixed ship operating profiles by quantifying the regulatory and environmental impacts of flexible strategies, such as slow steaming and multi-fuel blending, on GHG intensity and compliance costs.