Periodic Reporting for period 1 - AID4GREENEST (AI powereD characterization and modelling for GREEn STeel technology)
Periodo di rendicontazione: 2023-09-01 al 2025-02-28
The European steel industry must remain globally competitive while transitioning to sustainable, circular production. This transformation is vital to meet the European Green Deal, reduce emissions, and secure critical raw materials. Current steel development is slow, costly, and heavily dependent on physical testing, with fragmented adoption of digital and AI tools delaying Industry 4.0 progress.
The AID4GREENEST project tackles these issues by integrating advanced AI and materials modelling to accelerate sustainable steel innovation. It delivers six AI-based tools covering the entire development pipeline from alloy and process design to product performance, thus enabling faster, cost-effective, and greener steelmaking. Key developments include:
• AI tools linking chemistry, process, and microstructure for low-CO2 steel design.
• ML models predicting creep performance from accelerated tests.
• Microstructure evolution modelling in meter-scale forged components.
• A digital platform aligned with European standards (EMMC, EMCC, EMMO).
The project unites academic, industrial, and tech partners to combine experiments, AI modelling, and digital infrastructure. Impact of the project is driven by:
• Scientific innovation through data-driven, faster R&D.
• Industrial validation with partners like OCAS and RFC.
• Development of an open, interoperable platform.
• Policy alignment with EU climate and digital goals.
AID4GREENEST aims to cut steel development time and cost by over 50 %, reduce emissions and raw material use, and deliver replicable, industry-ready tools to support Europe’s digital and green industrial transition.
The AID4GREENEST project tackles these issues by integrating advanced AI and materials modelling to accelerate sustainable steel innovation. It delivers six AI-based tools covering the entire development pipeline from alloy and process design to product performance, thus enabling faster, cost-effective, and greener steelmaking. Key developments include:
• AI tools linking chemistry, process, and microstructure for low-CO2 steel design.
• ML models predicting creep performance from accelerated tests.
• Microstructure evolution modelling in meter-scale forged components.
• A digital platform aligned with European standards (EMMC, EMCC, EMMO).
The project unites academic, industrial, and tech partners to combine experiments, AI modelling, and digital infrastructure. Impact of the project is driven by:
• Scientific innovation through data-driven, faster R&D.
• Industrial validation with partners like OCAS and RFC.
• Development of an open, interoperable platform.
• Policy alignment with EU climate and digital goals.
AID4GREENEST aims to cut steel development time and cost by over 50 %, reduce emissions and raw material use, and deliver replicable, industry-ready tools to support Europe’s digital and green industrial transition.
AID4GREENEST achieved significant progress in developing AI-powered tools for rapid characterization and performance prediction of advanced steel components across multiple work packages.
In WP2 for AI-based microstructure–process characterization, the team generated an extensive dataset including 14 martensitic steel variants produced by OCAS, with 864 SEM images and 42 EBSD scans from UGENT. They developed SE-to-EBSD prediction models and tested segmentation techniques, though limited dataset size requires expansion for optimal performance. The work included creating diffusion-based generative models that produce realistic micrographs from processing parameters. ePotentia established a flexible database architecture supporting large-scale imaging data and metadata, while 8K+ historical image dataset served as a pilot test case for platform evaluation.
WP3 focused on accelerated creep testing and ML-based prediction, where researchers developed strategies for testing using Stepped Isotress Method (SSM) and LCF-based method. ULiege implemented the Morch law simulation framework while IMDEA created the Bayesian calibration tool (ACBICI). ULieige initiated experimental creep testing on 30CrMoNiV5-11 steel after comprehensive literature analysis.
For WP4, researchers captured detailed thermal histories during meter-scale shaft forging using real-time thermocouple monitoring. They performed physical simulations including Gleeble hot compression tests and developed CCT diagrams through dilatometry expertiments. The work included creating a mean-field model for microstructure evolution prediction during forging and characterizing microstructure variations across shaft locations, correlating phase fractions with cooling history, which is currently being fine-tuned.
WP5 established an AI-based online platform by launching the microstructuredb.com landing website and deploying a beta platform version. The team focused on standardization for interoperability across datasets and AI models following EMMC guidelines, implementing database architecture that supports diverse data types with future compatibility.
In the first period, AID4GREENEST established scientific foundations for predicting and optimizing microstructure and performance in advanced steels through combined experimental simulations and AI-driven models, promising reduced development time and cost while improving prediction accuracy for critical properties.
In WP2 for AI-based microstructure–process characterization, the team generated an extensive dataset including 14 martensitic steel variants produced by OCAS, with 864 SEM images and 42 EBSD scans from UGENT. They developed SE-to-EBSD prediction models and tested segmentation techniques, though limited dataset size requires expansion for optimal performance. The work included creating diffusion-based generative models that produce realistic micrographs from processing parameters. ePotentia established a flexible database architecture supporting large-scale imaging data and metadata, while 8K+ historical image dataset served as a pilot test case for platform evaluation.
WP3 focused on accelerated creep testing and ML-based prediction, where researchers developed strategies for testing using Stepped Isotress Method (SSM) and LCF-based method. ULiege implemented the Morch law simulation framework while IMDEA created the Bayesian calibration tool (ACBICI). ULieige initiated experimental creep testing on 30CrMoNiV5-11 steel after comprehensive literature analysis.
For WP4, researchers captured detailed thermal histories during meter-scale shaft forging using real-time thermocouple monitoring. They performed physical simulations including Gleeble hot compression tests and developed CCT diagrams through dilatometry expertiments. The work included creating a mean-field model for microstructure evolution prediction during forging and characterizing microstructure variations across shaft locations, correlating phase fractions with cooling history, which is currently being fine-tuned.
WP5 established an AI-based online platform by launching the microstructuredb.com landing website and deploying a beta platform version. The team focused on standardization for interoperability across datasets and AI models following EMMC guidelines, implementing database architecture that supports diverse data types with future compatibility.
In the first period, AID4GREENEST established scientific foundations for predicting and optimizing microstructure and performance in advanced steels through combined experimental simulations and AI-driven models, promising reduced development time and cost while improving prediction accuracy for critical properties.
The AID4GREENEST project is advancing the digital transformation of green steel development through AI. In its first reporting period, it achieved major technical milestones, introducing new approaches to steel characterisation, modelling, and performance prediction. Key developments include AI tools linking steel chemistry, processing, and microstructure to enable faster, model-based alloy design with reduced environmental impact. Generative models produce realistic microstructures from processing parameters, advancing digital steel design.
A breakthrough is also underway in long-term performance prediction. The project is developing an accelerated creep testing method supported by simulations and Bayesian calibration, crucial for components like high-temperature turbine shafts. Simulations and experiments on meter-scale parts covering thermal tracking, dilatometry, and high-temperature testing support a mean-field model for microstructure evolution during forging and quenching, aiding in energy-efficient process optimisation.
A landing website for the project’s digital platform, microstructuredb.com was launched. It will manage large datasets from advanced characterisation and integrate AI tools, following FAIR data principles and aligning with standards. The platform aims to support data sharing, reproducibility, and model deployment in materials science.
AID4GREENEST promises to accelerate steel R&D, improve AI predictability, and drive EU decarbonisation and industrial competitiveness. To ensure lasting impact, ongoing research must expand tool applicability, with industrial-scale validation critical in forging and heat treatment. Commercial success will require exploitation plans, industry engagement, and platform sustainability.
The project is working closely with ESTEP and EMMC to ensure alignment with industry and standards, and is committed to regulatory and standardisation integration, positioning Europe as a leader in digital steel innovation.
A breakthrough is also underway in long-term performance prediction. The project is developing an accelerated creep testing method supported by simulations and Bayesian calibration, crucial for components like high-temperature turbine shafts. Simulations and experiments on meter-scale parts covering thermal tracking, dilatometry, and high-temperature testing support a mean-field model for microstructure evolution during forging and quenching, aiding in energy-efficient process optimisation.
A landing website for the project’s digital platform, microstructuredb.com was launched. It will manage large datasets from advanced characterisation and integrate AI tools, following FAIR data principles and aligning with standards. The platform aims to support data sharing, reproducibility, and model deployment in materials science.
AID4GREENEST promises to accelerate steel R&D, improve AI predictability, and drive EU decarbonisation and industrial competitiveness. To ensure lasting impact, ongoing research must expand tool applicability, with industrial-scale validation critical in forging and heat treatment. Commercial success will require exploitation plans, industry engagement, and platform sustainability.
The project is working closely with ESTEP and EMMC to ensure alignment with industry and standards, and is committed to regulatory and standardisation integration, positioning Europe as a leader in digital steel innovation.