Temporal regulation of renewable supply for electrolytic hydrogen

Electrolytic hydrogen produced using renewable electricity can help lower carbon dioxide emissions in sectors where feedstocks, reducing agents, dense fuels or high temperatures are required. This study investigates the implications of various standards being proposed to certify that the grid electricity used is renewable. The standards vary in how strictly they match the renewable generation to the electrolyser demand in time and space. Using an energy system model, we compare electricity procurement strategies to meet a constant hydrogen demand for selected European countries in 2025 and 2030. We compare cases where no additional renewable generators are procured with cases where the electrolyser demand is matched to additional supply from local renewable generators on an annual, monthly or hourly basis. We show that local additionality is required to guarantee low emissions. For the annually and monthly matched case, we demonstrate that baseload operation of the electrolysis leads to using fossil-fuelled generation from the grid for some hours, resulting in higher emissions than the case without hydrogen demand. In the hourly matched case, hydrogen production does not increase system-level emissions, but baseload operation results in high costs for providing constant supply if only wind, solar and short-term battery storage are available. Flexible operation or buffering hydrogen with storage, either in steel tanks or underground caverns, reduces the cost penalty of hourly versus annual matching to 7%–8%. Hydrogen production with monthly matching can reduce system emissions if the electrolysers operate flexibly or the renewable generation share is large. The largest emission reduction is achieved with hourly matching when surplus electricity generation can be sold to the grid. We conclude that flexible operation of the electrolysis should be supported to guarantee low emissions and low hydrogen production costs.


Introduction
Governments worldwide are seeking to scale up the production of green hydrogen to reduce emissions from sectors of the economy where direct electrification is challenging.In its 2022 REPowerEU strategy [1], the European Commission raised its target for domestic renewable hydrogen production in 2030 to 10 million tonnes per year, with an additional 10 million tonnes per year to be imported.India announced a target in 2022 to produce 5 million tonnes of hydrogen per year by 2030 [2].The 2022 Inflation Reduction Act in the United States established a production tax credit (PTC) of up to 3 $/kg H2 for clean hydrogen.In July 2023, the German government published its update of the national hydrogen strategy [3], which aims to produce 28-53 TWh H2 green hydrogen per year in Germany and import further 45-90 TWh H2 per year.
To qualify hydrogen for subsidies, to meet quota requirements and to maintain consumer confidence, a transparent system is required to certify that hydrogen is 'green' , i.e. produced from renewable electricity.Several definitions for green hydrogen have been suggested.The strictest would be to require that only electricity from newly-built renewable generators directly connected to the electrolyser can be used to produce hydrogen.While this definition is unambiguous, it forces hydrogen production to be located at the site of generation, which may be far from hydrogen demand, and it prevents flexible operation of the assets from adapting to electricity market prices.The definition of green hydrogen could be relaxed in three ways: through (i) relaxing additionality, i.e. allowing generation from existing sources; by (ii) easing location requirements, permitting spatial separation within the same region, electricity market bidding zone, or continent; and by allowing (iii) relaxing temporal matching requirement, i.e. allowing for renewable generation and electricity demand for hydrogen production to match on sub-hourly, monthly or an annual basis.
While requiring additionality and locational matching within the same bidding zone are less controversial, there has been discussion about the need to require temporal matching.Hourly matching ensures that electricity is from renewable sources but can be technically hard to enforce and costly when renewable sources are scarce.In contrast, annual matching is technically easier.However, it implies that electrolysis operator relies on the local market electricity mix at times with low renewable feed-in, which may contain some share of the fossil-fueled electricity.This can result in higher emissions for electrolytic hydrogen compared to alternatives like 'blue' hydrogen, which is produced from natural gas through a process called methane reforming while capturing parts of the CO 2 emissions.
In February 2023, the European Commission published a regulation for green hydrogen production in the Delegated Act for the regulation on Union methodology for renewable fuels of non-biological origin (DA) [4].These rules mandate geographical correlation and additionality, with hourly matching phased in by 2030 and monthly matching in the transition phase.The US decided in December 2023 for a similar regulation as in the EU, with annual matching in a transition phase and hourly matching from 2028 onwards [5].While the industry appreciates the clarity this regulation provides, manufacturers warn that the strict rules may hinder the rapid scaling of hydrogen infrastructure [6] and indicate that continuous supply needs and the lack of hydrogen storage in most industrial sectors [7] could lead to high costs with hourly matching.While the DA does not recognize nuclear power as a renewable resource, countries with a grid carbon intensity below 64.8 grams of CO 2 per kWh (currently met by Sweden and France [8]) are exempted from fulfilling the additionality requirement for renewables.The PTC does not make a distinction between renewable sources and nuclear power, since the criterion is solely based on the carbon intensity of the produced hydrogen.
Several papers have considered temporal regulation in the literature.Ricks et al [9] examined procurement strategies in the US, assuming different offtake prices incentivised by the PTC introduced in the 2022 Inflation Reduction Act [10].Unlike our study, which assumes a constant hydrogen demand profile, they employed a setup with a high offtake price and a flexible electrolyser operation, leading to the electrolyser running when production costs are below offtake prices.They find that hourly matching adds minimal costs while lowering emissions unless competing for limited high-quality renewable resources.Our approach further differs by assuming no competition for renewable resources between hydrogen and electricity production, and employing a two-step optimization.Giovanniello et al [11] show that these distinct methodologies yield varying results for the emissions impact of temporal regulation.However, Giovanniello et al [11] model the US system without a CO 2 price.In Brauer et al [12], various additionality, location and temporal requirements were studied assuming a constant hydrogen demand in Germany for the year 2030.Their method involved using lowcost hydrogen storage from liquid organic hydrogen carriers (LOHC), whereas our study explores various storage options and the impacts of different electrolysis operation modes and background grid systems, which can substantially impact the results.Similar to the US study [9], the authors found that hourly matching has a small cost premium but lowers emissions, both from the perspective of emissions attributed to the electricity consumption and from the impact on system emissions of the hydrogen production.Ruhnau et al [13] also considered hourly versus annual matching for a baseload hydrogen demand in Germany but focused on the impact on the existing power system using historical marginal emission factors.Their study, limited to hydrogen storage in steel tanks and wind power, differs from ours in terms of the variety of storage options, inclusion of solar PV, and consideration of future grid cleanliness.They found significantly higher hourly matching costs, which is likely because only hydrogen storage in steel tanks was modeled.In their study, annual matching slightly reduced system emissions because of the freedom to have renewables feed in when prices and emissions are high.Due to the reliance on historical marginal data, the authors were unable to evaluate the effect of a larger hydrogen volume or future cost developments.
Previous literature has sampled particular combinations of background power systems and hydrogen flexibility options to get a wide array of different, singular results.From these diverse findings it is difficult to provide any conclusions about the causality between the assumptions and the resulting impact of temporal regulation on emissions and production costs of green hydrogen.We address this gap in the literature by considering a much wider range of background systems.Our scenario range encompasses various countries, with their unique electricity generation mixes, national energy and climate goals and assumptions on hydrogen production flexibility.We factor into the analysis a time horizon of 2025 and 2030 to reflect the development of background systems with updated national targets and technology costs.This approach contrasts with existing literature, which often limits its focus to specific regions or singular storage technologies.In this study, we aim to resolve the existing ambiguity in the literature by investigating how different regulatory frameworks for green hydrogen production can maintain low system emissions and simultaneously ensure cost-effective hydrogen production.
We contrast the case of running an electrolyser of a given capacity using grid electricity only (i.e.no additional procurement of renewable resources) versus additional procurement of wind, utility-scale solar and batteries in the same bidding zone whose supply is matched on either an annual, monthly or hourly basis.We assume that the hydrogen demand is constant following the requirement of European industry [7] but examine several options for hydrogen storage to buffer the hydrogen production, including scenarios with no storage, (rather expensive) steel tanks, (relatively low-cost) underground cavern storage and zero-cost storage.The case of zerocost storage is equivalent to having a time-flexible hydrogen demand.The German system in 2025 is taken as a base scenario.In a sensitivity analysis, we explore a case of a less clean grid than our benchmark scenario (the Netherlands in 2025, renewable generation share of 49%) and a case of a cleaner grid (Germany in 2030, renewable generation share of 80%).The appendix provides further examples for Poland, the Czech Republic, Portugal and Spain.

Model structure
This study uses the European power system model PyPSA-Eur [14].Total annualized system costs in a given year are minimized by optimizing power generation and storage capacities under relevant engineering (such as linearised unit commitment following the formulation of Hua et al [15]) and policy constraints (such as National Energy and Climate Plans (NECPs)).
The geographical scope of the model is set to Germany or Netherlands and its neighboring countries.We model the ENTSO-E 220 kV and 380 kV transmission infrastructure clustered to the individual bidding zones [16].We focus on the mediumterm planning horizon, modeling two individual years, 2025 or 2030, with an hourly resolution.The years differ in technology cost assumptions, the existing fleet of legacy power plants, CO 2 prices and NECP additions.
The modeling is performed in two optimization steps.In the first step, the capacities and dispatch of power plants and storage facilities in the power sector are optimized without any hydrogen production.In the second step, the optimized capacities of step one are exogenously fixed, and the hydrogen demand and production site are added.The optimization is rerun, allowing capacity expansion of wind, utilityscale solar and battery storage at the hydrogen production site only, and any hydrogen storage allowed in the scenario.Since Giovanniello et al [11] find a significant impact on emissions of different temporal matching regulations depending on how additionality is modeled, we conduct a co-optimization of the electricity background system and the hydrogen production as well, which is discussed in the appendix.

Temporal matching
Hydrogen production is co-optimized with the operation of the electricity system.In addition to purchasing grid electricity, the hydrogen producer can procure new renewable generators, batteries and hydrogen storage in the local market zone to meet any imposed policy requirement on temporal matching.The optimization finds a cost-optimal portfolio of onshore wind, utility solar PV, electrolysis, battery and hydrogen storage to produce hydrogen.
Annual and monthly matching: In scenarios with annual or monthly matching requirements, the sum of all dispatch g r,t of contracted renewable generators r ∈ R in hour t over a time span T is equal to the sum of the electricity demand d t of the electrolysis in this period Depending on the temporal regulation, T corresponds to a year or a month.The contracted renewable generators must be new (i.e.additional to the system) and sited in the local market zone.Purchases from the grid can cover electrolysis demand when renewable generation is low, as long as it is matched with sales to the grid when procured renewable resources exceed hourly electrolysis demand.The hourly price in the local market is derived from the dual variable of each zone's energy balance constraint.
Hourly matching: The hourly matching requirement is modeled with a constraint (2), enforcing the hydrogen producer to match electricity consumption with clean electricity on an hourly basis.Clean electricity can come from procured renewables or battery storage charged with procured renewables.No grid electricity is allowed to serve the electrolysis demand or procured storage.
Thus, the hourly generation from the procured renewable resources r ∈ R and the discharge and charge from the procured battery storage s ∈ S, minus hourly sales to the grid ex t must be equal to the hourly electricity demand of electrolysis: The excess hourly generation from the procured renewable resources can be sold to the grid or curtailed.Note that for the base scenarios discussed in the main part of the manuscript, the excess (after the curtailment) is set to zero in the hourly scenarios or 20% in the hourly scenarios with allowed excess generation.

Scenarios
We analyse five regulatory scenarios of hydrogen production (see figure 1): (i) On-grid production (grid), where the electrolysis is powered by grid electricity without any additional procurement.
(ii) Additional renewable capacities in the local bidding zone, whose generation has to match the electrolysis consumption annually (annually).Grid electricity purchases and sales are allowed if this constraint is fulfilled.(iii) Additional renewable capacities in the local bidding zone, whose generation must match the monthly electrolysis consumption (monthly).Grid electricity purchases and sales are allowed if this constraint is fulfilled.(iv) Additional renewable capacities in the local bidding zone, whose generation has to match the electrolysis consumption hourly (hourly).(v) Additional renewable capacities in the local bidding zone, whose generation has to match the electrolysis consumption hourly, while excess generation of 20% of yearly electrolysis demand can be sold to the grid (hourly excess 20%).
An advantage of surplus generation is that it provides a hedge against inter-annual variability in renewable feed-in.For example, in a year with low winds, the renewable production would be sufficient to cover the electrolysis demand.In other years, the surplus electricity can be sold.
An additional reference scenario without hydrogen production and associated electricity demand is also computed, and its results are used as a benchmark for our analysis.In order to account for the electricity trade, we model all neighboring countries in addition to the selected one.In all the modeled countries, renewable generation must meet the political targets as defined in the NECPs or by more recent national policy targets (such as the Easter package in Germany), see table 1 in appendix.In all scenarios, we assume a fixed hydrogen demand of 28 TWh H2 /a (0.84 million tonnes produced hydrogen per year) in the selected country.This demand corresponds to the minimum target of domestic green hydrogen production targets for 2030 ranging from 28-53 TWh H2 in Germany.The demand for hydrogen is continuous throughout the year, following the needs of European industry [7].In a sensitivity analysis, we examine the effects of a higher hydrogen production of 53 TWh H2 .The same demand is used in the Netherlands to allow results to be compared.The price of carbon dioxide emission certificates is set to 80 €/t CO2 in 2025 and 130 €/t CO2 in 2030.
We implement five variations of hydrogen storage for each policy scenario to represent different degrees of flexibility for hydrogen: (a) zero-cost storage (flexibledemand), corresponding to a time-flexible hydrogen demand.(b) storage in underground salt caverns (underground), low-cost storage, could be accessible via hydrogen pipeline network, (c) storage in medium pressure steel tanks (mtank), medium-cost storage, (d) storage in high pressure steel tanks (htank), relatively expensive storage, (e) no storage (nostore), inflexible hydrogen demand.

Results
In the following, we address two aspects of hydrogen production for the different regulatory scenarios.We first highlight the system impacts by analyzing the carbon dioxide emissions from hydrogen production for each scenario based on the consequential emissions (section 3.1).Second, we look at the costs (section 3.2) of hydrogen production.In a further analysis, we consider the impact of the degree of decarbonisation of the background system (section 3.3).

Consequential emissions
Consequential emissions reflect the total system emissions associated with hydrogen production.They are calculated as the difference between total system emissions and the reference scenario without hydrogen production.
The results show that additionality is required to prevent increased emissions.Consequential emissions are up to nearly three times the CO 2 intensity of gray hydrogen produced via steam methane reforming (10 kg CO2 /kg H2 ) in the grid scenario.They are greatly reduced in annual, monthly and hourly matching scenarios due to additional procurement.Emissions decrease through annual and monthly matching in case of flexible demand and through hourly matching in all scenarios where surplus generation sales are permitted (see figure 2).
Producers can buffer low renewable feed-ins and optimise electrolysis operation by shifting the production to times of low electricity prices due to affordable hydrogen storage or elastic demand.With inflexible demand, electrolysis operates continuously at full capacity (see figure S2 in appendix).In this case, without the additionality requirement, up to 29 kg CO2 are emitted per produced kg H2 .The emissions are particularly high because the electricity system cannot adapt to the new hydrogen demand according to the study design.
The effects of annual and monthly matching on emissions are nuanced.Annual and monthly matching increases demand in some hours when the electrolyser is running and RES are scarce, while it decreases demand for conventional generation in hours with plentiful RES.If the increase in demand is met with coal while gas is displaced at other times, emissions increase.On the other hand, if the increase is met by nuclear and otherwise curtailed renewable generation while coal is displaced, emissions sink.The precise impact depends on the background system mix and the electrolyser operation mode.
With inflexible hydrogen demand and continuous full-load electrolysis operation, annual matching can yield emissions up to 4 kg CO2 /kg H2 .In absolute numbers, this leads to an increase in emissions from the German power sector of 3.2 million tonnes of CO 2 , corresponding to about 1.5% of power sector emissions in Germany in 2021.Compared with annual matching, monthly matching results in lower CO 2 emissions, up to 2 kg CO2 /kg H2 if hydrogen storage is expensive; this yields 1.6 million tonnes of additional carbon dioxide emissions in the absolute terms.However, if cheaper storage is available, emissions decrease, enabling annual and monthly matching to reduce total system emissions by −5 kg CO2 /kg H2 and −2 kg CO2 /kg H2 , respectively, with flexible demand (see left columns on annual and monthly panels in figure 2).This is because having flexible electrolyser demand and variable generation, only constrained by the annual matching, provides many degrees of freedom.Capacity factors range with flexible demand between 62%-68% for monthly and annual matching.
Electrolysers can use both procured and excess renewable electricity from the grid.The procured renewable energy can also be fed into the grid when electricity prices and emission intensities are high.Costly storage scenarios limit these flexibilities.In these cases, mean capacity factors of the electrolysis increase to at least 79%.The constant demand for electrolysis is met by additional fossil generation, which increases emissions.
In hourly matching scenarios without surplus electricity sales, total emissions remain unaffected by the additional electricity demand for hydrogen production because renewable generation meets the electrolysis demand every hour.Electrolysis capacity factors range between 45%-52% if any hydrogen storage is available.The most substantial reduction in total system emissions of up to -9 kg CO2 /kg H2 occurs in hourly matching scenarios with a possible sale of surplus generation.This corresponds to a total reduction in system emissions of 7.2 million tonnes CO 2 .In this case, the additional renewable generation sold to the grid reduces the operation of coal-fired plants and decreases system emissions.Unlike annual matching, hourly matching with allowed excess sales consistently reduces emissions in every hydrogen storage scenario.

Hydrogen production costs
The costs for the production of hydrogen are lowest independently of the storage options in the case of annual matching followed by monthly matching, with costs ranging between 3.35-3.70€ /kg H2 and 3.48-4.35€ /kg H2 respectively (see figure 3).They are below the costs of the grid scenario because additional renewable capacity is built, which lowers the electricity prices.The costs for hourly matching compared to monthly matching are only 7%-8% higher if demand is flexible or low-cost storage in the form of salt caverns is available.In the case of inflexible demand, hydrogen production costs with hourly matching are 11.27 € /kg H2 and, therefore, 2.6 times higher than the production costs with monthly matching.These high costs result from transforming variable renewable electricity generation profiles into constant electrolyser output.This transformation is partly provided by battery storage (41% of production costs) and partly by overbuilding renewable capacities, which are then partially curtailed (see figure S9 in appendix).This underscores the importance of supporting low-cost storage or demand flexibility when implementing hourly matching regulations to prevent elevated production costs.
The cost of hourly matching with allowed excess is 3.20 € /kg H2 , which is 4%-8% below the cost of annual or monthly matching in the case of flexible demand.This is caused by the additional profit that can be made by selling electricity in hours of high feed-in of renewable generation.

Degree of decarbonisation of the power system
Hydrogen production emissions and costs are influenced by how countries generate electricity.To investigate the impact of power generation mix on our results, we set up an example of a dirtier grid, i.e. a grid with a higher share of fossil fuels (Netherlands 2025) and an example of a cleaner grid (Germany 2030 with coal power plants being phased out) compared to our Germany 2025 scenario.The country-specific shares of renewable generation are applied for the respective years (see table 1 in the appendix).An overview of the shares of the individual technologies in electricity generation for the respective year and country of the reference scenario is in the appendix (see figure S23).Hydrogen demand is assumed constant at 28 TWh H2 /a to keep the results comparable.
Progress in the decarbonisation of the power sector significantly impacts hydrogen production emissions for the grid, annual and monthly scenarios.Since hourly matching has zero or negative emissions in the case of allowed excess, we discuss only grid, annual, and monthly matching scenarios below.Suppose the renewable share in the overall electricity mix is lower, as in the case of the Netherlands in 2025, with a share of 49%.In that case, the consequential emissions of hydrogen production in the grid scenario can increase to 38 kg CO2 /kg H2 (see figure 4), nearly more than four times the carbon intensity of gray hydrogen production (10 kg CO2 /kg H2 ).Annually and monthly matching reduces emissions in scenarios with low-cost storage or flexible demand since local procurement sales reduce the coal power plant generation.In a background grid with a lower renewable share (see also Poland and Czech Republic in figure 5), the CO 2 emissions reduction with annual and monthly matching in the case of flexible operation can be stronger compared to countries with a higher RES share.This is because the flexibility allows the electrolysis to run when electricity prices are low and replace fossil generation with local additional renewable generation at times of high feed-in.Conversely, a baseload operation of electrolysis can also lead to significantly higher consequential emissions compared to countries with a higher RES share, e.g. in the Netherlands emissions increase by up to 6 kg CO2 /kg H2 , since electrolysis runs at hours with a higher share of coal generation.
We find that scenarios without additional local procurement increase system emissions even when there is a higher share of renewable generation and coal power plants are phased out, as in our scenarios for Germany 2030.The consequential emissions are negative with annual and monthly matching for every storage type, i.e. hydrogen production further reduces total system emissions.The reduction in emissions results from the fact that the purchased grid electricity has only a small share of fossil energy sources, and the sold electricity contributes more to the decarbonisation of the grid.Coal power plants in Germany are decommissioned, but additional emissions are reduced from coal generation in Poland for annual and monthly matching.A share of 80% renewable is below the set threshold of 90% in the DA [4].However, our results show that even with inflexible demand, total emissions decrease with annual and monthly matching.Overall, the scenario with a cleaner grid illustrates that strict regulation of temporal matching hydrogen production with renewable electricity plays a minor role with increasing decarbonisation.

Further analysis
In the appendix, we discuss the limitations of this study more extensively.In addition, we explore a variety of sensitivity scenarios in order to generalize our findings above the specific model assumptions.In particular, we drop the assumption that the background system cannot adapt to the new hydrogen demand, increase the hydrogen demand volume, analyse hydrogen production in four other European countries in further detail, alter the volume of excess electricity sales, alter the price of natural gas, and alter the share of renewable electricity in the background systems.All sensitivity analyses show that a flexible operation of the electrolysis reduces consequential emissions and that the generation mix in the background system has a large influence on the emissions of hydrogen production.

Discussion
We first discuss the interplay between costs, emissions and flexibility in green hydrogen production in our results and policy implications that can be drawn Figure 5. Consequential emissions depending on hydrogen production costs for the countries analysed, various regulation and flexibility options.Highlighted are those points that correspond to regulation options which result in low emissions and low cost ((i) hourly matching with low cost storage, (ii) restrict capacity factors below 70% or (iii) high renewable share (>80%) in the background system).Scenarios in which the electrolysis is directly connected to the grid without additional local procurement (grid) with hydrogen production costs larger than 20 € /kgH 2 are not plotted in this graphic.from it (section 4.1).Second, we compare our result to existing literature (section 4.2).

Balancing cost, emissions, and flexibility in green hydrogen production: Insights and policy implications
The rules for green hydrogen must balance the impact of production on carbon emissions with the additional cost burden on producers.Additional costs for producers may hinder the scale-up of hydrogen production necessary to meet long-term climate targets.Our results indicate that flexible operation is key in systems which do not have a high share of renewable generation (>80%) (see figure 5).In the case of a non-clean background system, electrolysers have cost-optimal capacity factors in the range of 45%-68% to adapt to hours of high wind and solar production.This flexibility is made possible by flexible hydrogen demand or low-cost hydrogen storage in underground caverns to buffer the variable hydrogen production.We show that electrolysis production running at high capacity factors either causes high emissions (in the case of annual or monthly matching) or low emissions but high costs (in the case of hourly matching).
Examples where flexible hydrogen demand is possible in industry include ammonia production via the Haber-Bosch process or methane production via the Sabatier process.Both of these processes can be flexible with must-run part loads down to 30%-50% [17][18][19].
Low-cost hydrogen storage in salt caverns relies on the availability of suitable geological salt deposits.Fortunately, there are abundant salt layers and domes in Europe [20].These salt deposits are mainly concentrated around the North Sea, where abundant wind power resources are available.Hydrogen storage in steel tanks is feasible outside these locations, but this has a significant cost penalty on the hydrogen.Storing the hydrogen in LOHC may alleviate this cost penalty [12].Our results show that in the case of inflexible hydrogen demand, hydrogen production systems will instead be run with steel tank storage than without any storage.Steel tanks can easily be deployed at hydrogen production or industrial sites, resulting in lower average production costs than no storage.A hydrogen pipeline network could also make underground storage accessible to a broader area.
Hourly matching is the only matching scheme that provides strong incentives for demand flexibility and storage since the cost differences between constant and flexible electrolyser operations are so high.Moreover, inflexible operation with hourly matching results in oversizing of renewable capacities, potentially amplifying supply chain and renewable deployment challenges.For annual matching, the differences are much more minor.Incentives for flexible electrolyser operations are desirable since the flexible operation is seen in top-down system cost optimizing studies [21].The difference between the emissions of annually, monthly and hourly matched green hydrogen reduces with a cleaner background electricity system (see the change in German emissions from 2025 to 2030 in figures 2 and 4).However, hourly matching always results in low emissions, regardless of the background system, and provides a hedge against the case where ambitious targets for expanding renewable electricity are not met (see figure 5).
It is sometimes asked why strict rules are applied to hydrogen but not other new electricity consumers such as electric vehicles or heat pumps.One reason is that rules are required only for certification for producers seeking to get the label 'green' and associated subsidies.Another reason is that it is easier to regulate hydrogen production because it is done centrally at a large scale.This study shows a third reason: if hydrogen is produced without additionality or temporal matching, its carbon emissions impact can be worse than that of gray hydrogen.Numerous studies have shown that electric vehicles and heat pumps reduce emissions compared to fossil-based alternatives even with today's electricity mix [22][23][24][25].
It has also been argued that additionality requirements cannot affect system emissions in a system like Europe, where an emissions cap applies in the form of an Emissions Trading System (ETS) [26].We argue that the large volume of planned hydrogen production in Europe by 2030, 10 million tonnes of hydrogen per year, means that additionality is a valuable precaution to ensure that renewable production keeps pace with electrolysis demand.Without this safeguard, emission certificate prices could rise to politically unsustainable levels and endanger the entire ETS.It would also lead to higher electricity prices, affecting all consumers.
The European Commission is considering competitive tendering as a support mechanism for hydrogen uptake and switching from natural gas-based to renewable hydrogen production for industrial processes [27].Financial subsidies received through such competitive tendering (e.g. via Contracts-for-Difference (CfDs)) can enable hydrogen producers to stabilize their electricity procurement cost at a certain level (the Strike Price) for the duration of the contract.These subsidies will naturally affect green hydrogen production profiles and, consequently, the energy system impacts of hydrogen production.The impacts will largely depend on the design of tendering procedures and contracts.For example, an essential feature of the CfDs is the Reference Price.In the absence of a functioning market for hydrogen, different indexation options are being considered, such as the electricity price, gray hydrogen cost, and available commercial cost indexes, among others [28].If the hourly electricity price is used as a reference point of a CfD, this would incentivise baseload operation of hydrogen producers by providing compensation against high electricity prices.In the context of our analysis, this would imply high attributional emissions (unless an hourly matching requirement is imposed).If a CfD is based on a time-fixed index, the subsidy will function like an offtake price.The latter can also facilitate a constant operation if the subsidy level is high compared to the market electricity prices.Taken together, the envisaged support for hydrogen projects makes the baseload operation scenario in our study even more relevant.Implementing a CAPEX subsidy for electrolysers could address some challenges and promote more economically viable operations.The subsidy should only be available for electrolysis with capacity factors below 70%.However, it is important to note that our investigation does not explore the impact of a capacity factor limitation.

Comparison to other studies
We now compare our results to other studies in the literature.A more extended literature review is given in the Supplementary Material.This study shares the findings with a US study by Ricks et al [9] that annual matching with electrolysis running on high capacity factors increases emissions and hourly matching results in low emissions.In contrast to their results, we find a larger cost premium for hourly matching and that emissions can decrease with annual matching if the electrolysis is operated flexibly.Different methodologies and background systems cause these varying results.Ricks et al [9] do not consider hydrogen storage and assume that electrolysis can shut down when production costs exceed the offtake price.Further, in their model, local renewable expansion competes for resources with system-wide expansion, while our primary results assume no such competition thanks to our two-step optimization.
Giovanniello et al [11] show that these different methodologies of modeling the competition significantly impact the consequential emissions and lead to higher emissions in case of annual matching within the compete framework.However, in a sensitivity analysis, we consider the impact of competition and the possible adjustment of the background system (see section 1.7 in the appendix).Our results show that emissions from annual and monthly matching can be lowered with flexible electrolysis operation in scenarios within a competitive framework.We see other factors besides the competition modeling as crucial, such as the high European CO 2 prices and a higher share of renewables in the European electricity mix compared to the US.
Like the current study, Brauer et al [12] model baseload hydrogen demand and observe a small cost premium for hourly versus annual matching due to low-cost hydrogen storage.However, their study analyses the effects of relaxing spatial requirements for renewable procurement, which was outside the scope of this study.The current study expands the study by Brauer et al and explores various storage options, the impact of electrolysis capacity factors and different grid cleanliness levels.We therefore find scenarios with largely decarbonised grids or flexible operation of the electrolysis in which annual matching reduces emissions.
Ruhnau and Schiele [13] find similar costs for hourly and annually matched hydrogen as our study for hydrogen storage in steel tanks, and a slight reduction in system emissions with annual matching.However, their study differs in using historical marginal emission factors, focusing only on wind power, and not exploring other hydrogen storage options or solar PV.This limits its scope compared to our study's expectation of a cleaner grid by 2030 and consideration of a broader range of energy sources and storage options.

Conclusion
Many countries have set targets for clean hydrogen production to reduce fossil fuel dependence and decarbonise hard-to-electrify sectors.Regulations are needed to make sure that hydrogen production contributes to decarbonisation and does not increase greenhouse gas emissions.
In this work, we investigated different ways of regulating green hydrogen production.We analyzed scenarios where the electrolysis operates directly with grid electricity without additional renewable generation and with additional local procurement.The local procurement matches the demand for electrolysis annually, monthly or hourly.
Our results reveal three low-emission and lowcost options for hydrogen production.Additional local renewable generation is necessary in all three cases to avoid increased emissions from hydrogen production.The first option is hourly matching with flexible demand or low-cost storage, which smooths out the variable feed-in of renewable generation.Flexible demand might be possible for some initial consumers like ammonia producers, who can switch easily between green and Gray hydrogen.The second option is annual or monthly matching either with flexible demand such that electrolysis capacity factors are limited to around 70% or an upper limit on the electricity price when the electrolyser is allowed to operate.The third option is annual or monthly matching if the grid already has a high share of renewable generation and coal is phased out.In our scenarios, a share of 80% of renewable generation is sufficient for negative consequential emissions with annual matching.All three options are already provided in some form by the legislation from the European Union [4].However, the order of implementation of monthly matching (transitional phase) and hourly matching (up to 2027) in the proposed legislation is inconsistent with our results.In order to limit emissions impacts, our results suggest a stricter regime in the short term that relaxes once system targets are met.For example, one could impose hourly matching or upper limits on electrolysis capacity factors until country-wide renewable targets are met and coal is phased out.Alternatively, one could relax the rules in the short term for the scale-up while volumes are small and then impose them in the medium term until targets are reached.
Compared to annual matching, hourly matching offers several benefits.Hourly matching has significantly lower attributional emissions based on the average grid mix when electricity is consumed for electrolysis.Hourly matching is the only case that provides incentives for demand flexibility and storage, which are typically cost-optimal in deep decarbonisation scenarios.If renewable electricity targets are not met, hourly matching provides a valuable hedge by guaranteeing low emissions, even in this case.Policy support mechanisms to boost green hydrogen production should be designed to promote flexible electrolysis operations to avoid increased emissions and system-friendly operation.
The regulation of green hydrogen production is often described as a trade-off between strict rules with higher costs or looser rules with potentially higher emissions.This work shows that regulations with low emissions and a small cost premium are possible.

Figure 1 .
Figure 1.Five different regulatory scenarios are modeled.A (i) grid scenario without any additional renewable generation requirement in which electrolysis is powered by the grid as well as scenarios in which additional renewable energy sources (RES) have to match the electrolysis consumption on an (ii) annual, (iii) monthly or hourly basis (iv) without and (v) with allowed excess generation of 20%.

Figure 2 .
Figure 2. Consequential emissions of hydrogen production in Germany 2025, calculated as the difference in total system emissions per produced kgH 2 compared to a reference scenario without any hydrogen production.

Figure 3 .
Figure 3. Cost of hydrogen production in Germany 2025.

Figure 4 .
Figure 4. Consequential emissions for two selected cases representing a less clean system, Netherlands 2025 (left), and a cleaner system, Germany 2030 (right), than the reference case of Germany 2025.