Enhancing level 2 performance with AI: The path to better stability and production improvement in steelmaking and casting

The landscape for level 2 systems in steelmaking and casting is continuously evolving. Today's demands surpass traditional needs for process operation visualization, process tracking and reporting. To keep up with these developments and in order to remain competitive companies must focus on utilizing their production capacities efficiently while reducing throughput times. Heats and slabs must be produced with ever-greater precision and quality – and demands on aspects such as quality assurance, minimal downtimes to enhance production efficiency, and rapid system adaptation to new production requirements and function extension are growing significantly.

Additionally, software services and according, operational support are becoming increasingly vital, since steel producers aim to continuously optimize level 2 systems and software systems long after their initial implementation. A good example are performance-based service contracts, where functionality enhancements are assessed based on their measurable contribution to improve production outcomes.

SMS group’s level 2 X-Pact^® Process Guidance Platform, a software solution for process automation, establishes the critical foundation. Combined with SMS group’s unique Digital Ready Standard, it provides the essential prerequisites for enabling the learning steel plant. With our digital ready standard, all automation data is described homogenously using contextual information (metadata) and data access information and published in a standardized way via the supported interface technologies. Individual, independently operating software modules communicate via ProconTEL event bus and are supplied with process data from a central process data lake called Information Hub. State-of-the-art frameworks, .NET programming languages, and Web-HMI user interface are implemented, ensuring that these remain the modern standard for years to come. Data is structured and stored according to the same rules thanks to the digital ready standard. 

Building on this solid foundation, the integration of artificial intelligence (AI) becomes remarkably powerful. The harmonized, modular architecture and standardized data storage provided by the level 2 Process Guidance Platform act as an essential enabler for deploying AI-driven functional modules. By leveraging structured and consistent datasets alongside modern communication frameworks, AI can seamlessly analyze critical processes and optimize operations in real-time.

AI modules can, for example, analyze and optimize software input parameters. By continuously evaluating operating data and comparing it with software functions, AI can dynamically optimize critical parameters, thus ensuring optimal performance under varying production conditions. This is applied, for example, in the commissioning of metallurgical models to determine empirically determined parameters more quickly and accurately. This is very helpful because the models are highly complex, which makes conventional linear optimization approaches very time-consuming and does not always lead to satisfying results.

This approach enhances the precision and reliability of process models, leading to more efficient resource utilization and reduced deviations in production quality.

Such AI-driven tuning stabilizes critical functionalities and empowers companies to achieve higher consistency and control over production outcomes. With the robust infrastructure of the level 2 Process Guidance Platform, the application of AI has the potential to be a game-changer, driving productivity, improving the accuracy of process models, and ensuring competitiveness in a demanding industry.

Metallurgical models provide a mathematical description of processes and operations. In the context of a steel plant, these models describe, for example, the processes occurring in a basic oxygen furnace (BOF) or argon oxygen cecarburization (AOD). They are used to calculate an optimal and realistic process flow based on predefined process targets and specifications, such as target temperature, chemical composition, or input materials, ultimately aiming to meet these specifications as accurately as possible. The requirements for a model can therefore be distilled into two main characteristics. 

• First characteristic: the model should describe reality and thus the process as accurately as possible.
• Second characteristic: the model should be capable of optimally implementing external requirements – meaning high accuracy, minimal resource usage, in the shortest possible time, etc.

But how can the precision of the model be increased? One approach is to represent the underlying process more accurately by considering more influencing factors, as well as the interconnections and dependencies between sub processes within the model. The more information that is taken into account, the more complex the description of the entire process becomes. A direct and inevitable consequence is that the associated model description also becomes significantly more complex, which can be reflected, for example, in a greater number of parameters or nonlinear relationships. As the number of input parameters increases, the effort required to find the right set of parameters to meet the second requirement increases exponentially. Similarly, the number of special cases to be considered also rises, which can be counterproductive to the first requirement regarding the precision of the model. To effectively resolve this conflict, AI can be utilized. It can be used to optimize existing metallurgical models by enhancing their accuracy, adaptability, and efficiency. 

By integrating AI into metallurgical models, steel plants can achieve higher precision, lower costs, and greater operational efficiency, ultimately improving both process outcomes and product quality.

To effectively integrate AI into metallurgical models, several preparations are required across data, infrastructure, expertise, and processes. 

The efforts described above, which have already been initiated, yield two significant outcomes. First, SMS group is positioned to effectively respond to and adapt to the rapidly evolving landscape of Level 2 systems in steelmaking and casters, ensuring continued competitiveness. Second, these efforts enable the long-term enhancement of existing models, focusing on improved process accuracy, dynamic parameter tuning, and faster simulations. 

The potential for possible applications is huge. For example, AI can be used during commissioning. This way, the number of manual steps, like analyzing the last heat and adjusting control and model parameters, can be reduced and at the same time, the quality of the process model can be increased. This is possible because AI can reveal underlying patterns that are seemingly unclear to human operators.

The greatest potential lies in the areas of steelworks and casters since there is a lot of potential for optimization in terms of energy, material and environmental conditions. Moreover, SMS group’s experts can cover the whole production process and have deep knowledge and insights about challenges and opportunities in these areas.

Results and findings can also be applied, with adjustments, to other areas such as forging plants, since many of the issues are of a similar nature. Conclusively the choice of Process Guidance as a platform and the introduction of Digital Ready have proven to be a big success, as they significantly simplify the work. Finally it can stated that SMS group creates a robust foundation for integrating AI into metallurgical models – unlocking the potential for improved efficiency, reduced costs, and enhanced product quality.