H₂ and CO₂ network strategies for the European energy system

The transition to a climate-neutral European economy is a pressing challenge that demands coordinated action across various energy sectors. Whereas management of both carbon dioxide (CO₂) and hydrogen (H₂) is considered a critical component of this transition, a gap exists in understanding how new hydrogen infrastructure effectively interacts with carbon management technologies, including carbon capture, transport, storage, utilization and sequestration. Hydrogen is being considered in several industries that are not eligible for electrification. It offers an efficient way to transport and store energy over long distances and periods and can be produced from electrolysis or steam methane reforming (SMR), for example. Captured carbon can be stored in geological formations permanently, a process known as carbon sequestration (CS).

Additionally, carbon can be combined with hydrogen to produce carbonaceous fuels, such as synthetic kerosene for aviation, synthetic methanol for shipping or synthetic methane for industrial feedstock or synthetic naphtha as a feedstock for high-value chemicals. The production of these carbonaceous fuels is jointly referred to as carbon utilization (CU). Carbon can be effectively captured from industrial processes and the combustion of biomass, fossil fuels or synthetic carbonaceous fuels through point source carbon capture (CC) or harvested from the atmosphere using direct air capture (DAC). If carbon from biomass or DAC is sequestered, there is net carbon dioxide removal from the atmosphere. Both CO₂ and H₂ networks are likely to play a crucial role in the cost-effective integration of these carbon management technologies to enable net-zero economies.

In line with the climate neutrality target by 2050 of the European Union under the European Green Deal¹, numerous programmes, financial models and initiatives have been launched to support the ramp-up of the hydrogen and carbon economy. The Renewable Energy Directive III² aims to produce 10 Mt hydrogen from renewable energy sources domestically in the European Union. The Net-Zero Industry Act³ proposes a CO₂ injection target of 50 MtCO₂ per year by 2030 for sequestration within the European Union. The European Commission’s 2024 industrial carbon management strategy⁴ plans to capture 450 MtCO₂ per year by 2050 in the European Union, of which almost 250 MtCO₂ per year is sequestered. The European Innovation Fund⁵ supports carbon management technologies to enhance the global competitiveness of European industries.

At the same time, collaborative industry initiatives such as the European Hydrogen Backbone⁶ or the Hydrogen Infrastructure Map⁷ showcase the potential of hydrogen as a fuel and energy carrier. Business models from companies such as Tree Energy Solutions⁸, Carbfix⁹ and Equinor¹⁰ advertise carbon management hubs that provide green hydrogen, methane and synthetic fuels on the one hand and offer to purchase CO₂ on the other hand. The Northern Lights project in Norway¹¹ is planning a transport and sequestration capacity of 1.5 MtCO₂ per year to be operative in 2024, expanding to a targeted scale of 5 MtCO₂ per year of sequestration by 2030. The market potential for carbon capture and sequestration in Europe is underscored by the Capture Map’s estimated potential of 1.7 MtCO₂ annually from point sources¹², accounting for about half of the continent’s emissions, and estimated sequestration capacity potentials of up to 3 MtCO₂ per year (ref. ¹³). Whereas certain aspects of sequestration technology are well established, particularly in offshore depleted oil and gas fields and saline aquifers, ongoing research and pilot projects continue to expand our understanding of different geologic storage options and monitoring techniques to ensure long-term containment^14,15,16. To advance the carbon economy, the Clean Air Task Force emphasizes the importance of developing a carbon transport system in Europe alongside a hydrogen network¹⁷. Carbon pipelines, considered a mature technology, have seen widespread installations in the United States and Canada, primarily to supply CO₂ to enhanced oil recovery^18,19.

Despite the abundance of policies and initiatives, up to this point, it remains unclear how the two transport systems of hydrogen and carbon may complement or even compete with each other. Both can bridge the misalignment of sources and sinks for carbon and hydrogen. A hydrogen network can supply regions with geographically fixed hydrogen demand, such as for steel production, with hydrogen from regions with the best renewable resources and enable CU at the site of point source CC. On the other hand, the carbon transport system allows transporting captured carbon to regions with sequestration potentials or high-quality renewable resources, the latter enabling cost-effective CU.

In the literature, the two network approaches and underlying technologies have been discussed in several publications^{20,21,22,23,24,25,26,27,28,29,30,31}. Such techno-economic models, in contrast, for example, to integrated Assessment Models, can account for the spatial distribution of carbon sources and sinks, which are crucial to provide a holistic view of the energy system and its technological interactions. However, all the techno-economic studies have dealt only with isolated aspects of hydrogen and carbon networks. Neumann et al.³¹ demonstrated the extent to which hydrogen and electricity grid expansion is interchangeable in a climate-neutral European energy system at the expense of higher investments. The underlying, highly resolved model encompasses the most relevant sectors. However, it neglects the infrastructure required to transport carbon across regions. Morbee et al.²¹ optimizes the topology and capacity of a CO₂ network in Europe but considers only the power sector without co-optimizing renewable deployment. This limited sectoral scope cannot capture important dynamics of carbon management because it neglects sectors such as industry that will need to handle most CO₂ in the future. The European Commission’s Joint Research Center analysed the possible topology of a CO₂ network in 2024³². However, it did not include the interaction of CO₂ sources and sinks with the rest of the energy system. Becattini et al.³⁰ present a mixed-integer model to optimize the time evolution of a CO₂ transport system in Switzerland, connected to a remote sequestration site in Norway. However, this limited spatial scope fails to consider other sequestration sites and the co-benefits from connecting the CO₂ network to neighbouring countries. Other high-resolution energy system models do not feature detailed CU technologies and CO₂ transport. In Pickering et al.³³, the Euro-Calliope model, for example, features neither carbon nor hydrogen transport and thus forces synthetic fuel production to rely on local electrolysis and DAC in the same region.

Studies that comprehensively assess and consider the co-optimization and comprehensive assessment of both CO₂ and H₂ networks in a fully sector-coupled energy system are lacking. Such assessments are greatly needed to identify realms of competition and synergy between hydrogen and carbon management technologies. In this Article, we present a detailed study of the European energy system for 2050, which includes high geographical resolution and a comprehensive representation of carbon management technologies. The study is conducted using the PyPSA-Eur energy system model³⁴, which optimizes operations and investments in all relevant energy sectors. It is assumed that Europe is self-sufficient in energy and does not import any fuels. This drives carbonaceous fuel production at sites where both CO₂ and H₂ can be provided at low cost.

To investigate the competing transport dynamics between CO₂ and H₂ networks, we contrast four scenarios (Table 1) with different sets of networks available to lower costs of supplying projected energy demands for 2050 (Fig. 1).

Derived from various sources^{41,42,43,44,45,46}, the figure shows total annual energy demands for each energy source, which determine the model’s endogenous investments and operation. Endogenous processes can lead to higher total production volumes of some energy carriers; for example, the demand for methanol requires more hydrogen and carbon as secondary (energy) inputs, which are not considered here. In the model, demands are defined per region and time stamp.

Full size image

We analyse how the option to deploy carbon and hydrogen networks affects different decarbonization strategies in the scenarios and how investments in none or only one of the transport technologies lead to inefficient infrastructure layouts. Initially, we focus on a net-zero carbon emissions target, limiting carbon sequestration to 200 MtCO₂ per year. This constraint ensures that carbon sequestration is reserved for the hardest-to-abate emissions while avoiding reliance on sequestration where other mitigation options are feasible. We give a detailed description of the different technological and geographical impacts of the transportation systems. Subsequently, our analysis demonstrates the extent to which different deployment strategies remain robust under a tightened target with net CO₂ removal.

All scenarios share key characteristics but differ in system costs and technology use (Fig. 2). Costs in the Baseline scenario are the highest at 764 billion € per year. Scenarios with carbon and hydrogen networks are more cost effective, reducing costs by 3.1% in the CO₂–grid, 4.7% in the H₂–grid and 5.3% in the Hybrid scenario.

Whereas in the Baseline scenario, the model has neither a carbon nor a hydrogen network, it can expand both in the Hybrid scenario. The gas infrastructure combines gas facilities for transport, power and heat production. CO₂ infrastructure and H₂ infrastructure combine transport and storage for each carrier. Carbon capture at point sources refers to all technologies with integrated carbon capture, including the cost of the main facility (for example, combined heat and power units) and the carbon capture application.

Full size image

System cost differences among the scenarios are driven by investments in wind and solar and by carbon capture technologies. In scenarios with carbon and hydrogen networks, the model primarily reduces the reliance on costly DAC, from 365 MtCO₂ per year carbon capture in the Baseline scenario to 113 MtCO₂ per year in the Hybrid scenario (Fig. 3). At the same time, carbon capture from biomass combustion, gas-based industry processes and biogas upgrading increase. The latter lowers the need for synthetic methane production.

Positive values indicate carbon capture, and negative values indicate carbon consumption. By integrating hydrogen and carbon networks, the predominant method for carbon removal shifts from DAC to bio-energetic processes with capture. At the same time, the reliance on methanation decreases.

Full size image

In the Baseline scenario, where there is no carbon and hydrogen transport, hydrogen must be produced where it is consumed. For CU, either hydrogen must be produced where carbon is cheap at point sources or DAC must be used at sites with low-cost hydrogen. Where fossil carbon from point sources cannot be sequestered, it has to be compensated elsewhere by negative emissions from DAC or biomass capture with sequestration. To provide hydrogen in some regions with poor renewable resources, the model resorts to SMR, which uses either green or fossil methane transported within the gas grid. All of these options are suboptimal from a cost perspective.

Both a carbon network and a hydrogen network can help alleviate these inefficiencies. On the one hand, a carbon network allows carbon from point sources to be transported to low-cost hydrogen and sequestration sites. On the other hand, a hydrogen network allows low-cost hydrogen to be transported from regions with good renewable resources to point sources for CU. This results in fundamentally different network flows in each scenario (Fig. 4).

a–d, Carbon (a,b) and hydrogen (c,d) flow patterns in the CO₂–grid (a,c) and the H₂–grid scenario (b,d). For each region, upper semicircles show the average production per technology, lower semicircles the consumption and colours the average marginal prices. Lines and arrows show the interregional transportation. Carbon sequestration offshore is drawn in full circles. Basemaps from Natural Earth.

Full size image

In the CO₂–grid scenario, there are three main transport purposes of the low-cost carbon captured in Central and Eastern Europe (Fig. 4a). A large part of the carbon is transported to the Iberian Peninsula (average 9.8 kt h) and the British Isles (average 11.8 kt h⁻¹), where it is used along with low-cost electrolytic hydrogen to produce Fischer–Tropsch fuels and methanol. In Italy and Greece, a portion of the carbon (average 3.3 kt h⁻¹) is directed to methanation plants, where the synthetic methane is fed into the gas grid at key terminals that receive fossil gas imports as well. Finally, a third part of captured carbon is transported directly to sites in the North Sea, Baltic Sea and Mediterranean Sea, where it is sequestered. Almost none of the captured carbon is used locally at the point source. In addition to electrolysis in regions with good renewable resources to supply CU (~125 GW electric (GW_el)), the model places smaller amounts of electrolysers where hydrogen is needed for industrial processes across Central Europe (~8 GW_el) (Fig. 4c). Hydrogen prices vary widely across the continent, with low price valleys at around 60 € MWh⁻¹, where the CU is located, and high price peaks in Central Europe at around 95 € MWh⁻¹. Despite the transport system, the marginal cost of carbon capture still varies from 105 € t⁻¹ in Central Europe to 160 € t⁻¹ at sequestration sites.

In the H₂–grid scenario, the main hydrogen transportation route goes from Spain (66 GW) to Central Europe to supply local carbonaceous fuel production with captured carbon from point sources since CO₂ can no longer be transported (Fig. 4d). Also, the hydrogen network supplies spatially fixed hydrogen demand for industry in Central Europe (2.5 GW in total) at much lower costs than local electrolysis or SMR. Carbon sequestration takes place at several coastal sites with local DAC facilities. The largest carbon sink is located in Portugal (3.5 kt h⁻¹; Fig. 4b). Marginal costs of carbon capture are low in Central Europe (110–120 € t⁻¹) and highest in sequestration regions (160 € t⁻¹). Hydrogen prices have flattened out at a relatively low level of 60 € MWh⁻¹ in all regions.

In both scenarios, CU is placed where material inputs, CO₂ and H₂, are provided at minimal cost, that is, in price valleys of hydrogen and carbon in the CO₂–grid and H₂–grid scenario respectively. In the CO₂–grid scenario, CU co-locates with low-cost hydrogen production and uses transported carbon from the inland. In the H₂–grid scenario, CU co-locates with carbon point sources and uses transported low-cost hydrogen.

We conclude that there are individual cost–benefits of the transport networks. On the one hand, the carbon network enables the transport of low-cost captured carbon to sequestration sites, reducing the dependence on high-cost DAC near the coast, which is present in the H₂–grid scenario. On the other hand, the H₂-–grid can supply the spatially fixed hydrogen demands in Central Europe with low-cost hydrogen imports. At the same time, the model in the CO₂–grid scenario is forced to deploy electrolysis in regions with poor renewable resources. The latter aspect finally leads to lower system costs in the H₂–grid scenario than in the CO₂–grid scenario.

The Hybrid scenario combines the advantages of hydrogen and carbon networks, resulting in the highest system cost reduction of 5.3% (41 billion € per year) compared to the Baseline scenario. Again, the model transports low-cost hydrogen to the point sources in Central Europe to supply CU but also transports low-cost carbon from point sources close to the shore to sequestration sites (Fig. 5). This provides low-cost hydrogen across the regions and reduces reliance on more expensive DAC.

a, Carbon sector. b, Hydrogen sector. For each region, upper semicircles show the average production per technology, lower semicircles the consumption and colours the average marginal prices. Carbon sequestration offshore is drawn in full circles. Lines and arrows show interregional transportation volume. Carbon from point source in the inland either supplies local CU with imported hydrogen or facilitates sequestration in nearby offshore regions. Maps for carriers other than CO₂ or H₂ (electricity, heat and gas) are shown in Supplementary Figs. 1 and 2. Basemaps from Natural Earth.

Full size image

For the carbon network, the model employs short and direct pipeline routes to transport carbon from inland point sources to nearby sequestration sites, for example, in Northern Germany, Italy and Greece (Fig. 5a). In addition, pipelines transport relatively small amounts of carbon from spatially fixed point sources to CU facilities. A notable outlier is the route from the Netherlands via the United Kingdom and France to large methanolization and Fischer–Tropsch plants in northern Spain. The hydrogen routes are similar to those in the H₂–grid scenario with transports from western regions (the Iberian Peninsula and the British Isles) to distributed CU production sites across the continent (Fig. 5b). Whereas the topology of the hydrogen network (1.2 billion € per year) roughly corresponds to that of the H₂–grid scenario and is primarily driven by CU, the carbon network (0.6 billion € per year) plays a secondary role and is driven by CS (Supplementary Fig. 3).

In this set-up, the networks enable all carbon-emitting technologies operating at a high average capacity factor, that is, above 70%, to capture most of their carbon (Fig. 6). More than 90% of emissions are captured for most technologies. The exceptions are SMR and gas-fired CHP plants, which operate only a few weeks annually, making carbon capture units economically unattractive for these applications. Due to the better integration of point sources, only a few DAC facilities are installed, such as those used for methanolization in Spain. This, in turn, leads to reduced power consumption and less wind and solar power deployment than in the other scenarios.

The size of the dots corresponds to the average capacity factor of the respective technology. While peak load technologies such as gas-fired CHP plants, which are only in operation a few weeks a year, are not expanded for carbon capture, point sources with base load characteristics such as biomass CHP and process emissions from industry are fully (partially) developed as soon as the transport of carbon (hydrogen) is permitted.

Full size image

In a further set of scenarios, we apply a stricter CO₂ emissions reduction target and, thereby, force the model to remove 460 MtCO₂ from the atmosphere in one year (equivalent to 10% of 1990 CO₂ emissions) while allowing a maximum sequestration of 660 MtCO₂. There are additional system costs of 78–82 billion € per year in all scenarios, of which the largest proportion is the investment in DAC and associated solar and wind power and heat supply (Fig. 7). In the Hybrid scenario, this share is 78%, whereas new carbon capture at point sources accounts for only 4% of incremental system costs. This shows that point source carbon capture is already almost exhausted in the net-zero case. The additional sequestration is mainly provided by large DAC facilities near the coast. To a much lesser extent, carbon captured from inland bioenergy sources, which would provide CU under a net-zero assumption, is instead transported and sequestered (Supplementary Fig. 4). Some of the CU located in Central Europe in net-zero scenarios is instead diverted to Spain, where it uses carbon from new DAC facilities.

Each cost bar is split into contributions of the same technology groups as used in Fig. 2, except for carbon sequestration costs, which are now displayed separately. For all models, DAC contributes the most to additional CO₂ removal, requiring further solar, wind and heat pump capacities for electricity and heat input.

Full size image

These changes have little effect on the layout of the CO₂ and H₂ networks (Supplementary Figs. 4 and 5). The slight shift in CU deployment results in less hydrogen pipeline construction in some parts of the system. Nevertheless, 80% of the hydrogen topology built under the net-zero emissions assumption is also built under the net-negative target. The CO₂ network partially extends routes from Central and Eastern Europe to the coasts. Compared with the other scenarios, the relative benefits of CO₂ transport and sequestration with tighter emissions targets are evident. The carbon network in the net-negative CO₂–grid scenario is more prominent than in the net-zero CO₂–grid scenario and transports more carbon from point sources to sequestration sites than to CU facilities. In the H₂–grid scenario, where all carbon for CU and CS must come from DAC, the hydrogen transport topology remains almost identical to the net-zero scenarios (detailed cost contributions in Supplementary Figs. 3–10).

Overall, this leads to the net cost increases of 78 billion € per year in the CO₂–grid scenario, by 80 billion € per year in the H₂–grid scenario and by 78 billion € per year in the Hybrid scenario. The resulting cost–benefit of the Hybrid scenario in comparison to the other scenarios is 6 billion € per year.

Our study assesses the roles of hydrogen (H₂) and carbon dioxide (CO₂) networks in Europe’s future energy system under net carbon neutrality and net-removal targets. We show that both networks strongly impact the optimal deployment of carbon capture, carbon utilization and sequestration. A CO₂ network reduces costs by transporting low-cost carbon from distributed point sources to sequestration sites and sites with low-cost hydrogen production to produce carbonaceous fuels. A hydrogen network provides greater cost savings as low-cost hydrogen can be transported to regions with high demand, allowing carbon utilization at point sources. A combination of both networks emphasizes the strengths of each infrastructure. In the hybrid configuration, the H₂ network supplies low-cost hydrogen to spatially fixed hydrogen demand and carbon utilization at point sources across regions. At the same time, the CO₂ network transports low-cost carbon captured from point sources near the coast to sequestration sites, reducing the reliance on direct air capture. This results not only in the lowest system cost but also in a system robust to tightening the target to net-negative emissions.

Limitations of the study, such as the neglect of synthetic fuel imports and assumptions on biomass potentials, are discussed in Supplementary Note 1. Supplementary Notes 2–5 provide an analysis of some model sensitivities. In the hybrid scenario, increasing the CO₂ sequestration limit from 200 to 800 MtCO₂ per year reduces system costs by 9.1% (Supplementary Fig. 11). This reduction is achieved by replacing the production of synthetic e-fuels with fossil fuels that are either captured and sequestered or left unabated and compensated by carbon dioxide removal. Investments in renewable power assets and hydrogen infrastructure, which serve the synthetic fuel production for the aviation, shipping and industry sectors, drop by roughly a third. A similar effect on synthetic fuel production in Europe would be expected if synthetic fuels were instead imported from outside Europe. Smaller effects can be seen from reducing or increasing capital costs for CO₂ and H₂ pipelines and costs of electrolysers (Supplementary Figs. 12–16).

For policymakers, these results show the need for coordinated planning across sectors such as carbon, hydrogen and synthetic fuels because they are strongly interacting and across borders because of the localized nature of good renewable resources and sequestration potentials. Deploying multiple networks offers some robustness should unforeseen problems arise with one of them, and our results show that the system can even cope with neither a CO₂ nor H₂ network for a cost penalty. Many technologies have not yet been deployed at scale before and will need support for financing, regulation and gaining the acceptance of local populations.

In summary, our study highlights the benefits of strategic planning and integration of both CO₂ and H₂ networks in the European energy landscape. The flexibility and cost efficiency offered by these networks can make a substantial contribution to achieving the EU’s climate targets, particularly by reducing dependence on technologies that are still in the early market phase, such as direct air capture. Leveraging the synergies between H₂ and CO₂ transportation and management, Europe can effectively reduce system costs while ensuring a robust and sustainable energy system.

Overview of European energy system model PyPSA-Eur

The study is conducted based on the open-source capacity expansion model PyPSA-Eur (derivative of v0.9.0)^34,35,36. The model uses the open-source software tools PyPSA³⁷ and Linopy³⁸ for modelling and cost optimization. The model optimizes the design and operation of the European energy system, encompassing the power, heat, industry, agriculture and transport sectors, including international aviation and shipping. All technology cost assumptions are taken for the year 2040 and sourced from the ‘technology-data’ repository (v0.6.0)³⁹. The database applies a discount rate of 7% for most technologies, accounting for the weighted average costs of capital of investments. Many assumptions are based on the technology catalogues published by the Danish Energy Agency⁴⁰. Endogenous model results include the expansion of renewable energy sources; storage technologies; power-to-X conversion and transmission capacities; heating technologies; peaking power plants and the deployment of grey, blue or green hydrogen production facilitates, among others. The model considers various energy carriers such as electricity, heat, hydrogen, methane, methanol, liquid hydrocarbons and biomass, together with the corresponding conversion technologies.

In our configuration, the model’s time horizon spans one year, with a 3-hourly temporal resolution and a spatial resolution of 90 regions across Europe. Each region has a complex subsystem with technologies for supplying, converting, storing and transporting energy. Exogenous assumptions on energy demand and non-abatable emissions are taken from various sources^{41,42,43,44,45,46} (Fig. 1). The energy demand for electricity, transport, biomass, heat and gas is defined per region and time step. Land transport demand, including heavy-duty vehicles, is exogenously assumed to be fully electrified. Demands for kerosene for aviation, methanol for shipping and naphtha for industry are not spatially resolved and assumed to be constant throughout all time steps. Heat demand is regionally subdivided into urban, rural and industrial site shares. The system exogenously produces 633 MtCO₂ per year from industry, aviation, shipping and agriculture, 153 MtCO₂ per year of which are fossil-based process emissions. Locations for industrial clusters are taken from Manz and Fleiter⁴¹. Energy demand for industries is calculated from the JRC-IDEES 2015 energy balances⁴³.

Low-carbon electricity potentials and H₂ supply chains

Low-carbon electricity is provided by wind, solar, biomass, hydro-electricity and nuclear power plants. Hydroelectric and nuclear plants cannot be extended beyond their currently installed capacities. Weather-dependent power potentials for solar, wind and hydro-electricity are calculated from the reanalysis and satellite data sets, ERA5⁴⁷ and SARAH-2⁴⁸ per region and time stamp, using the open-source tool Atlite⁴⁹. Solar and wind power can be expanded in alignment with land-use restrictions calculated, taking into account land usage classes and natural protection areas^50,51. We restrict the total volume of power transmission expansion to 20% of its current capacity, acknowledging the challenges in inaugurating new transmission projects. For the use of biomass, we consider only residual biomass products and no energy crops. We limit regional biomass use to the medium-level potentials from the JRC-ENSPRESO database^52,53. Interregional biomass transport is permitted, with transport costs of ~0.1 € MWh⁻¹ km⁻¹ considered.

In our model, we consider green, blue and grey hydrogen from electrolysis and steam methane reforming (SMR), the latter of which may be equipped with CC technology. The geographical distribution of underground hydrogen storage potentials in salt caverns is derived from Caglayan et al.⁵⁴. Re-electrification of hydrogen is possible via fuel cells.

Energy and carbon transport

The topology and capacities of the electricity transmission system are taken from the ENTSO-E transparency map⁵⁵ and selected Ten Year Network Development Plan projects. The topology of the gas network is based on the SciGrid gas model⁵⁶. The hydrogen network topology is based on the combined routes of existing gas and electricity infrastructure. Potential CO₂ pipeline routes are based on the existing electricity network topology.

The power flow in the electricity system is based on the linearized power flow approximation. For HVDC lines, we apply static losses of 2% at the substations and length-dependent losses of 3% per 1,000 km. The losses of high voltage alternating current (HVAC) lines are estimated by a piecewise linear approximation for resistive losses. For hydrogen pipelines, an average loss of 1.2% per 1,000 km is assumed, aligning with industry estimates that account for hydrogen’s high permeability and small molecular size, which can lead to higher losses compared to other gases. The electrical demand for compression stations is assumed to be 1.8% per 1,000 km of the transported energy. For CO₂ pipelines, an average loss rate of 0.8% per 1,000 km reflects a generally low leakage rate observed in supercritical or dense phase CO₂ transport^57,58. The compression energy needed is set to 250 kWh per 1,000 km and tonne CO₂ to maintain a supercritical state for efficient transport. For natural gas pipelines, the model assumes a 1% loss rate per 1,000 km, accounting for typical leakage and operational losses in well-maintained natural gas infrastructure as reported in ref. ⁵⁹. The electrical demand for compression is set at 1% per 1,000 km of the transported energy, which falls within the typical range for natural gas pipeline operations⁶⁰.

Liquid fuels such as oil, methanol and Fischer–Tropsch (FT) fuels are not spatially resolved because transport costs per unit of energy are negligible due to their high energy density. Throughout this study, we focus on CO₂ and H₂ networks because of their high relevance to infrastructure investments and, thus, public policy decisions. The electricity and gas networks are both included in the model with full geographical detail but are not further analysed. Electricity networks are already in place, and we restrict further extension to 25% due to public acceptance concerns. Gas networks also already exist and will most likely experience decreased use in the future, removing any bottlenecks or constraints on the optimal system buildout or operation. If enabled, hydrogen can be transported via pipelines between regions, which can be expanded without limits, considering costs for pipeline segments and compressors. Pipeline flows are modelled using net transfer capacities and without flow dynamics, pressure valves or energy demand for compression. Whereas decisions to repurpose existing gas pipelines for H₂ or CO₂ transport are not explicitly modelled, the impact of potential pipeline cost reductions that could be achieved, for instance, by retrofitting is illustrated in Supplementary Note 3.

Carbon utilization

We argue that it is crucial to model CC, CU and CS as separate technologies rather than grouping them under the common but imprecise umbrella term carbon capture, utilization and storage to allow for independent optimization and adequate representation of each function in the carbon management system. Our model features three drop-in fuel production technologies for CU: methanation, methanolization and Fischer–Tropsch synthesis. The processed fuels are not limited in their total quantity of use or production. Methane is transported through the gas network, whereas methanol and FT fuel can be transported between regions without additional costs or capacity constraints. Synthetic methane substitutes natural gas or upgraded biogas, serving combined heat and power plants, residential heating gas boilers or industrial process heat. Synthetic methanol decarbonizes marine industry fuel demands, and FT fuels replace fossil oil for naphtha production for high-value chemicals, aviation kerosene or agricultural machinery oil.

Carbon supply

To supply carbon needed for CS and CU, the system can choose to deploy carbon capture (CC) technologies at various point sources of fossil or biogenic origin (below) or through DAC facilities. The concept of a merit order for capturing carbon plays a pivotal role in optimizing the deployment of CC technologies based on their economic feasibility. This merit order ranks all CC technologies according to relative costs to capture an additional marginal ton. The cost value for a technology depends on the exact system configuration and operation. It may vary by scenario and even by region because many technologies produce other (primary) outputs, such as electricity and heat. The costs to capture an additional marginal ton of carbon may also differ from the average capture cost. At the lower end of the cost spectrum, CC technologies applied to process emissions, such as those from cement, offer a cost-effective starting point. Following this, biomass combined heat and power (CHP) systems provide the dual benefit of energy production and carbon capture. Moving up the scale, gas used in industrial applications and biomass employed in industrial processes represent more costly yet viable options to capture carbon. DAC, an emerging technology capable of extracting CO₂ directly from the atmosphere, stands at a high-cost level due to its current high capital costs and energy demand. Finally, based on our model results, biogas upgrading, a process that refines biogas to natural gas quality, incurs the highest costs in the merit order per marginal ton of captured CO₂. Biogas input is a high-cost fuel that based on endogenous modelling decisions, is not used in large quantities. To the extent that biogas-to-gas facilities are used by the model to supply additional (carbon-neutral) gas as fuel, adding CC infrastructure incurs low costs (and thus ranges on the left-hand side of the merit order curve). However, the price of capturing additional CO₂ from biogas-to-gas upgrading is high because a substantial amount of the biogas fuel costs factors into the marginal cost of captured CO₂. This merit order framework is crucial for strategically deploying necessary carbon capture solutions while balancing economic considerations. If we consider the spatial aspect of distributed carbon capture potentials, the CO₂ network can exploit comparative cost advantages between CO₂ bidding zones to utilize the lowest-cost CC facilities across the continent. We assume a capture rate of 90% for CC on process emissions, SMR, biogas to gas and gas and biomass used in industry and 95% for CC on biomass and gas CHPs. Supplementary Table 1 summarizes the technology readiness levels, capture rates and deployment status of carbon capture technologies used in the model. Supplementary Note 6 provides an extended description of the integration of carbon capture technologies in the model.

Carbon storage and sequestration

To store CO₂, we differentiate between short-term storage in steel tanks for intermediate containment and long-term, permanent sequestration in underground formations such as porous rock or depleted gas reservoirs. For short-term storage in CO₂ tanks, annualized costs are 242 € t⁻¹ CO₂ per year (investment 2,584 € t⁻¹ CO₂, 1% fixed operational and maintenance cost, 25-year lifetime). For long-term sequestration, we apply uniform costs of 10 € t⁻¹ CO₂ per year across all sites, aligning with literature estimates for offshore sequestration ranging from US$4 to 23 t⁻¹ CO₂ (ref. ⁶¹). For carbon sequestration, we consider only offshore sites as potential sinks (Supplementary Fig. 17). We make this choice because offshore sequestration sites typically have a larger capacity compared to onshore sequestration sites in saline aquifers and due to concerns over public safety for infrastructure near populated areas. Our estimates for carbon sequestration potentials are restrictive, limiting the annual sequestration potential to 25 MtCO₂ per region and constraining it further to provide enough volume to allow for 25 years of sequestration.

Furthermore, we decided to limit the total amount of sequestered CO₂ to 200 MtCO₂ per year for our net-zero CO₂ emissions target and 660 MtCO₂ per year for our net-negative emissions target. This constraint is deliberately implemented to avoid over-reliance on carbon capture and sequestration as a back-stop technology, which could otherwise offset fossil emissions instead of prioritizing the reduction of fossil fuel use where technologically feasible. By capping the sequestration at 200 MtCO₂ per year, we aim to represent a balanced energy transition, where sequestration is used to offset hard-to-abate fossil process emissions (for example, calcination of limestone in cement manufacturing) but does not become a substitute for more sustainable mitigation measures. The implications of this choice are discussed in Supplementary Note 2.