Assessing an Optimal Green Hydrogen Strategy for an Inland Refinery ^†

Abstract

This study assesses four hydrogen production pathways (electrolysis, ammonia cracking, steam biomethane reforming, and methane pyrolysis) for an inland refinery under European Renewable Energy Directive III (RED III) goals. Using multicriteria decision analysis (MCDA), economic, environmental, technological, and implementation factors were evaluated. The results show that biomethane reforming offers the lowest cost, while electrolysis provides the best environmental and technological performance. Sensitivity analysis highlights electricity price as the key factor. The MCDA model proved to be effective for systematic comparison and informed strategic decision making. However, RED III regulatory requirements may favor ammonia or electrolysis for renewable fuel of non-biological origin production, emphasizing the need for long-term strategic planning to maintain competitiveness.

1. Introduction

The latest Renewable Energy Directive III (RED III) and the goal of carbon neutrality by 2050 require European industries to integrate renewable energy sources into their operations [1]. Refineries are recognized as one of the main emitters of greenhouse gases [2]. As highlighted by Popoola et al., strategies to address emissions include green hydrogen, electrification, and biofuels. Advanced biofuels show strong potential for decarbonizing sectors such as aviation [3], yet few studies evaluate their economic feasibility for refineries [2]. Hydrogen production represents one of the main sources of CO₂ emissions in refineries [4]; therefore, replacing fossil-based hydrogen with low-carbon alternatives may be a solution for refinery decarbonization [5,6]. Atia et al. [7] demonstrated the potential of green hydrogen by replacing hydrogen from steam methane reforming with electrolytic hydrogen. A Finnish study further confirmed the feasibility of green hydrogen for refineries using wind energy [8]. Romero-Piñeiro et al. [9] presented a techno-economic assessment of hydrogen production via electrolysis and residual biomethane gasification for an existing refinery, showing that both pathways are viable. Studies also emphasize the role of hydrogen transport through dedicated pipelines or hydrogen carriers such as ammonia and methanol [10,11]. The development of such networks is expected to expand green hydrogen markets [12,13]; however, it is significantly impacted by several technical, regulatory and financial factors [14,15].

The aim of the present study is to provide a comprehensive techno-economical evaluation of four strategic hydrogen pathways for an inland refinery to respond effectively to the European Union’s sustainability requirements. The study examines the technical, economic, environmental, and implementation dimensions of each strategy. To account for all (often conflicting) criteria, multicriteria decision analysis (MCDA) is used. In this study, the analytic hierarchy process (AHP) calculation algorithm is selected as the most commonly used method [16]. The algorithm allows strategies to be compared based on several areas simultaneously and it can also reduce the influence of subjective approaches. Another contribution relates to the renewable solutions for an inland refinery in Central Europe, which has less favorable conditions for the development of renewable energy sources compared to coastal or southern European regions.

2. Methodology

Four hydrogen production alternatives (A) are compared in MCDA. The first alternative (A1) includes investing in an electrolyzer with renewable energy supply. This produced hydrogen can be labelled as a renewable fuel of non-biological origin (RFNBO) based on the RED III requirements. High hydrogen prices, which arise due to high energy prices in the EU and challenges in consistent renewable electricity supply, can be resolved by importing green ammonia (A2) produced in countries with lower electricity prices, transported into the refinery, and subsequently cracked to produce hydrogen categorized as RFNBO. Biomethane with similar properties to natural gas can be used in existing steam reforming units (A3), where natural gas is currently utilized to produce hydrogen. The use of biofuels is supported in EU targets established under RED III. Depending on the origin of the biomethane supplied, the overall carbon footprint of hydrogen produced via steam reforming can be reduced. Thus, integration of biomethane is considered part of the refinery’s renewable hydrogen strategy. Another option is to install a pyrolysis unit that produces low-carbon hydrogen from natural gas as no CO₂ is generated directly in the process (A4). The processes are considered as a black box, excluding steam reforming. The existing SMR unit in the studied refinery is treated using operating data.

2.1. Criteria

Hydrogen production technologies were compared based on four criteria defined by one or more sub-criteria (Figure 1). The sub-criteria were quantified through binary scoring, ordinal ranking, or assessment based on available data.

2.1.1. Economical Criterion

Prices of the most relevant utilities, by-products, or feedstocks used in the economic evaluation are listed in

Table 1. Price of the most relevant utilities, by-products, and feedstocks.

Hydrogen Price Component	Price
Electric energy (EUR/MWh)	108
Renewable electric energy (EUR/MWh)	118
Natural gas (EUR/MWh)	38
Biomethane (EUR/MWh) 1	80
CO2 ETS (Emission Trading System) (EUR/t)	70

¹ Including natural gas and biopremium, advanced category.

. First, direct utilization of biomethane in the existing SMR unit was assumed. Since no new investment was needed, the average production price of SMR H₂ derived from operating data was used in the analysis.

For clarity, the operating expenses (OPEX) were classified into six main categories (

Table 2. Categories of operating expenses used for SMR economic analysis. CO₂ Emission Trading System costs apply.

Utilities	Steam	Fixed OPEX	Biomethane	Natural Gas	CO2 ETS
Mixbed water	Excess 3.5 MPa steam 1	Catalysts and adsorbents	Feed	Fuel for furnace
Nitrogen	Excess 0.4 MPa steam 1	Turnaround
Instrumental Air
Cooling Water
Electricity
Off-gas export 1
Desulfurization H2
Reforming gas

¹ Treated as a by-product of steam methane reforming unit.

). The unit itself was analyzed in detail in the underlying research; therefore, a detailed assessment is beyond the scope of this study.

To enable objective comparison of hydrogen prices for electrolysis, ammonia cracking, and methane pyrolysis, the first-year levelized cost was calculated. A Proton Exchange Membrane (PEM) electrolyzer was considered in this study; it is designed for continuous operation with average annual utilization of 8200 h. For electrolysis, the first-year levelized cost of hydrogen was evaluated as:

$${\mathrm{L}\mathrm{C}\mathrm{O}\mathrm{H}}_1^{\mathrm{E}\mathrm{l}}={\mathrm{d}}_{\mathrm{p}}+\mathrm{O}\mathrm{P}\mathrm{E}\mathrm{X}+{\mathrm{C}}_{\mathrm{E}}+{\mathrm{C}}_{\mathrm{S}}$$

(1)

where d_p represents depreciation, ${\mathrm{C}}_{\mathrm{E}}$ electricity costs, and ${\mathrm{C}}_{\mathrm{S}}$ stock-related cost balance.

Linear depreciation over a 20-year period was assumed, with one stack replacement during this timeframe based on its expected lifetime. Electricity costs were calculated from the specific energy consumption. Renewable energy was assumed to be used to produce hydrogen. OPEX and ${\mathrm{C}}_{\mathrm{S}}$ were determined as a percentage of the electrolyzer’s capital cost. Parameters used for the ${\mathrm{L}\mathrm{C}\mathrm{O}\mathrm{H}}_1^{\mathrm{E}\mathrm{l}}$ determination are summarized in

Table 3. PEM electrolyzer parameters [17,18,19].

Parameter	Value
Lifetime stack (h)	80,000
Specific energy consumption (kWh/kg H2)	56.7
CAPEX (EUR/kW) 1	2000
OPEX (% of CAPEX)	3
Stack replacement costs (% ISBL)	50

¹ 50% inside battery limits (ISBL), 50% outside battery limits (OSBL).

.

Furthermore, ammonia was selected as the hydrogen carrier. Green ammonia was produced, shipped to a European port, transported by rail, and stored at 25 bar before being cracked to hydrogen at 73% efficiency [20]. Costs associated with NH₃ production and logistics were determined in collaboration with the refinery. However, only the resulting hydrogen cost prior to the cracking process (${\mathrm{L}\mathrm{C}\mathrm{O}\mathrm{H}\prime }_1^{\mathrm{N}{\mathrm{H}}_3}$) is reported in

Table 4. Green ammonia cracking parameters [20].

Parameter	Value
${\mathrm{L}\mathrm{C}\mathrm{O}\mathrm{H}\prime }_1^{\mathrm{N}{\mathrm{H}}_3}$(EUR/kg H2) 1	8.78
Total installed costs (EUR/kg)	0.40
Specific energy consumption (kWh/kg H2)	0.315

¹ Hydrogen cost prior to the cracking process.

, together with the required parameters. Linear depreciation over a 20-year period was assumed. During the ammonia cracking, renewable energy was assumed to be used. The levelized cost of hydrogen for the first year of operation was then calculated as:

$${\mathrm{L}\mathrm{C}\mathrm{O}\mathrm{H}}_1^{\mathrm{N}{\mathrm{H}}_3}={\mathrm{d}}_{\mathrm{p}}+{\mathrm{C}}_{\mathrm{E}}+{\mathrm{L}\mathrm{C}\mathrm{O}\mathrm{H}\prime }_1^{\mathrm{N}{\mathrm{H}}_3}$$

(2)

Finally, the first-year levelized cost of hydrogen from electricity-driven methane pyrolysis was calculated as:

$${\mathrm{L}\mathrm{C}\mathrm{O}\mathrm{H}}_1^{\mathrm{M}\mathrm{P}}={\mathrm{d}}_{\mathrm{p}}+{\mathrm{C}}_{\mathrm{E}}+{\mathrm{C}}_{\mathrm{F}}+{\mathrm{C}}_{\mathrm{M}}$$

(3)

with ${\mathrm{C}}_{\mathrm{F}}$ representing feed costs and ${\mathrm{C}}_{\mathrm{M}}$ maintenance costs. Input data used in the calculation are summarized in

Table 5. Methane pyrolysis parameters [21].

Parameter	Value
CAPEX (EUR/kW)	10,000
Maintenance (% of CAPEX)	2.5
Specific methane consumption (kg CH4/kg H2)	4.3
Specific energy consumption (kWh/kg H2)	15

. Depreciation was assumed over a 20-year period. Expenses or potential revenues associated with the carbon by-product were not considered in this analysis.

2.1.2. Technological Criterion

Technological criterion was assessed via binary scoring, simple ranking, or sub-criterion values (

Table 6. Technological criterion—values of sub-criteria [20,22,23,24,25,26,27,28,29,30,31].

Ai	H2 Compressor	Minimum Load	Start-Up Time	Waste Heat	TRL	H2 Purity
A1	0	10%	1	0.5	9	A
A2	1	20%	2 1	1	7	C
A3	0	30%	3	1	9	C
A4	1	30% 1	2 1	1	8	C

¹ Refinery’s internal experts’ opinion.

). Hydrogen compressors were evaluated due to cost and maintenance using binary system. Within the reforming process in the refinery, no hydrogen compressor was used. The PEM electrolyzer generated high-pressure hydrogen even with an inlet pressure of 1 bar due to electrochemical compression. The PEM electrolyzer used a proton exchange membrane impermeable to hydrogen, leading to hydrogen accumulation on the cathode side and an increase in pressure. Therefore, a hydrogen compressor was not required [22]. Methane pyrolysis operated at low pressure due to equilibrium shift [23] and ammonia cracking [20], indicating the need for a hydrogen compressor.

The minimum load depended on the process design, e.g., the reforming unit was limited by the compressor inlet flow. PEM electrolyzers had a minimum operating load of 10% [24,25]. The minimum load of 20% was assigned to ammonia cracking [20,26] and 30% to methane pyrolysis based on the recommendation.

Alternatives were ranked according to start-up time. Electrolysis with the fastest response (<20 min) is ranked first [24,25] while large-scale SMR requiring one to two days was placed last. As ammonia cracking and methane pyrolysis were expected to operate at a lower scale than the reforming unit, they were consequently ranked second. High-temperature processes (A2–A4) allowed waste-heat recovery and were therefore rated 1. Electrolysis, with potential heat-pump integration [27], was rated 0.5. The next sub-criterion characterized the maturity of technology based on the technology readiness level (TRL), adopted from literature research [28,29,30]. Hydrogen purity without an additional separation unit was also assessed. Electrolysis produces fuel-cell-grade hydrogen (A level), whereas other technologies required an additional purification step to reach >99% purity (C level) [20,31].

2.1.3. Environmental Criterion

Within this criterion, direct CO₂ emissions were calculated. As this was valid only for the reforming process, A1, A2, and A4 were assigned the value of 0. Emissions from steam reforming using biomethane were estimated based on operating data of the existing SMR unit, while the carbon footprint of feedstock was replaced. The carbon footprint of biomethane was −5.7 g CO₂eq/MJ, as determined through an ISCC audit based on a full life-cycle assessment (LCA) of the biomethane plant. Emissions calculated from measured data considering biomethane as feedstock were 7.97 kg CO₂/kg of H₂. A detailed analysis of the SMR unit was conducted in our previous work and exceeds the scope of this article, meaning that only the resulting emission values are reported here.

In local legislation, the term renewable gas includes not only renewable hydrogen, but also gaseous fuels produced from biomass, such as biogas or biomethane. Renewable hydrogen is defined as hydrogen with an energy content that originates from renewable energy sources and meets the criteria specified in a separate regulation (not further detailed). A sustainability certificate can be issued for fuels including hydrogen produced from biomass, renewable sources of non-biological origin, or for fuels containing recycled carbon [32]. European RED III [32] targets a renewable fuel share of 22–29% and mandates the use of RFNBO. Only electricity generated from fully renewable sources may be used for RFNBO production. From 2030 onward, additional requirements will apply for renewable electricity, including the principle of additionality and a 70% reduction in greenhouse gas emissions compared with energy generated from fossil fuels.

To balance regulatory compliance with emission reduction goals, hydrogen derived from biomethane was included among the assessed alternatives. Although it does not meet the EU criteria for renewable hydrogen, biomethane is classified as renewable gas under national legislation and allows for significant emission reductions within the refinery boundary. Therefore, it allows the refinery to integrate renewable feedstocks and achieve reductions in direct CO₂ emissions.

The system boundary for calculating the carbon footprint in this study was limited to the refinery boundary. A full life-cycle carbon footprint was not included, as a complete life-cycle assessment of all hydrogen pathways falls outside the scope of this study. Such an assessment requires additional information on the origin of green ammonia, or terminal infrastructure, e.g., Akhtar et al. performed a full life-cycle assessment of hydrogen derived from green ammonia imported from Australia [33]. Another recent study compares carbon intensities of green hydrogen production for several countries, depending on their energy mix and marginal power sources over one year period [34]. Nevertheless, a comprehensive life-cycle assessment can serve as a valuable framework for future evaluation of the complete environmental impact of all hydrogen pathways.

2.1.4. Implementation Criterion

The final criterion evaluated practical experience with each technology. SMR, already operating, ranked first, followed by electrolysis technology already implemented within the refinery corporate group. Novel technologies (A2, A4) ranked last. Current availability for hydrogen production was a further criterion, with ammonia transport risks included in the binary evaluation. Values are listed in

Table 7. Implementation criterion—values of sub-criteria.

Ai	Experience	Logistic Risks	Availability
A1	2	0	2
A2	3	1	4
A3	1	0	1
A4	3	0	3

.

2.2. MCDA

Evaluating alternatives often requires considering multiple criteria simultaneously, as focusing on a single criterion can produce conflicting results. Exploring all possible interactions between criteria is possible but time-consuming. To address this, MCDA was applied, using the AHP in this research. Details of the algorithm are described in previous papers [33,35]. A comprehensive description of the MCDA algorithm falls outside the scope of this study, and full details can be found in [36]. The main advantage of this approach is its complexity and objectivity. Evaluation of alternatives considering all criteria is possible without needing the decision-maker to rank or prioritize individual criteria, requiring only the selection of relevant criteria for consideration.

3. Results

3.1. Criteria Evaluation

The whole process starts with the quantification of the four selected criteria. A summary of the data required for criteria calculation are provided in previous subchapters. For the evaluation, the economic criterion uses linear interpolation, while the other criteria are assessed via simple ranking.

To ensure maximum objectivity, the alternatives are first evaluated using the implementation criterion within the MCDA algorithm, yielding a score for further analysis. A simple ranking method is employed, with no preferences assigned. Consequently, all possible combinations of the three monitored sub-criteria are considered. This results in 139 suitable combinations used for the analysis. From these, a percentage success rate is calculated, with specific positions (1st–4th) representing the respective weights, which can be referred to as the average weighted position. The resulting scores reflect the relative success of each alternative with the total of all percentage scores summing up to 100% (

Table 8. Calculated criteria of selected hydrogen technologies.

	Criterion
Ai	Economic (EUR/kg H2)	Environmental (kg CO2/kg H2)	Technological	Implementation
A1	7.97	0	0.9167	0.3000
A2	9.21	0	0.4306	0.1000
A3	4.14	7.97	0.5972	0.4000
A4	5.07	0	0.4306	0.2000

).

This approach is not applicable to the technological criterion, as the MCDA model is currently limited to four criteria, since including more would substantially increase the calculation time. Thus, the technological sub-criteria are normalized to values between 0 and 1, and equal weighting is used, as no particular sub-criterion is prioritized. For the six sub-criteria, an equal weight of 0.167 is assigned to each. The overall score of an alternative for the technological criterion ($\mathrm{s}\mathrm{c}\mathrm{o}\mathrm{r}{\mathrm{e}}_{\mathrm{t}\mathrm{e}\mathrm{c}\mathrm{h}})$ is then calculated as the sum of the weighted scores of all sub-criteria ($\mathrm{s}{\mathrm{c}\mathrm{o}\mathrm{r}\mathrm{e}}_{{\mathrm{S}-\mathrm{C}}_{\mathrm{i}}})$:

$$\mathrm{s}\mathrm{c}\mathrm{o}\mathrm{r}{\mathrm{e}}_{\mathrm{t}\mathrm{e}\mathrm{c}\mathrm{h}}=\sum _{\mathrm{i}=1}^6{\mathrm{w}\mathrm{e}\mathrm{i}\mathrm{g}\mathrm{h}\mathrm{t}}_{\mathrm{S}-{\mathrm{C}}_{\mathrm{i}}}·\mathrm{s}{\mathrm{c}\mathrm{o}\mathrm{r}\mathrm{e}}_{{\mathrm{S}-\mathrm{C}}_{\mathrm{i}}}$$

(4)

where ${\mathrm{w}\mathrm{e}\mathrm{i}\mathrm{g}\mathrm{h}\mathrm{t}}_{\mathrm{S}-{\mathrm{C}}_{\mathrm{i}}}$ is the weight of the ith sub-criterion. An example of the MCDA criterion calculation can be found in previous work [35].

The economic criterion, expressed as either the first-year levelized cost of hydrogen or the hydrogen production price, is determined from operating data for the reforming unit, or calculated using Equations (1)–(3) (Section 2.1.1). Evaluation of the environmental criterion is discussed in Section 2.1.3.

Table 8 provides calculated values for individual H₂ technologies. Based on economic and environmental criteria, A3 offers the lowest cost but highest emissions, while A1 shows the best environmental performance at moderate cost. Including technological and implementation factors, the results are less distinct, requiring MCDA evaluation in the following step.

3.2. MCDA Results

For four monitored criteria, 16,395 consistent combinations of criteria importance were identified. Data regarding the individual combinations are provided in an excel file as a Supplementary Material to be downloaded. Figure 2a presents a percentage matrix showing how frequently each alternative occupies each rank, with rows for ranking positions and columns for alternatives. Figure 2b displays weighted positions and the influence of each criterion, with total scores summing up to 100%. Weighted positions reflect each alternative’s overall performance based on rank frequency and position weight (1st–4th). Figure 2a shows clear preference for A1 and A3, while ammonia (A2) consistently ranks last. Ammonia cracking remains uncompetitive under current conditions due to low maturity and utilization. SMR (A3) ranks first in 59% of cases (9673 combinations). Electrolysis (A1) dominates the top two positions in 84% of cases with no occurrence at the 4th position. With biomethane reforming placed 1st or 2nd in 86% of cases, the final rankings are nearly identical—33.36% for A1 and 32.37% for A3 (Figure 2b). A1 leads in the technological criterion due to its operational flexibility and high hydrogen quality, whereas A3 benefits from the lowest production cost. Although methane cracking (A4) is economically comparable and produces solid carbon, this does not significantly improve its ranking.

Evaluating all consistent preference combinations enables their objective comparison. However, prioritizing or ordering criteria can significantly affect outcomes. When one criterion is prioritized, consistent weight combinations decrease to 3970. When hydrogen price is prioritized, SMR (A3) ranks first in nearly all scenarios (Figure 3a). Although methane pyrolysis (A4) has the second-lowest cost, it never ranks in first place. Under environmental preference, A4 ranks first in only 3% and second in 60% of cases, while A3 drops to 4th in 44% (Figure 3b). Limiting the impact of hydrogen price reduces the advantage of large-scale production revealing slower start-up and load constraints of the reforming unit (Figure 3c). In the final case (Figure 3d), rankings remain stable across preferences: SMR (A3) consistently leads, while A4, as a newer technology, trails behind. Lastly, setting a preferred order of criteria reduces consistent combinations to 624 for four criteria, creating a personalized yet complex comparison.

Figure 4a shows results for the order economic > technological > implementation > environmental. Despite using feedstock twice as expensive as natural gas, biomethane reforming (A3) still achieves the lowest hydrogen production cost. Electrolysis (A1), performing best technologically, ranks no higher than second (8% of combinations), with hydrogen costing 1.9 times more than steam reforming. While A1 scores 1.5 times higher in technology (Table 8), it cannot offset the dominant economic criterion.

Sensitivity analyses consider ±60% changes in electricity or methane-rich gas prices. While higher methane gas prices favor electrolysis, the largest ranking shift occurs with a 60% drop in electricity price, giving hydrogen costs of 3.96, 9.19, 4.12, and 4.10 EUR/kg H₂ for A1–A4, respectively. Under this scenario, A1 ranks first or second across all combinations (Figure 4b). The impact on the price and ranking of A2 is less significant.

Sensitivity analysis confirms that the electricity price strongly influences the competitiveness of hydrogen production technologies. Therefore, the analysis was extended to evaluate the effect of percentage changes in the price of electricity and methane-rich gases on the percentage score of individual hydrogen alternatives (Figure 5).

When the electricity price decreases by approximately 20%, electrolysis (A1) surpasses steam biomethane reforming (A3) in its overall percentage score (Figure 5a). With decreasing electricity prices, A1 benefits from its strong technological performance, stemming from its operational flexibility, high hydrogen purity, and rapid dynamic response. A similar trend can be observed for methane pyrolysis (A4). As the price of electricity decreases, the production cost of pyrolysis-based hydrogen approaches that of SMR-based hydrogen, leading to an increase in the A4 score. In contrast, the score of steam biomethane reforming (A3) gradually decreases relative to the other alternatives. This indicates that electricity price affects methane pyrolysis more than steam biomethane reforming.

Conversely, when the price of methane-rich gases increases, the score of A3 declines (Figure 5b). For A3 to outperform A1, an approximate 70% increase in methane-rich gas prices is required. The price of hydrogen produced via methane pyrolysis (A4) is less affected by increasing methane prices compared to SMR. However, for A4 to outperform A3, an even larger increase of around 90% is necessary. Up to this level of increase, SMR maintains a competitive advantage due to its high score in the implementation criterion and the second-highest score in the technological criterion as a well-established and mature process.

However, several factors beyond energy prices influence investment decisions, with the European Union playing a key role through policies favoring renewable projects. The RED III directive affects hydrogen classification, and only electrolysis can produce RFNBO, which refineries must supply to avoid regulatory penalties. Meeting RED III requirements for continuous renewable electricity is challenging, which makes the ammonia cracking scenario a potential solution. While ammonia cracking (A2) ranks lowest in techno-economic terms, its ability to contribute to RFNBO adds strategic value. Overall, choosing the company’s direction is a complex process (Figure 6) requiring detailed evaluation and the long-term monitoring of market trends and EU policies, which is beyond the scope of this study.

4. Conclusions

This study compares four hydrogen production pathways for an inland refinery in Central Europe using MCDA. The results show that biomethane reforming (A3) currently offers the lowest production cost, while electrolysis (A1) provides the best environmental and technological performance. Sensitivity analysis confirms that electricity price strongly influences competitiveness of the alternatives. Lower electricity costs significantly improve electrolysis economics and, thus, its performance in the MCDA analysis. However, the final business strategy extends beyond the scope of the MCDA as it is influenced by both market dynamics and evolving regulations. For example, despite ranking lowest in techno-economical performance, ammonia cracking may be required under RED III for RFNBO target. Overall, the company should balance economics with regulatory and sustainability goals.