The impact of the energy crisis on the UK’s net-zero transition

Summary

Recent drastic increases in natural gas prices have brought into sharp focus the inherent tensions between net-zero transitions, energy security, and affordability. We investigate the impact of different fuel prices on the energy system transition, explicitly accounting for the increasingly coupled power and heating sectors, and also incorporate the emerging hydrogen sector. The aim is to identify low-regret decisions and optimal energy system transitions for different fuel prices. We observe that the evolution of the heating sector is highly sensitive to gas price, whereas the composition of the power sector is not qualitatively impacted by gas prices. We also observe that bioenergy plays an important role in the energy system transition, and the balance between gas prices and biomass prices determines the optimal technology portfolios. The future evolution of the prices of these two resources is highly uncertain, and future energy systems must be resilient to these uncertainties.

Graphical abstract

Highlights

• •Fuel prices matter for heat, less so for power
• •Identifying low-regret options is key
• •Biomass valuable to replace firm gas-fired generation at higher gas prices
• •Heat pumps as main abatement technology attractive at higher gas prices

Energy policy; Energy management; Energy modeling; Energy systems; Biofuel

Introduction

An increasing number of administrations have committed to ambitious decarbonization targets to meet the climate goals laid out in the Paris Agreement.¹ More than 70 countries, responsible for 76 % of global emissions, have set net-zero targets.² Current decarbonization strategies were developed under the premise of low and stable gas prices. For example, the modeling work underpinning the UK net-zero strategy assumes a central scenario with a gas price of 19.5 £/MWh.³ Natural gas has historically played an important role in reducing the carbon intensity of power grids by displacing coal and can contribute further to net-zero energy systems if deployed with carbon-capture and storage (CCS). However, natural gas prices have recently experienced drastic increases in large parts of the world. In Europe and the UK, this price surge was initially driven by lower wind availability and thus a higher gas demand for power generation, increased natural gas demand from Asia, and higher oil prices.⁴ In 2022, the situation was exacerbated by Russia’s war on Ukraine and related political decisions on gas supply from Russia. US natural gas prices are also affected by the surging prices in Europe due to a higher demand for liquified natural gas (LNG) exports. Additionally, warmer weather caused a higher cooling demand and thus a higher natural gas demand for power generation.⁵ Similarly, prices in Asia have increased due to higher global LNG prices and increased demand from the region.⁶ These price surges raise the question whether current decarbonization strategies are still adequate.

More broadly, the challenge for policymakers and advisors is to design efficient, robust, and equitable decarbonization strategies while facing highly uncertain, volatile, and potentially high fuel prices. Energy policy has been conventionally framed as having to address the trilemma of energy security, affordability, and sustainability.⁷ Energy security is essential, as widespread blackouts have potentially devastating consequences, including a breakdown of water and food supply and the healthcare system.⁸ Access to affordable energy for citizens and businesses is of critical relevance, as increased energy consumption is linked to increased quality of life,⁹ and energy poverty is a concern.^10,11 Environmental sustainability is becoming increasingly important as net-zero carbon emission targets are being adopted by more and more countries. Increased energy costs make it harder to strike the balance among these aspects. We appear to be transitioning to a new paradigm, and it is thus important to understand how the trilemma can be resolved in this new context.

Figure 1A shows the development of natural gas prices in key markets over the last 25+ years, while Figure 1B shows a characterization of the energy systems of different countries based on gas price and consumption, using values from 2019. Until mid-2021, the prices of natural gas futures were low and showed only relatively small variability. In the UK, used as a case study in this work, the average price between June 2017 and June 2021 was 15 £/MWh, with a standard deviation of 5.4 £/MWh. However, over the last year (June 2021 to June 2022), the average price increased to 72 £/MWh, while the standard deviation was more than five times higher at 27 £/MWh. UK gas futures prices peaked at over 170 £/MWh on the seventh of March 2022 during Russia’s attack on Kyiv. The gas supply from Russia to Europe through the three main pipelines in June 2022 was 48% lower than that in June 2021 and 57% lower than that in June 2019.¹²

Figure 1.

Graphs showing the development of natural gas futures prices in key markets since the 1990s; and the characterization of energy systems of selected countries based on natural gas price and consumption, total energy consumption, and availability of renewable energy sources in 2019

(A) The shown natural gas prices are Dutch title transfer facility (TTF) Futures, a commonly used indicator for EU prices; UK Futures; Henry Hub Futures, a reference for North American prices; the Japan Korea Marker (JKM), indicative of LNG prices in Asia; and Gulf Coast LNG Futures, a reference for US LNG prices. (B) Gas prices and total energy supply of countries from the IEA,¹³ natural gas consumption from BP,¹⁴ average availability of variable renewable energy (VRE) sources from the Renewables Ninja tool.^15,16,17 VRE availability is hereby the average capacity factor of onshore wind, offshore wind, and solar PV over a year. Countries are labeled according to their ISO-3166-1 country code. All values are from 2019.

The UK’s energy system has historically been characterized by low natural gas prices and a high share of natural gas in total energy consumption, as shown in Figure 1B. It is therefore exceptionally vulnerable to price shocks. Similarly, many other European countries such as the Netherlands, Italy, Belgium, Germany, and Spain face significant risks from natural gas price changes due to their high consumption. The United States and Canada also have low prices and high consumption, but the risk of price shocks is smaller due to high domestic production and the limited possibility to participate in physical global trading. North American countries can only trade natural gas with the rest of the world in form of LNG, domestic production of pipeline gas is large unaffected by global prices (see Henry Hub price in Figure 1 (A)). Countries with low natural gas consumption, such as the Scandinavian countries, Switzerland, France, and Chile, are also less exposed to price variability.

Given the uncertainty around energy prices, it is tempting to delay action on the energy system transition. However, many studies^{18,19,20,21,22,23,24,25} have shown that this increases the risk of missing climate targets and may ultimately result in a more expensive system. This poses the challenge of designing the energy transition and supporting policies, such that significant action is already taken soon while still maintaining sufficient flexibility to adapt to potential future price changes. Low-regret decisions, which are valuable in every scenario, should be made now. The quick deployment of technologies up to a level that provides value in each scenario should be supported.

Various energy decarbonization pathways have been proposed in the literature. While many studies focus on a fully renewable energy supply,^{26,27,28,29,30} mainly based on wind and solar energy, the feasibility and cost-effectiveness of such approaches are debated.^{31,32,33,34,35} In any case, natural gas plays an important role in many energy systems today and will continue to do so throughout the energy system transitions, and potentially also in net-zero energy systems, when coupled with CCS. Therefore, considering the price uncertainty is important.

Though it has been identified as a critical uncertainty for decarbonization pathways,^36,37 studies considering fuel price uncertainties are rare. Additionally, most published studies either do not consider net-zero targets, only consider certain parts of the energy sector, use highly simplified models optimizing only technology portfolios, and/or consider only small ranges for fuel price uncertainties.

Brown et al.³⁸ used a scenario-based approach to assess the responses of Canadian, Mexican, and US energy systems to gas price shocks, defined as a sudden price increase by 24% or decrease by 19%. Seven different energy system models were used in the analysis. The results show that mainly wind, solar, and coal generation is increasing or decreasing in case of a high or low gas price shock. The work also shows that increased penetration of renewables reduces price increases in high price shock scenarios, but also reduces cost benefits from low price shocks. It, therefore, acts as a damper on energy system costs.

Moret et al.³⁹ performed a global sensitivity analysis on uncertain input parameters for strategic energy planning models, using a snapshot model of the whole-energy system of Switzerland. Over 400 uncertain input parameters were considered, including gas prices, which were assumed to vary between −47% and +90% from the baseline value. Results show that interest rates and resource costs are the most influential uncertain input parameters. Further, it is concluded that a different characterization of input uncertainties can substantially change the results. Cunico et al.⁴⁰ developed a fuzzy optimization model for investments in the Argentinian energy sector, considering uncertainties in oil and gas prices via triangular fuzzy numbers. Results were shown to be different compared to a deterministic approach, with the fuzzy approach resulting in larger overall installed capacities to hedge against risk. Suo et al.⁴¹ applied an interval-programming approach to a model of the Chinese energy system. This approach only considers the upper and lower bounds of uncertain input parameters, thus reducing the complexity. Results show that the power generation mixes in the lower and upper scenarios are similar.

Several studies using a Monte Carlo approach to address uncertainties in energy system models have been published. Lehtveer and Hedenus⁴² performed a Monte Carlo analysis with a global capacity-expansion model, with a focus on nuclear power. The model is rich in technology options, but very simplified in terms of technical detail. Baseline, +50 % and −50 % gas price scenarios were considered, in addition to CCS, renewables, and nuclear capital cost uncertainties. Pilpola and Lund⁴³ performed a Monte Carlo analysis on a snapshot model of the Finnish energy system in 2050, considering uncertainties on demand levels, renewable potential, and costs. Results show a significant range in annual costs, carbon emissions, and technology capacities of the optimal energy systems. Energy demand, wind power capacity factors, biomass potential, and carbon price were identified as the most influential input parameters. Pye et al.³⁷ used a Monte Carlo approach and sensitivity analysis to analyze the influence of uncertainties on UK energy transition pathways, enforcing an 80 % emission reduction target. The energy system modelling environment (ESME) UK whole-energy system model was employed for the analysis, and gas prices of 14 to 36 £/MWh were considered, as well as various investment costs, build rates, and biomass availability uncertainties. The gas price and biomass availability were identified as the most influential uncertain input parameters. Li and Trutnevyte⁴⁴ also applied a Monte Carlo analysis in a UK context, focusing on power sector investments. Additionally, a modeling-to-generate-alternatives approach was used to generate 800 near-optimal electricity sector transition pathways. Results show that investments in the UK power sector until 2030 likely need to match or exceed historical precedents regardless of climate policy. To achieve climate targets, these high investments then need to be sustained until 2050, while without climate targets investments may decrease.

Ioannou et al.⁴⁵ developed a model combining a Monte Carlo approach and a scenario tree for stochastic optimization of power system capacity expansion under uncertainty. It was applied to the Indonesian power system, considering uncertainties in electricity demand, renewable capital cost reduction, as well as coal, natural gas, and diesel prices. Especially the installed gas-fired generation capacity showed a large variation, while for other technologies the variation in optimal capacity was significantly smaller. Hu and Hobbs⁴⁶ also used stochastic modeling with a decision tree to assess optimal investments in the US power sector under uncertainty. The market and allocation (MARKAL) power system model was used for the analysis, and the expected cost of ignoring uncertainty as well as the expected value of perfect information was estimated. Considered uncertainties were policies, demand growth, and natural gas prices, modeled as baseline, +40 % and −40 % scenarios. The authors concluded that robust decision-making can improve expected performance.

Nahmmacher et al.⁴⁷ analyzed different strategies to cope with shocks to power systems, applying robust decision-making to a european union (EU) power sector investment optimization model. The authors considered a variety of power system shocks, including fuel supply shocks with a sudden gas price increase of 300% as well as a complete stop of gas imports from Russia. Similarly, sudden biomass and coal price spikes of 200% and a massive decrease in biomass availability are considered. It was concluded that the robustness of the system can only be increased significantly with additional reserve capacities. The overall capacity of the robust system was higher, but the technology portfolio and generation did not change qualitatively. Further, it is recommended for countries to build national reserve capacity, which enables them to always satisfy 95% of their demand domestically. Riepin et al.⁴⁸ investigated whether modeling uncertainty in electricity and gas systems is worth the effort. They applied stochastic optimization to a European electricity and gas market model, assuming that capacity investment decisions have to be made under uncertainty while dispatch decisions can be made after the uncertainty is revealed. The objective was to minimize expected total costs. It was found that the cost of ignoring uncertainties can be as much as 20% of total investment costs.

A summary of the literature is presented in short in Table S1 in the supplementary material S1. Yue et al.⁴⁹ provide a further review of approaches to deal with uncertainty in energy system models. The study presented here, though employing a scenario-based approach rather than stochastic optimization, adds to the existing scientific literature by (i) using a comprehensive technology-rich energy system model, including power, heating, hydrogen, and industry sectors; to investigate (ii) national net-zero energy system transition strategies under the light of recent price spikes. Since no stochastic optimization is performed in this study it should not be viewed as a tool for robust decision-making, but rather an exploration of the value of different technologies for the energy system transition under uncertain fuel prices.

Further electrification of the heating and mobility sectors of the energy system is widely anticipated to play a key role in achieving a net zero transition. The Climate Change Committee (CCC) estimates that the UK’s electricity demand will more than double by 2050 owing to the electrification of the economy.⁵⁰ Thus, modeling approaches considering individual sectors of the energy system in isolation are increasingly inadequate. Heat electrification not only changes absolute demand values but also the shape of the load duration curve, with higher peaks on winter mornings and evenings.^51,52 Additionally, air conditioning in summer becomes an increasingly important element of thermal management. It is therefore essential to perform an integrated assessment of the whole-energy system when investigating net-zero strategies.

In this work, we use the newly developed ESO-HeSO model (see Figure 2), a capacity-expansion and unit-commitment optimization model that identifies least-cost whole-energy system transitions, to examine the impact of fuel price changes on net-zero strategies. The model, based on the electricity system optimization (ESO) model developed by Heuberger et al.,^53,54 performs a simultaneous optimization of investment decisions and dispatch profiles to obtain the least-cost energy system transition to 2050. This study extends the ESO framework beyond the electricity sector to include the emerging hydrogen sector, and also the domestic, commercial, and industrial heating sectors. In this work, the model is applied to the UK energy system. The UK is an ideal region to study, as it has committed to a net-zero emissions target by 2050 and is exceptionally vulnerable to gas price shocks. However, insights are relevant for similar regions with decarbonization targets and high gas consumption as well.

Figure 2.

Overview of the ESO-HeSO optimization model that optimizes the energy system transition until 2050, explicitly accounting for power, hydrogen, domestic and commercial heating, and industrial heating sectors

Results and discussion

Low versus high natural gas price worlds

We begin our investigation by comparing optimal energy system transitions for scenarios with sustained low and sustained high natural gas prices. Gas prices are assumed to be constant at 15 £/MWh and 72 £/MWh, the average values between June 2017 – June 2021 and June 2021 – June 2022 as shown in the introduction. In all scenarios, unless stated otherwise, a linear decarbonization trajectory to reach net-zero in 2050 is enforced. We also assume carbon prices for electricity, hydrogen, and industrial heating sectors are in line with projections from the Department of Business, Energy & Industrial Strategy (BEIS).⁵⁵ Additionally, the targets of at least 40 GW of offshore wind capacity⁵⁶ as well as 5 MtCO₂/a of negative emissions⁵⁷ by 2030 set out by the UK government are enforced in the model.

The optimal energy system transition for the low-gas price scenario is shown in Figure 3. In the low-gas price scenario, rapid deployment of significant power generation capacity is required. Starting from 95 GW in 2020, the total installed capacity reaches 210 GW in 2040. This increase in required generation capacity is driven by increasing demand arising from the electrification of transport and some heating, as well as growing shares of non-dispatchable renewable energy technologies, which in the UK only reach average annual capacity factors of 26% (see Figure 1B).

Figure 3.

Optimal energy system transition for a low gas price world (15 £/MWh)

Shown are the evolution of (A) power generation capacity; (B) power generation; (C) hydrogen generation; (D) system emissions by sector; (E) domestic heat generation; (F) commercial heat generation; (G) low-temperature industrial heat generation; and (H) high-temperature industrial heat generation from 2020 to 2050.

Onshore wind, offshore wind, and solar photovoltaics (PV) account for the biggest share of additional capacity. Maximum build-rate constraints are hit during the first years, meaning that these technologies are deployed as fast as possible. UK policy in this area is complex. Even though 70% of British adults generally support more onshore wind turbines (10% oppose),⁵⁸ including 72% of survey participants who live within 5 miles of a wind farm,⁵⁹ political support from the UK government^60,61 and local authorities⁶² is lacking. Thus, the maximum capacity of onshore wind is set to 40 GW, which is about three times as high as the currently installed capacity. Model runs without this constraint show that onshore wind capacities of up to 130 GW provide system value, displacing mainly offshore wind generation capacity. Due to lower investment costs, onshore wind is preferred over offshore wind despite having lower capacity factors.

In addition to wind and solar, the nuclear power generation capacity increases from 2035. At the same time, combined-cycle gas turbines (CCGTs) fitted with carbon-capture and storage (CCS) systems and bioenergy plants with CCS (BECCS) are deployed. The latter provide negative emissions to offset residual emissions from the remaining gas-fired electricity generation assets, as well as other sectors of the energy system. Unabated CCGT and open-cycle gas turbine (OCGT) capacity remains in the system until 2050 to provide valuable flexibility and system inertia. A comparison between installed capacities and annually generated power, as well as analyzing dispatch profiles (see Figures S1 and S2), reveals that nuclear power plants and BECCS plants provide baseload generation, renewables generate power whenever available, using also pumped-hydro storage and interconnections, and gas turbines act as peaking plants.

In the cost-optimal solution for the low gas price scenario, large amounts of natural gas (>360 TWh/a) continue to be burned in gas boilers throughout the time horizon to provide domestic and commercial heating. Only about half of the commercial heating demand is electrified by 2050, while hydrogen is irrelevant for domestic and commercial heating in this scenario. Low-temperature industrial heat generation switches from natural gas to heat pumps by 2040, while for high-temperature industrial heating hydrogen boilers are used as transition technology from 2040, until in 2050 most demand is electrified. The decarbonization of industrial heating is partly driven by the carbon price applied to this sector. Hydrogen boilers provide value during the transition while the power capacity is ramping up. The required hydrogen, up to 45 TWh/a, is generated from steam-methane reformers (SMRs) with CCS. Finally, in 2050 when the net-zero emission constraint applies, most industrial demand is electrified using carbon-negative power.

The emissions from burning natural gas for domestic and commercial heating are offset by the significant deployment of BECCS, which provides 75 MtCO₂/a of negative emissions in 2050. This is associated with a total biomass consumption of 290 TWh/a in 2050, which can be achieved in the UK with biomass pellet imports totaling 94 TWh/a.⁶³

We have also considered a context where biomass supplies are relatively scarce. A scenario in which the annual biomass consumption is limited to 98.4 TWh (No biomass imports are allowed, and availability of domestic sources is assumed to be half of the baseline values) is shown in Figure S3, it comes at only 1.9 % higher total system costs. That scenario includes the electrification of 40% of the domestic heating demand via air-source heat pumps (ASHPs) by 2050, and the use of predominantly blue hydrogen for another third of the domestic heating demand. This indicates significant flexibility in designing cost-effective net-zero energy system transitions, as two very different systems yield similar total costs. This aligns with recent energy systems transition literature, such as Neumann and Brown⁶⁴ and Pickering et al.⁶⁵ A further investigation is provided in a later section.

The cost-optimal energy system transition identified by the model is notably different in the high natural gas price scenario, see Figure 4. About half of the domestic heating demand and 57% of the commercial heating demand are electrified by 2050, using ASHPs. Heat pump deployment starts in 2025, with maximum build-rate constraints being hit in the first period. As in the previous scenario, low-temperature industrial heat is also electrified via heat pumps. For high-temperature industrial heat, electric heating is used throughout. The higher natural gas price results in SMR hydrogen as transition fuel being less competitive.

Figure 4.

Optimal energy system transition for a high gas price world (72 £/MWh)

Shown are the evolution of (A) power generation capacity; (B) power generation; (C) hydrogen generation; (D) system emissions by sector; (E) domestic heat generation; (F) commercial heat generation; (G) low-temperature industrial heat generation; and (H) high-temperature industrial heat generation from 2020 to 2050.

The increased heat electrification results in a 15% higher total electricity demand in 2050 compared to the low gas price scenario. This causes the installed solar capacity in 2050 to double to a total of 106 GW, while the nuclear capacity increases from 25 GW to 39 GW. The installed BECCS capacity does not change significantly, but the annually generated power from BECCS is smaller since fewer negative emissions are required. Significant CCGT and OCGT capacities (15 GW and 5 GW in 2050, respectively) remain in the system despite the high gas prices. They provide peak power generation, flexibility, and system inertia. The annual power generation from these technologies however is negligible. Therefore, unabated gas plants are preferred over CCGTs with CCS deployed in the previous scenario, since they have lower capital costs and emissions from the little power generation are insignificant. Biomass generation without CCS is relevant especially in the transition years from 2025 to 2045, displacing firm gas-powered generation due to the higher gas price. Some of the capacity gets retrofitted with CCS in 2050.

Hydrogen only plays a marginal role in this scenario. Blue hydrogen from SMRs is unattractive due to the high gas price, and heat pumps are more efficient than the combination of electrolyzers and hydrogen boilers. Nevertheless, small quantities of electrolytic hydrogen (3.5 TWh/a in 2050) are generated to provide some flexibility to the electricity system. In terms of total system costs, the high gas price scenario is 94% more expensive than the low gas price scenario. Considering that the price of natural gas, the main fuel used in the low gas price scenario, is almost five times as high, this substantial increase in total system costs is not surprising.

Sensitivity of technology roles and energy system transition strategies toward different natural gas prices

To further analyze the impact of changing natural gas prices on optimal energy system transitions, we perform the system optimization for natural gas prices from 10 £/MWh to 200 £/MWh. The prices are constant in each scenario, and the same for every energy sector considered in the model. A summary of the results is provided in Figure 5, while additional results are shown in Figures S4 and S5.

Figure 5.

The optimal net-zero energy systems in 2050 for natural gas prices ranging from 10 £/MWh to 200 £/MWh

Shown are (A) power generation; (B) hydrogen generation; (C) domestic heat generation; and (D) commercial heat generation by technology. The dashed lines correspond to the low- and high-gas price scenarios shown in Figures 3 and 4, while the dotted line represents the maximum price seen in the UK so far.

The first thing to note is that the optimal degree of heat electrification depends heavily on the gas price. While for prices below 20 £/MWh, all domestic heat in 2050 is provided from gas boilers, for prices above 130 £/MWh all domestic heat is electrified via ASHPs. A similar trend, though less extreme, is observed in the commercial heating sector. For a gas price of 10 £/MWh, 48% of the commercial heat demand in 2050 is provided by ASHPs, while for a gas price of 200 £/MWh, this value increases to 94%. Regardless of the gas price, low-temperature industrial heat is provided by ASHPs in 2050. High-temperature industrial heat is provided from hydrogen boilers only for gas prices below 20 £/MWh, and from electric heating otherwise.

The increasing heat electrification with higher gas prices results in both higher peak and higher total annual electricity demand. This increased demand is mainly met by an increase in solar, offshore wind, bioenergy, and nuclear generation capacity. The deployed solar capacity at a gas price of 200 £/MWh is 4 times higher compared to the capacity at a gas price of 10 £/MWh, while the nuclear capacity increases by a factor of 2.3. Onshore wind already reaches its capacity limit for low gas prices; therefore, an increase in capacity to meet additional electricity demand is not possible.

CCGTs and OCGTs are part of the optimal technology portfolio in 2050 regardless of the gas price, indicating once more the value these technologies offer as dispatchable assets for demand peaks, as backup capacity, and to provide system inertia. The actual amount of power generated from CCGTs and OCGTs is small, only of the order of 1% of total generation for high gas prices, indicating that they solely run during peak hours. These small capacity factors are the reason that unabated CCGTs are preferred over CCGT plants with CCS, as discussed above. At gas prices below 40 £/MWh on the other hand, CCGTs with CCS are preferred, as capacity factors are higher.

A similar trend is observed for bioenergy plants, in which BECCS is preferentially being deployed for low gas prices and bioenergy plants without CCS playing a larger role at high gas prices, albeit for different reasons. At low gas prices, residual emissions from gas boilers used for heating are high, meaning that substantial amounts of carbon removal are required. At gas prices of 10 £/MWh, this is achieved by running 11.4 GW of BECCS capacity at a capacity factor of 100%, generating 100 TWh/a of electricity and 79 MtCO₂/a of negative emissions in 2050. Conversely, when gas prices are high, fewer negative emissions are required (11 MtCO₂/a at a gas price of 200 £/MWh), hence the cheaper and more efficient bioenergy plants without CCS are deployed more as low cost, reliable renewable power is more important than carbon removal.

Hydrogen only plays a small role in the scenarios considered here. Only at gas prices below 20 £/MWh, where blue hydrogen from SMRs is competitive for high-temperature industrial heating, the total hydrogen generation in 2050 exceed 20 TWh/a. At gas prices over 30 £/MWh, virtually no blue hydrogen is being used by 2050. The production of electrolytic hydrogen increases as the gas price increases, up to 20 TWh/a for a gas price of 200 £/MWh. This hydrogen provides some flexibility to the electricity sector. It can be generated at times with high availability of renewables and then used as a dispatchable, carbon-free power generation source in hydrogen turbines. In other words, it is essentially an energy storage option.

It is, however, important to note this result is predicated on the assumption that all industrial heat demand can, in fact, be electrified. In reality, hydrogen may be required for certain industrial processes. The model also does not account for potentially required electricity grid reinforcements or the deployment of hydrogen infrastructure. Thus, local differences such as potentially preferred use of hydrogen in/near hydrogen hubs, cannot be captured. Another significant assumption, which is supported by a recent report,⁶³ is that all domestic properties are suitable for ASHPs.

The total system costs are 3.4 times higher at gas prices of 200 £/MWh compared to the baseline at 15 £/MWh. The marginal increase in costs decreases slightly with higher gas prices. The degree of heat electrification increases rapidly between 30 and 120 £/MWh, where most of the domestic and commercial heating is electrified. Natural gas use in 2050 shows the opposite trend. It declines steeply from 10 to 20 £/MWh due to SMR hydrogen disappearing from the system and then declines further as heat is electrified. The required negative emissions follow a similar trajectory, though the disappearing SMR generation only has a small effect since most emissions from SMRs are captured by the CCS system.

At gas prices of up to 50 £/MWh, all biomass is used in BECCS plants to provide power and negative emissions. For higher gas prices, however, biomass consumption is not coupled to the demand for negative emissions anymore. Biomass is also being burnt in plants without CCS to provide dispatchable, low-carbon power. At 200 £/MWh, biomass consumption is still at over 70% of the maximum of 300 TWh/a, while negative emissions are only 14% of the maximum.

The cost of decisions and policies — How expensive are political preferences?

In an ideal world, policy interventions in the energy transition ought to help guide the system along the least-cost pathway and should be resilient to uncertainty regarding energy costs. Figure 6 shows the impact of selected technology-deployment and policy decisions on total system costs. The optimal solutions for different natural gas prices, shown in the previous section, serve as a baseline.

Figure 6.

Added total system costs due to different technology-deployment and policy decisions for different natural gas prices

The dashed lines correspond to the low- and high-gas price scenarios shown in Figures 3 and 4, while the dotted line represents the maximum price seen in the UK so far.

The results show that for most electricity generation technologies, the power sector can cope well with the decision not to deploy it. This indicates flexibility in selecting the desired technology portfolio without a large impact on costs. The decisions not to build any more nuclear power plants or not increase installed onshore wind capacity result in a maximum of 5% increase in total system costs regardless of the natural gas price. Decisions not to increase the installed solar capacity or not to build any new CCGTs or OCGTs have an even smaller impact on total system costs. Only if no additional renewable capacity is installed at all (wind or solar), the energy system transition becomes significantly more expensive. While the added costs are only 3% at a gas price of 15 £/MWh, this value increases to 15% at 72 £/MWh and reaches a maximum of 29% higher total system costs at 200 £/MWh.

On the heating side, both phasing out gas boilers and not allowing any heat electrification are associated with cost risks. Enforcing a phaseout of gas boilers in 2035 can add up to 9% to the total system costs at a gas price of 10 £/MWh. This value decreases to less than 5% for gas prices above 50 £/MWh. Not deploying any electric heating on the other hand results in over 5% higher total system costs for gas prices above 20 £/MWh, and over 10% higher costs for gas prices above 90 £/MWh. The maximum increase in total system costs is 21% at 200 £/MWh. To properly account for risk associated with certain decisions, the expected distribution of gas prices would have to be known. However, for a wide range of studied prices, the cost risk of not deploying heat electrification is higher than the risk of phasing out gas boilers.

The amount of biomass used annually by the energy system can be limited to 98.4 TWh (no biomass imports are allowed, and availability of domestic sources is assumed to be half of the baseline values) at only little additional costs. At maximum, the total system costs increase by 4%. This shows that initially, reducing the need for negative emissions and thus biomass consumption via some heat electrification comes at a relatively low price. However, some negative emissions will always be required to reach net-zero emissions.

The decision not to deploy any gas CCS (CCGT or SMR with CCS) is cost-neutral over a wide range of gas prices. For high gas prices, unabated CCGT, OCGT, and bioenergy plants are preferred to provide flexibility, with residual emissions being offset by BECCS. Additionally, electrification is preferred over blue hydrogen for industrial heating. Only at a gas price of 10 £/MWh, where both CCGTs with CCS and blue hydrogen from SMRs with CCS are used in the optimal scenario, the decision not to deploy these technologies adds 1.3% to the total system cost.

Preventing any BECCS deployment results in 1 to 2% higher total system costs regardless of the gas price, since BECCS must be replaced with the more expensive biomass gasification with CCS hydrogen production to provide negative emissions. If both BECCS and biomass gasification are prohibited the model becomes infeasible, since it has no way of providing negative emissions. Direct air capture is an alternative that could be explored in the future.

The target of 5 MtCO₂/a of negative emissions by 2030, set out by the UK government, has no impact on total system costs. Regardless of the gas price, the required negative emissions in the cost-optimal energy system are higher; thus, the policy does not impact the system design. Relaxing the offshore wind deployment target of 40 GW by 2030 on the other hand yields lower total system costs compared to the baseline at gas prices below 60 £/MWh. Offshore wind is more expensive than onshore wind or solar PV, and the higher capacity factors are not sufficient to counteract this. Therefore, prioritizing offshore wind deployment over other renewables results in higher costs. Only at higher gas prices, when aggressive deployment of all renewable technologies is required, the offshore wind deployment target has no impact on total system costs.

It should be noted that overall, changes in total system costs are rather small. This can partially be explained by the fact that there is limited leverage during the first years. Today’s technology portfolios are given, and any change takes time due to build rate limitations. Discounting of future costs also significantly limits the impact of different energy system designs toward the end of the time horizon on total system costs. It should also be highlighted that Figure 6 only shows the effect of singular decisions. The effect of combining different decisions, e.g., no heat electrification and limited biomass use, may result in more severe cost increases or even infeasible systems.

In the no heat electrification scenario, all heat pump and resistive heating options are disabled. Boiler phaseout refers to the phaseout of natural gas boilers. In the limited biomass use scenario, the annual biomass consumption is limited to 98.4 TWh (see above). No gas CCS means that CCGT + CCS for power generation and SMR + CCS for hydrogen generation are prohibited. The no offshore wind target and no greenhouse gas removal (GGR) target scenarios refer to the UK government’s targets of 40 GW of offshore wind and 5 MtCO2/a of greenhouse gas removal by 2030.

Gas price is not the only risk — biomass prices are also uncertain

In the previous sections, we analyzed the role of biomass for different natural gas prices and discussed the complementary nature of these two technologies: BECCS can offset residual emissions from burning natural gas, and both technologies can prove firm generation capacity and system inertia. So far, the price of biomass has been assumed to follow the supply curve identified by Bui et al.⁶⁶ In the following, we extend our investigation of the impact of fuel prices on the energy system transition by also considering different biomass prices. The biomass prices from the supply curve are multiplied by factors of 0.5 to 2, while the natural gas price is varied between 10 and 200 £/MWh as before. Since electricity and hydrogen production are explicitly modeled, the only other fuel prices which are model externalities are the price of coal and nuclear fuel. Coal-based power generation is scheduled to be phased out by 2025, and nuclear fuel is subject to tight regulations and long-term contracts; therefore, the impact of varying coal and nuclear fuel prices is not considered here.

Figure 7 shows key indicators of the optimal energy system in 2050 for varying biomass and natural gas prices. The total system costs show a high sensitivity toward the gas price but only change marginally with varying biomass prices. A large share of biomass consumption can be avoided at only slightly higher costs by increasing heat electrification and reducing residual emissions from the power sector. On the other hand, natural gas consumption in the first years cannot be avoided due to a large share of heating depending on it today. Limited build rates of alternative technologies slow down the potential replacement of natural gas as heating fuel. Additionally, due to cost-discounting the impact of the first years on the total system costs is the highest.

Figure 7.

Key characteristics of the optimal net-zero energy systems in 2050 for different natural gas and biomass prices

Shown are (A) total system cost; (B) the degree of heat electrification; (C) natural gas consumption; (D) biomass consumption; (E) negative emissions; (F) hydrogen production; (G) sum of onshore wind, offshore wind, and solar PV capacity; and (H) sum of gas turbine capacity.

The degree of heat electrification in 2050 also shows a high dependence on the gas price, with over 70% of heat demand being electrified for gas prices of 100 £/MWh and above, regardless of biomass prices. At gas prices of 50 £/MWh and below, the degree of heat electrification increases significantly with increasing biomass prices. Higher biomass prices result in BECCS as negative emission technology being less competitive; therefore, it is beneficial to avoid emissions by electrifying more and more heating.

The annual biomass consumption is the highest at a low gas price and low biomass prices. In that case, negative emissions from BECCS are cheap.

At low gas prices, the sensitivity of biomass consumption to biomass prices is low. For a gas price of 10 £/MWh, the annual biomass consumption only decreases by 19% when the biomass price multiplier increases from 0.5 to 2 (a 4-fold increase in prices). This indicates once more that natural gas prices are more relevant to overall costs. Even if biomass is expensive, it is still worth burning the cheap natural gas and offsetting any emissions with biomass.

At high gas prices, on the other hand, the annual biomass consumption depends strongly on the biomass prices. This is not related to the need for negative emissions, but rather to the cost of power generation from bioenergy plants without CCS.

Bioenergy plants become more competitive with gas-fired power plants with increasing gas prices and decreasing biomass prices. For this reason, the installed CCGT and OCGT capacity shows an inverse pattern. Installed capacities increase with increasing biomass prices. Only at a gas price of 10 £/MWh, where CCGTs (with CCS if biomass prices are high) provide a bigger share of the annual power generation, do we see relatively high gas-fired power plant capacities throughout. Interestingly, we observe a second peak in CCGT and OCGT capacities at medium gas prices of about 100 £/MWh. At this price, a large share of the heating demand is already electrified, increasing total power demand as well as seasonality and winter peaks. Therefore, dispatchable gas-fired capacity is valuable. At higher gas prices, however, these assets become more costly to run; consequently, installed capacity decreases.

Regardless of the gas and biomass prices, the optimal system comprises significant gas-fired power generation capacity and uses significant amounts of biomass, indicating that both are essential for cost-effective net-zero energy systems.

More than 40 TWh/year of hydrogen is produced at a gas price of 10 £/MW, where SMRs to generate blue hydrogen are competitive, regardless of the biomass price. For higher gas prices, the share of electrolytic hydrogen increases with increasing biomass prices, since burning biomass in BECCS plants to provide negative emissions becomes less attractive and hydrogen as heating fuel can reduce emissions compared to gas boilers. Additionally, hydrogen production also increases with higher gas prices, as also seen in Figure 5, as more and more hydrogen is used for firm power generation.

The installed capacity of renewable power generation, onshore and offshore wind as well as solar PV, increases with both increasing gas prices and increasing biomass prices. These technologies have no operational costs and have, therefore, become more competitive as the fuel prices of other technologies increase. However, even at a gas price of 10 £/MWh and biomass prices of only half of the baseline values, more than 130 GW of wind and solar generation capacity are installed in the cost-optimal system in 2050. This highlights the value of these technologies. For a gas price of 200 £/MWh and doubled biomass prices, the installed wind and solar capacity reaches 385 GW.

It should be noted that the feasibility of such high renewable capacities needs to be investigated. A recent study estimated the total potential for large-scale solar farms to be 263 GW to 2,460 GW depending on grid connection constraints.⁶⁷ The technical potential for onshore wind in the UK was estimated to be 690 GW,⁶⁸ while the estimated potential for offshore wind ranges from 308 GW at 110 £/MWh levelised cost of energy (LCOE) to 2,888 GW at 160 £/MWh.⁶⁹

Natural gas and biomass prices are varied independently here to explore a wide range of potential future scenarios. The two markets are very different. Natural gas is a globally traded commodity with a large market with good liquidity. It can easily be transported in large quantities over long distances. Therefore, it is easily affected by global politics and international trade policies. Biomass on the other hand is a smaller, more local market that is still emerging. Nonetheless, it is possible that the two prices may be somewhat linked in the future, as ultimately, both fuels can fulfill a similar demand in decarbonized energy systems. Even though the majority of natural gas is used for non-electricity generation purposes, a higher reliance on biomass for power generation will reduce the natural gas demand of the power sector, which in turn may lead to lower natural gas prices. The results show that policymakers are to some degree able to decide how much to rely on either commodity in the net-zero transition. The different markets, and uncertainties and risks associated with these markets, should affect this decision.

Conclusions

The results shown in the aforementioned section lead to some key observations regarding the impact of different fuel prices on optimal net-zero energy system transition strategies.

Fuel prices matter for heat, less so for power

The degree of heat electrification in the cost-optimal net-zero energy system transition depends strongly on the price of natural gas.

In a world with low gas prices, straight-forward combustion of natural gas remains integral to the provision of heating, whereas, in a high gas price paradigm, almost all heat is electrified, mainly via ASHPs. The average gas price between June 2021 and June 2022, 72 £/MWh, lies right in the transition zone: about half of the domestic heating demand is electrified in the optimized energy system in 2050.

The impact of this heat electrification on the power sector is dampened by the high efficiency of heat pumps, if they are installed and sized correctly. Field trials of ASHP installations in the UK found seasonal performance factors of about 2.5.⁷⁰ Nevertheless, the annual power demand is as much as 200 TWh higher in high gas-price scenarios owing to heat electrification. Additionally, the heat pump efficiency is worst during the coldest times when the demand is the highest. Therefore, the peak electricity demand increases substantially due to heat electrification. The required installed capacity can be more than twice as high in high gas-price scenarios compared to low gas-price scenarios. However, even though the size of the power sector increases significantly with increasing gas prices, the deployed optimal generation technology portfolio does not change fundamentally.

Onshore and offshore wind, and solar PV account for most of the installed capacity and annual power generation in 2050, regardless of the gas price. Quick and substantial deployment of these technologies is required in cost-optimal energy system transition pathways. Renewable generation is complemented by nuclear power and bioenergy. Gas-fired power plants also remain in the system regardless of gas price. The amount of power generated from these plants is small and decreases with increasing gas prices; nevertheless, they provide valuable peaking capacity and contribute to system stability.

Heat electrification contributes significantly to higher electricity demand peaks. However, heat pumps can potentially provide valuable demand shifting when coupled with a thermal store or using the thermal inertia of buildings. This option is not considered in this work but warrants consideration when designing future energy systems. Combined heat and power plants and heat networks are also potentially attractive options for net-zero energy systems that should be explored in future work.

Identifying low-regret decisions is key

Restricting and preventing the deployment of different technologies has been shown to increase total system costs by a few percent only for most technologies. This indicates that many different system designs can reach costs close to the optimum, giving the flexibility to cater to public or political preferences and other concerns. For example, political support for onshore wind is low in the UK. The results from this study show that further deployment of this technology can be avoided at only marginally higher costs if offshore wind and solar PV are deployed more extensively instead.

Similarly, many cost-optimal solutions rely on the use of large quantities of biomass, more than 300 TWh in some cases, which can only be satisfied with imports. However, if the consumption of that much biomass is a concern, it was shown that it can be reduced by more than two-thirds at less than 2% higher costs by electrifying heat and/or using alternative power generation technologies. The need for negative emissions, which are exclusively provided by BECCS, can be reduced in the same way.

However, the increased use of renewable energy sources is associated with a high space demand. At 200 W/m², a typical value for PV panels, the 230 GW of solar capacity seen in extreme scenarios corresponds to a space requirement of 1,150 km² or 0.5% of the total area of the UK. Striking the right balance between the different decarbonization options and limiting the potential harm from technologies is a challenge for policymakers. A balanced approach to the energy system transition that makes use of all available, reasonable technologies appears most appropriate.

The analysis also suggests that there’s no silver bullet for the heating sector. Both phasing out gas boilers and not electrifying any heating are associated with significant cost risks. For gas prices below 30 £/MWh, forcing a gas boiler phaseout results in higher total system costs than not electrifying any demand, while for higher gas prices it is the other way around. However, looking at individual end users, it appears that a phased approach to heat electrification seems to be the best option. While some users have low switching costs and already benefit from heat electrification at gas prices of 30 £/MWh, others continue burning gas even at prices as high as 130 £/MWh. Forcing a certain technology for all end users is therefore evidently not economical.

The deployment of renewable power generation capacity is a low-regret option. While the analysis has shown that every single renewable technology is dispensable, not deploying any new renewable capacity results in significantly higher total system costs regardless of the gas price. At the same time, there is value in keeping existing gas-fired power plants in the system and potentially constructing new ones, as these plants are present in the optimal electricity generation portfolio for every gas price. Similarly, the deployment of new nuclear capacity provides value in each scenario, especially if significant amounts of heat are to be electrified. Bioenergy, especially BECCS providing negative emissions, is also present in every scenario. The UK does not currently have any BECCS capacity or significant CCS infrastructure, but our results show that recent investments in these technologies appear to be justified. Reaching net-zero without significant deployment of negative emission technologies appears unrealistic.

Hydrogen only plays a small role in most scenarios studied in this work. However, it may be useful or even necessary for certain sectors not covered here, such as particular industrial processes or mobility. Blue hydrogen is competitive at very low gas prices, while green hydrogen becomes increasingly valuable with increasing gas prices. The switch from predominantly blue to green hydrogen happens as gas prices increase from 20 £/MWh to 30 £/MWh.

The role and value of gas and biomass will be tightly coupled

Biomass can offset emissions from natural gas when used in BECCS plants, and both bioenergy and gas-fired power plants can provide flexible, dispatchable power. The prices of both resources are the key fuel price uncertainties for future energy systems. The balance between both has a large influence on the optimal net-zero energy system design. High prices of both result in a high amount of heat electrification and deployment of large wind and solar capacities. Low prices on the other hand result in large amounts of natural gas still being used in 2050, and high utilization of biomass in BECCS plants to offset emissions. Generally, the impact of the gas price on the optimal net-zero energy system transition is more significant than that of biomass prices.

While the results presented in this paper are specific to the UK energy system, conclusions can be drawn for the decarbonization of energy systems of similar countries with electricity and heat demand. Especially the trade-off between natural gas and biomass use, and the gradual switch from gas boilers to ASHPs for heating with increasing gas prices are relevant. Additionally, it is evident that both the power sector and the heating sector will likely require a diverse mix of technologies, and some capacity of negative emission technologies will be required to reach net-zero.

The results agree with existing literature in that natural gas prices have the largest impact on energy system transition costs,³⁷ that a diverse mix of generation technologies is required,^44,47 and that different pathways can lead to similar transition costs.^64,65 Further, it was shown that the optimal power generation mix does not change qualitatively with changing gas prices. This aligns with the results of the stochastic optimization of Nahmmacher et al.,⁴⁷ who identified similar technology portfolios for the default and robust strategy.

The scenario-based cost-minimization approach assuming perfect foresight comes with a set of limitations and does not accurately reflect real-world decision-making. However, the results still show the effect of fuel prices on idealized decarbonization pathways, the role of different technologies, and the cost of decarbonization. Therefore, the results are relevant for policymakers as well as industry stakeholders.

Limitations of the study

This study uses an integrated whole-energy system optimization model to study the impact of varying natural gas and biomass prices on the cost-optimal energy system transition in the UK and similar countries.

The energy system optimization model comes with a range of limitations. Firstly, it assumes the perfect foresight of a central decisionmaker. Electricity grid reinforcement and expansion is not modeled. Transport sector electrification is considered as exogenous demand, thus not optimized. Further, the industrial heating sector is modeled in a simplified way, as low- and high-temperature demand, with the assumption that all demand can in principle be electrified. International electricity trade is also modeled in a simplified way. Interconnections between the UK and mainland Europe are considered, but no international markets are modeled.

Fuel price changes are exogenous to the model. Potential causes for price changes, such as resource scarcity or political instability cannot be captured with this model. Global energy markets are not modeled. The study does not aim to explain or predict fuel price changes, but rather to investigate the impact of exogenous changes on optimal net-zero transitions and the robust design of future energy systems.

The study assumes fixed capital costs of technologies, rather than applying learning rates. A previous study⁵⁴ using only the power sector model has shown that learning rates are highly uncertain, and different learning rates may affect results quantitatively, but not qualitatively.

This study explicitly focuses on the effect of uncertain fuel prices. However, capital costs, operational costs, and performance, especially of emerging technologies such as renewables and CCS, are also associated with significant uncertainties, which are not captured in this work. Changes in these parameters are not expected to be as drastic and unpredictable as the historical turbulences in fuel prices. However, uncertain technology costs also have an impact on optimal net-zero energy system transitions.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Deposited data
Availability of renewable energy sources	Renewables Ninja	https://www.renewables.ninja/
Baseline fuel prices	Department for Business, Energy & Industrial Strategy (BEIS)	https://www.gov.uk/government/publications/fossil-fuel-price-assumptions-2019
Carbon emission factors	Department for Business, Energy & Industrial Strategy (BEIS)	https://www.gov.uk/government/publications/greenhouse-gas-reporting-conversion-factors-2019
Carbon price projections	Department for Business, Energy & Industrial Strategy (BEIS)	https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/794188/2018-short-term-traded-carbon-values-for-modelling-purposes.pdf
Library of price and performance data of domestic and commercial technologies for low-carbon energy systems	Zenodo	https://doi.org/10.5281/zenodo.4692648
Power generation and storage technical and economic data	Department for Business, Energy & Industrial Strategy (BEIS)	https://www.gov.uk/government/publications/beis-electricity-generation-costs-2020
Power generation and storage technical and economic data	European Commission	https://energy.ec.europa.eu/data-and-analysis/energy-modelling/eu-reference-scenario-2020_en
Power generation and storage technical and economic data	International Renewable Energy Agency (IRENA)	https://www.irena.org/Statistics/View-Data-by-Topic/Costs/Global-Trends
Power generation and storage technical and economic data	McKinsey & Company	https://www.mckinsey.com/∼/media/mckinsey/dotcom/client_service/epng/pdfs/transformation_of_europes_power_system.ashx
Typical interest rate, uptake of electric vehicles	Climate Change Committee (CCC)	https://www.theccc.org.uk/wp-content/uploads/2020/12/The-Sixth-Carbon-Budget-Methodology-Report.pdf
UK commercial building and heat demand data	UK Government	https://www.gov.uk/government/statistical-data-sets/live-tables-on-energy-performance-of-buildings-certificates
UK domestic building and heat demand data	Department for Business, Energy & Industrial Strategy (BEIS)	https://www.gov.uk/government/publications/cambridge-housing-model-and-user-guide
UK energy system data	This paper (supplemental information)	S3
UK hourly electricity demand	National Grid	https://www.nationalgrideso.com/industry-information/industry-data-and-reports/data-finder-and-explorer
Weather data, typical meteorological year	European Commission	https://re.jrc.ec.europa.eu/pvg tools/en/
Software and algorithms
UK energy system model	This paper (supplemental information)	S3
Other
Summary of existing literature on the impact of fuel price uncertainties on optimal energy system transitions	This paper (supplemental information)	S1
Conversion between commonly used fuel price units	This paper (supplemental information)	S4

Resource availability

Lead contact

Further information and requests for resources should be directed to and will be fulfilled by the lead contact, Niall Mac Dowell (niall@imperial.ac.uk).

Materials availability

This study did not generate new unique reagents.

Method details

UK energy system optimization model

The UK energy system optimization model is a linear-programming (LP) capacity-expansion and unit-commitment model that simultaneously optimizes long-term investment decisions and intra-annual hourly technology dispatch schedules. It is based on the Electricity System Optimisation (ESO) model,^53,54 which has been updated and extended with explicit representations of the domestic, commercial, and industrial heating sectors as well as the hydrogen sector. The model optimizes the energy system transition from 2020 to 2050, minimizing total system costs. Various emission targets and policy measures can be considered in the model. An overview is provided in Figure 2.

The model treats the UK as a single node. It optimizes investment decisions in 5-year periods until 2050, thus optimizing future technology portfolios. Build-rate constraints are enforced to limit the deployment of technologies to technically feasible rates. At the same time, hourly technology dispatch profiles are optimized for each 5-year period in order to calculate and minimize operational costs. Using full-hourly time series (8760 h per year) results in a very large model, therefore, to balance between model accuracy and computational complexity, a typical-day approach has been selected for this study. This approach was shown to yield accurate results for energy system models.⁷¹ Typical days are determined from annual time series using the hierarchical clustering algorithm provided by the tsam package.^72,73 The clustering is based on electricity demand, domestic heat demand, commercial heat demand, import electricity price, onshore wind availability, offshore wind availability, and solar availability, using equal weights for each of these time series. By default, the model uses 12 typical days determined from the clustering method and 3 extreme days. The extreme days are selected by maximum electricity demand, maximum domestic heat demand, and maximum commercial heat demand. They ensure that the system is capable of meeting energy demands at every time during the year.

The model objective is to minimize discounted total system costs, defined as the sum of capital costs, operational and maintenance costs, as well as additional costs such as carbon prices, over all considered sectors. A fixed discount rate of 3.5%/a is assumed in the model. Carbon prices are applied to the electricity, hydrogen, and industrial heating sectors, using the future values predicted by the Department for Business, Energy & Industrial Strategy (BEIS), linearly extrapolated to 2050.⁵⁵ The optimization model is implemented in GAMS and solved by the barrier algorithm of the Gurobi solver. All data pre- and postprocessing, as well as a graphical user interface, is implemented in Python. Details of each sector are provided in the following sections, and all model equations as well as parameter values are provided in the supplementary material S3.

Electricity

The target of the electricity sector is to satisfy demand at every time step. The electricity demand for the base year, 2020, is taken from National Grid figures.⁷⁴ Since future demand is difficult to predict, it was assumed that the pure electricity demand remains constant. However, since the transport sector is exogenous to the model, assumptions are required on future electricity demand from transport electrification. The annual electricity demand from transportation is taken from the Climate Change Committee’s (CCC) most recent carbon budget report,⁵⁰ and the hourly profile from a charging behavior study.⁷⁵ Hourly electricity demand from heat electrification and electrolysers is calculated directly in the model. Electricity exports are not considered.

A comprehensive selection of electricity generation and storage technologies are considered in the model. Available thermal power plants are coal-fired power plants, gas-fired and hydrogen-fired CCGTs and OCGTs, gas-fired CCGTs with post-combustion CCS, nuclear power plants, bioenergy, and BECCS plants. Additionally, onshore wind, offshore wind, and solar PV are available as renewable generation technologies. Interconnections with continental Europe, which allow the import of power, are also considered in the model. Modeled electricity storage technologies are pumped hydro storage, Li-ion batteries, and interconnectors. Existing interconnections with mainland Europe are modeled, and moderate expansion of interconnection capacity is allowed. Assumed cost and performance characteristics of the various technologies are provided in the supplemental information. The availability of renewable electricity generation is taken from the Renewables Ninja tool.^15,16,17

Key constraints of the electricity sector are hourly demand matching, as well as system inertia and backup capacity requirements. Additionally, operational constraints such as minimum up- and downtimes, as well as technology lifetimes, are enforced and the technology deployment is limited by build-rate constraints and maximum total capacity constraints. It was not possible to enforce minimum part-load constraints due to the LP model formulation that does not allow for integer variables. Calculated costs include capital costs of technology assets, fuel costs, fixed and variable operational and maintenance costs, startup costs, costs of CO₂ transport and storage for CCS technologies, and carbon costs.

Domestic heating

The approach to model the domestic heating sector differs from the electricity sector because of the decentralized nature. Assets cannot be shared between multiple users; hence, the energy balance has to be closed for each individual household at each time step.

In our model, we consider five representative households, which were selected from the Cambridge Housing Model⁷⁶ database using a k_medoids clustering algorithm. For each of these household categories, the model optimizes the share of users which use certain heating technology portfolios. Then, for each combination of household category and heating technology portfolio, the hourly dispatch is optimized. Heating technology portfolios hereby consist of one or two generation technologies (e.g., heat pump and electric backup heater), and optionally a thermal store. Unreasonable combinations, such as an ASHP and a ground-source heat pump, are excluded to simplify the model. The same is true for infeasible technology portfolios, such a heat pump without any thermal store. Values for the whole country are then aggregated from the different representative households.

Domestic heating assets are assumed to have fixed sizes. This assumption is necessary to allow the optimization of user shares for each technology portfolio in an LP model, and also reflects the limited options available on the market. Considered technologies are natural gas boilers, hydrogen boilers, two sizes of ASHPs, ground-source heat pumps, electric resistive heaters, and solar-thermal collectors. Thermal storage options are 150 L, 300 L and 600 L water tanks. Technology cost and performance data are taken from recent market data.⁷⁷ The performance of the heat pumps is calculated on an hourly basis according to the ambient or ground temperature, assuming a heat sink temperature of 55°C. The availability of heat from the solar-thermal collector is calculated from irradiation data. To account for inter-annual weather variation, temperature and irradiation data are taken from a typical meteorological year, generated by the PVGIS tool,⁷⁸ at the UK population⁷⁹ center.

The Cambridge Housing Model⁷⁶ provides annual demand values for the representative domestic households. These are translated into hourly demand profiles based on the ambient temperature, using the methodology of Watson et al.⁸⁰

Commercial heating

Commercial heating is modeled in a similar way as domestic heating, with six representative commercial properties being considered. These were selected based on commercial Energy Performance Certificate (EPC) records.⁸¹ The commercial properties were first grouped by (estimated) occupancy profile and office & retail, restaurants & cafes and industrial & warehouses were selected as representative types, covering about 90% of the commercial demand included in the EPC records. Each type was then divided into a small and a large instance, using a k_means clustering algorithm. This results in the aforementioned six representative commercial properties. Since the EPC records only include less than half of the commercial demand reported in the Energy Consumption in the UK (ECUK) dataset,⁸² the number of instances in each category is scaled to match the reported total annual commercial demand. The implicit assumption is that the EPC records are representative of the total commercial building stock.

Since commercial properties can vary vastly in size and are more flexible in installing multiple heating technologies, it was chosen to optimize the size of generation assets for each representative property in the model. This comes at the cost of not being able to restrict feasible technology portfolios, or to enforce minimum sizes of assets, in the LP model. Considered technologies are natural gas boilers, hydrogen boilers, ASHPs, and electric resistive heaters, as well as water tanks as thermal stores.

Hourly industrial heating demand profiles are calculated based on the ambient temperature using a delta-T/heating degree hours approach. For each hour of the year, the difference between ambient temperature and no-heat temperature, set to 15°C, is calculated. These values are then used to distribute the annual demand. Occupancy profiles are assumed and the heating demand outside these active hours is set to zero. It is assumed that office & retail properties are occupied 6 am to 8 pm on weekdays, restaurants & cafes 10 am to 12 am every day, and industrial properties & warehouses 24 h every day.

Industrial heating

Industrial heating demand is split into low-temperature and high-temperature demand according to ECUK data.⁸² Space heating, drying & separation and ‘unknown’ are hereby assumed to be low-temperature demand. High-temperature demand can be satisfied by natural gas boilers, hydrogen boilers and electric heating. For low-temperature demand, ASHPs are considered as additional option. Industrial heating is assumed to have a flat demand profile throughout the year.

Hydrogen

The hydrogen sector is modeled in a similar way as the electricity sector, with different generation assets feeding into a shared pool, the network. The key constraints, hourly demand matching, operational constraints, and technology deployment constraints, are similar, though electricity-specific constraints such as inertia requirements are missing.

No exogenous hydrogen demand is considered in the model, all demand is either due to hydrogen CCGTs or OCGTs for electricity generation or due to hydrogen boilers for heating. Considered hydrogen generation technologies are SMR with CCS, electrolysis, and biomass gasification with CCS. Salt caverns are available as hydrogen storage option.

Biomass

Biomass supply is modeled based on the work of Bui et al.,⁶⁶ who identified available quantities and prices of different biomass sources in the UK. Considered biomass types are waste wood, forest residue, crop residue, miscanthus, municipal solid waste, and imports from the EU and the US. Biomass is assumed to have embodied emissions of 50 gCO₂/kWh that cannot be captured by CCS.

Published: March 23, 2023

Data and code availability

• •The attached supplemental information includes all datasets generated or analyzed to perform this study.
• •This study does not report original code.
• •Any additional information is available from the lead contact upon request.