A unified European hydrogen infrastructure planning to support the rapid scale-up of hydrogen production – Nature Communications

Balmorel energy system optimization model

Balmorel is an open-source⁴⁰, deterministic, partial equilibrium model for optimizing an energy system assuming perfect markets and economic rationality²⁷. Similarly to other energy system models, it builds on a bottom-up approach and computes the least-cost solution for the energy system to satisfy various energy demands.

Balmorel is a technology-rich energy system model with a comprehensive representation of energy technologies and infrastructures. Energy sources are converted to energy vectors, which through transmission can be used to satisfy demands or be used by conversion technologies in different energy sectors. In parallel, it optimizes both investment planning and operational dispatch.

The modeling framework has been developed extensively by an open-source community since its first release in 2001⁴⁰. The mathematical formulation and results have recently been compared against four other well-known open-source energy system models^41,42, with conclusions emphasizing the model’s validity.

The model has been used to evaluate various energy transition scenarios throughout the years and is being developed to enable holistic energy system evaluations. It was also used to perform deep-dive investigations of certain components of the energy system on various geographical scales and scopes. For example, Balmorel was applied to analyze the role of hydrogen in the future North European power system in 2060⁴³. It was applied to provide decarbonization pathways for the Northern European integrated power and district heating system⁴⁴ with a focus on the role of renewable gas. Recently, it was utilized for assessing the future opportunities for offshore hydrogen production in Northern Europe¹⁹ or diving deeper into the production of renewable transport fuels, including Power-to-X (PtX) and sector coupling opportunities^45,46.

Balmorel modeling and data advancements

In comparison to previous model versions, the one used in this study demonstrates new developments and notable improvements. We expanded considerably the geographical coverage of the model to include the EU 27, the United Kingdom, Norway, Switzerland, and the remaining Balkan nations. The model encompasses a representation of all major energy sectors (see Fig. 1) and allows comprehensive sector coupling to investigate potential synergies between energy vectors and sectors. Furthermore, we improved the coverage of hydrogen-related elements such as geospatial allocation of prospective European hydrogen demands, hydrogen network development consisting of repurposed natural gas or new pipelines, adequate modeling of hydrogen underground storage in salt caverns, and interaction with third nations for trading hydrogen. In addition, we validate the technical renewable energy investment potentials (Supplementary Method 9) across multiple resources^23,47,48.

We build upon previous modeling advancements to update the model’s geographic coverage. The power sector coverage (Supplementary Method 8) is expanded, and the modeling of electricity flow between regions utilizes the techniques (i.e., net transfer capacity) defined in ref. ⁴⁹, assuming similar capital expenditures⁵⁰ (see Supplementary Fig. 4a and 4b). Data related to final electricity consumption are extracted from Eurostat⁵¹. Existing electricity interconnection data and prospective plans are consistent with the most recent pan-European electricity infrastructure development plan (TYNDP 2022)⁵².

The heating sector is divided into individual users (residential and tertiary sectors), process heating (low, medium, and high-temperature) in the industry⁵³, and district heating⁵⁴. To expand the geographical coverage of the model, relevant data for district heating and individual consumers are based on the most recent report on renewable space heating under the revised renewable energy directive⁵⁵. Additionally, the industrial heat consumption by country is updated according to ref. ⁵⁶.

Furthermore, we revise the future hydrogen demand per country (Supplementary Method 10) in accordance with the European Hydrogen backbone report⁵. A comparison of European hydrogen demand projections across multiple studies for 2030 and 2050 can be found in the Supplementary Method 11. We downscale industrial and transport country-level hydrogen demand projections to the geographical granularity of Balmorel by utilizing geographical information mapping of European industrial⁵⁷ and long-haul truck activities⁵⁸ (see Supplementary Fig. 8a and 8b). The rest of transport activities, such as buses and coaches, passenger cars, light commercial vehicles, and rails, towards 2050 are assumed to be decarbonized through direct electrification. Country-level demand projections for electrifying the transport sector are extracted from the EU Reference Scenario 2020⁵⁹.

The main hydrogen-related mathematical modeling is available in Supplementary Method 1. To evaluate the effect of synthetic fuel exogenous demands on the optimal sizing and hydrogen pipeline network topology, a spatial demand shift module is developed. New decision variables (see Supplementary Method 2) allow for exogenous assigned synthetic fuel demand (hydrogen derivatives) to endogenously shift spatially to other model regions. Furthermore, a myopic modeling approach is also used due to the high complexity of the optimization problem. The differences between myopic, limited, and perfect foresight modeling methodologies are examined to assess the blue hydrogen lock-in effect. There are minor differences in the results (see Supplementary Method 3).

Modeling hydrogen production and sector coupling

Hydrogen can be used for various purposes, e.g., 1) directly in the industrial sector for providing high-value heat or transport sector, or as peak power production, 2) to produce liquid PtX fuels, or 3) to produce synthetic methane, which can substitute natural gas. In Balmorel, demands for direct hydrogen in the industrial and transport sectors and liquid PtX fuels are defined exogenously. The final exogenous demand ranges from 326 TWh in 2030 to 931 TWh in 2040 to 1530 TWh in 2050. Furthermore, the need to use hydrogen for peak power production is endogenously calculated.

Hydrogen can be produced using different pathways, with the most prominent being 1) via alkaline water electrolysis, 2) using steam methane reforming (SMR) (gray hydrogen) and 3) using SMR with CCS (blue hydrogen). From the production facilities, hydrogen can be stored and transported via transmission infrastructure to its point of use. Currently, a hydrogen transmission infrastructure does not exist, but Balmorel is allowed to invest in new hydrogen infrastructure, such as hydrogen pipelines and storage facilities.

Furthermore, cross-sectoral synergies are incorporated into the modeling framework, e.g., by enabling excess heat from electrolytic hydrogen production to efficiently supply heating demands through district heating. The electrolyzer fleet can provide flexibility to the power system, and the operation is optimized endogenously in Balmorel. The downstream PtX production is less flexible, supplying a more stable demand. Thus the flexibility of the electrolyzer is subject to investment in and operation of storage facilities. Further details for hydrogen mathematical modeling description can be found in Supplementary Method 1.

Carbon capture and storage

Carbon capture and storage (CCS) is a possibility for new investments in generation technologies. Due to the model complexity and the focus of the current study, we permit investments for CCS in technologies generating electricity, hydrogen, and heat. Due to the large influence of economies of scale for CCS, the CCS is allowed only for large-scale CHP and non-CHP plants such as (steam turbines, gas turbines, combined cycle, or engines)⁵⁴. The CO₂ management is developed to account for transportation and storage costs (€ 20 \({{{{{{{{\rm{tCO}}}}}}}}}_{2}^{-1}\))⁶⁰, similar average cost is provided by ref. ³⁷. The capture rate is assumed to be 90%. A sensitivity analysis on the cost of transportation and storage as well as on the capture rate and limit on build-out rates of carbon storage potential is conducted (see Results section). Additional electricity consumption is accounted for in the capturing process (371 MWh \({{{{{{{{\rm{tCO}}}}}}}}}_{2}^{-1}\) captured⁶¹), which reduces the net efficiency of the unit. Furthermore, we evaluate the technically accessible CO₂ storage resources based on the European Commission project CO2StoP⁶². The reanalysis results provide probabilistic estimates of resources for underground storage (saline aquifers and hydrocarbon fields). The European Commission recently published the Net Zero Industry Act⁶³ highlighting that a key bottleneck for the carbon capture investments is the lack of operating CO₂ storage sites. The European Commission sets a Union target of 50 Mt of annual operational CO₂ injection capacity by 2030 with a potential estimate of 550 Mt by 2050⁶³. Our second scenario, GH2E, is motivated by the uncertainty in CCS deployment and the likelihood of extended natural gas consumption for low-carbon hydrogen production, blue hydrogen.

Hydrogen infrastructure expansion network and storage

Hydrogen transport follows the same level of geographical aggregation as the electricity network and is modeled with transmission pipelines assuming linear bi-directional flow. In Supplementary Method 6, we include a cost comparison of hydrogen transport methods, highlighting pipelines as the most competitive option for European cross-border volumes trading. Based on pipeline size, the specific investment capital expenditures for hydrogen transmission pipelines and compressors are derived from the most recent European Hydrogen Backbone (EHB) report⁸. In accordance with the EHB report, pipeline investment expenditures are classified as either repurposed or new. For both types, the distance in a straight line between the centers of the modeled regions is estimated. Later, a weighted investment cost (€ MW⁻¹) per pipeline is computed based on the characteristics of the required infrastructure, such as onshore, offshore, new, or repurposed. Moreover, due to the relatively low demand for hydrogen in Europe, we assume that only medium-sized lines will be repurposed or newly invested until 2030. Meanwhile, economies of scale and learning rates for large cross-border pipelines can be accounted for beginning in 2040, lowering the expected capital expenditures (see Supplementary Table 1). In addition, costs and assumptions are made for the energy required to compress the hydrogen produced by water electrolysis, the expected pipeline lifetime, and hydrogen transmission energy losses for further information, see Supplementary Method 5.

Although the EHB characterizes which lines are classified as repurposed (first type), it does not provide information regarding their existing capacity. Therefore, we utilized the geographical information mapping of the existing methane European grid based on the SciGRID project⁶⁴. We note that due to the small existing capacity, repurposing existing methane pipelines may still necessitate the construction of new hydrogen transmission in a few instances (e.g., cross-border connections between Spain and France). The proportion corresponding to the repurposed length is adjusted to reflect these specifics. In addition, the length split into offshore and onshore pipeline distance is determined based on the EHB reports. The investment costs of new pipelines (second type) are calculated using a similar methodology and breakdown costs. The final computed costs per pipeline can be seen in Supplementary Fig. 3a and Fig. 3b.

In this study, hydrogen can be stored in steel tanks or underground salt nearshore and onshore caverns⁶⁵. While the hydrogen steel storage could, for the sake of simplicity, operate at the same pressure as the future hydrogen grid¹⁹, the salt caverns’ operational status could affect the internal gas pressure. To adequately capture the pressure differences when expanding hydrogen from caverns, we use the software REFPROP/NIST⁶⁶ to calculate the density of hydrogen at a given pressure and temperature. These parameters are incorporated into a simulation operational model (see Method 7, Supplementary Table 2) of 1 TWh of hydrogen underground storage to determine maximum discharge volumes per time period. For simplicity, the volume of the cavern is assumed as constant. We allow the cavern to operate between 180 and 105 bar at a constant temperature of 39 ^∘C. A maximum drop of 10 bar is permitted due to concerns about geotechnical safety⁶⁵ limiting the maximum daily volume for discharge. The simulation tool provides the total amount of hours per charging or discharging cycle used later as input to Balmorel.

Importing hydrogen from third nations

The Balmorel framework is expanded further to permit importing hydrogen flows from third-party nations outside the examined energy system borders. We notice two distinct modeling approaches. The first would require simulating the whole energy system of those countries, as well as the associated hydrogen transmission and transportation alternatives. The European modeling framework would incorporate additional investment expansion decisions. As a result, the problem’s objective function will be revised to account for investment decisions of a larger and interconnected energy system. Yet, assuming perfect competition and rational decisions, this technique would lead to studying and addressing the question of the possible imported hydrogen volumes from other nations, realized by minimizing the total cost of an extended system not only limited to the European framework.

However, nations such as Algeria, Tunisia, Morocco, and Ukraine have already stated exporting ambitions and targets. With the second approach, we question whether those targets are competitive with domestic hydrogen production, and how imports would impact the development of the European future hydrogen infrastructure. We apply an external planning and operation optimization framework with the objective of minimizing system costs while meeting a yearly demand target for hydrogen generation. The optimization problem results in investments in technologies such as utility solar PV, onshore and offshore wind turbines, and hydrogen transmission pipes. The levelized cost of producing and transporting hydrogen via dedicated repurposed natural gas pipes to system boundaries is then estimated. The problem is addressed sequentially for the years of the announced targets (i.e., 2030, 2040, and 2050) from the three potential importing choices (i.e., Algeria and Tunisia, Morocco, and Ukraine). As exogenous input, the European modeling framework is updated with expected imported prices, pipeline capacities, and available yearly volumes. Those details act as the contact limits between the system boundaries and the third countries. More details regarding technological costs and outcome results can be found in Supplementary Method 12.

Scenario choice and description

We conduct three modeling scenarios based on a least-cost optimization with a focus on examining the future hydrogen production pathways and infrastructure in Europe. The scenarios are based on the most promising and technologically feasible options for producing and storing hydrogen, as well as options for importing green hydrogen from other nations. The following sections provide extensive information for each scenario.

Hydrogen Europe (H2E)

The hydrogen Europe (H2E) scenario addresses the study’s main question on where, when, and how to produce hydrogen in a European energy setup. We want to offer insight into the competition between hydrogen-producing technologies in Europe, import possibilities, and information on potential hydrogen infrastructure. We allow the model to use all available technologies, including electrolysis (alkaline cells), conventional steam methane reforming (SMR), and steam methane reforming with CCS (SMR-CCS). Besides those production technologies, the model can invest in hydrogen storage, such as underground nearshore and onshore salt caverns or steel tanks. The subsurface formations are located in certain geographical areas and have a large potential for underground hydrogen storage⁶⁵. In addition to internal European production, the H2E scenario also allows the importing of hydrogen through third-party nations (Morocco, Algeria and Tunisia, and Ukraine) based on the methodology described above. National electrolysis capacity targets up to 2030 are considered to depict a plausible short-term hydrogen market evolution⁴.

Furthermore, over the long term, it is expected that the future hydrogen grid will complement the electricity grid reinforcements¹³. We pay specific attention to the electricity grid development in our scenario setting. According to ENTSO-E’s Ten Year Network Development Plan (TYNDP) 2020, more than 300 transmission projects are expected to be completed by 2040⁵². Despite this, 60% of the projects are delayed or altered in some way⁶⁷. Because those TYNDP projections are proving ambitious, we limit the electricity expansion grid to TYNDP across neighboring European countries until 2035. After 2035, the model co-optimizes power and hydrogen networks. The main model input parameters for hydrogen grid infrastructure, technological investment costs (Supplementary Note 8), and assumptions are those discussed in Methods and in the Supplementary Information. This approach and restrictions are applied across all scenarios.

Green H2 Europe (GH2E)

Fit for 55 packages² and the more recent RePowerEU³ initiative aim at accelerating renewable hydrogen production while phasing out the dependency on fossil fuels. The latter plan necessitates a large European Electrolyzer capacity of around 64 GW by 2030⁶⁸. The carbon tax pricing and fossil fuel price projections (Supplementary Note 7) in this study are based on the World Energy Outlook 2022 (Net Zero Emissions scenario, NZE)⁶⁹. According to the NZE scenario, high CO₂ taxation assumptions ranging from 140 € ton⁻¹ in 2030 to 250 € ton⁻¹ in 2050 will result in lower demand for fossil fuels and, consequently, lower market prices. Therefore, low-carbon hydrogen produced from natural gas with CCS is projected to be economically viable. However, whether hydrogen produced from SMR-CCS can be considered low carbon is debated in the literature^{36,70,71,72,73}. Some studies focus on methane leakages and life-cycle emissions, causing additional warming effects, whereas others assume higher capture rates, resulting in lower overall emissions. Another emerging challenge is that large-scale deployment of underground carbon storage leakage rates must be kept to less than 0.1% a⁻¹ on average, but methods for monitoring and confirming storage to this precision have yet to be established⁷⁴. Furthermore, CCS, an immature technology with minimal public awareness, may face social acceptance challenges. However, there is evidence that it is possible to encourage social acceptance of CCS and perhaps avert demonstrations and opposition by presenting information on its environmental benefits⁷⁵. Lastly, although blue hydrogen provides an alternative pathway⁷⁶, it conflicts with the European Commission’s ambitions to accelerate the face out of natural gas and dependency on fossil fuels. These factors encourage us to explore the effects of a large-scale electrolysis expansion on the European energy system in the absence of blue hydrogen. We develop a policy scenario called Green H2 Europe (GH2E), in which we exclude the potential for investing in SMR-CCS starting in 2030. Although there is no direct restriction in the model that electrolytic hydrogen is produced with renewable electricity, with increasing CO₂ quota prices (Supplementary Note 6) assumed, the electricity production will increasingly become green, and so will the hydrogen. Yet, we continue to permit renewable hydrogen imports from third countries as RePowerEU proposes.

Self Sufficient Green Hydrogen Europe (SSGH2E)

Another contentious discussion is the risk of relying on the import of hydrogen from other countries^24,77,78. In the final scenario, we subtract this opportunity. The model must determine the best approach to meet hydrogen demand using solely water electrolysis technology. Blue hydrogen investments are not permitted so as to decouple hydrogen production from conventional fuels for the concerns described in the GH2E scenario. This scenario strains the energy system and sheds light on European countries’ competition for renewable energy resources for hydrogen generation while shaping an alternative hydrogen network without the effect of imports. Finally, a picture of a future European energy system that is self-sufficient in domestic green hydrogen generation is provided in this scenario.