May 2023 Volume 5

MATERIALS

Advancements in Metal Forming Technologies By Rahman Rostami

As both a student and researcher in Germany, I was fortunate to work alongside experts in the field of materials science and engineering. Over the past three years I have immersed myself in a dynamic academic environment at the Institute of Metal Forming, TU Bergakademie Freiberg which deals with research projects focusing on forming technologies and the interaction between forming technology and property development of formed materials; from forging, flat and wire rolling to numerical simulation of materials and processes. The institute works in close cooperation with German metal industry, automotive industry and European steel companies. Freiberg University is also the oldest University of Mining and Technology, and it continues to operate to this day. Several ongoing projects include studying the effect of deformation on the microstructure of metallic materials to better understand the complex relationships between material behavior and deformation processes, and to optimize material properties and performance. One of the latest research projects is carrying out about the microstructure of open-die forged parts. Open-die forging has emerged as an important processing technology for the production of large complex parts made of special alloys (high alloy steel, nickel/ superalloys and titanium alloys) for heavy industry, such as cranks, turbine shafts and heavy rings or tubes. In conventional open-die forging of steels, a billet is hot deformed at the temperature range of 800 to 1250 °C. Final part geometry is obtained after many steps of incremental deformation. After the last deformation, the forged part is heat treated by hardening and tempering to a required hardness in the state of completely tempered martensite. Despite the high strength, the workpiece exhibits poor ductility, which limits its ability for energy absorption before failure. Recent developments in heat treatment technology have allowed for conducting partial hardening by interrupted cooling with quasi isothermal holding, yielding duplex or complex microstructures, which offer advanced strength-failure-strain combinations. When the quasi-isothermal holding is executed in the bainite region, the resulting microstructure contains partly bainite and partly martensite. This combination gives ultra-high strength and a very good ductility while maintaining relatively low material costs. Moreover, a specific chemical composition of the steel can be selected for large open-die forgings that enables to form a favorable nano-bainitic microstructure by austempering (a heat-treatment technique to obtain bainite). The main objective of this project is to develop an online calculation system (a digital twin of material flow due to the deformation), which in a real time controls furnaces, a robot/manipulator, a forging press and a measurement system to form the nano-bainitic microstructure in a forged part directly from the forging heat.

In another research project – which ultimately led to the creation of a start-up; our team focused on developing an autonomous system for microstructure analysis of forged steels as well as rolled products. Through the use of Artificial Intelligence, we developed an application to be able to accurately and quickly characterize microstructures, obtaining repeatable and consistent results in mere seconds, in comparison to the time-consuming process of manual metallographic examination. This breakthrough allows steel manufacturers to significantly speed up the quality control process while concurrently enhancing production efficiency. Other key features of our web application included crack detection and grain size measurement, two important components of the metal forming industry. To ensure that our AI algorithm had sufficient data for self training, we compiled a dataset of over 2.5 million metallographic images taken via digital optical microscope. This extensive dataset encompassed a wide range of heat treated steels, including all possible phases and their combinations. As a co-founder in this start-up, I was responsible for laboratory management and creation of this comprehensive dataset, which was essential for training of the Convolutional Neural Network (CNN) utilized in this project. CNN is a class of neural networks in deep learning which is widely used for image analysis and computer vision purposes. We are currently writing an article that highlights the innovative application of AI for microstructure analysis to share our findings and insights with the broader scientific community, while also providing a valuable resource for forging industry professionals interested in AI application to enhance their own quality control processes. Moving to United States, I am excited to promote the skills and knowledge I gained through my previous work experience in Germany to further develop my expertise in the fields of materials engineering, materials characterization, research and development and project management in the metal industry. I look forward to leveraging these skills to contribute to the success of my future endeavors and to continue growing as a professional. Rahman Rostami Email: rahman.rostami@mivia.ai

FIA MAGAZINE | MAY 2023 48