Periodic Reporting for period 1 - INEVITABLE (Optimization and performance improving in metal industry by digital technologies)
Reporting period: 2019-10-01 to 2021-03-31
The INEVITABLE project is targeting at resource and energy intensive sectors of the process industry, with focus on the steel and nonferrous metals sector that have an enormous impact on energy and resource consumption. Consequently, the major targeted results of the project are positive impacts on process and environment sustainability, while improving the environmental footprint and increasing energy and material efficiency. Production processes in steel production plants are similar to significant extent. This means that developed digitalization and decision support solutions can be adopted by many similar processes in various steel production plants with only limited modifications. Moreover, the newly developed data-based technologies have a big potential to be translated also to various other industry branches and fields.
The general objective of the INEVITABLE project is to improve the performance indicators of the targeted industries by retrofitting existing production sites using digital and innovative control and decision support technologies. The adopted approach will elevate the overall digitalization and cognition level of the production processes, while targeting the following goals:
• Improve energy efficiency and in this way reduce negative impacts to the environment,
• Improve product quality, reduce scrap and in this way improve energy efficiency,
• Improve process repeatability,
• Optimize production processes by providing accurate process models to test the effect of process parameters and setting prior the production process,
• Provide model based soft sensors for estimation of unmeasured process variables to better control the process,
• Provide methods for process diagnostics and condition monitoring, in order to detect process errors before they become critical and cause process interruption.
The mentioned goals are being achieved at several selected processes within the INEVITABLE use cases that cover big part of steel making chain. Use cases are used to demonstrate, test and validate the developed digital enabling technologies.
Project started by defining requirements and specifications for all use cases, as well as defining KPI measures to assess the impact of INEVITABLE solutions. Later the work was focused on digital retrofitting and basic digitisation upgrade of the INEVITABLE use-cases. The upgraded production processes are now ready for collecting production data needed for analysis, and for development and validation of developed solutions. The focus of the activities is now moving from the Data collection & sensor technologies, towards the preparation of specialised Tools for data analysis, control and optimization and preparation of the Digitalization platform. The whole spectrum of these basic enabling technologies is needed to reach the INEVITABLE objectives (Figure 1). Most of the ongoing activities are currently focused on the analysis of production data and the development and tuning of cognitive solutions. Several analytical tools are being developed to monitor process and equipment and to optimize production processes (Figure 2). These tools are based on soft sensors, predictive models and developed digital twins. In the frame of the digital platform preparation, design principles for the communication and data infrastructure are being developed. The general approach follows the standard structure, using data from different sources. Infrastructure concept of the Industrial IoT and cloud solution has been defined for one use-case and is in the final design for the remaining two use cases.
• Process optimization and monitoring by understanding the relation between process parameters and performance indicators (quality, efficiency, energy consumption, etc.),
• Decision support for setting the process parameters to achieve performance indicators,
• Estimation of unmeasured process variables (soft sensors),
• Diagnostics and condition monitoring,
• Prediction of product properties as a function of process parameters, input material, etc.
The mentioned functionalities are a subject of academic research for many years, but practical implementation does not follow the theoretical results from several reasons. Exemplarily, advanced algorithms are demanding in terms of computational power (optimization, identification, machine learning) and data storage capacity (big amounts of data to be stored and processed). By increasing the computational power of industrial controllers and by the possibility of computing on decentralized platforms (e.g. cloud platforms, edge processing units), one of the obstacles for implementation of advanced control systems is significantly reduced.