Steelmaking remains one of the world’s largest industrial sources of CO2 emissions. This study presents a sustainable valorisation pathway by converting blast furnace and basic oxygen furnace off gases into synthetic methane a storable energy carrier.

We introduce a novel hybrid approach that couples rigorous thermodynamic modelling with advanced machine learning (ML) to optimize methane production while minimizing carbon formation. Over 5000 scenarios using Gibbs free energy minimization are performed to map optimal operating conditions. Results show methane yield peaks between 150–250 °C and above 20 bar, with CO2 conversion exceeding 98% and solid carbon suppressed to <1 wt%.

To accelerate catalyst discovery and reduce experimental load, we develop ML models trained on a Organized dataset of 2777 catalysts. These models use features derived from atomic structure, formation energy, and synthesis metadata. XGBoost and CatBoost achieve R2 values above 0.93 for methane yield and CO2 conversion. SHAP analysis identifies synthesis method and surface atomic structure as dominant drivers.

This is the first integrated framework that combines thermodynamics and ML to directly link catalyst design with methane yield from real steel plant emissions. It offers a scalable, data-driven solution for transforming CO2-rich industrial gases into valuable low-carbon fuels.

By turning steelmaking emissions into resources, this work supports circular carbon strategies and aligns with global decarbonization targets for heavy industry.

Read the full abstract here.

The decarbonisation of the steel industry – responsible for 8% of global CO2 emissions – is essential for achieving our net-zero targets by 2050. Studies have shown that using steel scrap instead of virgin ore in steelmaking can reduce emissions by 58% and water consumption by 40%, while also conserving natural resources. The Electric Arc Furnace (EAF) route, which can achieve up to 100% scrap recycling, compared to the 20% maximum for the traditional Basic Oxygen Furnace (BOF) route, is central to realising a complete Circular Economy for steel.

The rapid increase in reliance on electric arc furnace (EAF) steelmaking faces a critical bottleneck: the quality and availability of steel scrap. However, approximately 60% of available scrap is classified as obsolete scrap. This material typically contains elevated concentrations of residual elements (e.g., Cu, Sn, and Sb), which are detrimental to mechanical properties and therefore limit its use in high-performance applications. Consequently, it is necessary to expand the tolerance of high-strength steels to residual elements in order to increase the recycling rate of obsolete scrap and ensure a sustainable supply of recycled steel resources for EAF-based steel production.

Additionally, high-strength, low-alloy steels used for hydrogen transmission and storage present another challenge, hydrogen embrittlement. This is because these elements can interact with microalloying elements to segregate at interface of precipitates and matrix, or grain boundaries, potentially increasing susceptibility to hydrogen-related failures.

This research will provide valuable insights into the influence of residual elements and hydrogen on the micromechanical embrittlement mechanisms. This work will also seek to refine tolerance limits of residual elements and to aid in the development of residual element tolerant and hydrogen resistant microstructures for use in hydrogen transmission and storage. Through improved understanding and enhanced utilisation of recycled obsolete scrap, the steel industry will be able to produce more grades of steel via the scrap-EAF route, enhancing competitiveness in green steelmaking and reduce reliance on clean steel imports.

To achieve this objective, the effects of residual elements on phase transformation behaviour and elemental segregation will first be simulated to provide guidance for residual element selection. This will be followed by experimental validation to determine acceptable residual element contents in high-strength low-alloy steels. The influence of residual elements on microstructure and the corresponding mechanical properties will then be systematically investigated using electron microscopy, together with tensile and toughness testing. Ultimately, optimal compositions for high-strength low-alloy steels will be established by defining allowable windows of residual element content in relation to microstructural characteristics and mechanical performance.

View the full abstract here.

Digital twins offer a powerful route to accelerating innovation in steel processing by providing a verified virtual counterpart to real-world production, testing, and in-service conditions. At WMG, we develop digital twins that simulates the behaviour of physical equipment such as rolling mills, run-out table cooling rigs, continuous annealing lines, and mechanical test systems. These models operate across multiple physics domains: thermal, mechanical, and electromagnetic; and are coupled with metallurgical models capable of predicting phase transformation, grain size evolution, recrystallisation, and final material properties. Together, they form a comprehensive simulation chain that can replicate each stage of steel processing and feed validated initial conditions into subsequent models, enabling end-to-end process prediction.

The objective of this work is to establish a digital ecosystem that allows users to interact with these models, explore fundamental metallurgical behaviour, and apply the tools to industrially relevant scenarios. This ecosystem will support both open-source dissemination and confidential collaboration with industrial partners, ensuring that models can be deployed effectively across research, development, and educational contexts. The integration of experimental facilities with their digital representations provides deeper insight into process sensitivities, material–process interactions, and the effects of parameter variation.

Current activity includes a major HVM Catapult programme to develop the long-term digital twin platform at WMG. The ecosystem will host digital twins that span hot and cold rolling, run-out table cooling, continuous annealing, and mechanical testing.

View the full abstract here.

Microplastics are of significant growing concern, having been identified in every environment across the Planet. While there is a wealth of research being performed on the identification and collection of microplastics from marine environments, there is a lack of information regarding their release from coatings, which are known to degrade through various mechanisms during service life. This degradation causes a reduction in chemical and physical properties, as well as a thinning of the system. This work combines a novel encapsulated approach to accelerated weathering exposure, combined with assessment of the solid samples pre- and post-weathering and micro-FTIR analysis on the filtered components, as a unique method to identify and assess microplastic artifacts that are released as a consequence of weathering degradation.

The study also includes an assessment of the reproducibility and reliability of the utilised methodology to investigate the inherent variability of the approach.

View the full abstract here.

The urgency of climate mitigation is reshaping global industry and calling for transformative changes towards decarbonisation, electrification and carbon recycling for achieving a sustainable and circular economy. One promising approach involves utilising CO₂ emissions from hard-to-abate industries as a carbon alternative for synthesising valuable chemicals and fuels. Electrochemical carbon dioxide reduction (EC-CO2R) has emerged as a key technology that offers the direct conversion of captured CO₂ into a range of chemicals using renewable electricity. Operating at low temperature and pressure, and using sustainable environmental materials, the scale-up of EC-CO2R holds great promise for tailored and modular CO2 recycling. Despite this, implementation of EC-CO2R is hindered at present by the short term stability of electrodes when operated at relevant conversion rates. The crux of EC-CO2R electrolysis is the gas diffusion electrode (GDE) employed for the cathodic process. The GDE must function as a barrier between the CO2 and electrolyte compartments of the electrolyser, whilst being permeable to gaseous CO2 and product diffusion to/from the catalyst. Unfortunately, GDL wetting and hydrophobicity loss during prolonged operation is a well-known issue that causes the electrolyte to penetrate beyond the catalyst layer resulting in adverse performance. The ‘electrolyte flooding’ of EC-CO2R reactors is a common problem that limits the vertical size of GDEs because of a low pressure barrier to flooding and the hydrostatic pressure gradient of the electrolyte. Work by our group has tackled this problem and developed a novel approach for eliminating the GDE pressure gradient. By use of a 32 cm tall prototype electrolyser and differential pressure probing, we demonstrate for the first time a GDE with vertically homogenised pressure, and we explore the design aspects for prolonged GDE operation and efficient catalysis at scale. Our electrochemical engineering presentation will elucidate the unintuitive fluid physics of operation and highlight the persisting challenges with GDEs for EC-CO2R.

View the full abstract here.

A new pyrometallurgical route has been developed to treat and valorise the solid waste from iron- and steel-making processes. The technology enables the removal of zinc (dezincing) from iron-rich sludges from steelmaking and improves the makeup of the iron phase, allowing its reintroduction into the steel production cycle as a sinter feedstock for the blast furnace. The new technology was introduced and the life cycle assessment (LCA) was carried out to evaluate the environmental impacts arising in the treatment of seven steelmaking sludges from various stages of the steel production. The technology involves preparation and roasting of sludges in a top-blown rotary converter (TBRC) at 1200°C for 60 minutes resulting in high degree of dezincing (68-97%) and high retention of iron oxide. The inventory for the LCA was collected in situ and scaled up to emulate industrial conditions, consistent with the modelling of emerging technologies in LCA. The impacts are calculated for eight categories of the CML impact assessment method. The assessment was carried out to determine the main process hotspots and compare the dezincing of sludges with the disposal to the landfill, as the current end-of-life practice. The impacts of dezincing route were calculated for the process-only burdens scenario and avoided burdens scenario (incorporating credits of the downstream valorisation of the dezinced product). The findings suggest that the dezincing of all sludges is beneficial in majority of the impact categories compared to the landfill. Valorisation of dezinced product results in offsetting net impacts of dezincing by 20-30% in most of the impact categories. In the global warming category, dezincing route emits on average 0.27 and 0.3kg CO2eq. in comparison to 0.36kg CO2eq. estimated in the landfilling scenario. Energy use for TBRC and drying, and treatment of off-gas dust are the most impactful process stages and several process-design opportunities are identified to further reduce and mitigate these impacts.

View the full abstract here.

Hydrogen embrittlement (HE) poses a significant challenge to the integrity and safety of materials in hydrogen-rich environments, particularly as the hydrogen economy gains momentum as a cornerstone of global clean energy initiatives. It occurs when hydrogen atoms diffuse into the metal, leading to a reduction in ductility and toughness, ultimately causing premature failure under stress. This problem is particularly critical in high-stress environments such as aerospace, automotive, and energy sectors, where the reliability and safety of materials are paramount. This study aims to show the critical role of hydrogen barrier coatings (HBCs) as a primary strategy to combat HE in susceptible steels, focusing on the state-of-the-art advancements in barrier coatings particularly in polymer coatings, an underexplored yet promising solution.

View the full abstract here.

The UK steel industry is undergoing a structural shift away from traditional blast furnace basic oxygen furnace (BF-BOF) routes toward a more electrified scrap-based system under tightening climate policy, rising electricity costs and the legally binding 2050 net zero target. This review assembles a long-run dataset from 1800 to 2025 that links crude steel output, production processes, and product mix with estimates of primary energy use, electricity demand, and direct CO2 emissions. Then it uses these trends to evaluate low-energy pathways for 2025-2050. Four route families are assessed, including scrap-fed electric arc furnaces (EAF), hydrogen-based direct reduction with EAF, blast-furnace-based routes equipped with carbon capture and storage, and cross-cutting options such as low-temperature efficiency retrofits and digital twins to increase process efficiency. Period-averaged energy and carbon intensities show that historic transitions from Bessemer and open hearth to BF and BOF, and more recently to EAF, have progressively lowered primary energy and emissions per tonne while increasing reliance on low-carbon electricity. Scenario analysis indicates that a scrap-maximising EAF strategy, anchored in abundant UK scrap and supported by efficiency upgrades and digital optimisation, can deliver most near-term abatement, provided grid carbon intensity and industrial power prices continue to fall. These technology choices are mapped to current and emerging instruments including the design of the UK Emissions Trading Scheme and its free allocation reforms, carbon price support (CPS), the planned UK Carbon Border Adjustment Mechanism (CBAM), UK emissions trading scheme (ETS) and the steel strategy process, which together shape carbon cost exposure, investment signals and leakage risk. The review therefore proposes a low-energy sequencing in which scrap EAF deployment and efficiency gains are prioritised, while hydrogen-ready DRI options are preserved for later phases, as hydrogen networks, the National Wealth Fund and electricity market reform enable deeper decarbonisation of UK steel.

View the full abstract here.

An electromagnetic (EM) sensor array system consisting of four independently operated sensor heads was developed for in situ, non-destructive monitoring of phase transformation in a speciality steel narrow strip mill. The array was installed between mill rollers at the pyrometer location immediately upstream of the coiler, enabling direct temporal correlation between EM responses and surface temperature measurements. Mechanical modification of the apron plate, incorporating a thick ceramic insert, provided a thermally stable and electromagnetically transparent sensing window compatible with industrial operating conditions.

EM sensor responses were acquired during processing of multiple steel grades and selected metallic alloys under varying cooling regimes. Signal analysis focused on transformation-sensitive parameters, including zero-crossing frequency, and the real and imaginary components of inductance, as well as low-field impedance behaviour. These features were evaluated as potential indicators of microstructural evolution during cooling. To support interpretation of sensor responses, complementary resistivity modelling was performed based on alloy chemical composition, providing insight into the relationship between electromagnetic properties and phase transformation dynamics.

The study demonstrates that multi-sensor EM array deployment is feasible in a high-noise industrial environment without observable cross-sensor interference. The results indicate that EM-derived parameters exhibit measurable sensitivity to transformation-related property changes, enabling characterisation of spatial and temporal variations in transformation behaviour. These findings establish a framework for applying EM sensing as a quantitative tool for real-time microstructural monitoring in thermomechanical steel processing.

View the full abstract here.

The transition towards circular steel is not only a technical and environmental challenge, but also an organisational one. Unifying the steel supply chain around circularity goals requires a holistic sustainability approach. Yet, organisations continue to face both time pressures and uncertainty when preparing for disruption. This ongoing study examines whether organisational resilience can enable steel circularity, and under which circumstances organisations may be compelled to invest in, adopt, or demand circular steel processes.

Building on prior research, the study applies four levers of organisational resilience: resiliency management, business continuity, organisational learning capacity, and operational flexibility. These levers are used to explore how organisations respond to disruptions that may affect the circular steel transition. The poster presents a web-based research tool with which participants are tasked with prioritising investment in these levers, with a view to tackle a potential disruption under time pressure and uncertainty constraints.

These potential disruptions may occur at the political level (e.g., tariffs, tighter origin compliance, and CBAM implementation), at the supply chain level (e.g., closed-loop dependence, climate-related incidents, and critical material constraints), or at the product level (e.g., missed design specifications during circularity-driven upgrades). The poster serves as both a research presentation and a live data-collection opportunity, inviting practitioners to contribute to an emerging evidence base on resilience and circularity in the steel circularity transition.

View the full abstract here.

Precise control of microstructure during hot strip processing is central to achieving targeted mechanical properties while improving process efficiency and reducing over-processing. Current industrial practice relies primarily on temperature-based models, empirical correlations, and grade-specific calibration, which limits transferability under evolving cooling strategies.

This work presents the development of a physics-based digital twin framework for electromagnetic (EM) sensor monitoring of phase transformation on the run-out table. The approach links microstructure evolution modelling with multi-scale magnetic homogenisation and full Maxwell-based sensor forward modelling to predict EM signal response directly from evolving phase fractions.

Measured EM signals are used to update the internal microstructure state of the model via state estimation, enabling a transition from monitoring to predictive capability. Initial results demonstrate sensitivity to phase transformation behaviour across cooling conditions (e.g., ferrite vs bainite/martensite formation).

The framework provides a structured route toward microstructure-informed cooling control, reducing dependence on grade-specific empirical calibration while improving robustness and interpretability.

View the full abstract here.

In the context of climate change, steel production accounts for ~7% of global CO2 emissions according to the IEA, making it a major contributor whilst being a vital technology for humanity. Many countries are prioritising the development of emission-reduction technologies in the steel industry, and there is a growing trend towards reform and innovation within steel companies. Specific technologies include molten salt and molten oxide electrolysis (MSE/MOE), which enable the direct production of metal from oxide feedstock. A typical MSE/MOE cell is composed of four main elements: an anode, a cathode, an electrolyte, and an energy source. However, the carbon-neutral production of metals depends on the use of an inert anode, which allows the reaction by-product to be O2 and achieves a virtually carbon-zero-emission process (replacing consumable graphite electrodes, with associated C emissions). Finding a suitable anode material remains a limitation in this industry, with challenging requirements: resistance to high temperatures, durability, corrosion resistance under anodic polarisation, and electronic conductivity, while remaining economically feasible.

In this project, Cr-based alloys are explored as a possible alternative to produce inert anodes, building on the promising properties they have shown towards concentrated solar power. For that, several Cr alloys will be characterised and tested in lab-scale cells to be compared with the benchmark used by Allanore, A. et al. (Cr-Fe). These alloys were produced by powder metallurgy and provided by Plansee Composite Materials, a company experienced in producing alloys that are challenging to manufacture. The alloys have been characterised using advanced techniques, such as scanning electron microscopy (SEM), electron backscattered diffraction (EBSD), and x-ray diffraction (XRD), to understand their microstructure & grain size. To evaluate the more promising candidates, preliminary testing is being performed, exposing the alloys to the molten salts/oxide used by the companies in which the lab-scale tests will occur. After testing, the interaction between the alloy and the corrosive medium is evaluated by SEM, with Energy Dispersive Spectroscopy (EDS) as well as EBSD.

This analysis allows us to evaluate how the different alloys perform in contact with the extremely corrosive environment, giving crucial information to the scientific community that can be useful for different applications (e.g., concentrated solar cells) and support the decision-making in lab-scale cell testing.

View the full abstract here.

Increasing scrap use in steelmaking raises levels of residual elements such as Cu and Sn, which are difficult to remove and could influence microstructure evolution during reheating and thermomechanical processing. This study systematically investigates how residual Cu and Sn influence austenite grain growth in a C–Mn steel between 900–1150°C. Grain size measurements show that both elements suppress grain growth, but Sn is more effective than Cu. Microscopy reveals clear Sn segregation at austenite grain boundaries, while Cu segregation is not detected. Solute-drag model was used to quantify segregation tendency and its impact on grain boundary mobility. When compared to conventional microalloying elements, the retardation effect follows the order: Nb > Mo > Sn > Cu.

View the full abstract here.

With the exponential expansion of secondary steelmaking currently, production of aluminium from secondary sources is following suit, specifically in the automotive industry. This shift is also fuelled by the increased demand for aluminium as an alternative to steel in certain applications as a result of lightweighting and alloy advancements. There are many similarities between the challenges facing secondary aluminium production and secondary steelmaking.

Whilst the steel industry’s sourcing and recycling of scrap is more mature, with existing infrastructure and supply chain in place in the UK, secondary aluminium processing is still new in comparison. Accumulation of residual elements in recycled steel, their effect on microstructure and processing, may mirror those in aluminium, and originate from known locations or specific industries.

Since many automotive components are subject to high cycle fatigue (HCF) in service, it is imperative to understand the effects of increased recycled content on mechanical and fatigue behaviour of aluminium. Increased content of alloying elements such as iron, silicon and manganese is known to decrease fatigue life of aluminium alloys at HCF. Mitigating the effects of these residuals will involve amendments to processing parameters and tolerances, a challenge already being tackled in secondary steelmaking. An understanding of common scrap aluminium sources and their alloying content is a crucial first step in scaling up to an industrial level.

View the full abstract here.

Evolving vehicle technologies, supply chains and environmental considerations are aligning ELV sectoral practices more closely with wider circular economy (CE) principles. The ferrous scrap outputs from ELV processes represent an important material feedstock for EAF steelmaking. This paper supports SUSTAIN GCRA1 Carbon Neutral Iron and Steelmaking by exploring the delivery of cleaner ELV-derived scrap for circular EAF production, aligned with the conference theme The Iron Cycle, Foundations for Circular Steelmaking. The paper establishes practical technologies and business models for optimising ELV dismantling and selective separation, aiming to increase component-value recovery and improve the quality of ferrous scrap supplied to EAF steelmakers.

Authorised Treatment Facilities (ATFs) undertake the depollution, dismantling and crushing or baling of ELVs. Shredding at specialist facilities follows, with mechanical separation of ferrous and non-ferrous metallic fractions and non-metallic residues for recycling or disposal. This paper focuses on identifying and prioritising items for removal from ELVs, whether for recycling, repurposing or re-use, with particular emphasis on optimised ferrous scrap chemistry. Targeted removal of electronics and other mixed-material assemblies is considered as a route to reducing cross-contamination in steel streams. The work forms part of the SUSTAIN e-RAMS programme on managing residual elements in recycled steel and proposes a research plan demonstrating how enhanced ELV processing knowledge, tools and techniques can deliver greater CO2 emissions reductions and improved CE outcomes.

Whilst large-scale destructive separation of ELVs improves material recovery rates at scale, the removal of smaller components remains challenging for this technology. The manual dismantling of ELVs to harvest components for resale also delivers positive CE outcomes, albeit subject to labour costs. When combined with targeted skills and knowledge developments, smaller scale destructive separation technology represents an opportunity to remove additional ELV items, enhancing direct value as well as that of residual ferrous scrap supplied to EAF steelmakers. Process innovations emerging from EV dismantling, especially battery-first separation routines, offer transferable benefits for internal combustion ELVs. Additionally, potential benefits are identified in the application of newer technologies to ancillary process tasks.  Across these technology options, data-driven identification of an optimum level of dismantling / separation for specific ELVs is key. Process selection factors include costs, ELV condition, resource availability and material market prices. Manufacturing and in-use stakeholders (OEMs, suppliers) are largely disconnected from ELV stakeholders (ATFs, recyclers), weakening feedback loops.

The study therefore combines enhanced OEM and supplier engagement, and the identification of new outlets for recovered components and materials, with experimental work at ATFs. Timed process trials of manual dismantling and destructive separation, including assessment of component condition on removal, generate data that can be used to model costs and revenues for different dismantling strategies. On this basis the paper develops a framework that links ELV process choices to stakeholder outcomes: for recyclers, higher value per ELV and clearer guidance on viable process models and target components; for OEMs, richer feedback on design for disassembly and wider circular economy goals; and for steelmakers, cleaner and more predictable ferrous scrap feedstock suitable for circular EAF-based steelmaking pathways.

View the full abstract here.

Electric Arc Furnaces (EAFs) present a low-carbon alternative to traditional Basic Oxygen Furnaces (BOFs), reducing emissions from 1.7 tCO₂/tonne to 0.2 tCO₂/tonne of steel produced. However, challenges such as thermal non-uniformity and cold spot formation continue to affect process efficiency and steel quality. These inefficiencies lead to increased energy demand and inconsistent melting. While oxy-fuel burners are widely deployed to mitigate this, their dependence on natural gas continues to contribute to CO₂ emissions.

This research investigates a novel approach that combines computational fluid dynamics (CFD) modelling with biomass-based fuel injection to improve thermal flow and reduce environmental impact. A 3D CFD model of an EAF is being developed in ANSYS Fluent to simulate heat transfer and fluid flow with and without biomass co-injection. Cold spot regions will be identified in baseline oxy-fuel simulations and subsequent biomass injection studies will optimise parameters such as flow rate, injection location and combustion efficiency.

The study expects to achieve up to a 10% reduction in energy consumption and significantly more uniform heat distribution within the furnace, thereby improving overall melting efficiency. Experimental validation trials are planned to compare CFD predictions with real-world data. A feasibility study will also assess the potential for scaling the biomass injection method to industrial scale EAFs.

This work aims to provide a CFD-driven optimisation framework that not only improves EAF energy efficiency and steel quality but also supports the transition towards net-zero steelmaking through integration of renewable biomass fuels.

View the full abstract here.

Refractory wear in gas-stirred vessels is strongly influenced by plume behaviour at the plug. In this study, computer vision was used to analyse high-speed video and measure plume width over time. The results show that plume fluctuations, more than average size alone, play a key role in wear. This approach supports data-driven design and operation improvements.

View the full abstract here.

Reliable heat treatment is critical to achieving the required strength–toughness balance and performance consistency in advanced high-strength steels (AHSS). However, industrial verification of tempering and annealing processes still relies largely on destructive post-process testing or indirect thermal analysis, limiting real-time quality assurance. This work presents an electromagnetic (EM) sensing approach integrated into commercially relevant heat-treatment platforms to enable in-situ, non-destructive monitoring of microstructural evolution.

First, a miniaturised high-temperature EM sensor was embedded within a commercial Differential Scanning Calorimeter (DSC) to characterise tempering behaviour in AISI 4340 steel. Simultaneous heat-flow and electromagnetic measurements were performed during stepwise tempering between 200°C and 600°C with isothermal holds. Progressive increases in electromagnetic impedance were observed during tempering, corresponding to martensite decomposition, carbide precipitation, and microstructural stabilisation. The electromagnetic response showed strong correlation with mechanical softening measured via microhardness testing.

Second, a multi-frequency EM sensor system was deployed within a high-temperature annealing rig to monitor phase transformations in DP600, DP1000, and interstitial-free (IF) steels under thermal cycles representative of continuous annealing. The system successfully detected austenite formation, Curie transitions, recrystallisation behaviour, and transformation-dependent magnetic evolution during heating and cooling.

Together, these results demonstrate that electromagnetic sensing enhances commercial heat-treatment systems by enabling true in-situ, real-time monitoring directly within the thermal processing environment. The sensors operate during heating and cooling, continuously tracking tempering progression and phase transformations without interrupting the thermal cycle or requiring destructive inspection. By embedding electromagnetic sensing within commercial DSC and furnace platforms, microstructural changes are detected as they occur rather than validated post-process. This capability establishes a pathway toward intelligent, microstructure-informed heat-treatment verification and supports the development of future closed-loop process control strategies in advanced steel manufacturing.

View the full abstract here.

Steelmaking process generates substantial quantities of waste but resourceful materials, a significant portion of which are stored in lagoon systems or landfills. These landfills can pose significant environmental risks due to excessive quantity of heavy metals and other hazardous constituents. This study presents a comprehensive physicochemical and environmental analysis of steelmaking waste of Llanwern Steelworks, UK, collected from two lagoon depths: the top layer (TL: 0.15-0.25 m) and the bottom layer (BL: 0.70-0.80 m). Analytical techniques including Inductively Coupled Plasma Optical Emission Spectroscopy (ICP-OES), Scanning Electron Microscopy/Energy-Dispersive X-ray Spectroscopy (SEM/EDS), Brunauer-Emmett-Teller for surface area determination (BET), X-ray Diffraction (XRD), Fourier Transform Infrared spectroscopy (FTIR), Raman spectroscopy, Thermogravimetric Analysis (TGA) and Pyrolysis-Gas Chromatography-Time-of-Flight mass spectrometry (PY-GCToF) revealed significant heterogeneity in steelmaking waste. Iron dominated all samples (TL: 789-803 ppm; BL: 816-866 ppm), while calcium was 4.5 times higher in BL (13.37-13.86 ppm vs. TL: 2.98-3.73 ppm). Toxic elements (As, Cd, Sb, Be, Co) were below detection limits (<0.05 ppm), however Cr (up to 0.35 ppm), Cu (up to 0.24 ppm) and Pb (up to 0.17 ppm) exceeded Water Framework Directive (WFD) drinking water standards. TL exhibited higher porosity, organic content, and BET surface area, whereas BL showed greater mineral crystallinity and thermal stability. PY-GCToF identified volatile organic compounds (benzene, toluene, naphthalene) in TL, suggesting the presence of more recent organic matter deposits. These results highlight resource recovery potential for environmental remediation application and underscore the need for depth-resolved risk management in lagoon systems.

View the full abstract here.

The global shift toward sustainable steel production has accelerated the adoption of Electric Arc Furnace (EAF) technology, where steel scrap replaces traditional raw materials. However, the presence of residual elements in recycled steel, such as tin (Sn) and copper (Cu), can significantly influence microstructural evolution during thermomechanical processing. These effects include alterations in grain growth, recrystallisation during hot deformation, and phase transformations during cooling. A series of double-hit compression tests were employed by the Gleeble-HDS-V40 to determine the fraction of recrystallisation under varying residual contents such as Sn and Cu. Tests were conducted under varied strains and temperatures with a specific interpass time, leading to partial or full static recrystallisation. The austenite recrystallisation fraction was determined by evaluating the softening fraction observed during the second deformation pass relative to the first hit. The Solute Retardation Parameter (SRP) was used to evaluate the solute drag effect introduced by the residual elements, allowing for a comparative assessment against conventional microalloying elements. Additionally, the effect of residual element addition on the austenite grain size after full recrystallisation was studied using reconstructed prior austenite grain maps from EBSD analysis.

View the full abstract here.

In this work, we present an integrated research package aimed at developing a computer vision–based system for steel scrap quality monitoring, an essential step powering the UK’s transition to scrap‑fed electric arc furnaces (EAFs)-based steelmaking. As the steel industry moves toward circular, low‑carbon production, robust scrap characterisation becomes critical to control tramp elements such as copper, which directly affect steel quality and process compatibility. Our research tackles this challenge through two coordinated routes: (1) the development of a segmentation‑guided vision system and scalable dataset for estimating tramp copper content in fragmentised Grades 3A/3B scrap, and (2) the creation of a machine unlearning framework that supports data‑efficient, confidentiality‑compliant model training and adaptation across industrial partners.

In the first route, we design a computer vision pipeline to estimate copper percentage directly from RGB images of commercial 3B‑derived scrap mixtures. Using batch‑level mass measurements and pixel‑wise annotations distinguishing ferrous and copper‑bearing components (including wires and “meatballs”), a Vision Transformer (ViT) architecture is trained to segment and classify individual scrap pieces. By combining the resulting segmentation masks with RGB images, the downstream classifier distinguishes between clean, intermediate (∼3A/3B‑grade), and copper‑rich classes, achieving 86.67% accuracy, significantly higher than an RGB‑only baseline. To validate the method at industrial scale, we perform a large‑scale experiment using a 7‑ton dummy furnace replicating EAF charging practice. Cameras mounted above the basket capture thousands of images during scrap loading, resulting in a curated dataset exceeding 7,500 annotated images across twelve charge compositions. These results demonstrate the technical feasibility of deploying a vision‑based copper estimator in industrially realistic conditions.

The second route builds a complementary framework for data confidentiality-protecting and adaptive learning. We develop a verified machine unlearning method for knowledge distillation, enabling training on distributed datasets sourced from steel plants, recyclers, and research institutes while maintaining full data ownership for each participant. The framework ensures that any partner’s data can be incorporated and later removed from a shared model without compromising data confidentiality or performance guarantees. When integrated within the vision pipeline, the unlearning method also supports semi‑supervised learning by refining pseudo‑labels and reducing manual annotation requirements, thereby accelerating system development and improving adaptability across diverse scrap streams. This approach makes possible efficient, federated‑like collaboration within the steel supply chain while ensuring confidentiality and accountability in model evolution.

Together, these two routes establish a practical and transferable computer vision framework for automated tramp copper monitoring. The combination of segmentation‑guided perception and machine unlearning provides both technical performance and governance transparency, crucial requirements for AI‑enabled process control in the steel sector. Ultimately, this work demonstrates computer vision can support the UK’s scrap‑based steelmaking transition, improve scrap utilisation and advance sustainable metallurgy.

View the full abstract here.

The effects of Sn additions up to 0.21 wt% to a C-Mn base steel have been investigated in this study to reveal its influence (in the range consistent with high scrap content steelmaking) on austenite decomposition using the same initial prior austenite grain size (PAGS) distribution under continuous cooling. The initial austenite to ferrite transformation (5% transformation) is delayed to lower temperatures with Sn additions during continuous cooling at rates of 0.1-10°C/s, consistent with delayed nucleation at prior austenite grain boundaries (PAGBs) due to Sn segregation at those sites increasing the critical barrier for nucleation. No evidence was found for significant changes in the subsequent diffusional growth of allotriomorphic ferrite. Pearlite formation at low cooling rates was enhanced by Sn additions, due to reduced Mn micro-segregation in the higher Sn content steel. Bainite formation at a higher cooling rate of 50°C/s is retarded by Sn additions, due to suppression in bainitic ferrite nucleation and bainitic cementite formation. Martensite start temperature (Ms) is independent of Sn content, although Sn modifies the extent of auto-tempering on quenching and the short-term low-temperature tempering responses. Overall, this study demonstrates the effects and potential benefits from residual Sn additions in (i) modifying ferrite nucleation to refine ferrite grain sizes and (ii) delaying bainite formation and tempering of martensite in low carbon low alloy steels.

View the full abstract here.