Industry 4.0 is revolutionizing the decision-making processes within the manufacturing industry. Among the technology portfolio that has made this revolution possible, the latest research has highlighted the potential of data analysis to improve the different stages of the production cycle, from resource provisioning to scheduling, delivery and warehousing.
The reuse of scrap in the foundry sector, in particular in large steelworks, is a critical aspect in the sustainable and efficient production of metal with the quality required by the industry.
The state of the art in optimization systems for melting raw materials, lies in the characterization and classification of the different families of materials (scrap, internal returns, etc.), based on information provided by the suppliers. In the best of cases, this information is verified with an approximate sample of material received, which is melted and analyzed in small furnaces.
The reliability of the results given by the optimization system depends directly on the representativeness of this information and this is where the problem arises: if the scrap has good quality, the final results match quite well with the expected ones; on the contrary, if the scrap is heterogeneous or its logistic control is more complex, there is a big risk of chemical deviations in the final metal. This would mean the rejection of an important part of the melted heats, with the consequent negative impact on costs and productivity.
As a consequence, low quality scrap, which is also very abundant and low cost, is underused in favor of virgin raw material (primary aluminum, pig iron…), which is much more expensive and difficult to produce in a sustainable way.
To try to solve this difficulty there are systems that offer 3.0 solutions with Operational Research algorithms that, together, collect almost all the variables that influence the melting process: complete traceability of materials, environmental variables, preparation of furnace charges, additives, melting parameters (times, temperatures, electrical variables…), by-products, final metal, chemical composition or yields.
The next natural step is to apply all this knowledge, and add Big Data techniques, predictive Machine Learning algorithms and anomaly detection algorithms. This allows us to know as accurately as possible the behavior of the different scraps according to operating conditions and provide real time information to the decision-making systems. This means a turning point in efficiency and quality parameters for metal manufacturing.
Strategic products such as steel can reduce their manufacturing costs in the range of 2% – 3%, improving their sustainability by increasing the use of low-quality scrap by more than 10%. In a steel plant with a production of 1 million tons / year, it would mean a cost reduction of more than 10 million euros per year.
This technology is also applicable to iron foundries (grey / ductile) in such key sectors as the automotive and wind power industries.
We are already working with those results