Efficiency stagnation in global steel production urges joint supply- and demand-side mitigation efforts

The study addresses how the global steel industry has progressed in reducing greenhouse gas emissions over the last century and whether current trajectories align with climate targets such as the Paris Agreement and the 1.5 °C pathway. Steel is highly energy- and carbon-intensive, demand is projected to grow with population and affluence, carbon-based inputs are essential for high-temperature processes, and long-lived facilities risk carbon lock-in. Traditional decarbonization strategies have emphasized production-side efficiency (energy efficiency, technology innovation, fuel switching), but their effectiveness for absolute emissions reduction is questioned. The authors aim to quantify historical emissions, efficiency trends, and the interplay between material flows and technology efficiency, and to evaluate future scenarios to identify the necessity of combined supply- and demand-side measures.

Prior work has focused on specific production technologies and energy efficiency measures (e.g., blast furnace improvements, basic oxygen furnace replacing open-hearth, pelletizing vs. sintering, electric arc furnaces, and fuel switching). Studies and roadmaps have highlighted the difficulty of decarbonizing heavy industry and questioned whether production-side efficiency alone can reduce absolute emissions given rising demand and infrastructure lock-in. Analyses often overlooked the system-wide interplay between material flows (primary vs. secondary routes, scrap availability, stock dynamics) and process efficiencies. Benchmarks indicate large regional differences in energy and CO2 intensities, and prior scenario studies (e.g., IEA) explore technological and material efficiency pathways but may underestimate the combined challenge posed by growing demand in emerging economies.

The authors integrate dynamic material flow analysis (MFA) with process-based life cycle assessment (LCA) to estimate annual production, efficiency, and GHG emissions across 19 dominant processes in the global steel supply chain (mining, material preparation, ironmaking, steelmaking, and steel finishing) for 1900–2015. Total emissions each year are computed as the sum over technologies of activity (annual throughput) times technology-specific GHG intensity (Scopes 1–3). Key steps: (1) Identify production technologies and system boundaries from historical literature. (2) Perform dynamic MFA to quantify yearly production of each technology and track stocks and flows (manufacturing, in-use, end-of-life), expressing flows as average Fe content. (3) Compile annual technology GHG intensities from ecoinvent v3 and literature, representing global averages, including historical evolution and considering global-average electricity and heat mixes; group emissions into Scopes 1–3. (4) Apply LMDI decomposition over five-year periods to separate effects of production volume and intensity on total emissions for primary and secondary routes. (5) Conduct uncertainty analysis using the pedigree-matrix approach to assign uncertainty to inputs, and Monte Carlo simulation (100,000 iterations) to quantify result uncertainties. (6) Validate MFA outputs against World Steel Association statistics (correlation 0.9997; differences <19% in most years) and compare GHG intensities with literature (correlation 0.856 post-1950). Regional analysis: Partition global stocks and flows into eight regions (Europe, North America, Developed Asia & Oceania, China, India, Developing Asia & Middle East, Latin America & Caribbean, Africa) for retrospective (1995–2015) and prospective (2016–2050) analysis using World Steel and IEA projections (Stated Policies and Sustainable Development scenarios). Trade is not modeled; analysis is indicative. Scenario analysis: Six scenarios test combinations of demand growth and efficiency assumptions under a 1.5 °C carbon budget allocated to steel. S1: BAU with stagnating intensity and IEA SPS demand; S2: technical efficiency to 30% intensity reduction by 2050 (IEA SDS trend); S3: material efficiency with 12% demand reduction (IEA SDS); S4: radical technical intensity reduction averaging 0.85 t CO2-eq/t steel per decade to carbon-neutral by 2047; S5: additional 34% demand reduction vs. IEA SDS. Breakthrough technologies (37 identified) grouped into seven categories (hydrogen-based, electrolysis-based, CCUS with direct/smelting reduction, biomass-based, blast furnace improvements, carbon-free EAF, low-carbon rolling) and material efficiency strategies are cataloged for feasibility discussion.

• Cumulative emissions and production (1900–2015): ~147 Gt CO2-eq emitted (~9% of global GHG over the period) while ~45 Gt of steel products were produced; emissions mass is ~3× steel mass.
• Emissions by stage: Ironmaking contributed ~50% (~58 Gt CO2-eq); steelmaking emitted ~33 Gt (about half from open-hearth); steel finishing emitted ~27 Gt due to high throughput despite lower intensity; mineral treatment emitted 18.7 Gt CO2-eq.
• Production routes: Primary (ore–blast furnace) emitted ~132 Gt CO2-eq (>90% of total), whereas the secondary (scrap–EAF) route is ~1/8 as carbon-intensive and contributed ~5% of annual GHG emissions in 2015.
• Stocks and flows: 1900–2015, inputs were ~46 Gt iron ore and ~31 Gt of scrap (home, new, old) to yield ~45 Gt steel products. ~25 Gt remain in-use (over half), with ~16 Gt in buildings; ~83% of in-use stocks were built after 1990. Old scrap generation rose from ~45 Mt/yr (1950) to ~427 Mt/yr (2015); EoL recycling rate ~70%.
• Secondary share decline: Secondary production share fell from 30% (1995) to 21% (2015), contributing to higher total emissions amid demand growth and young stocks in emerging economies that limit scrap supply.
• Efficiency trends: GHG intensity reduced by ~67% since 1900. Major drops before 1940 (blast furnace energy efficiency) and 1970–1995 (from ~4.5 to 2.6 t CO2-eq/t steel) via technology shifts (pelletizing, BOF adoption) and electricity decarbonization.
• Net effect: A 44-fold increase in annual steel production caused a 17-fold increase in annual emissions despite efficiency gains. Global GHG intensity stagnated from ~1995 to 2015 due to expansion of carbon-intensive primary production in emerging regions.
• Regional dynamics (1995–2015): Crude steel production from the most carbon-intensive regions (Tier 3) expanded ~8-fold from 129 to 914 Mt/yr, while Tier 1–2 shares decreased from 83% to 43% of global production, offsetting regional efficiency improvements.
• Carbon budget implications: ~37% of the steel sector’s 2050 GHG budget has already been used. Under BAU (S1), the 1.5 °C-aligned steel budget is exhausted by ~2035; even with S2 (30% intensity reduction) or S3 (12% demand reduction), the budget is exhausted before 2040. Meeting 1.5 °C likely requires either radical intensity reduction averaging 0.85 t CO2-eq/t per decade to reach carbon-neutrality by 2047 (S4) or an additional 34% demand reduction relative to IEA SDS (S5).
• Technology readiness: 37 breakthrough technologies grouped into seven categories could theoretically achieve required reductions, but most are 10–25 years from wide availability; deployment pace is a critical constraint.
• Strategic priorities: Recommendations include deploying low-carbon technologies in emerging markets (near-term CCUS with direct/smelting reduction), early retirement of primary capacity in China (−170 Mt by ~2035 and −500 Mt by 2050), moving developed regions toward closed-loop scrap-based EAF systems powered by renewables, integrating material efficiency (potentially ~40% of cumulative reductions), and fostering a global green steel market and policies (e.g., certification, carbon pricing).

The integrated MFA–LCA and decomposition reveal that while process efficiency improved substantially, rapid demand growth—especially primary-route expansion in emerging economies with limited scrap availability—has driven absolute emissions up and stalled global intensity reductions since the mid-1990s. This directly answers the research question by showing efficiency alone has been inadequate to deliver absolute emissions cuts. Regional structure and the primary vs. secondary route balance crucially determine global outcomes; the recent shift toward Tier 3 regions with more primary steel has offset technical gains elsewhere. Scenario analysis underscores that neither moderate technical improvements nor modest material efficiency can keep the sector within a 1.5 °C-compatible budget; only a combination of rapid intensity reductions (approaching carbon neutrality by 2047) and significant demand-side reductions can. The findings emphasize the need for coordinated, region-specific portfolios that synchronize technology rollout with evolving steel flows, accelerate retirement of carbon-intensive assets, expand scrap-based production with clean electricity, and mainstream material efficiency across end-use sectors, supported by global cooperation and market mechanisms for green steel.

Historical analysis shows that improvements in steel production efficiency (~67% lower intensity since 1900) have been overwhelmed by a 44-fold increase in output, causing a 17-fold rise in annual GHG emissions and recent stagnation in global intensity due to growth in carbon-intensive regions. Future pathways consistent with 1.5 °C require rapid, large-scale deployment of breakthrough low-carbon technologies and substantial demand reductions; either approach alone is insufficient within the given carbon budget. The study contributes a century-scale, system-wide quantified view of steel emissions, the interplay of flows and technology, and scenario-based requirements. It recommends that nations and producers develop ambitious, regionally tailored roadmaps combining supply-side decarbonization (including CCUS, hydrogen/electrolysis where feasible, carbon-free EAFs, and BF improvements) with demand-side material efficiency and circularity, alongside policy and market instruments to accelerate adoption. Future research should refine regionalized LCA intensities, incorporate trade flows and product-quality constraints in secondary steel, evaluate policy mixes for technology diffusion, and assess dynamic interactions between material efficiency, product design, and service demand.

• Technology intensities are modeled as global averages; regional differences are not fully captured due to limited comprehensive datasets.
• Material flows are expressed as average Fe content, which may slightly affect quantitative results.
• Regional retrospective/prospective analyses omit international trade of steel and products, so regional trends are indicative rather than actual flows.
• Scenario analysis is explorative and depends on assumed carbon budget allocations to steel and IEA scenario projections; real-world policy, market, and technology adoption dynamics may differ.
• Breakthrough technology timelines and potentials are based on literature and reports; uncertainties in cost, scalability, and integration remain.