Fuel Switching Strategies for Decarbonising the Glass Industry Using Renewable Energy and Hydrogen-Based Solutions

Abstract

This study addresses the decarbonisation of the glass industry from an integrated energy system perspective, analysing the role of renewable electricity, furnace electrification, and hydrogen in meeting the high and continuous thermal demands of glass melting. While direct electrification represents the most energy-efficient option, its implementation is challenged by the intermittent nature and limited operating hours of renewable generation, scale constraints, and technological limitations in replacing fossil-based processes, highlighting a potential complementary role for hydrogen. A general methodological framework is first developed and then applied to a representative oxyfuel glass furnace using mixed-integer linear programming (MILP) optimisation that minimises melting costs while accounting for variable solar and wind generation, battery storage, and hydrogen production and storage. The results show that high levels of furnace electrification combined with wind-dominated renewable supply yield the lowest decarbonisation costs, which can become negative at moderate decarbonisation levels. Under the current solar–wind capacity expansion mix, the integration of battery and hydrogen storage extends achievable emission reductions from around 50% to 80%, with hydrogen acting as a complementary solution to electrification. Sensitivity analysis of energy and carbon prices, as well as technology investment costs, identifies the economic conditions in which storage-based solutions become cost-effective, highlighting the strategic role of hydrogen under conditions of low electricity prices and high fuel prices. The findings demonstrate viable pathways for deep decarbonisation of the glass sector and provide a transferable methodological framework for optimal renewable energy integration in other hard-to-abate industrial sectors facing similar constraints.

1. Introduction

Heavy industry accounts for around 70% of industrial emissions [1] and is classified as hard-to-abate due to structural and economic constraints [2]. Energy-related emissions, both direct and indirect, constitute nearly 80% of total emissions [3], highlighting the central role of energy-focused decarbonisation strategies across industrial subsectors. While process-emission-intensive sectors such as steel and cement focus on chemical and material innovations to address intrinsic chemical reactions that release CO₂, energy-emission-intensive sectors such as chemicals, aluminium, glass, and paper prioritise energy efficiency and fuel switching to reduce fossil fuel consumption.

Focusing on energy-related emissions, energy efficiency has contributed transversally to substantial gains in industrial decarbonisation driven by the large impact of energy on production costs. For instance, the Specific Energy Consumption (SEC) of steel production decreased from 50 GJ/tonne in 1960 to 20 GJ/tonne in 2020 [4], while that of glass manufacturing decreased from about 20 GJ/tonne in 1930 to around 6 GJ/tonne in 1990 [5]. Nonetheless, as thermodynamic limits of material production are approached [6], complementary strategies to reduce the carbon intensity of energy use are increasingly needed, with renewable sources playing a major role.

Renewable energy integration is essential to the sustainability of fuel switching options, such as hydrogen combustion and direct electrification, but requires the alignment of intermittent renewable electricity with the high and continuous thermal demands of heavy industry. Regardless of whether direct or indirect electrification pathways are considered [7], the magnitude and temporal mismatch between supply and demand requires careful strategic planning of renewable generation plants, energy infrastructure, and industrial systems. Industrial decarbonisation studies are increasingly shifting from treating renewable energy as exogenous to optimising its integration. While renewable energy discussions largely centre on electricity generation, the integration of renewables into hard-to-abate sectors is constrained by the limited feasibility of direct process electrification. In this context, green hydrogen offers a promising pathway to bridge this gap by converting renewable electricity into a flexible energy carrier capable of supplying high-temperature industrial thermal demand [8,9,10].

Table 1 summarises the most relevant contributions on renewable energy integration in hard-to-abate industrial sectors, reporting the industrial sectors, plant sizes and site locations considered, together with the modelling approaches adopted, the inclusion of sensitivity analyses, and the renewable energy sources, fuel switching options, and load-matching technologies examined. While these studies provide valuable insights into renewable energy integration in hard-to-abate industries, they are characterised by a high degree of heterogeneity in terms of the industrial sectors addressed, the technological solutions considered, and the scale of the analysed plants.

Reflecting its dominant contribution to industrial emissions, the metallurgical industry emerges as the most extensively analysed sector, whereas other energy-intensive industries, including glass manufacturing, remain comparatively underrepresented and often investigated under less homogeneous assumptions. Among the studies, plant capacities reach about 1 Mtonne/year for steel, cement, and aluminium, while they only reach around 100 ktonne/year for glass. Regional coverage spans different continents, including Europe, Asia and North America, with solar and wind energy analysed depending on local availability and often studied in combination to identify optimal generation mixes. Hydrogen is frequently analysed as a comprehensive strategy to cover plant thermal energy demands, while batteries are included when direct electrification is possible or when electric loads are considered. Technical evaluations are conducted either with simulation or optimisation approaches [11], sometimes complemented by sensitivity analyses of economic and environmental parameters.

Among hard-to-abate industrial sectors, the glass industry is particularly relevant to investigating alternative fuel switching pathways, due to its limited plant size and energy intensity, as well as the potential to combine direct electrification and hydrogen combustion in hybrid glass furnaces. Indeed, while electrification of glass production appears the most straightforward pathway, it is constrained by small nominal pull rates, restrictions in cullet use, lower flexibility, and shorter furnace lifetime [12]. On the other hand, hydrogen can offer more operational flexibility for renewable energy integration while imposing fewer constraints on existing melting furnaces.

Table 1. Key studies on renewable energy integration in hard-to-abate industrial sectors.

Ref.	Year	Sector	Plant Size	Site Location	Modelling Approach a	Sensitivity Analysis	Renewable Source b		Fuel Switching c		Load Matching d
							S	W	HC	DE	HS	BS
[13]	2012	Aluminium	780 ktonne/year	China	SIM	●		●
[14]	2018	Cement	2.3 Mtonne/year	Jordan	OPT		●	●				●
[15]	2021	Aluminium	1 Mtonne/year	UAE	OPT		●	●			●	●
[16]	2021	Glass	36.5 ktonne/year	Germany	OPT		●	●	●		●
[17]	2022	Steel	1.5 Mtonne/year	Spain	SIM		●	●	●		●
[18]	2023	Steel	100 ktonne/year	Italy	SIM	●	●		●		●
[19]	2024	Steel	1 Mtonne/year	United States	OPT	●	●	●	●		●
[20]	2024	General	-	Spain	OPT		●		●		●
[21]	2024	General	-	Five sites	OPT	●	●	●	●		●
[22]	2024	Ceramics	-	-	OPT	●	●	●	●		●
[23]	2024	Paper	-	Italy	SIM	●	●		●		●
[24]	2024	Steel	1.5 Mtonne/year	Finland	OPT	●		●	●		●
[25]	2025	Steel, ammonia	1 Mtonne/year, 300 ktonne/year	US	OPT	●	●	●	●		●	●
[26]	2025	Glass	110 ktonne/year	Italy	OPT		●		●	●	●
[27]	2025	Glass	110 ktonne/year	Italy	OPT		●		●	●	●	●
[28]	2025	Multiple	-	US	OPT	●	●		●		●

^a OPT, Optimisation; SIM, Simulation. ^b S, Solar; W, Wind. ^c HC, Hydrogen Combustion; DE, Direct Electrification. ^d HS, Hydrogen Storage; BS, Battery Storage.

Table 1. Key studies on renewable energy integration in hard-to-abate industrial sectors.

Ref.	Year	Sector	Plant Size	Site Location	Modelling Approach a	Sensitivity Analysis	Renewable Source b		Fuel Switching c		Load Matching d
							S	W	HC	DE	HS	BS
[13]	2012	Aluminium	780 ktonne/year	China	SIM	●		●
[14]	2018	Cement	2.3 Mtonne/year	Jordan	OPT		●	●				●
[15]	2021	Aluminium	1 Mtonne/year	UAE	OPT		●	●			●	●
[16]	2021	Glass	36.5 ktonne/year	Germany	OPT		●	●	●		●
[17]	2022	Steel	1.5 Mtonne/year	Spain	SIM		●	●	●		●
[18]	2023	Steel	100 ktonne/year	Italy	SIM	●	●		●		●
[19]	2024	Steel	1 Mtonne/year	United States	OPT	●	●	●	●		●
[20]	2024	General	-	Spain	OPT		●		●		●
[21]	2024	General	-	Five sites	OPT	●	●	●	●		●
[22]	2024	Ceramics	-	-	OPT	●	●	●	●		●
[23]	2024	Paper	-	Italy	SIM	●	●		●		●
[24]	2024	Steel	1.5 Mtonne/year	Finland	OPT	●		●	●		●
[25]	2025	Steel, ammonia	1 Mtonne/year, 300 ktonne/year	US	OPT	●	●	●	●		●	●
[26]	2025	Glass	110 ktonne/year	Italy	OPT		●		●	●	●
[27]	2025	Glass	110 ktonne/year	Italy	OPT		●		●	●	●	●
[28]	2025	Multiple	-	US	OPT	●	●		●		●

^a OPT, Optimisation; SIM, Simulation. ^b S, Solar; W, Wind. ^c HC, Hydrogen Combustion; DE, Direct Electrification. ^d HS, Hydrogen Storage; BS, Battery Storage.

Several studies have addressed decarbonisation pathways in the glass industry. At a sectoral level, technological roadmaps were defined for the UK under scenarios of reasonable action and radical transition [29]; for Germany, highlighting possible compliance with carbon dioxide budgets targets of 2 °C [30]; and for Italy, contrasting green fuels with carbon capture [31]. However, none of these roadmaps is supported by quantification of renewable capacity addition requirements. Plant-specific analysis relies on the assessment of representative furnaces, such as a small 100 t/day container glass furnace [16], a large 300 t/day container glass furnace [12], or a large 800 t/day flat glass furnace [32]. Renewable energy integration has been based on existing nearby renewable plants [26], optimised based on economic and environmental performance [26] or on the achievement of specific decarbonisation targets [27].

This study pursues a twofold objective: first, to provide a general perspective on the decarbonisation opportunities and challenges of the glass sector; second, to explore and assess potential renewable energy integration strategies through a representative case study. As a unique original contribution, the study combines a detailed techno-economic assessment of electrification and hydrogen pathways with a comprehensive sensitivity analysis of key economic parameters, including energy and carbon prices as well as technology capital investments. The results aim to guide the sector’s energy transition by identifying broadly applicable, cost-optimal decarbonisation pathways.

The work is organised as follows. Section 2 presents the issue of decarbonising the glass industry, outlining energy and carbon profiles. Section 3 presents technical constraints on the decarbonisation of glass melting, focusing on electrification and hydrogen combustion as the most promising decarbonisation pathways. Section 4 describes a general methodological framework, including the system layout, energy balances and optimisation approach, as well as key performance indicators and detailed component modelling. Section 5 analyses the impact of design choices on decarbonisation costs and conducts a comprehensive sensitivity analysis under varying economic parameters. Lastly, Section 6 pinpoints the main findings and discusses their implications for deep decarbonisation of the glass industry through renewable energy.

2. Decarbonising the Glass Industry: Status and Prospects

Glass manufacturing is among the most energy-intensive industrial processes and has traditionally relied on fossil fuels, primarily natural gas. As a hard-to-abate sector, it faces persistent decarbonisation challenges, particularly in production contexts where high-temperature thermal demand remains dominant.

The glass industry comprises several product categories, including container glass, flat glass, glass fibres, and specialty glasses, each characterised by different production volumes and energy requirements. Among these, container and flat glass represent the largest shares of production, accounting respectively for 44% and 29% at the global level, 50% and 25% in the US, and 58% and 23% in the EU [33].

Figure 1 schematically illustrates the glass production process, showing the main steps together with their characteristic temperature ranges and typical SEC values and highlighting the magnitude of the thermal and energy demands. The glass production process can be divided into four main steps: batching, melting, forming, and finishing. Batching and melting are common to all glass types, while forming and finishing differ depending on the product segment. Melting is the step that reaches the highest temperatures, typically in the range of 1500–1600 °C, and it is the most energy-intensive step of the process. In container glass manufacturing, it accounts for 80% of the total energy consumption, of which about 90% is supplied by fossil fuels, primarily natural gas. The extreme peak temperatures required determine the critical challenges associated with modifying or decarbonising this process. The schematic represents a general overview, with temperatures subject to variation based on production type and recycled material content.

From a sectoral perspective, the glass industry is characterised by significant environmental impact and overall energy consumption, with a strong reliance on natural gas due to its high thermal requirements. Several studies in the literature have quantified these impacts and energy uses at regional and global scales. The data summarised in

Table 2. Carbon and energy profiles of the glass industry across regional scales. Data from [34,35,36,37,38,39,40,41].

	CO2 Emissions (Mtonne)	Total Energy Consumption (PJ)	Natural Gas Share (%)
World	86	>800	75–85
US	15	200	75
EU	22	350	85
Italy	4	44	81

, arranged by the authors and constructed using data from the literature, provide a consistent picture of the industry’s current energy and carbon profiles.

The discrepancies observed between the regional data for the US and Europe and the global figures likely reflect the use of alternative, more carbon-intensive fuels in certain production contexts, in addition to natural gas. This highlights that energy mixes can vary significantly across regions, influencing the environmental profile of glass production.

Beyond general sector-level figures, which can only provide a broad indication, it is useful to examine energy use and CO₂ emissions relative to actual production. To this end, more detailed data on specific glass types help illustrate how energy and carbon intensity vary across products and regions. Flat and container glass require moderate energy (around 2–3 kWh/kg), whereas specialty products can reach higher energy intensities of up to 5 kWh/kg. Median specific emissions in the EU range from 0.36 kgCO₂/kg for coloured container glass to 0.50 kgCO₂/kg for flat glass, while in the US values are slightly higher at 0.40 kgCO₂/kg for container glass and 0.54 kgCO₂/kg for flat glass [42,43,44]. Energy-related emissions dominate across all glass types, accounting for 75–95% of total emissions, with process emissions contributing the remainder [45]. Fuel combustion represents the largest share (around 60–80%), while electricity accounts for 10–30%, varying with geographic location and production context [45].

As a result, several glass companies have committed to reducing their carbon emissions, defining short- and long-term decarbonisation targets for Scope 1, 2, and 3 emissions. Regarding EU container glass companies, Scope 1 and 2 emission reduction targets ranging from 25% to 50% have been established for 2030, with more ambitious values of up to 64% set for 2035 and carbon neutrality by 2050 [46]. For Scope 3 emissions, the most ambitious target identified corresponds to a 37.5% reduction by 2034 [46]. The targets set correspond to average annual reduction rates of around 4% for Scope 1 and 2 emissions and around 2% for Scope 3, testifying to the relevance of the commitments made.

3. Decarbonising the Glass Industry: Technical Framework

Several decarbonisation strategies have been proposed for the glass industry [45,47,48,49,50,51,52,53,54], most of which focus on addressing the key challenge of supplying the very high thermal power required for melting in glass furnaces with low-carbon energy. Broadly, these strategies can be grouped into three categories:

(i)

General implementation of energy efficiency and savings measures;

(ii)

Replacement of conventional fuels through electrification of the melting process, supported by renewable electricity;

(iii)

Process-oriented combustion modifications, such as oxyfuel combustion and the use of green hydrogen in blended combustion. Together, these approaches aim to maintain process performance while significantly reducing carbon emissions.

Examining the entire process provides insight into where targeted interventions can maximise energy use and carbon savings. As illustrated in Figure 2a, glass melting requires a very specific temperature profile to ensure proper elimination of gas bubbles and thermal homogenisation of glass properties. As a result, glass melting furnaces share common structural features, including a melting tank, refining zones, and delivery systems, but differ in energy supply, firing direction, and heat recovery mode, as shown in Figure 2b. The most common furnace technologies, recuperative, regenerative, oxyfuel, and all-electric, differ in efficiency, scale, and operating costs, which makes them suitable for specific glass types.

Recuperative furnaces, used for smaller or cost-sensitive production, exchange heat continuously through metal or ceramic recuperators that preheat combustion air up to about 750 °C, allowing stable firing, fine-tuning of temperature profiles, low investment costs, and relatively low NO_x emissions, but at the expense of higher fuel consumption and lower overall thermal efficiency. In contrast, regenerative furnaces, either cross-fired or end-port, use paired checker-brick chambers that alternately store and release heat, preheating the combustion air to 1100–1300 °C for significantly higher efficiency. Cross-fired regenerative furnaces, predominant in float glass production, place burner ports on opposite sides to tailor temperature profiles for high-quality fining, though they are less energy-efficient and more expensive to build. End-port regenerative furnaces, the most common for container glass, use two ports on one side to form a long U-shaped flame and are characterised by excellent energy utilisation but limited longitudinal energy control, high investment costs, and high NO_x emissions. Oxyfuel furnaces eliminate nitrogen from combustion, achieving high specific loads, reduced fuel use, and lower NO_x emissions, with the drawback of oxygen generation costs and condensation issues in flue gases. All-electric furnaces melt the batch using high-current molybdenum electrodes, enabling compact, low-emission operation ideal for high-purity glasses, though they require uninterrupted power supply and are limited for melts with strong infrared absorption or strongly oxidising conditions.

Figure 3 provides a schematic overview of typical input power and pull rates for different furnace types. While the figure is illustrative rather than exhaustive, it allows identification of plausible furnace sizes for various technological solutions. All-electric furnaces are generally limited to smaller scales (below 10 MW) and lower productivity (below 250 tonnes/day of production), whereas hybrid designs combine the high efficiency of electric and oxyfuel melting with the size advantages of regenerative furnaces.

Figure 4 illustrates representative changes in heat transfer modes associated with hybrid furnaces operating at substantially increased levels of electric boosting, as well as with hydrogen-fired configurations. While a limited electric contribution is already present in many conventional glass furnaces, further electrification alters the balance between radiative and convective heat transfer, modifying convection patterns within the glass melt. Similarly, hydrogen combustion introduces distinct flame characteristics. These changes can influence product quality, helping to explain the cautious uptake of technological innovations in the glass industry. High electric boosting reduces radiative heat transfer above the melt while enhancing convective heat transfer within the melt, effects that can be mitigated through appropriate furnace design [55]. Hydrogen combustion, by contrast, features longer, hotter flames with lower emissivity; however, numerical [56] and experimental studies [57] confirm the possible interchangeability with natural gas. From an infrastructure perspective, all major decarbonisation options entail a substantial increase in system power capacity and renewable energy input.

Electrification entails a substantial increase in the use of electricity within the plant, not necessarily through dedicated on-site generation, but by leveraging industrial facilities as flexible and large-scale consumers capable of valorising electricity produced by renewable power plants. This shift raises challenges related to power quality and grid integration. Hydrogen-based pathways follow a similar strategy, enabling the conversion of renewable electricity into a storable energy carrier that can be used to supply thermal demand when direct electrification is constrained. This approach, however, introduces additional energy losses due to electrolysis, currently operating at 65–70% efficiency, and requires compatible burners, adapted pipelines, and enhanced safety systems.

4. Methodological Framework for Assessing Renewable-Based Electrification and Hydrogen Pathways in the Glass Industry

The decarbonisation of hard-to-abate industrial sectors such as the glass sector ultimately relies on the use of renewable electricity to supply high-temperature thermal energy. This can be achieved either through direct electrification of industrial processes or indirectly through energy storage vectors such as hydrogen, which provides a clear link between variable renewable power generation and continuous thermal energy requirements.

In the glass industry, this challenge is compounded by the large scale of industrial furnaces, with thermal power requirements of several tens of megawatts, and by the fact that process modifications and additional plant components inevitably increase system complexity and costs. Under these conditions, assessing decarbonisation options requires not only a technical evaluation, but also a consistent economic comparison of alternative strategies. Considering these aspects, this section proposes a methodological framework to systematically assess renewable-based electrification and hydrogen pathways in the glass industry. Beyond identifying cost-optimal solutions for a single configuration, the proposed methodological approach is designed to explore the conditions under which different decarbonisation strategies become viable. By systematically varying key design, dimensional, and operating parameters, the framework enables the identification of the regimes in which direct electrification, hydrogen-based solutions, or hybrid configurations represent the most suitable option for glass melting decarbonisation.

The framework is then applied to a representative case study to translate general decarbonisation concepts into a coherent quantitative analysis. Several plant configurations are examined to assess the respective margins and trade-offs associated with direct electrification, hydrogen use, and their combined application in hybrid solutions. The analysis explicitly acknowledges the complexity of the problem, given the large scale of industrial glass furnaces and the need to preserve strict product quality requirements. Accordingly, the modelling framework integrates system boundaries and energy balances with optimisation methods and performance indicators, supported by a detailed representation of the relevant process components.

4.1. System Layout and Energy Balances

Figure 5 illustrates three alternative integrated system layouts, selected as representative of the main renewable energy integration options for the glass industry: direct, battery-based, and hydrogen-based integration. In the direct integration configuration, Figure 5a, electricity generated by solar and wind plants is supplied directly to the furnace. In the battery-based configuration, Figure 5b, surplus electricity is stored in batteries and later used on-site to offset electricity demand. In the hydrogen-based configuration, Figure 5c, excess electricity is routed to an electrolyser, for example, a proton exchange membrane (PEM), where it is converted into hydrogen that is either supplied to the furnace burners or compressed and stored for later use, thereby reducing natural gas consumption.

The electric balance of the integrated system under different configurations is described by Equation (1), where $W_{PV}^t$ and $W_{wind}^t$ are the hourly solar and wind power generation, $W_{gr,in}^t$ and $W_{gr,out}^t$ are the hourly electricity import from and export to the electricity grid, $W_{fur}^t$ is the furnace electrical input (including oxygen generation requirements), $W_{BSS,ch}^t$ and $W_{BSS,dis}^t$ are the hourly charge and discharge of the battery storage, $W_{PEM}^t$ is the power consumption of the electrolyser, and $W_{H_{2}C}^t$ is the power consumption of the hydrogen compressor.

$$\left\{\begin{array}{ll}{W}_{PV}^t+W_{wind}^t+W_{gr,in}^t-W_{gr,out}^t-W_{fur}^t=0& \mathrm{Direct} \mathrm{integration}\\ W_{PV}^t+W_{wind}^t-W_{BSS,ch}^t+W_{BSS,dis}^t+W_{gr,in}^t-W_{gr,out}^t-W_{fur}^t=0& \mathrm{Battery} \mathrm{integration}\\ W_{PV}^t+W_{wind}^t-W_{PEM}^t-W_{H2C}^t+W_{gr,in}^t-W_{gr,out}^t-W_{fur}^t=0& \mathrm{Hydrogen} \mathrm{integration}\end{array}\right.$$

(1)

The thermal balance of the integrated system under different configurations is described by Equation (2), where $Q_{gr,in}^t$ is the thermal energy input from the natural gas grid, $Q_{fur}^t$ represents the furnace thermal input, $Q_{PEM}^t$ is the energy content of hourly electrolyser hydrogen generation, and $Q_{H_{2}S,ch}^t$ and $Q_{H_{2}S,dis}^t$ are the hourly charge and discharge of the hydrogen storage tank.

$$\left\{\begin{array}{ll}{Q}_{gr,in}^t-Q_{fur}^t=0 & \mathrm{Direct} \mathrm{integration}\\ Q_{gr,in}^t-Q_{fur}^t=0& \mathrm{Battery} \mathrm{integration}\\ Q_{PEM}^t-Q_{H2S,ch}^t+Q_{H2S,dis}^t+Q_{gr,in}^t-Q_{fur}^t=0& \mathrm{Hydrogen} \mathrm{integration}\end{array}\right.$$

(2)

4.2. Optimisation Formulation and Key Performance Indicators

Among the possible optimisation criteria, cost minimisation is selected in this study to reflect the central role of economic feasibility in guiding investment and decisions in energy-intensive industrial processes. Optimal hourly energy supply for the hybrid glass furnace is determined using mixed-integer linear programming (MILP) optimisation. The annual simulation is performed by solving independent weekly optimisation problems with an hourly time resolution ($N_{h,week}=24·7=168$). A cyclic constraint is imposed on the initial and final states of the storage components.

The objective function, $f_{obj}$, is defined as the melting cost, calculated by accounting for the costs of natural gas and electricity, obtained by multiplying the purchase volumes, $Q_{gr,in}^t$ and $W_{gr,in}^t$, by their specific prices, $c_{ng}$ and $c_{el}$; the costs of direct carbon emissions, obtained by multiplying the natural gas input by the CO₂ emission factor, $E{F}_{ng}$, and the price of CO₂, $c_{C{O}_2}$; and the cost of water, obtained by multiplying the water purchase volume, $H_2{O}_{in}^t$, by its price, $c_{H_{2}O}$ [58].

$$f_{obj}={\sum }_t^{N_h}\left[Q_{gr,in}^t·c_{ng}+W_{gr,in}^t·c_{el}+Q_{gr,in}^t·E{F}_{ng}·c_{C{O}_2}+{H_{2}O}_{in}^t·c_{H_{2}O}\right]·∆t$$

(3)

The economic performance is evaluated by calculating specific melting costs, ${SC}_{melt}$, as the sum of natural gas, electricity, carbon, and water costs, $C_{ng}^t$, $C_{el}^t$, $C_{C{O}_2 }^t$, and $C_{H_{2}O}^t$, divided by the amount of glass produced, obtained by multiplying the furnace size, ${sz}_{fur}$, by the number of operative days, $N_d$, considered ($N_{d,year}=365)$.

$${SC}_{melt}={\displaystyle \frac{C_{melt}}{{sz}_{fur}·N_d}}={\displaystyle \frac{{\sum }_t^{24·N_d}{C}_{ng}^t+C_{el}^t+C_{C{O}_2}^t+C_{H_{2}O}^t}{{sz}_{fur}·N_d}}$$

(4)

The environmental performance is assessed by calculating specific melting emissions, $S{E}_{melt}$, as the sum of direct emissions from natural gas combustion, obtained by multiplying natural gas imports, $Q_{gr,in}^t$, by an emission factor, $E{F}_{ng}$, and indirect emissions from electricity consumption, obtained by multiplying electricity imports, $W_{gr,in}^t$, by a variable grid emission factor, $E{F}_{el}^t$ [59], divided by the amount of glass produced.

$$S{E}_{melt}={\displaystyle \frac{E_{melt}}{{sz}_{fur}·N_d}}={\displaystyle \frac{{\sum }_t^{24·N_d}{Q}_{gr,in}^t·{EF}_{ng}+W_{gr,in}^t·{EF}_{el}^t}{{sz}_{fur}·N_d}}$$

(5)

Beyond operational considerations, the costs of CO₂ avoided, or levelised decarbonisation costs, $LCOD$, are defined as the sum of capital (CAPEX), annualised operational (OPEX), and replacement expenditures (REPEX) for the various components involved, $CAPE{X}_j$, $OPE{X}_j$, and $REPE{X}_j$, minus the annualised operational energy cost savings compared to the conventional configuration, $C_{melt}-C_{melt}^0$, divided by the overall carbon reductions compared to the conventional configuration, $E_{melt}^0-E_{melt}$.

$$LCOD={\displaystyle \frac{{\sum }_j^{N_c}\left(CAPE{X}_j+{\sum }_n^{N_y}\frac{OPE{X}_j}{{\left(1+d\right)}^n}+REPE{X}_j\right)+{\sum }_n^{N_y}\frac{C_{melt}-C_{melt}^0 }{{\left(1+d\right)}^n}}{(E_{melt}^0-E_{melt})·N_y}}$$

(6)

where $N_c$ represents the number of components, $N_y$ represents the number of years in the evaluation period, and $d$ represents the discount rate for cash flow discounting.

Obviously, the economic model proposed does not explicitly capture all scale- and implementation-related effects, but it remains useful for a first methodological analysis. Comparison of the decarbonisation costs of different design configurations can provide information on the optimal decarbonisation pathways to be pursued.

5. Case Study and Results

To translate the general decarbonisation strategies discussed into a quantitative framework, this section focuses on a representative case. The analysis considers an oxyfuel furnace with a size $s{z}_{fur}$ of 300 t/day, operating at constant pull rates throughout the entire year, and a nominal SEC of 3.8 GJ/t (about 1 kWh/kg) plus 250 kWh per ton of oxygen [60].

A comprehensive sensitivity analysis is performed on furnace electric boosting levels, ${fur}_{bst}$; renewable energy coverage, $co{v}_{ren}$; renewable energy mix, $mi{x}_{ren}$; battery storage and hydrogen electrolyser coverage, $co{v}_{BSS}$ and ${cov}_{PEM}$; and electricity, natural gas, and CO₂ prices, $c_{el}$, $c_{ng}$, and $c_{CO2}$. Renewable energy coverage is defined relative to the total energy demand, accounting for demand reductions associated with increased furnace boosting and providing insights into the overall energy self-sufficiency of the system. Battery storage and hydrogen electrolyser coverage are defined relative to the renewable capacity, obtained as the sum of solar and wind plants.

The thermal and electric SEC of the glass furnace are assumed to be linear with boosting levels, as in [27]. For the case study, energy system data representative of the European context is used, with a specific focus on Italy. Nation-wide solar and wind hourly load profiles for 2023, $l{p}_{PV}$ and $l{p}_{wind}$, are obtained from ENTSO-E [59], while the capacity factors, $c{p}_{PV}$ and $c{p}_{wind}$, are derived from TERNA [61]. The electrolyser load factor, $l{f}_{PEM}$, is assumed to range between 15 and 100% of nominal power, with efficiency varying linearly with the load factor, as in [62], and a standby power consumption of 1.5%, as in [58]. For storage components, constant charge and discharge efficiencies, ${\eta }_{BSS,ch|dis}$ and ${\eta }_{H2S,ch|dis}$, are considered, with constraints obtained from [63]. Battery storage is sized to cover intra-day variability, while hydrogen storage is limited to intra-week flexibility. More details are provided in Appendix A.

Table 3. Main cost assumptions and reference energy and CO₂ prices for the techno-economic analysis.

Component	CAPEX	OPEX	Lifetime	REPEX
Unit of measure	(EUR/kW) or (EUR/kWh)	(%CAPEX/y)	(y)	(%CAPEX)
Furnace electrodes [66]	350	2	10	100
Furnace burners [67]	100	2	10	100
PV plant [68]	820	2	20
Wind plant [68]	1370	2	20
Battery storage [58]	300 (200)	2.5	10	50
PEM electrolyser [58,69,70]	1200 (700, 300)	5	10	35
Hydrogen storage [69]	14	2	20
Hydrogen compressor [58,69]	4500	2	20

reports the main cost assumptions for the components, with a discount rate of 4% and a 20-year evaluation period. For the basic techno-economic assessment, reference prices are set at 150 EUR/MWh for electricity, 50 EUR/MWh for natural gas, and 75 EUR/t for CO₂, reflecting current Italian price levels [64,65].

5.1. Operational Behaviour

To capture the system behaviour under a range of design and operating conditions, multiple integration scenarios were simulated. As an illustrative example, Figure 6 shows the electric and thermal energy balances for a representative week of the year under direct, battery, and hydrogen integration scenarios, assuming 50% furnace electric boosting, 75% renewable energy coverage, a balanced renewable mix (50% solar–50% wind), and battery and electrolyser coverage both set at 40%.

In the direct integration scenario, excess renewable generation is entirely exported to the grid, necessitating electricity purchases during periods of low renewable output and natural gas purchases throughout the entire period. Battery integration allows surplus electricity to be stored, minimising external electricity purchases, while hydrogen integration converts excess electricity into hydrogen, used to reduce natural gas consumption.

5.2. Techno-Economic Assessment

This section presents the results of the techno-economic assessment of renewable energy integration in glass furnaces, focusing on the combined effects of furnace electrification, renewable energy coverage, and the solar–wind generation mix. The analysis focuses on decarbonisation costs, supporting industrial decision-making with quantitative design criteria regarding renewable energy plants, furnace electrification levels, battery storage, and hydrogen infrastructure.

5.2.1. Direct Integration

Figure 7 reports levelised decarbonisation cost maps for low, medium and high boosting levels. Each curve represents combinations of renewable energy coverage and furnace electrification that yield the same levelised cost of CO₂ avoided, explicitly illustrating the trade-off between renewable availability and electric boosting in minimising decarbonisation costs.

The achievement of negative values means that the required progress in terms of emissions reduction is profitable when contrasting the increased capital expenditures and the resulting operational cost savings. This is possible because the levelised cost of electricity for solar and wind power can be lower than the cost of electricity from the grid, as also noted in [26]. For electrification rates of 20%, the lowest cost of CO₂ avoided is equal to −130 EUR/tCO₂. At a level of 50%, the optimal value is −65 EUR/tCO₂, whereas at boosting levels of 80%, the optimal value is −40 EUR/tCO₂ for a renewable coverage of 75%. Higher electrification levels progressively shift the cost-optimal region towards higher renewable penetration, reflecting the increasing reliance on low-carbon electricity.

Figure 8 provides a graphical interpretation of the trade-off between emission reductions and decarbonisation costs for the different system configurations analysed. Each point in the figure represents a feasible system configuration, defined by a specific combination of furnace electric boosting, renewable energy coverage, and renewable mix. The horizontal axis quantifies the achieved CO₂ emission reduction relative to the reference case (natural gas), while the vertical axis reports the levelised cost of CO₂ avoided.

The distribution of points highlights how technological choices shift the balance between environmental and economic performance. Configurations characterised by higher furnace boosting and greater renewable coverage tend to cluster towards higher emission reductions, whereas cost-effective solutions are favoured by higher shares of wind energy in the renewable mix. This reflects the ability of wind generation to provide a more continuous low-carbon electricity supply better aligned with the steady thermal demand of glass furnaces. Overall, the figure shows that deep decarbonisation of glass melting can be achieved without disproportionate cost penalties when electrification and renewable integration are coherently designed, and it visually identifies the regions where favourable compromises between cost and emission reduction are attained.

5.2.2. Advanced Integration

A realistic representation of the economic and environmental suitability of advanced integration scenarios for the case study can be obtained by referring to the current capacity addition mix of Italy. In 2023, this was correspondent to 10% wind and 90% solar [71]. Detailed maps of cost and emission reductions achieved by battery and hydrogen integration compared to the conventional furnace are discussed in Appendix B.

Battery integration achieves higher cost and emission reductions for large levels of renewable coverage and medium and large levels of furnace boosting. Conversely, hydrogen integration yields maximum cost and emission reductions for low boosting levels. The analysis reveals the importance of proper sizing, since optimal values of battery coverage are around 30–40%, while electrolyser coverage is only around 20–30%. Figure 9 shows maps of the LCOD for battery and hydrogen integration under varying furnace boosting, renewable coverage, and battery and electrolyser coverage conditions.

No negative values are obtained in these cases, highlighting cost increases compared to direct integration and conventional fired furnaces. In the battery integration scenario, the minimum costs of CO2 avoided are in the range of 100–200 EUR/t. The higher the boosting levels, the higher the renewable coverage for which these optimal values are obtained. At 50% boosting, optimal renewable coverage is around 50%, while it reaches 75% at 80% boosting. In the hydrogen integration scenario, optimal values of decarbonisation costs are in the range of 200–300 EUR/t. Except for the region of low furnace boosting levels and renewable coverage in the range of 75–100%, the lower the electrolyser coverage, the lower the decarbonisation costs, highlighting how the costs of the technology significantly impact its cost-effectiveness. The relationship between decarbonisation costs and emission reduction under a realistic renewable capacity addition mix are shown in Figure 10a for direct, battery, and hydrogen integration. Both batteries and hydrogen extend maximum emission reductions from around 50% up to 80%. For battery integration, solutions that allow for emission reductions higher than 55% correspond to boosting levels above 60% and renewable coverage above 75% and yield minimum decarbonisation costs of 140 EUR/t, with average values of 215 EUR/t. For hydrogen integration, these correspond to renewable coverage above 125% and boosting levels below 70%, yielding minimum decarbonisation costs of 315 EUR/t, with average values of 385 EUR/t. This reveals that, at current energy and carbon prices, batteries represent a more cost-effective solution for the deep decarbonisation of the glass industry.

5.3. Sensitivity Analysis

The sensitivity of deep decarbonisation cost options (>55% emission reduction) for battery and hydrogen integration was analysed under varying electricity, natural gas, and carbon prices, as shown in Figure 11. The comparison was performed on a consistent basis, with average emission reductions of about 60% for both configurations.

Decarbonisation costs exhibit different sensitivities to energy prices depending on the integration strategy. Battery integration costs are only weakly affected by electricity prices, as batteries primarily shift the timing of renewable electricity use rather than increasing overall electricity demand. Consequently, their economic performance is mainly driven by natural gas and carbon prices, which determine the value of avoided fuel consumption and emissions. Hydrogen integration, by contrast, shows a pronounced sensitivity to electricity prices, reflecting the additional electricity demand and conversion losses associated with electrolysis. In this case, electricity price variations directly affect the cost of hydrogen production and, therefore, the cost of CO₂ avoided.

In both configurations, lower electricity prices combined with higher natural gas and carbon prices improve economic performance. Minimum decarbonisation costs range from −425 to 240 EUR/t for battery integration and from −445 to 420 EUR/t for hydrogen integration. Despite these differences, optimal design parameters remain largely stable across price regimes and consistently require high levels of electrification (80% boosting). Battery-based solutions are favoured at moderate renewable coverage (75%), whereas hydrogen integration requires higher renewable energy availability (125%), confirming large-scale electrification as a necessary condition for deep decarbonisation.

Figure 12 compares the cost-effectiveness of battery and hydrogen integration across combined fuel and electricity price regimes for different battery and electrolyser CAPEX values. At current electrolyser costs (1200 EUR/kW), battery integration is generally more cost-effective across a wide range of energy price and investment cost conditions.

Hydrogen integration emerges as a competitive option under price regimes characterised by high fuel and carbon prices combined with low electricity prices, reflecting its ability to decouple renewable electricity availability from continuous thermal demand. As electrolyser costs decline, the set of price regimes in which hydrogen becomes economically attractive expands substantially. This trend highlights hydrogen not merely as an alternative to batteries, but as a complementary flexibility option that becomes increasingly relevant in systems with high renewable penetration, limited electrical storage, and stringent decarbonisation targets. In this context, hydrogen enables the temporal extension of renewable electricity use beyond direct electrification, supporting continuous furnace operation. It should be noted that the optimisation framework relies on linear cost assumptions implicitly if energy prices and technology costs scale proportionally with system size and utilisation. Moreover, the availability of water for electrolysis may represent an additional location-specific constraint that is not explicitly captured in the model [72]. In practice, price formation mechanisms, grid constraints, contractual arrangements, and economies of scale may introduce non-linear effects that are not captured by the model. Accordingly, the identified price thresholds should not be interpreted as strict competitiveness boundaries, but rather as indicative markers of the conditions under which hydrogen-based solutions begin to deliver system-level value. Nevertheless, the analysis remains valuable in consistently identifying relative trends, trade-offs, and boundary conditions under which electrification-, battery-, and hydrogen-based strategies contribute to cost-effective deep decarbonisation.

6. Conclusions

Deep decarbonisation of hard-to-abate industrial sectors ultimately hinges on the ability to supply large and continuous thermal energy demands with low-carbon energy sources. In energy-intensive industries such as glass manufacturing, this challenge is exacerbated by the scale of industrial furnaces, the stringent requirements on process stability and product quality, and the intermittency of renewable electricity generation. Under these conditions, electrification, either direct or mediated through energy storage vectors such as hydrogen, emerges as an interesting opportunity, but its practical feasibility depends on the coordinated design of process technologies, energy systems, and economic boundary conditions.

In this context, the paper proposes a methodological approach to assess alternative decarbonisation strategies for hard-to-abate industrial sectors and demonstrates its application through a case study in glass manufacturing. A representative 300 t/day industrial glass furnace is used as a case study to exemplify how variable renewable electricity can be integrated into high-temperature processes through three alternative configurations: direct electrification, battery-supported electrification, and hydrogen-supported electrification via on-site electrolysis and storage.

The study quantifies operational behaviour and system performance, focusing on energy use, carbon emissions, and economic costs. Under current electricity, natural gas, and carbon prices (150 EUR/MWh, 50 EUR/MWh, and 75 EUR/t, respectively), results for the direct integration scenario show that high levels of furnace electrification combined with wind-dominated renewable supply achieve levelised decarbonisation costs of around –40 EUR/t, demonstrating the practical viability of this approach.

Battery solutions are generally favoured at moderate renewable penetration (~75%), hydrogen requires higher renewable availability (~125%), and high electrification (~80%) is consistently needed to achieve deep decarbonisation. For solutions that achieve deep decarbonisation targets (>55%), battery integration is associated with lower levelised decarbonisation costs under current cost and price assumptions, whereas hydrogen-based solutions require higher renewable overcapacity.

The comparison highlights structural differences between the two pathways without identifying a specific optimal solution. Battery storage supports short-term balancing, while hydrogen adds flexibility despite conversion losses. Large-scale furnace electrification is essential for deep decarbonisation. Hydrogen becomes more relevant with high renewables, limited electrical flexibility, and lower electrolyser costs.

The proposed analysis inevitably involves simplifications and does not explicitly capture all the dimensional, technical, and organisational aspects that may complicate the real-world deployment of the investigated solutions. Some technical constraints related to plant integration of battery and hydrogen, operational flexibility, and scale effects are treated implicitly or assumed to be manageable, whereas in practice they may play a key role. At the furnace level, additional technical factors not explicitly modelled include electrification limits, hydrogen-specific combustion effects, material durability, and auxiliary energy requirements. At the system level, factors such as seasonal hydrogen storage, economies of scale, water consumption, and infrastructure constraints may influence technology deployment and investment feasibility. Nevertheless, the proposed framework represents a valuable first step, as it allows the systematic comparison of alternative decarbonisation pathways under consistent assumptions, highlighting relative trends, trade-offs, and boundary conditions. In this sense, the analysis should be interpreted as an exploratory tool rather than a prescriptive design exercise, providing guidance on where deeper, technology-specific investigations are most warranted.