Forecasting Development of Green Hydrogen Production Technologies Using Component-Based Learning Curves

Abstract

Hydrogen energy is expected to become one of the most efficient ways to decarbonize global energy and transportation systems. Green hydrogen production costs are currently high but are likely to decline due to the economy of scale and learning-by-doing effects. The purpose of this paper is to forecast future green hydrogen costs based on the multicomponent learning curves approach. The study investigates the learning curves for the main components in hydrogen value chains: electrolyzers and renewable energy. Our findings estimate the learning rates in the production of PEM and AE electrolyzers as 4%, which is quite conservative compared to other studies. The estimations of learning rates in renewable energy electricity generation range from 14.28 to 14.44% for solar-based and 7.35 to 9.63% for wind-based production. The estimation of the learning rate in green hydrogen production ranges from 4% to 10.2% due to uncertainty in data about the cost structure. The study finds that government support is needed to accelerate electrolysis technology development and achieve decarbonization goals by 2050.

1. Introduction

Hydrogen might become one of the essential decarbonization tools, especially in sectors where the potential of other decarbonization methods is limited from a technical point of view. According to the International Energy Agency, approximately 40 countries have already approved strategies or roadmaps for developing hydrogen production, storage, transport, and end-use technologies, including the Hydrogen and Fuel Cells Program introduced in the United States in 2004, the Hydrogen Industry Development Plan (2021–2035) adopted in 2022 in China, the “Package for the Future—Hydrogen Strategy” in 2021 in Germany, the “Hydrogen Strategy 2030” in 2020 in Portugal, etc. Despite some differences in plans, all major economies expect hydrogen to play a significant role in decarbonization.

Green hydrogen is obtained from water electrolysis using renewable energy sources and has a minimal carbon footprint. However, today, the share of green hydrogen is insignificant owing to its high cost. Global hydrogen production reached 94 million tons (Mt) in 2021 [1], while low-emission hydrogen production is less than 1 million tons (less than 1%), with almost all of it coming from fossil fuel plants with carbon capture, utilization and storage (CCUS).

Nevertheless, the flow of low-emission hydrogen projects is increasing at an impressive rate [1]. In scenario forecasts for the development of the global energy sector, international analytical agencies assign a significant place to low-carbon hydrogen with approximately 10–20% in the final energy consumption by 2050. The IEA’s zero emissions scenario by 2050 (NZE scenario) assumes that the cost of hydrogen from renewable energy sources will fall to USD 1.3–3.5 per kg by 2030 in regions with good renewable resources and will be comparable to the cost of hydrogen from natural gas with CCUS. In the longer term, costs for hydrogen from renewable electricity will drop to USD 1.0–3.0/kg in some country scenarios, making hydrogen from photovoltaic solar cells cost-competitive with hydrogen from natural gas, even without CCUS in these regions [2].

With global events driving up oil prices and production subsidies, as well as the increase in gas prices in 2022, renewable hydrogen could be the cheapest hydrogen production option in many regions today if production capacity is available [1]. A three- to fourfold increase in hydrogen demand projected over the next decade requires a 30 percent annual growth in global electrolyzer installed capacity until 2050. Although very small, the amount of hydrogen produced by water electrolysis has increased by almost 20% compared with 2020. This reflects the increasing adoption of water electrolysis systems—the cost of every component of the hydrogen supply chain is decreasing. The cost of electrolyzers could easily fall by 40% in a few years, whereas some electrolyzer manufacturers predict an even faster decrease. The price of the renewable electricity required to produce green hydrogen is constantly decreasing as well. For example, the costs of solar utilities fell by 85% between 2010 and 2020 [3]. Thus, green hydrogen has the most significant potential for cost reduction as the technology evolves and scales. However, the estimates of the future cost of different components in the green hydrogen supply chain vary significantly in different sources [4,5,6,7,8], making it very difficult to predict the final cost of hydrogen.

The purpose of this paper is a forecast of future green hydrogen costs based on the learning curves approach. The research hypothesis is based on scientific data that suggest that the basis for the future development of green energy is hydrogen energy based on renewable energy sources (RES). As the total installed hydrogen production capacity increases, experience accumulates in both the construction of production facilities and hydrogen production. Growing experience leads to impressive cost reductions. The observed relationship between these cost savings and experience gained during deployment is commonly referred to as “learning by doing” [9,10,11]. Therefore, the objective of this study is to calculate component-by-component learning curves to predict the prospects for reducing the cost of hydrogen production.

The total hydrogen input can be divided into four main components: production, distribution, storage, and end-use costs, such as fuel cells. In this study, we focus on the first component and present an analysis of hydrogen production cost reduction using the learning curves approach. This observed cost reduction may be of interest in assessing the feasibility of realizing hydrogen cost reductions in the future and pointing to the viability of advancing the hydrogen economy for sustainable development.

The rest of the paper is organized as follows: Section 2 reviews the existing literature on green hydrogen production; Section 3 describes the basics of the learning curve methodology and its different applications for energy technologies. Section 4 is devoted to a meta-analysis of data on electrolyzer capacities, renewable energy cost, and green hydrogen cost in different locations and calculating learning rates. Section 5 briefly discusses our estimations of the learning rates compared to other studies and describes the study’s limitations and ways for future improvements. Section 6 concludes the research and offers some policy applications.

2. Literature Review

Green hydrogen production is critical for low-carbon energy development in the current global energy strategy. The main direction of technological change is to move away from the consumption of traditional fossil energy in the production of hydrogen or «carbon-free green hydrogen» [12].

Today, the use of hydrogen as an alternative energy source has several reasons, such as climate change and the depletion of fossil fuels. For the production of hydrogen, both traditional and alternative energy sources are used. However, the future development of modern hydrogen production technologies not only includes significant innovations for the storage, transportation and utilization of H₂, but, above all, suggests the possibility of using green hydrogen. This will help reduce a given country’s dependence on fossil fuel imports [13].

Green renewable hydrogen is produced using water electrolysis technology and electricity from renewable energy sources such as solar or wind energy. This is a critical component in accelerating the transition to clean energy. Electrolytic hydrogen is considered green when electricity is produced from renewable energy sources. The wide spread of electrolyzers could make green hydrogen more affordable and enable energy systems worldwide to undergo fundamental transformations to reduce emissions and their negative environmental impact. Simultaneously, electrolyzers can help integrate renewable electricity into power grids, as their electricity consumption can be adjusted to match wind and solar power generation [14].

The most important question in the development of the hydrogen industry is whether the price of green hydrogen produced from renewable energy sources can be attractive compared to fossil-fuel-derived hydrogen. In the production of hydrogen, the main components are the cost of electricity and electrolyzers, as well as the capacity utilization factor. According to the latest statistics, we see a rapid drop in the cost of electricity from photovoltaic solar panels and wind, as well as a decrease in the cost of electrolyzers [15].

The main cost component in producing green hydrogen is the cost of the renewable electricity required to power the cell. The second most significant factor is the cost of the electrolysis plants [16]. Hydrogen production costs are influenced not only by the chosen technology and external factors but also by the capital cost of the cell and the price of electricity. The plant power factor and longevity are the essential parameters affecting production costs [17]. Regarding sustainability and environmental impact, polymer electrolyte membrane (PEM) water electrolysis is considered the most promising method for high-purity, efficient hydrogen production from renewable energy sources and emits only oxygen as a byproduct without any carbon emissions [18].

A sensitivity analysis of the prices of various components showed that the price of the water electrolyzer has the most significant impact on the cost of producing green hydrogen [19]. Despite the different characteristics, the overall specific energy requirements of various alkaline electrolyzers (AE, AEL), PEM and solid oxide electrolyzer cell (SOEC) water electrolysis technologies are mainly similar. The choice of the type of electrolysis technology depends on external criteria, particularly location. AEL is the most suitable technology for low-carbon hydrogen production with unsustainable renewables owing to its lower investment costs and lower specific energy consumption. With reduced production costs, PEM will become a competitive alternative. Once SOEC becomes commercially available, it may become an attractive option in combination with renewable energy. Nevertheless, there is a need to improve all technologies further and expand the use of renewable energy sources [20]. The results of the research show that PEM electrolysis in the sea represents the best prospects for applications in the short term [21].

The capital expenditure on electrolyzers (CAPEX) is expected to increase from EUR 400/kWh to EUR 240/kWh by 2030 and EUR 80/kWh by 2050 [4]. The sensitivity analysis results show that the electrolyzers with solid oxide electrolysis coupled with waste heat sources (SOEC W.H) have the most competitive option in lowering capital costs and subsidizing electricity costs and tax rates to compete with grey hydrogen [22]. Membraneless electrolyzers have a hydrogen levelized cost of approximately USD 2/kg H₂, competing with steam reforming and showing much higher profitability than traditional electrolyzer technologies [5].

For growing global energy needs, the production of green hydrogen is an important solution since green hydrogen can be produced from constant solar energy and available water. Considering the known electrolysis technologies for solar hydrogen energy, the most suitable for integration with a photovoltaic system are alkaline water electrolyzers, which are recognized as the most mature today. Technologies for the production of solar hydrogen are quite advanced, but research shows that they need to be further improved so that they can become a significant part of the energy system [23]. Ceran [6] presented the results of a multi-variate comparative analysis of scenarios for the annual production of one million tons of pure hydrogen by electrolysis in Poland. The scenario in which hydrogen production uses only photovoltaic energy is the best option for the weighting factor of the environmental criterion at the level. The efficiency of solar-to-hydrogen photovoltaic electrolysis (PV-E) plants is a critical parameter in reducing the cost of green hydrogen production. These results demonstrate the potential of CPV technology for large-scale production of green hydrogen as a replacement for fossil fuels [21].

Offshore wind-generated green hydrogen is emerging as a promising solution for removing barriers to a carbon-free economy in Europe and beyond. Denmark built its first artificial offshore energy hub. Research has shown that offshore green hydrogen production can reduce costs by up to EUR 2.4 per kg, which is competitive with the costs of hydrogen currently produced from natural gas. In addition, the cost of wind power is reduced when the cell is installed offshore, which reduces peak loads [8].

The researchers note that wind–solar hybridization can reduce hydrogen production costs by a few percent when the effect of increasing the electrolyzer load factor outweighs the increase in electricity costs [24]. The energy efficiency of photovoltaic electrolysis reduces the cost of green hydrogen production. In addition, the combination of solar and wind power ensures better offline performance [25,26].

The potential of producing and using green hydrogen varies across countries. This depends on many factors, such as the availability of technology and competences; however, climate and geographical location play a significant role. For example, green hydrogen could play a significant role in the decarbonization of Mediterranean countries. Infrastructure and value chain development should be oriented towards the uses and country-specific differences in this context. Structural differences between EU countries open up the potential for a new European division of labor within a standard hydrogen network, both in production and consumption [27].

Studies by Chinese scientists show that the energy efficiency of hydrogen production from wind energy is much higher than the efficiency of hydrogen production from solar energy, and the potential for green hydrogen production in the northwest and north of China is much higher than in other regions [28].

With the launch of the National Hydrogen Mission in India, the industry transition to clean hydrogen seems possible soon, given the great opportunities for using solar energy [29]. India’s ever-increasing renewable energy capacity offers it an edge in producing hydrogen from clean sources such as solar and wind during periods of reduced demand [30]. There is a high potential for green hydrogen production using RES in Malaysia [31] and the Philippines [32]. Research has shown that solar-powered green hydrogen production in Oman is promising [19]. Abundant solar energy as a key renewable energy source estimates Turkey’s total hydrogen production potential to be 415.48–427.22 million tons (Mt), depending on the cell type. The results showed that some regions of the country have the highest potential for hydrogen production [33].

The transformation of the German energy system to fully renewable energy is expected to include the widespread use of green hydrogen [34]. Brazil could become a source of green hydrogen for the domestic market and potential exports to Germany and other European countries [35]. For the Russian economy, there are several combinations of using solar and wind energy to produce green hydrogen [36]. Qatar’s green hydrogen is expected to become as competitive as blue hydrogen in the future [37].

Modelling by Glenk, G. and Reichelstein, S. in Germany and Texas showed that renewable hydrogen is already cost-competitive in niche applications (EUR 3.23/kg), although not yet for industrial-scale supply. This conclusion, however, is projected to change over the course of a decade (EUR 2.50/kg), assuming that recent market trends will continue in the coming years [38].

A study at the regional level [39] showed that out of 109 hydrogen production regions (27 EU countries and the UK), 88 regions (81%) demonstrated RES production that exceeded the required annual electricity energy consumption in all industries and hydrogen production. It should be noted that 84 regions had more than half of the excess RES potential after taking into account all the energy required for the population, industry and water electrolysis.

The higher cost of green hydrogen compared to its competitors is the most critical barrier to its wider use. Although the cost of renewable electricity is considered a significant hurdle, the problems associated with electrolyzers are another major issue that has important implications for reducing the cost of green hydrogen [40].

In a study by Schoots et al. [41], three hydrogen production methods were considered: SMR, coal gasification, and water electrolysis. The construction of learning curves showed that to achieve the cost targets set for electrolysis, it is necessary to reduce both investment and electricity costs. A gradual transition from alkaline electrolyzers to PEM electrolyzers has also been predicted because the latter’s cost can be easily reduced to the desired level. Lane et al. [42] analyzed a learning curve methodology with uncertainty through Monte Carlo simulations to predict the market share of competing renewable hydrogen technologies up to 2050. The study results showed that higher learning rates and a long-term downward trend in renewable electricity prices led to an equal share of new installations in medium-term electrolyzers. The costs for electricity, biomass and upfront costs for gasification technology are the three main factors predicted to affect the structure of the renewable hydrogen market in 2050.

Detz et al. [43] analyzed the technologies required for the production of renewable fuels using learning curves associated with the individual components of the system to predict a possible reduction in investment costs and a reduction in fuel production costs. The results showed that competitiveness could be achieved between 2025 and 2048 for renewable energy technologies such as hydrogen, syngas, methanol and diesel. H₂ production via electrolysis and diesel production via the Fischer–Tropsch synthesis breaks costs up to 2050, even in a conservative base case, using solid oxide electrolysis, which has the advantage of rapid cost reduction and high efficiency.

In the cost of green hydrogen, the cost share of electrolyzers will be the dominant factor compared to renewable electricity supplies. The authors emphasize that the cost of green hydrogen will depend on the ability to reduce the main components of capital costs, as well as the availability of renewable electricity for hydrogen production, which in turn affects the power factor and the present value of hydrogen [44]. By 2050, hydrogen is expected to provide approximately two percent of the world’s energy needs [45].

In addition, most researchers have noted the role of international organizations and government support in developing green hydrogen production technologies. Lack of government support slows green hydrogen technological development and leads to a necessity for more than double the capital expenditure compared to 2021. This makes hydrogen production uneconomical even in the long term (after 2050). Simultaneously, the primary measure of state support is subsidizing electrolysis technologies [16]. Although many governments and private companies are investing significant resources in the development of hydrogen technologies, there are still many unresolved issues, including technical problems and economic and geopolitical consequences. The hydrogen supply chain includes many stages which lead to additional energy losses. Although much attention is paid to hydrogen production costs, its transportation and storage cannot be neglected [46]. Thus, in cooperation with the industry, politicians and government authorities are responsible for making this transition happen by adopting necessary legal and tax regulations. In the long term, supply security, job security and affordability can only be guaranteed by moving towards a sustainable economy [47,48].

Together, these studies provide important insights into the process of green hydrogen production and its expected improvements that can lead to the reduction in the final cost. However, as we have seen, the experimental data are rather controversial, and there is no general agreement about the expected rates of future cost reduction.

3. Methodology and Data

3.1. Meta-Analysis of Data

In this study, we performed a statistical meta-analysis of these data. Meta-analysis is a widely used method for forecasting in the case of a limited possibility of obtaining reliable statistical data for modelling. In the first stage, we examined most previously published studies on the cost of hydrogen production. In the second stage, we compared the results obtained by other authors to identify common trends in all studies. In the next step, we combined the data by converting them for our estimation.

3.2. Data and Assumptions

The capital costs associated with electrolyzers, which can be found in scientific articles, reports and energy reviews, are measured in different currencies and normalized by different years. Sometimes, these values include some other system components as well. Due to the revealed uncertainties, we agreed with [17,38] that the CAPEX parameters are easy to compare and understand. Glenk et al. [38] distinguished between the completeness and transparency of the methodology. Only original data sources were included in the review, with precise cost estimates and methodologies for deriving cost estimates. Capital cost calculations for electrolyzers have been updated in the Assessment of Hydrogen Production Costs from Electrolysis: United States and Europe [49].

Table 1. Initial data for calculations.

Data	Year	Units	Original Source
Installed electrolysis capacity	2015–2020	MW	[50]
Electrolyzer CAPEX Costs	2015–2020	(USD 2020/kW)	[38,49,51]
Electricity Generation Solar PV	2011–2020	GWh	[52]
Electricity Generation Wind Total (Onshore, Offshore)	2011–2020	GWh	[53]
Unsubsidized Solar PV LCOE	2011–2020	USD/MWh	[54]
Unsubsidized Wind LCOE	2011–2020	USD/MWh	[54]

Source: compiled by the authors.

presents the data from this source but includes other sources for completeness.

As one can see from Table 1, information from several sources was considered, including data from scientific publications, reviews, reports and websites of official energy associations (IRENA, Energydata, IEA Hydrogen, etc.). The cost of electricity is measured as the levelized cost of electricity (LCOE), which is calculated as the average estimated cost of electricity generation over the entire life cycle of the power plant. This study assumes that water electrolysis using AE and PEM electrolyzers is used for green hydrogen production. Calculations were not performed for the SOEC because we could not obtain sufficiently reliable annual installation data for this technology and determine the evolution of its application over time.

3.3. Learning Curves

Learning curves methodology is based on the fundamental laws of the learning theory that as the cumulative (total) volume of production increases, the cost of production decreases owing to the accumulation of production experience (learning-by-doing effect). The learning curves express the hypothesis that the technology costs decrease by a constant fraction (%) for each doubling of the installed capacity. Therefore, on a log–log scale, the ratio between these technology costs and total production or production is a downward-sloping straight line [9,10,11,55]. Often, instead of the cost of production, indicators such as the price per unit of production are considered, because under normal conditions, it is proportional to the cost. The cost can also be called unit costs (per unit of output) or unit costs.

Mathematically, the basic law of learning theory is described by the following formula:

$${\displaystyle \begin{array}{c}C\left(X_t\right)=C_0{\left(\frac{X_t}{X_o}\right)}^{-r},\\ LR=1-2^{-r},\end{array}}$$

(1)

where

• C₀—initial unit costs (costs per unit of output);
• C(X_t)—unit costs at time t;
• X_t—cumulative (for the entire period) volume of production;
• LR—estimated rate of learning-by-doing in the manufacturing process.

LR (learning rate) is usually expressed as a percentage and shows how much the cost of production decreases with each doubling of the cumulative production volume. By calculating the learning rate according to the available statistical data, it is possible to predict how the price/cost of production changes further with an increase in production volumes.

In this study, we used a component-by-component learning curve methodology, that is, based on the available data on the production and cost of AE and PEM and the production and cost of wind and solar energy. Learning curves were built, and learning rate estimates were obtained separately for each component (LR_i). The mathematical model of the multicomponent learning curve can be represented as follows [11,55]:

$$C\left(X_t\right)={\displaystyle \sum }_i{C}_{0i}{\left(\frac{X_{ti}}{X_{0i}}\right)}^{-r_i},$$

(2)

where index i refers to each component in the supply chain. In simplified form (relating all individual learning curves to the cumulative productions of the overall system only), Equation (2) can be rewritten as

$$C\left(X_t\right)={\displaystyle \sum }_i{C}_{0i}{\left(\frac{X_{ti}}{X_0}\right)}^{-r_i}.$$

(3)

Although this model excludes some factors, such as spillover effects between components, Böhm et al. [56] prove that Equation (3) is still adequate and more practical for early learning rate estimations at a component level. If a system has two main cost-intensive components, then Equation (3) can be rewritten as follows:

$$C\left(X_t\right)={\displaystyle \sum }_i{\alpha }_i{C}_0{\left(\frac{X_{ti}}{X_0}\right)}^{-r_i}=C_0\left({\alpha }_1{\left(\frac{X_{t1}}{X_0}\right)}^{-r1}+{\alpha }_2{\left(\frac{X_{t2}}{X_0}\right)}^{-r2}\right),$$

(4)

where α_i is a share of the ith component in the total cost of the system. Assuming that X_ti—cumulative (for the entire period) volume of production of each component keeps the same share in the total production of the system, we can present Equation (4) as follows:

$$C\left(X_t\right)={\displaystyle \sum }_i{\alpha }_i{C}_0{\left(\frac{X_t}{X_0}\right)}^{-r_i}=C_0\left({\alpha }_1{\left(\frac{X_t}{X_0}\right)}^{-r1}+{\alpha }_2{\left(\frac{X_t}{X_0}\right)}^{-r2}\right).$$

(5)

Considering that the ratio $\frac{X_t}{X_0}$ is equal to 2 (cumulative production doubles), we can calculate the ratio $\frac{C\left(X_t\right)}{C_0}$ and find the rate of the progress (1-LR) of the entire supply chain.

Finally, if the system contains not only the components which involve learning (e.g., the production process) but also those that do not (e.g., labor costs and material costs), the formula for the cost of the product at the time t can be written as follows:

$$C\left(X_t\right)={\displaystyle \sum }_i{\alpha }_i{C}_0{\left(\frac{X_t}{X_0}\right)}^{-r_i}=C_0\left({\alpha }_1{\left(\frac{X_t}{X_0}\right)}^{-r1}+{\alpha }_2{\left(\frac{X_t}{X_0}\right)}^{-r2}+{\alpha }_3\right),$$

(6)

where α₃ is the share of the component, which does not involve learning in the initial cost of the product.

4. Results

4.1. Green Hydrogen Production Technologies

Today, a relatively large number of hydrogen production methods are known, including electrolysis, reforming, gasification, and biomass conversion. Most of the hydrogen is produced by steam reforming of methane (SMR) of natural gas or through coal gasification processes. Only insignificant amounts are currently produced with low carbon emissions.

According to the IEA, in 2021, there were 14 operating plants to produce low-carbon hydrogen from hydrocarbons worldwide and 40 similar projects under development, of which four are under construction (in China and the United States), 35 of which are planned to use natural gas, and 19 which are in Europe, primarily in the Netherlands and the United Kingdom. One of the most significant projects, H21 North of England, is planned to produce approximately 3 million tons of hydrogen and bury about 20 million tons of CO₂ annually by 2035. To achieve climate goals in the Net Zero scenario, according to the IEA, by 2030, it is necessary to build about 230 plants to produce blue hydrogen with CCUS (or retrofit CCUS to existing facilities to produce “grey” hydrogen).

Several potential technologies can produce very-low- or no-carbon hydrogen, but many are still in the early stages of development or face inherent disadvantages [57]. Currently, electrolysis is the most common technology used for producing green hydrogen. One promising area of research is the production of hydrogen from biomass. This may be a means to achieve harmful CO₂ emissions, but it is unlikely to play an essential role because of the overall limited sustainable biomass resources [58].

Table 2. Hydrogen production technologies by importance for achieving net-zero emissions and readiness levels.

Production Technology	Importance for Net-Zero Emissions	Readiness Level (TRL)
Electrolysis > Electrolyzer design > Polymer electrolyte membrane	Very high	9
Electrolysis > Electrolyzer design > Alkaline	Very high	9
Partial oxidation with CCUS	High	6
Methane reforming > Autothermal reforming with CCUS > With gas heat-reformed	High	5
Methane reforming > Autothermal reforming with CCUS > Single reformer	High	5
Methane reforming > Steam reforming with CCUS > High capture rates	High	5
Thermochemical water splitting > Nuclear	Moderate	3
Electrolysis > Seawater electrolysis	Moderate	3
Electrolysis > Electrolyzer design > Anion exchange membrane electrolyzer	Moderate	6
Electrolysis > Electrolyzer design > Solid oxide electrolyzer cell	Moderate	7

Source: Compiled according to [59].

presents technologies for hydrogen production with high and moderate impacts on net-zero emissions with an assessment of their level of readiness.

According to the IEA, electrolysis using PEM and AE is the most important step in the production of low-carbon hydrogen. It should be noted that these two types of electrolyzers are the most mature.

Electrolysis requires more electricity than other technologies. Therefore, it is the most sensitive to the source of the electricity consumed. The carbon intensity of the hydrogen produced by electrolysis depends on the carbon intensity of the electricity used in operation and the carbon intensity of the production process of the electrolyzer. It could become zero if all electricity used comes from zero-carbon sources.

The cost of clean hydrogen depends mainly on two factors: the cost of zero-carbon electricity and the capital cost of the electrolyzers. Electricity is approximately 30–60% of the cost of producing hydrogen from electrolyzers with a capacity of more than 20 MW, so LCOH is highly dependent on the cost of available sources of electricity. Next, the cost of the electrolyzer is expected to decrease because of the rapid growth of industry and technological progress [54].

There are four types of water electrolysis technologies that are used to create green hydrogen depending on the electrolyzer used. First is alkaline electrolysis (AE), which is by far the most developed and commercialized process. PEM electrolysis is the next most mature process with increasing commercialization; it has advantages over alkaline electrolysis due to its smaller footprint, ability to follow load due to faster start-up and system response, lower minimum load requirements and greater load flexibility. There are also solid oxide electrolysis with anion exchange membranes (AEM), which are in the pilot/development phase. These processes are not expected to be widely commercialized until the mid-2020s, and solid oxide electrolyzers (SOEC) are still in the demonstration phase, but, according to some estimates, are claimed to be a breakthrough in the industry. In recent years, PEM electrolyzers have been leading in new inputs compared to AE electrolyzers.

We analyzed the projects included in the IEA database [60]. The database includes all projects to produce hydrogen that have been commissioned since 2000, as well as those that are planned or in the process of commissioning. The projects are classified by production technology (electrolysis, fossil fuels with carbon capture, utilization and storage, and other technologies), hydrogen-based fuel produced (hydrogen, methanol, ammonia, methane, and synthetic hydrocarbons) and use of the resulting fuel. The analysis revealed that electrolysis is the most popular technology for producing low-carbon hydrogen is electrolysis. In total, 1467 projects use electrolysis, of which 1302 are already in operation. The projects use various types of electrolyzers (

Table 3. Electrolyzers Used in Hydrogen Generation Projects.

Electrolyzer Type	Number of Projects
Alkaline electrolyzer	192
PEM	290
SOEK	38
Other Electrolysis	790
Total	1310

Source: [60].

).

Electrolyzer power has shown constant growth in recent years. At the same time, SOEC electrolyzers are in the initial path, and PEM still lags alkaline electrolyzers in terms of power (Figure 1).

Although alkaline electrolyzers have a more extended history of use, like PEM electrolyzers, they are at a high technology readiness stage (TRL9) and have already been commercialized. SOEC electrolysis is currently a demonstration technology (TRL7). Anion exchange membrane (AEM) electrolyzers are in the early stages of development and are being assessed by the International Energy Agency (IEA) as a complete prototype to scale (TRL6) [61]. By 2023, the electrolysis capacity could grow ten times if the projects are implemented on time. If all projects under development are implemented, the global electrolysis capacity can reach 134–240 GW by 2030 [61].

In total, 652 projects in the IEA database [60] use renewables: 29 use or plan to use hydropower, 138—solar PV, 101—offshore wind, and 70—onshore wind. Among the 176 projects active by 2021, 58 are related to renewable energy sources. Hydropower is used by 7 projects, solar PV—20 and onshore wind—22 [60].

If the electrolyzer projects under development are implemented and a planned increase in production capacity occurs, by 2030, cell costs could be reduced by approximately 70% compared to now. Combined with the expected reduction in the cost of renewable energy, this could reduce the cost of renewable hydrogen to USD 1.3–4.5/kg H₂ (equivalent to USD 39–135/MWh). The lower end of this range lies in the regions with good access to renewables, where renewable hydrogen may already be structurally competitive with fossil fuels [1]. Promising prospects in this direction are attributed to the countries with good natural climatic conditions, such as Chile, India, Turkey, Mexico, South Korea, Vietnam, etc., which have high rates of solar and wind energy production.

Electrolyzer costs are currently ~33–45% of the total capital costs, and electricity costs are approximately ~30–60% of the present value of green hydrogen. Policies (such as carbon prices or incentives) should also influence the relative cost of green hydrogen compared with fossil fuel alternatives. The competitiveness of green hydrogen in the future will also depend on the interaction between end use and proximity of production, which in turn determines the transportation and storage costs associated with this application [62]. Notably, the cost of electrolyzers fell by 40% between 2015 and 2019, and the cost of producing green hydrogen decreased by 50% over the same period. The cost of electrolyzers is expected to be halved by 2030, owing to the increased scale and standardization of production [63].

Producers worldwide aim to increase their production capacity to more than 100 GW of electrolyzers annually by 2030—up from today’s 2 GW. These new facilities continue to drive competition and economies of scale in the industry, resulting in significant cost reductions [64].

To date, RES-powered electrolysis is still the most expensive technology for hydrogen production—up to three times more expensive than steam reforming of methane, and the goal of all national hydrogen programs is a sharp reduction in cost. At the same time, the almost-zero carbon footprint and the lack of need to combine electrolysis with CCUS technology are essential advantages.

4.2. Estimating Learning Rates Using Component-Based Learning Curves

It was mentioned earlier that our analysis is based on the opinion of most researchers and experts that “green” hydrogen will become the primary fuel and the main energy vector for a future 100% sustainable energy and industrial system. IEA expects that low-carbon hydrogen production will continue to grow as shown in Figure 2.

In hydrogen production by electrolysis based on RES, the central part of the expenditure is electricity (55 kWh/kg of hydrogen) and the cost of electrolyzers, which is high because of insufficient scaling. With a decrease in the cost of renewable energy (most frequently wind and solar), the cost of hydrogen production will decrease.

The analysis of the dynamic of the cost of unsubsidized solar and wind energy was based on open data sources (as opposed to confidential company data) (Figure 3). According to a report by the United Nations and Bloomberg (BNEF), the average “present value” or total cost of producing solar energy worldwide before the pandemic fell by 17%, the costs for onshore wind energy fell by 18%, and the costs for offshore wind wing energy fell by 28% [66].

For learning rates calculation, we used data on unsubsidized LCOE of wind and solar energy in 2011 and 2020 [54] and Electricity Generation Solar PV and Wind [52,53] (Figure 4 and Figure 5).

Because the LCOE of wind and solar energy varies by country and location, we used three values—minimum, median and maximum. Therefore, the learning rate estimates were obtained in three variants (

Table 4. Learning rates estimation (LR) for wind and solar energy.

Wind Energy			Solar Energy
Lower Limit	Median	Upper Limit	Lower Limit	Median	Upper Limit
7.35%	8.01%	9.63%	14.28%	14.35%	14.44%

Source: compiled by the authors.

).

The obtained estimates of learning rates show that doubling the production will decrease the cost of wind energy by an average of 8.01%, ranging from 7.35% to 9.63%. Regarding solar energy production, the learning rate is almost two times higher (14.35%) with a range from 14.28% to 14.44%, which is narrower than wind energy costs.

The assessments of the electrolyzers’ cost in the literature also vary greatly. Recent IEA report [2] asserts that electrolyzer installations cost USD 1000–1750 per kilowatt (2020). At the same time, the IEA indicates (referring, however, to the Hydrogen Council [67]) that by 2030, the cost will fall sharply in all scenarios and will be somewhere in the range of USD 400–600, which is close to the Fraunhofer ISE estimate [51]. An important assumption made in this work is the large scatter of data in the evaluation of the cost of electrolyzers found in various sources. Therefore, we considered the cost of electrolyzers as the second component for assessing the learning rate, which, according to all forecasts, will decrease in the future. It should be noted that AE pots are more common than PEM, but the future still belongs to PEM pots, as they are less sensitive to loading, do not require a stable load profile, allow fluctuations, and, in principle, allow operation in a variable load mode. Calculations were performed for both types of electrolyzers based on CAPEX costs in 2015 and 2020 [38,49,51]. The cumulative production volume is calculated from data of installed electrolyze capacity [50] (Figure 6 and Figure 7). The results are presented in

Table 5. Learning rates estimations (LR) for electrolyzers.

Electrolyzer AE	Electrolyzer PEM
4%	4%

Source: compiled by the authors.

.

The resulting calculations show that doubling the electrolyzer production for both types of AE and PEM will decrease their CAPEX by 4% on average. It should be noted that the learning rate of the costs of water electrolysis equipment has large margins of error but is within the learning range reported in the literature for other technologies in the energy sector.

According to IRENA data [62], the share of the cost of energy in the final cost of green hydrogen range from 30% to 60%, the share of operation and maintenance is about 10%, and the rest is the share of the electrolyzer. Therefore, we can calculate the future cost of green hydrogen at time t following two scenarios:

Scenario I (the share of energy in the final cost is 60%):

$$C\left(X_t\right)={\displaystyle \sum }_i{\alpha }_i{C}_0{\left(\frac{X_t}{X_0}\right)}^{-r_i}=C_0\left(0.6\ast {\left(\frac{X_t}{X_0}\right)}^{-r1}+0.3{\left(\frac{X_t}{X_0}\right)}^{-r2}+0.1\right).$$

(7)

Scenario II (the share of energy in the final cost is 30%):

$$C\left(X_t\right)={\displaystyle \sum }_i{\alpha }_i{C}_0{\left(\frac{X_t}{X_0}\right)}^{-r_i}=C_0\left(0.3\ast {\left(\frac{X_t}{X_0}\right)}^{-r1}+0.6\ast {\left(\frac{X_t}{X_0}\right)}^{-r2}+0.1\right).$$

(8)

Next, each scenario can be divided into two scenarios according to the type of green energy with different learning rates:

Scenario I (the share of wind energy in the final cost is 60%):

$$C\left(X_t\right)={\displaystyle \sum }_i{\alpha }_i{C}_0{\left(\frac{X_t}{X_0}\right)}^{-r_i}=C_0\left(0.6\ast {\left(\frac{X_t}{X_0}\right)}^{-0.122}+0.3{\left(\frac{X_t}{X_0}\right)}^{-0.06}+0.1\right).$$

(9)

Scenario II (the share of solar energy in the final cost is 60%):

$$C\left(X_t\right)={\displaystyle \sum }_i{\alpha }_i{C}_0{\left(\frac{X_t}{X_0}\right)}^{-r_i}=C_0\left(0.6\ast {\left(\frac{X_t}{X_0}\right)}^{-0.223}+0.3{\left(\frac{X_t}{X_0}\right)}^{-0.06}+0.1\right).$$

(10)

Scenario III (the share of wind energy in the final cost is 30%):

$$C\left(X_t\right)={\displaystyle \sum }_i{\alpha }_i{C}_0{\left(\frac{X_t}{X_0}\right)}^{-r_i}=C_0\left(0.3\ast {\left(\frac{X_t}{X_0}\right)}^{-0.122}+0.6\ast {\left(\frac{X_t}{X_0}\right)}^{-0.06}+0.1\right).$$

(11)

Scenario IV (the share of solar energy in the final cost is 30%):

$$C\left(X_t\right)={\displaystyle \sum }_i{\alpha }_i{C}_0{\left(\frac{X_t}{X_0}\right)}^{-r_i}=C_0\left(0.3\ast {\left(\frac{X_t}{X_0}\right)}^{-0.223}+0.6\ast {\left(\frac{X_t}{X_0}\right)}^{-0.06}+0.1\right).$$

(12)

Moreover, the four presented scenarios can be divided further into sub-scenarios corresponding to the lowest and highest estimations for learning rates for wind and solar energy. Then, assuming that $\frac{X_t}{X_0}$ = 2 (twofold increase in green hydrogen production), we can calculate the learning rates for green hydrogen production by electrolysis. The results are summarized in

Table 6. Learning rates in green hydrogen production by electrolysis.

	Share of Energy by Type	Low Limit	Median	Upper Limit
Share of energy 30%	Wind Solar	4% 7%	5% 7%	5% 8%
Share of energy 60%	Wind Solar	6% 10%	6% 10.1%	7% 10.2%

Source: compiled by the authors.

.

By the data obtained, the progress in production based on solar energy is developing faster: according to Scenario II where the share of the cost of the electrolyzer is lower, the learning rate is slightly higher than 10%; in Scenario IV where the cost of the electrolyzer is higher, the maximum learning rate is 8%. The situation is similar to that of wind energy. Therefore, we can conclude that the cost of hydrogen will decrease more rapidly in regions with suitable climatic conditions for developing solar energy.

5. Discussion

Our estimate is quite conservative as the electrolyzers market is still at an early stage of development: insufficient commercialization, insufficient number of suppliers, and limited capacity, and our results show insufficiently fast learning rates. According to NOW GmbH, measures to stimulate the demand for electrolyzers are much more important than investment in R&D and demonstration projects [68].

More optimistic estimates were provided by Rethink Energy which predict a learning rate for electrolyzer modules of 11%. A similar assessment of electrolyzers was also presented in the Hydrogen Council report [68]. Our estimate is more conservative, most likely due to the later period of the study. The score for the electricity component is also higher than that for electrolyzers, which is in line with the studies presented by other energy companies.

Our study shows that the cost of the production of green hydrogen can potentially be reduced. Electrolyzers have great potential, especially given the increasing commercialization of electrolyzers with proton membranes and the new level of readiness for solid oxide electrolyzers. As studies have shown, renewable energy will contribute even more to the cost of green hydrogen. This is especially true for solar energy because of its price reduction. Obviously, it must be considered that for some countries, this component will have a more significant contribution. This applies to countries that, by their geographical location, are in more favourable climatic conditions and have not yet used this potential. As for wind energy, in this case, the learning rate was somewhat lower, although still quite high (approximately 8%). To obtain higher learning rates, a higher level of incentives and government support is needed to attract large manufacturers [69]. Of course, our assessment is an average and may vary depending on the region and country, geography, climate and government policy.

Considering the limitation of this research, we can note that in this paper, the estimates were based on the fact that a significant share of the cost of hydrogen production is the electrolyzer and electricity. In our work, we carried out a meta-analysis of the data since the cost estimate for electrolyzers varies considerably among different sources. The learning rate for the electrolyzer component was determined for AE and PEM as they are at a high stage of readiness and already have a high distribution. SOEC electrolyzers have excellent prospects as they become commercialized, but they cannot yet be evaluated owing to the lack of reliable data.

The uncertainty of cost projections based on obtained estimations is still very high, but it can be lowered by continuously updating available data. New datasets generated by the projects that are realized during the upcoming years can substantially improve the reliability of these learning curve projections for the longer term.

Further research into the prospects for green hydrogen production will include not only production but also costs for transportation, conversion, infrastructure and end-use upgrades.

6. Conclusions

Green hydrogen is not yet cost-competitive compared with the conventional fuels it will replace. Inequality will decrease as the cost of renewable energy and electrolyzers decrease through improved technology, scaling and further commercialization. The study showed that the learning rates in the production of the main components in the final cost of green hydrogen are not sufficient enough and can be higher. In our opinion, state support can play a significant role in attracting investments in R&D for the development of the market for the main components in the green hydrogen supply chain. The involvement and interest of large businesses can have a significant impact on learning rates through the introduction of new, improved manufacturing technologies. Green hydrogen production has high potential, from both an economic point of view and considering the Net Zero policy pursued in most countries of the world.