A pin-style bottom anode employs conductive steel rods that serve as the pathway for the high electrical power through rammed refractory at the bottom of a DC electric arc furnace (EAF). Anode wear during operation is important to monitor, as anode replacement is expensive and impacts EAF productivity. Liquid steel penetration into the unsintered refractory layer can result from rapid electrical power ramp-up, dips in furnace temperature or operating the anode for too long between EAF campaigns. In extreme cases, the liquid steel may penetrate the bottom of the furnace when anode wear progresses too close to the bottom shell, which is extremely dangerous and must be avoided. The current state of the art for monitoring bottom anode wear employs thermocouples embedded in the anode pins at points in the anode. However, this approach is not sensitive enough to detect localized damage to the anode, especially when cracking occurs. The present work utilizes fiber-optic sensors to monitor the health of the anode by creating a real-time spatially distributed temperature map of the anode. Unlike traditional thermocouples, these sensors can be mounted at significantly greater depths, provide distributed temperature measurements, and can withstand temperatures of up to 900°C. Additionally, they are able to perform temper-ature measurements with a spatial resolution of 1.3 mm at a 5 Hz acquisition rate, providing unprecedented high-density real-time monitoring of anode health and increasing the efficiency, and safety of EAF operation.

This article presents a heuristic algorithm developed within a digital twin system to optimize scheduling in a complex steelmaking plant. The digital twin provides insights and anticipates bottlenecks, contributing to efficient scheduling. The algorithm addresses nonlinear constraints arising from technical, technological and financial limitations, ensuring a cus-tomized approach to production planning. The optimized scheduling aims to enhance steel production, minimize delays and maximize resource allocation within the supply chain. By leveraging mathematical optimization and digital twin tech-nology, this solution achieved significant gains in a real steel mill, increasing monthly production by 1% without the need for infrastructure investment.

For integration of artificial intelligence (AI) into critical decision-making processes, transparency is crucial. AI and other data-driven methods offer powerful alternatives to first principles modeling of complex metallurgical processes. However, they tend to be black boxes, leading to unreliable guidance especially in critical operational situations. This can be over-come by a transparency approach using explainable AI and considering meta information. This work demonstrates how advanced AI-based applications for sinter plants, blast furnaces and direct reduction plants can be seamlessly integrated into state-of-the-art decision support systems.

This study focuses on understanding the strip’s shape evolution during the hot rolling process. An off-line numerical sim-ulation tool was developed to predict the finishing mill setup for the entire roll campaign, including strip crown and flatness evolution. It explores the application of shape simulations to optimize the rolling campaign of a hot strip mill, specifically targeting work roll shifting strategies and initial roll profiles. The results highlight the importance of shifting strategies in reducing roll wear and improving strip flatness, with varying stroke and step strategies showing promising outcomes. Addi-tionally, optimizing initial work roll profiles led to significant improvements in strip crown and flatness, enhancing overall product quality and process stability. Industrial trials using the optimized initial work roll profiles validated the proposed approach, showing consistent process performance and improvements in quality parameters.

This project has developed a software system that enables steel manufacturers to rapidly develop and deploy site-specific virtual hazard recognition training. The system takes advantage of innovative digital technology to create an immersive, virtual reality training experience. The tool allows steel mill personnel to construct their own safety scenarios based on their specific environment, using their own 360° videos, images, text and video content. This provides a training experience that directly mirrors real-world conditions, enhancing the effectiveness of hazard recognition and response training. The team’s industry mentor facilitated multiple rounds of input and feedback with steel mill personnel to improve the system’s capabili-ties, usability and deployability, addressing practical issues along the way. This process has resulted in multiple iterations of the software being tested by industry personnel and led to technically robust software that is now being prepared for wider rollout within the steel industry.