Introduction

To prevent further deterioration of the global climate system, the transition of the energy system is becoming increasingly urgent. This transition requires phasing out fossil fuels1,2,3,4,5 and expanding sustainable energy sources such as solar and wind power6,7,8,9. However, this transition has substantial material implications, as the demand for materials, especially metals, is expected to increase dramatically10,11,12,13,14,15,16,17. A growing body of literature is dedicated to the material requirements of the renewable energy system, indicating that the demand for materials, especially major metals18,19,20 and critical materials13,21,22,23,24. Ensuring a stable supply of these materials has become a growing priority for both governments and industries25,26,27. Most previous studies have adequately considered and evaluated the materials needed to decarbonize electricity generation5,18,28,29,30,31 and transportation30,32,33,34. However, less attention has been given to the fate of fossil fuel infrastructure under the energy transition. The phase-out of fossil fuels will not occur immediately35; given their current 80% share of the global energy system36, there will be a gradual shift toward renewables with a declining dependence on fossil fuels. This transition will not only affect fossil fuel demand but also reshape the material use associated with fossil fuel infrastructure, including extraction, processing, storage, and transport37,38,39. In this study, “fossil fuel infrastructure” refers to physical assets across the fossil fuel supply chain, including coal mining infrastructure, oil and gas platforms, refineries, gas processing plants, vehicles (e.g., trucks, trains, ships, and tankers), pipelines, and other related structures and equipment40,41,42. A detailed classification of the infrastructure considered is provided in the Methods section (Fig. 1). Fossil fuel power plants are excluded, as their material composition and decommissioning have already been extensively studied30,43.

Fig. 1: Global material stocks in fossil fuel infrastructure from 1990 to 2050 under SSP2(BL) and SSP2(2D) scenarios.
figure 1

The material stocks of steel (a, b), copper (c, d), and aluminium (e, f) are displayed under both the SSP2 (BL) and SSP2 (2D) scenarios.

Full size image

Over the past century, a vast amount of fossil fuel infrastructure—including platforms, refineries, pipelines, ships, and tankers—has accumulated worldwide. As fossil fuels are phased out, these assets will become obsolete, making their embedded materials, particularly metals, available for reuse or recycling44. However, the quantity, composition, and availability of these materials remain poorly understood.

Currently, there is limited information on the total stock of materials embedded in fossil fuel infrastructure, as well as the flows of new materials entering the system and waste materials leaving it. Furthermore, the impact of the energy transition on these material stocks and flows remains largely unexamined. Despite these knowledge gaps, gaining a deeper understanding of fossil fuel material dynamics could be highly valuable for policymakers and industry stakeholders. Effective decommissioning planning can not only facilitate material recycling and optimize resource allocation but also contribute to the energy transition by mitigating material demand through the repurposing of materials from the existing fossil fuel system for diverse applications. At the very least, such assessments provide critical insights for enhancing global resource efficiency and advancing circular economy policies. To fill this gap, we developed a global dynamic fossil fuel material model (FUMA). It allows for incorporating material-energy feedback into Integrated Assessment Models (IAM), such as the Integrated Model to Assess the Global Environment (IMAGE). The implementation of this model will allow us to assess the dynamics of material stocks and flows within the fossil fuel infrastructure and examine its interaction effect with changes in fossil energy demand. Our analysis aims (1) to quantify the global stock of three major metals (steel, copper, and aluminium) embedded in the fossil fuel infrastructure, (2) to understand the dynamics of material inflows and outflows related to this stock, and (3) to assess the potential of these materials to mitigate the energy transition’s material needs. We compare a baseline scenario with a climate scenario, assuming an energy transition in line with the 2-degree Celsius (2D) target as agreed on in the Paris Agreement45, based on the IMAGE Shared Socioeconomic Pathway 2 (SSP2)46.

Results

Material stock patterns in the development of fossil fuel infrastructure

The material stocks and their development until 2050 in the fossil fuel infrastructure are presented in Fig. 1. Steel is a critical component of all fossil fuel infrastructure. Specifically, the steel stock measured in megatons is projected to grow. The steel stocks are three orders of magnitude larger than aluminium and copper stocks (For the detailed composition of the individual element stocks in the various infrastructures, please refer to Supplementary Fig. 7).

Under the SSP2(BL) scenario, the steady growth in fossil fuel demand drives a continuous increase in material stocks until 2050, reaching approximately 1700 Mt for steel, 14 Mt for aluminium, and 1.1 Mt for copper. Meanwhile, under the SSP2(2D) scenario, fossil fuel infrastructure stocks were projected to peak around 2022 before gradually declining. This decline is mainly due to a decrease in stocks associated with coal and oil infrastructure. In the SSP2(2D) scenario, natural gas is assumed to serve as a transitional energy source between coal and renewables. This leads to a substantial increase in gas-related infrastructure stocks, particularly for steel (Fig. 1b).

Corresponding to the total stocks, Fig. 2 details the relative share of steel, copper, and aluminium in the global fossil fuel infrastructure. The stock of steel in 2050 in the 2D scenario is slightly above its 2019 value, while the total stock of copper and aluminium shrinks by ~30% compared to 2019. Notably, in 2019, 60% of steel stocks are gas-related, which will further rise to 65% (SSP2(BL)) and 73% (SSP2(2D)) in 2050. Gas pipelines account for most of the steel stock ( ~70%) in gas-related infrastructure. Unlike steel, the stocks of copper and aluminium are primarily linked to vehicles (trucks, rail cargo and especially ships), with the share of coal stocks currently surpassing those of gas and oil.

Fig. 2: Breakdown of material stocks in different fossil fuel infrastructure.
figure 2

Breakdown of material stocks in different fossil fuel infrastructures in the base year (a–c), the year 2050 under SSP2(BL) (d–f) and SSP2(2D) (g–i) scenarios. Circle areas represent the size of the corresponding material stock. The base year stocks of steel (a), copper (b), and aluminium (c) are set with the index to 1 (stock for 2019 = 1). The inner layer of each circle shows the distribution of material stocks for coal (gray), oil (orange), and natural gas (blue). The middle layer represents the infrastructure for different phases of each fossil fuel type (including extraction, industrial handling, vehicles, and pipelines). The outer layer represents the subtypes for various infrastructure elements (e.g., offshore/onshore extraction, transmission/distribution gas pipelines, rails/trucks/inland/ocean ships, oil storage, refineries, etc.). Only the major composition elements (\(\ge\)1%) of the infrastructure stock as shown in the figure. The full information is provided in Supplementary Table 12.

Full size image

The future is expected to witness an upward trajectory of gas-related transportation material stocks, driven by the growing trade volume of liquid natural gas (LNG). The trade of LNG is already increasing, reaching a level comparable to that of gaseous natural gas by 202247. This trend is expected to continue in both SSP2(BL) and SSP2(2D)48. For steel, the share of gas in total fossil fuel infrastructure will grow in both scenarios, but most in the SSP2(2D) scenario.

The share of oil increases in both scenarios for copper as well as for aluminium. In absolute terms (Fig. 1), the amounts will grow in the SSP2(BL) scenario but go down under SSP2(2D) assumptions. This indicates that coal is expected to be rapidly phased out and switched to natural gas.

Material requirements, outflows, and surplus

Figure 3 depicts the material demand (inflows) and discarded materials (outflows), which could be regarded as potential secondary material supply under the two scenarios. Here, we present an in-depth examination of these trends, highlighting noteworthy observations regarding the inflows and outflows.

  • In both scenarios, the material demand for fossil fuel infrastructure remains substantial through 2050. Under the SSP2(BL) scenario, it continues to grow, whereas in the SSP2(2D) scenario, it declines but remains considerable. The reduction in CO₂ emissions in SSP2(2D) assumes a large-scale deployment of Carbon Capture and Storage (CCS), supporting the continued use of fossil fuel infrastructure. Additionally, further adding capital and material stocks used for those CCS installations, as covered by the analysis on material use for electricity generation28.

  • The SSP2(BL) scenario fossil fuel system requires more materials than SSP2(2D). The outflows remain smaller than the inflows throughout the whole assessment period. Even assuming all the waste materials will be recycled and used in the same applications, an inflow of virgin materials in the fossil fuel infrastructure is still needed.

  • Under the SSP2(2D) scenario, demand for both copper and aluminium peaked around 2022 and then gradually declined. At the same time, these materials will generate annual surpluses until 2050. The cumulative surplus of copper (2022–2050) is ~340 kilotons, which is comparable to the global copper demand for electric vehicles in 202249. Unlike copper and aluminium, steel is expected to experience only a brief surplus before demand rises again after 2030.

  • The gas industry will still require an increasing inflow of materials even under SSP2(2D), particularly for the construction of new pipelines, which rely heavily on steel. Over 90% of the cumulative gas-related steel demand from 2020 to 2050 is seen in pipelines. The cumulative demand for steel in natural gas pipelines between 2020 and 2050 is greater than the cumulative steel demand of the coal and oil industries combined over the same period.

Fig. 3: Global material flows and the material surplus in fossil fuel infrastructure during 1980–2050 under SSP2(BL) and SSP2(2D) scenarios.
figure 3

The material flows of steel (a, b), copper (d, e), and aluminium (g, h), along with their corresponding surpluses (c, f, i), are shown. Surplus is defined as the difference between outflows and inflows within the same year, where positive values in the shaded areas indicate that outflows exceed inflows.

Full size image

Therefore, the availability of secondary materials from the end-of-life fossil fuel infrastructure remains limited in both scenarios until 2050, minimizing its potential contribution to future material demand.

Discussion

Our findings indicate that global stocks of coal-related materials will decline, while stocks of natural gas-related materials will continue to increase until 2050. This shift in material stocks reflects the ongoing transition towards a more sustainable energy landscape. While natural gas has been viewed as a transition fuel due to its lower carbon emissions compared to coal50, its continued expansion may slow the decarbonization of energy systems51. In addition to methane leaks from gas infrastructure52,53,54,55,56, dependence on natural gas infrastructure may create a lock-in effect57,58,59, where entrenched investments and infrastructure hinder the transformation to renewable energy systems. The phasing out of fossil infrastructure will not advance far enough by 2050 to become a major source of secondary materials. Under SSP2(BL) assumptions, material stock will continue to grow, requiring more materials rather than supplying them. SSP2(2D) scenario, copper and aluminium stocks decline, making some materials available for recycling. However, the projected surplus (outflow minus inflow) in 2050 is only 10 kt of copper and 108 kt for aluminium. When compared to the material demands of renewable energy systems, this volume is marginal: it could supply only 8% of the global aluminium demand and 0.8% of the global copper demand required for wind and solar PV deployment under the SSP2-2D scenario28. In comparison, under the International Energy Agency’s (IEA) Net-Zero Emissions scenario, this surplus would meet only 0.2% of the projected 2050 copper demand for these technologies60.

Compared to the material demands of renewable energy technologies, the recoverable materials from end-of-life fossil fuel infrastructure are limited, as demonstrated in our analysis. Additionally, technical constraints further restrict the potential of decommissioned fossil fuel infrastructure in supplying secondary materials. Material losses during collection and recycling, along with technical recycling rates, influence overall material availability61,62,63. Unlike copper and aluminium, which can be recycled without substantial loss of quality, steel recycling is often constrained by variations in alloy composition and the lack of detailed chemical data, which hinder large-scale recycling efforts64. Most steel undergoes a downcycling process, meaning it often fails to meet specifications for high-performance applications due to alloying elements, corrosion, and strength limitations65,66. However, it remains a valuable secondary resource that can be recycled and repurposed for various applications in the global market, contributing to the broader circular economy.

The recyclability of fossil fuel infrastructure varies across different components. Above-ground infrastructure, such as ships and trains, is generally easier to recycle than underground pipelines. Decommissioned underground gas pipelines often remain as hibernating stock without entering the recovery process67,68. Similarly, onshore infrastructure is typically easier to recycle than offshore installations69,70. Offshore structures also require sacrificial anodes made of zinc and/or magnesium, which corrode over time due to seawater exposure and cannot be recycled. Even for the same metal, recycling availability varies widely depending on its form, quality, and location, factors that warrant further investigation.

Certain limitations must be acknowledged. The fossil fuel demand projections used in this analysis are based on IMAGE model data accessed as of January 202446 and subsequent updates are not reflected here. However, the FUMA model framework allows for the integration of the latest data inputs as needed.

Additionally, fossil fuel infrastructure in the real world is more complex than the model assumptions. Some facilities might be co-used by multiple sectors, making their material stocks difficult to attribute to the fossil fuel sector fully. In this study, refineries, processing plants, storage facilities, and pipelines are assumed to be used exclusively for fossil fuels. In reality, some of these assets could be repurposed for biofuels71, hydrogen72, or other industrial applications. For transport vehicles, material stocks are estimated based on the relationship between fossil fuel transport demand and vehicle material stock weight. Here, we assume that all transport materials are exclusively used for fossil fuel transportation. Vehicles used for gas and oil transport are specifically designed to withstand hazardous conditions, including high pressure, flammability, and contamination risks, making repurposing for other types of transportation significantly more challenging73. Additionally, strict safety regulations, material compatibility concerns, and the need for extensive decontamination further restrict their potential for reuse outside the fossil fuel sector74,75. However, some transport vehicles, particularly those used for coal transport, may be repurposed for other energy sectors or general freight.

Beyond recycling, repurposing fossil fuel infrastructure presents an alternative pathway. One example is converting decommissioned gas pipelines into hydrogen pipelines72,76,77, which could extend the lifetime of certain infrastructure while supporting the clean energy transition.

However, notable challenges remain due to the distinct properties of hydrogen compared to natural gas, including issues like steel embrittlement and leak detection, which are not investigated in this study78,79. Further research is needed to determine viable pipeline components for repurposing, along with appropriate technologies and costs. Moreover, repurposing comes with trade-offs, as it may delay material recovery through recycling. A careful evaluation of material efficiency and long-term sustainability is essential to balance these approaches80.

Ultimately, a substantial amount of material will continue to be required for fossil fuel infrastructure through 2050 under both the SSP2(BL) and SSP2(2D) scenarios. This indicates that much of the material embedded in fossil fuel infrastructure will remain in use and unavailable as a secondary resource during this period. Therefore, accelerating the transition to renewable energy and reducing fossil fuel dependence could not only decrease the demand for new fossil infrastructure but also improve the availability of secondary materials within the system.

Methods

System definition

FUMA is one of the dynamic MFA modules attached to the IMAGE model46 used to support climate policies81,82. Similar models have been developed for buildings83,84, for the electricity system28, and vehicles85. These models are summarized as IMAGE-MAT86, use socio-economic and technological information from the IMAGE model and translate that in several steps to material flows and stocks. FUMA does the same for the fossil fuel infrastructure.

The main novelty of FUMA is the incorporation of the IAM, such as IMAGE, into the dynamic MFA model. We use the material-specific data for fossil fuel infrastructure and make the connection with the fossil fuel demand projections from IMAGE. We employ FUMA to assess the use of several major metals: steel, copper, and aluminium. This study covers the main stages of the fossil fuel supply chain, from extraction and industrial processing to distribution40,41. Figure 4 provides a schematic overview of the modelling approach and system boundaries, where the dashed box represents the excluded processes, and the solid boxes indicate the included system components. Specific modelling steps and components are detailed in the Supplementary Method.

Fig. 4: System definition of the global fossil fuel infrastructure and the framework of FUMA.
figure 4

D&I domestic and international transport, C&P crude oil and products, T&D transmission and distribution.

Full size image

Model description

The analysis presented is based on the FUMA model framework, which starts with a base year (2019) of fossil fuel demand and proceeds to calculate the required scale and functioning of operational infrastructure (including extraction, industrial processing, and transport) to meet that demand.

A linear scaling approach is applied to project fossil fuel use in two scenarios from the IMAGE model, deriving the required global infrastructure stocks over time. Subsequently, by employing stock-driven modelling routines85,87,88, FUMA translates IMAGE’s projected fossil fuel demand data81 into a demand for in-use infrastructure stock. This process enables the production of new infrastructure and the demolition of obsolete infrastructure. Additionally, it involves creating connections with materials intensity data and involving the normally distributed lifetime functions, enabling the calculation of materials stock and flow.

Scenario construction

The annual fossil fuel demand under different scenarios was derived from the IMAGE model81. We present a baseline scenario (BL) that is consistent with the “Middle of the Road (Medium challenges to mitigation and adaptation)” second shared socioeconomic pathway (SSP2) - SSP2(BL), which describes a development consistent with intermediate challenges for both adaptation and mitigation82,89,90. Besides, to examine the impacts on fossil fuel material demand under different climate targets, we compared the SSP2(BL) scenario with a programmed mitigation pathway that meets the target of limiting global warming to below 2-degree above pre-industrial levels, still in line with SSP2: the SSP2(2D) scenario. We divided the final energy categories into eleven direct fossil fuel end-users. The related Sankey diagrams and interpretations to elaborate on this allocation are included in Supplementary Fig. 6.

Base year and future material stock and flow calculation

We access the material composition data and the lifetime data for the fossil fuel infrastructure from the background reports of Ecoinvent3.891. The base year data of fossil fuel demand is sourced from IMAGE. These datasets are used to calculate material intensity per unit of activity across different stages of the fossil fuel supply chain, such as fuel extraction, gas processing, and refining. Additionally, we derive the transport stock demand for fossil fuels in the base year by combining the transport turnover demand (ton-kilometer per unit of fossil fuel, tkm/kg or tkm/m³) with data on the weight of different transport vehicle modes. This allows us to estimate the total transport demand weight across various vehicle types85. Next, we determine the stock of materials in the vehicles based on the fraction of the weight of materials in the different vehicles85. In addition, for oil and gas, we calculate the pipeline stock in kilometers of pipeline length in the base year, 2019. This includes categorized length calculations for natural gas transmission and distribution pipelines and crude oil and petroleum product pipelines. (Supplementary Tables 9 and 10). To eliminate the geographical mismatch of energy production and consumption, the base year fossil fuel infrastructure stocks have been redistributed based on the regional share of global fossil fuel demand so that final fossil fuel consumption drives the requirement for fossil fuel infrastructure elements.

From the base year infrastructure stock and the projected energy demand in IMAGE46, we then derived the fossil fuel infrastructure stock yearly. The dynamics of fossil fuel infrastructure using a stock-driven approach are based on references85,87,88. We assess the demolition from the existing fossil fuel infrastructure stock using a lifetime model based on documented lifetime distributions92. Then, the construction can be calculated using the basic mass balance (inflow = outflow + ∆stock). Using material intensity data for specific infrastructure elements, we calculate the corresponding material flows and stocks. A detailed description of the model, the data used, and assumptions, as well as numerical results, is provided in the Supplementary Methods.