In recent years, remarkable advances have been made in statistical analyses based on deep learning techniques. Applied studies of deep learning have been reported in various industrial fields, and those of the iron and steel industry are no exception. The production of iron and steel requires a variety of processes, such as processing of ingredients, iron making, casting, and rolling. As a result, the data acquired from them are diverse, and various tasks exist for which deep learning algorithms can assist. Hence, providing a summary of the application is helpful not only for researchers specializing in information science to grasp the current trend of applied studies on deep learning techniques, but also for researchers specializing in each field of the iron and steel making industry to understand what types of deep learning techniques are being utilized in other specialized fields. Therefore, in this paper, we summarize current studies on the application of deep learning in the iron and steel making field by organizing them into several categories of processes and analytical methodologies. Furthermore, based on the results, we discuss future perspectives on the development of deep learning techniques in this field.
Materials Integration is the concept of accelerating materials development by linking processing, structure, property, and performance (PSPP) on a computer using any types of models such as theoretical, empirical, numerical-simulation, and machine learning models. In the first and second phases of Cross-ministerial Strategic Innovation Promotion Program in Cabinet Office, Japan, we have developed a system called MInt (Materials Integration for Network Technology), which links PSPP with computational workflows that combine modules implemented, in order to realize the concept of Materials Integration. MInt is equipped with an application programming interface (API) that can be called from various algorithms in the artificial intelligence (AI) field and one can use MInt-API together with the AI algorithms to inversely design materials and processes from desired performance. The target material systems have expanded to steel, aluminum alloys, nickel alloys, and titanium alloys, and the target processes have also expanded to welding, heat treatment, 3D additive manufacturing, and powder metallurgy. MInt is more than just software for materials design; it is designed to serve as a digital platform for industry-academia collaboration. The Materials Integration Consortium has been established with MInt as its core technology, based on the philosophy of sharing tools such as modules and workflows, while competing on how to use them. In materials research and development, which has traditionally been regarded as a competitive area, we hope that a digital collaborative area will be formed and that investment efficiency will be drastically improved.
Most production systems are operated in a human-in-the-loop fashion, and it is sometimes argued that the human decisions involved make it possible, or at least easier, for the systems to cope with various stationary and nonstationary variations. However, it has not been well-studied and understood how this positive contribution of human decisions work, what factors determine its effect, how the function should be supported or fostered, etc. This paper first briefly reviews conventional production systems simulation techniques and discusses why it is difficult for them alone to address aforementioned questions. This next points to some recent attempts, in production systems engineering and related areas, to study human decisions and their effects by complementally using gaming simulation and agent-based simulation and highlights the potential of combining such behavioral and computational scientific approaches. Then, the paper introduces a cognitive framework model composed of interface, interaction, and incentive dimensions. It can be used for formally characterizing the decisions made by an individual facing a problem situation in operating a production system, and functions as a bond connecting behavioral and computational analyses of the decision maker. The paper further presents some example ongoing research projects worked on by the author’s team in this direction and discusses some future perspective.
In continuous casting, molten steel is fed from the tundish into the mold through the immersion nozzle. In the immersion nozzle, inclusions mainly composed of alumina present in the molten steel adhere and accumulate, it causes limitation of continuous castings. To prevent the nozzle clogging, Ar gas is blown into the immersion nozzle. However, Ar bubbles flow into the mold along with the molten steel and become trapped in the solidifying shell, causing bubbling defects of the slab. To suppress bubbling defects, it is effective to keep Ar bubbles away from the solidification interface or to use molten steel to wash away Ar bubbles that have adhered to the solidification interface. The molten steel flow in the mold is greatly affected by the shape of the immersion nozzle. In this paper, we consider the optimization of the shape of the immersion nozzle to reduce Ar bubbles trapped in the solidifying shell. A numerical model of molten steel flow and heat transfer solidification in the mold is combined with an optimization method. In the optimization process, Ar bubbles trapped in the solidifying shell are evaluated by a neural network to improve the calculation speed. The application of this method to the search for immersion nozzle shape is also reported, and the effectiveness of the obtained nozzle shape in reducing Ar bubbles is discussed.
One of the objectives for the development of high-strength dual-phase (DP) steel is improving the stretch-flangeability. Large-strained sheared edges are deformed and frequently cracked during stretch-flange formation. Considering shearing as the first deformation, the stretch-flange deformation may be regarded as a secondary deformation. To improve the stretch-flangeability of the DP steels, many researchers have analyzed the microvoid formation. However, in these analyses, the shearing process was not considered. With this background, ex-situ mini-bending tests combined with scanning electron microscopy (SEM) monitoring of microvoid formation were conducted during the secondary deformation. Prior to the secondary deformation, several microvoids were observed on the sheared surface and fine subgrains formed in the ferrite. During secondary deformation, the preliminary microvoids present at the ferrite-martensite interface propagated into the ferrite phase. In contrast, this behavior was not observed for the reamed surface deformation, which was formed without preliminary deformation. Furthermore, microvoids were initiated on ferrite grains that were not present at the ferrite-martensite interface, and martensite islands were not cracked during secondary deformation. This result is noteworthy because martensite cracking was the main factor involved in microvoid initiation, in the absence of shearing. Electron backscattering diffraction analysis revealed that the work hardening of ferrite, prior to the secondary deformation, caused a deviation in the strain concentration sites from those found in the reamed surface deformation. Therefore, this study elucidated microvoid formation on preliminary deformed surfaces via shearing and provided insights for material development considering deformations on the sheared surfaces of materials.
Martensite-matrix dual-phase (DP) steel is increasingly used for high-strength automobile parts owing to its excellent compatibility, ductility, and tensile strength. However, its higher fracture strain, reflected by the hole expansion ratio, remains an issue hindering further adoption of this material. Therefore, this study conducted a microscale investigation of the ductile fracture behavior of 1180-MPa class martensite-matrix DP steel to obtain a guideline for microstructural design realizing improved fracture strain. In this investigation, in-situ tensile testing was conducted simultaneously with scanning electron microscope observations and crystal plasticity finite-element analysis (CP-FEA). The in-situ tensile test results indicated that microcracks initiated at particular martensite packets and did not propagate into other packets; the CP-FEA results revealed that the martensite crystal orientation caused this behavior to induce remarkable stress and strain localization at interfaces in the vicinity of ferrite islands, relaxing the stress and strain localization at distant martensite packets. Although the cracks observed around the ferrite–martensite interfaces were similar to those observed in conventional ferrite-matrix DP steel, such matrix-phase cracks have rarely been reported except immediately prior to final fracture. Thus, the optimization of ferrite island distribution to suppress the formation of stress and strain localization sites was identified as the key aspect of martensite-matrix DP steel microstructure design. This design aspect can be achieved using a combination of data science and CP-FEA.
In this study, deformation behaviors at the grain level of coarse-grained ultralow carbon steel subjected to uniaxial tension and simple shear were simulated by using a crystal plasticity finite-element method. Heterogeneity of strain distributions appeared at the early stage and remained almost unchanged in the following deformation. Localized strain bands occurred at the grain level, but the directions of the bands depended on the deformation mode. These trends agreed well with experimental results reported in a previous paper [Hama et al., ISIJ Int., 61 (2021), 1971]. The mechanisms that the direction of the localized strain bands depended on the deformation mode were studied on the basis of the slip activities. The activities of slip systems roughly followed the Schmid factor, and the slip directions of the most active slip systems were consistent with the directions of localized strain bands, suggesting that the direction of localized strain bands were determined primarily by the Schmid factor.