Learn from the past, to conquer the future
Modern automation and control solutions are key for successful processes. European Steel Industry has put great efforts into continuous improvement, research and piloting of novel automation solutions for decades. We analyse the previous work and extract a unique roadmap for future automation research, assisting to reach the goals of the green deal and to provide a vision for a new automation era in steel processing...
Research Fund for Coal and Steel, Grant Agreement No. 899208
Impact Mapping of Research Projects
Project Objectives
- To promote the dissemination of the knowledge gained and the technological solutionsintroduced in relevant projects on the application of “Advanced Automation and Control Solutions in Downstream Steel Processes” techniques (Achieved)
- To identify present merits and limitations of the various technological solutions, as well as the spread of their implementation in the European steel plants (Achieved)
- To evaluate scale-up possibilities for the analysed technologies within the steel production and, even more important, to identify potential transfer to other branches (Achieved)
- To evaluate the principal reasons for success and failure of past projects, considering scientific, technical and economic aspects (Partially achieved)
- To evaluate the impact of the results on the technological level and the competitiveness of the European steel industry (Achieved)
- To identify most promising and most useful emerging development lines and to encourage the use of best results and innovative solutions, taking into account possible technological barriers (Achieved)
- To develop, distribute and evaluate a questionnaire regarding the demands of steel producers for future automation related developments (Only achieved in parts due to the COVID-19 crisis)
- To develop a sequence of research topics for the next years based on the above described roadmap, the results of the former RFCS projects and the results of the evaluation of the questionnaire (Achieved)
Our project team
The ControlInSteel consortium is composed of the Scuola Superiore di Sant´Anna (SSSA, Italy), the University of Madrid (UPM, Spain), Rina Consulting (Centro Sviluppo Material, CSM, Italy) and the Betriebsforschungsinstitut (BFI, Germany) - a group of dedicated research organisations with longstanding expertise in automation and control.
Our motivation
We dedicated our research to modern and advanced automation systems, striving to bring Industry 4.0 to real applications. With the ControlInSteel dissemination project, we have the chance to revisit challenges of the past and find out, where crucial impact can still be achieved...
Objectives
Analyse two decades of Research Fund for Coal and Steel (RFCS) projects and establish a semantic relationship between the problem space, the encountered barriers and issues, the used scientific methodology and, most importantly the achieved impacts...
Our Mission
Disseminating RFCS project results regarding Advanced Automation and Control in Downstream Steel Processing, the proposal aims to maximize the impact of 46 research projects. After a previous dissemination project covering automation in secondary-metallurgy, a dedicated project valorising advanced control techniques throughout the solid phase is urgently needed. Dissemination will present existing project results to broad audience, including workshops on European level performed together with subcontractor ESTEP. The project evaluates used techniques, achieved impacts and the potential of transfer in a scientific way. Based on this evaluation, directions for future research will be identified and summarized in a guidance document.
Our method
How to analyse research and science projects in a structured and logical way? We use so-called controlled vocabularies of the most important dimensions: a) the problem space, b) the barriers and issues space, c) the solution space and d) the impact space. The following figure depicts an example of the relationship between solution and impact space. With out approach, we can systematically relate project outcomes with success, problems and benefits.
The extracted semantic data is used to construct taxonomies and based on these taxonomies with establish ontologies for the relationships. The systems will then not only allow to analyse past results but essentially also help us to predict future demand and the potential solutions considering a specific challenging impact that should be generated. The project just started, but we will report here also about our work and our results.
System Identification (SI)
Advanced control requires knowledge on the considered dynamical system. Sometimes, a rigorous first-principle modelling – also called white box modelling – can be applied, e.g. describing the intrinsic physical processes by theoretical models known from literature. But often, systems are far too complex to be modelled by ab initio approaches. Then, statistical methods are applied to build the mathematical models based on measurements. Such models are referred to as black box models. Grey box models rely, at least in some part, on system knowledge which is only supported by a set of measurements. Throughout the various RFCS projects, several different approaches of system identification have been used. Additionally, it will be evaluated how model precision and execution speed affect advanced automation and control systems, especially, which types of models can be regarded as the most successful ones.
Iterative Learning Control (ILC)
Repetitive operations such as factory batch processes can sometimes be improved by considering the results of previous batches. In such situations, iterative learning control adapts its control output by learning from previous control outputs. This concept allows good control performance, especially when the model is uncertain or when there is little information about the system structure and its nonlinearities.
Machine Learning Control (MLC)
General machine learning control is a rising topic, driven by the hype for artificial intelligence. In machine learning control, models applied by the internal model controller or the model-predictive control are based on trained paradigms like neural networks, decision trees or Bayesian networks. Especially neural networks feature tremendous potential for application in control context, as they allow for structured approach to reach the so-called Koopman space. The latter can be used to find naturally linear projections of complex control problems.
Model Predictive Control (MPC)
Model Predictive Control (MPC) uses a system model to predict the future system behaviour during its control loop. The prediction requires a significant computational effort per time step. These high computing times are one reason why MPC has been mainly introduced for large numbers of slowly changing variables. It depends on the quality of the underlying model and on a proper system identification. Novel aspect of the technique was neither the solution of an optimization task nor the inclusion of restrictions, but the principle of a so-called moving horizon, in which the optimization task is solved with restrictions for each time step.
Implementation and Integration of Control Solutions (I2CS)
A specific question concerning advanced control technologies regards to bringing these, mathematically rather complex approaches, to a successful real-world application. The project will therefore also evaluate which strategies were pursued to ensure a real implementation success and where the difficulties arise in introducing new control approaches in the plants. Of course, industrial automation and control covers far more facets, and above methods are only a very small subset of examples. Further topics (e.g. internal model control, quaternion-based control) will of course be also part of the dissemination work.
Objectives of the ControlInSteel project
The primary objective of this dissemination project is to revisit the most important European projects related to “Advanced Automation and Control Solutions in Downstream Steel Processes” technologies in the field of steel production carried out in the last 10 years. Subobjective is to identify which approaches were successful technologies with respect to impact.
In RFCS projects, a special emphasize resides on the consideration of transferring the research results to other application scenarios. Further subobjective of this dissemination project will rigorously analyse, how control methods with large impact can be effectively transferred within the same plant and also to other plants.
All of them can benefit from the research results of RFCS funded projects in the field of “Advanced Automation and Control Solutions in Downstream Steel Processes”.
The information about the transferability is available in many project reports, as it has always been vital part of the project considerations. Nevertheless, although other dissemination activities have taken place, this specific information has not been evaluated in much detail and in such a scope before.
The Consortium
RINA Consulting - Centro Svilluppo Materiali S.p.A. (CSM)
CSM was established in 1963 as a research centre for the state iron and steel industry. CSM carries out research and development in close collaboration with Government, European and International supporting institutions, agencies and industries. Moreover, CSM collaborates with many national and international universities. CSM has a long-standing European tradition dating back to 1970, when Italy participated in the first projects of CECA programme. The traditional presence of CSM as the Italian reference pole in European research (RFCS and Framework Programmes) results in hundreds of international collaboration agreements and contracts. CSM is an active player (founding father and participant) in the main PPP (Private Public Partnership) of Horizon 2020 programme and in other relevant materials initiatives as A4M and EMIRI. CSM is working continuously to increase the presence worldwide and particularly in emerging areas (KET, KIC, EIP) and in other initiatives (FoF, E2B).
Scuola Superiore di Studi Universitari e di Perfezionamento Sant'Anna (SSSA)
SSSA is an autonomous, special statute public university operating in the field of applied sciences with the aims of promoting the development of culture, scientific and technological research, and innovation. SSSA will participate to the project through the centre of ICT for Complex Industrial Systems and Processes (ICT-COISP) of the Institute of Communication Information and Perception Technologies (TeCIP). ICT-COISP is very active in the research field in both process and manufacturing industries and holds an extensive expertise in development and software implementation of mathematical models of industrial processes and application of traditional and advanced (also Artificial Intelligence (AI)-based) techniques to complex industrial processes and machineries simulation, monitoring and control, including analysis and mining of data coming from the same applications. ICT-COISP also holds a deep expertise in the advanced control techniques. The competencies of ICT-COISP match perfectly most of the topics, tools and techniques which are relevant from the dissemination point of view.
Universidad Politécnica de Madrid (UPM)
UPM is the Technical University of Madrid is the largest Spanish technological university. With two recognitions as Campus of International Excellence, it is outstanding in its research activity together with its training of highly qualified professionals. Heading the Spanish University participations in the European Framework programmes with more than 280 projects, they are internationally renowned and competitive. It has a remarkable track of works in the fields relevant advanced industrial control, including energy reduction, predictive systems based on artificial intelligence, quality control and intelligent supervision systems. Prof. Dr. J. Ordieres-Meré (m) is full professor at the Department of Industrial Engineering of the UPM, author and co-author for more than 60 scientific journals in JCR journals, more than 20 books and more than 150 conference papers. Moreover, he is member of the EU Technical Expert Group TGS8 in the RFCS programme. Prof Ernestina Menasalvas (f) is associate professor at the Department of Computer Systems Languages and Software Engineering. She has published three international books and multiple international journals including Data and Knowledge Engineering Journal, Information Sciences, Expert Systems with applications, Journal of Medical Systems and International Journal of Intelligent Data Analysis.
VDEh-Betriebsforschungsinstitut GmbH (BFI, Coordinator)
BFI is a non-profit research organisation with a staff of around 100 people. Two main work areas are process optimisation and automation. Especially in advanced control, industrial automation and quality optimization, the institute has longstanding expertise with a rich series of RFCS projects. BFI here mostly took the role as a coordinator, becoming a mayor driver of innovation in steel industry in Europe for the last decades. Works of BFI will be conducted at the Department for Automation Downstream. Here, Dr. Andreas Wolff is leading senior expert on advanced control technology and has successfully conducted several of the cited projects. Moritz Loos is currently working on his Ph.D. in the field of control and optimization. Dr. Marcus J. Neuer has longstanding experience with algorithms and mathematical optimization approaches for ensuring process stability and efficiency.