Periodic Reporting for period 1 - DiGreeS (Demonstration of Digital twins for a Green Steel value chain)
Periodo di rendicontazione: 2024-11-01 al 2025-12-31
The transition to low-carbon, eco-friendly European steel production requires a profound transformation of processes, including new steelmaking routes. Key enablers are needed to plan and manage the revolution, achieve near-zero emissions, and enable breakthroughs that reduce environmental footprints. Seamless digitalization of production processes and a skilled workforce are essential.
DiGreeS will tackle these challenges by implementing an integrated digitalisation approach throughout the steel value chain, to enable an enhanced use of the industrial data collected along the process chain and ensuring the uptake of human experiences for easier industrial integration. The aim of DiGreeS is to develop a user-friendly digital platform for networked production based on novel and soft sensors as well as related approaches and models, which will be demonstrated in three individual use cases targeting different segments of the steel value chain. Within DiGreeS comprehensive digital twins will be developed to support the efficient feedstock verification and real-time control of the production of crude steel with the electric arc furnace (EAF) and to increase the process yield while improving the quality of intermediate and final steel products. In this context the potential of artificial intelligence and machine learning technologies will be fully exploited to support the optimal use of process data, and various scenarios specific to each use case (UC) will be modelled.
DiGreeS will tackle these challenges by implementing an integrated digitalisation approach throughout the steel value chain, to enable an enhanced use of the industrial data collected along the process chain and ensuring the uptake of human experiences for easier industrial integration. The aim of DiGreeS is to develop a user-friendly digital platform for networked production based on novel and soft sensors as well as related approaches and models, which will be demonstrated in three individual use cases targeting different segments of the steel value chain. Within DiGreeS comprehensive digital twins will be developed to support the efficient feedstock verification and real-time control of the production of crude steel with the electric arc furnace (EAF) and to increase the process yield while improving the quality of intermediate and final steel products. In this context the potential of artificial intelligence and machine learning technologies will be fully exploited to support the optimal use of process data, and various scenarios specific to each use case (UC) will be modelled.
A requirement analysis of the use cases was carried out in terms of the current state of the use case environments, both with regards to physical and digital aspects. Data sources, interfaces, security and data sets pertaining to the use cases were reviewed and gaps and needs for implementing digital twins and process models were identified. Data structures and sensors were defined, and new hardware was ordered so that they meet the requirements and can be adapted and integrated in existing process lines and infrastructure in the respective use case. Missing information and IT/OT infrastructures were identified.
Significant progress has been made in mechanical design, lab testing, and sensor calibration:
• Customisation of LIBS and camera systems to characterize heavy melting scrap; Engineering concepts and sensor testing are well advanced.
• Customisation sensors to monitor EAF processes; Sensor placement studies and first measurements to optimise the sensor to the EAF operational environment were carried out.
• Customisation of magnetic sensors to measure residual stress in steel sheets inline; Initial lab tests using thistechnique were completed successfully distinguishing between different stress conditions in steel plates.
To ensure the complete dataflow within and between the use cases and a better use of industrial data, appropriate interfaces from and to the IT system attached to the various steel manufacturing processes that play a crucial role in all forms of data communication were defined. This includes a data lake instance, a dashboard and a computational environment. A centralized offline data architecture and data storage systems, which serve as a solid basis for modelling and incorporate the necessary feature-generation functionalities has been designed. Moreover, a methodology for the cybersecurity design has been adopted adhering to the principle of “secure by design”.
Regarding the security architecture design and utilization, cybersecurity risks are managed from the initial steps of the project. With the holistic security architecture, the introduced CSIRT and federated SOC operator the continuous monitoring of security posture will take place. Additionally, security incident management framework will be established and periodically tested for the purpose of further enhancing trust between participants.
To develop, adapt, and extend use case specific process models using physical/analytical as well as data-based/ML approaches with inputs from process data and novel sensor information folowing activities were carried out.
• UC1: Initial trials with test images of a VIS camera and LIBS to segment scrap pieces and investigation of methods to speed up segmentation
• UC2: Definition of process input data needed for adaptation and validation of the existing model for the EAF; Assessment and evaluation of the first historical process data (electrical energy supply) at the Ascoval EAF;
• UC3: Definition of available data and internal efforts to bring all the relevant data into one data-warehouse; Sharing of first data-batch and training materials helpful for the ML-model development; Clarification about documentation and programming languages
Within the project, data and innovation management a clear organizational framework, guidance and all support mechanisms to enable a smooth project workflow and to ensure that objectives and milestones are be met in time were provided. The projectIt also coordinated communication, dissemination, provided optimal visibility and a wide outreach to relevant stakeholders.
Significant progress has been made in mechanical design, lab testing, and sensor calibration:
• Customisation of LIBS and camera systems to characterize heavy melting scrap; Engineering concepts and sensor testing are well advanced.
• Customisation sensors to monitor EAF processes; Sensor placement studies and first measurements to optimise the sensor to the EAF operational environment were carried out.
• Customisation of magnetic sensors to measure residual stress in steel sheets inline; Initial lab tests using thistechnique were completed successfully distinguishing between different stress conditions in steel plates.
To ensure the complete dataflow within and between the use cases and a better use of industrial data, appropriate interfaces from and to the IT system attached to the various steel manufacturing processes that play a crucial role in all forms of data communication were defined. This includes a data lake instance, a dashboard and a computational environment. A centralized offline data architecture and data storage systems, which serve as a solid basis for modelling and incorporate the necessary feature-generation functionalities has been designed. Moreover, a methodology for the cybersecurity design has been adopted adhering to the principle of “secure by design”.
Regarding the security architecture design and utilization, cybersecurity risks are managed from the initial steps of the project. With the holistic security architecture, the introduced CSIRT and federated SOC operator the continuous monitoring of security posture will take place. Additionally, security incident management framework will be established and periodically tested for the purpose of further enhancing trust between participants.
To develop, adapt, and extend use case specific process models using physical/analytical as well as data-based/ML approaches with inputs from process data and novel sensor information folowing activities were carried out.
• UC1: Initial trials with test images of a VIS camera and LIBS to segment scrap pieces and investigation of methods to speed up segmentation
• UC2: Definition of process input data needed for adaptation and validation of the existing model for the EAF; Assessment and evaluation of the first historical process data (electrical energy supply) at the Ascoval EAF;
• UC3: Definition of available data and internal efforts to bring all the relevant data into one data-warehouse; Sharing of first data-batch and training materials helpful for the ML-model development; Clarification about documentation and programming languages
Within the project, data and innovation management a clear organizational framework, guidance and all support mechanisms to enable a smooth project workflow and to ensure that objectives and milestones are be met in time were provided. The projectIt also coordinated communication, dissemination, provided optimal visibility and a wide outreach to relevant stakeholders.
Sensor portfolio per use case was defined (UC1 LIBS+VIS; UC2 Off-Gas, AE sensors, Rogowski coils; UC3 3MA-X8).
The combination of LIBS and visual (RGB) camera systems for the HMS characterisation, addressing additional challenges coming with the size and shape, and variations in object surface tilt angle was succesfully tested.
Novel inline-capable sensors to characterize the residual stress level in steel plates was succesfully tested in operational environment.
Sensors to monitor the processes at the EAF were succesfully tested.
Networking concepts particularly networked secured data lakes were developed. A scalable, unified data architecture featuring a central Data Lake was implemented in the SZTAKI cloud. Robust communication interfaces have been designed. The exchange of data is predominantly file-based.
The combination of LIBS and visual (RGB) camera systems for the HMS characterisation, addressing additional challenges coming with the size and shape, and variations in object surface tilt angle was succesfully tested.
Novel inline-capable sensors to characterize the residual stress level in steel plates was succesfully tested in operational environment.
Sensors to monitor the processes at the EAF were succesfully tested.
Networking concepts particularly networked secured data lakes were developed. A scalable, unified data architecture featuring a central Data Lake was implemented in the SZTAKI cloud. Robust communication interfaces have been designed. The exchange of data is predominantly file-based.