Pohang Steelworks Establishes 'Process Automation' System for Molten Steel Pre-Treatment

Advanced Management of Steelmaking Process Using AI and Video Recognition Technology

POSCO Pohang Steelworks' No. 3 Steelmaking Plant has succeeded in automating the molten iron pretreatment process.

The automated process that was successfully implemented this time fully automates the removal of impurities from molten iron in the blast furnace using AI and video recognition technology, bringing Pohang Steelworks one step closer to completing its smart factory.

The steelmaking process, which adjusts the composition of molten iron produced in the blast furnace, is one of the key processes that determine the quality of steel products. The pretreatment process, which is the beginning of the steelmaking process, primarily removes slag, an impurity in the molten iron from the blast furnace, and adjusts the sulfur content to make the steel easier to break.

The preliminary treatment facility (skimmer) is automatically removing slag. [Image source= Pohang Steelworks]

The core of the pretreatment process is to remove the slag that floats like foam on top of the molten iron. If the slag is not properly removed, impurities sink into the molten iron, ultimately leading to quality degradation.

To remove the slag, a device called a ‘skimmer,’ which looks like a giant shovel, is used. On-site workers check the molten iron through monitor screens and operate the equipment directly to scrape off the slag.

Since this work relied solely on the workers’ eyes, hands, and senses, there was inevitably variation between operators. However, the newly developed pretreatment automation system has completely resolved this issue.

The pretreatment automation system is a system in which artificial intelligence learns the operator’s sense of scraping off slag and automatically operates the equipment. Pohang Steelworks’ Steelmaking Department jointly developed this system with the EIC Technology Department and POSCO DX.

Using a video recognition system, the AI directly analyzes the state of the molten iron, identifies the amount and location of slag, then learns the working methods of on-site operators to devise the optimal path to remove the slag up to the target amount.

Unlike in the past, when operators had to hold a stick in front of a monitor and control the equipment, now the entire pretreatment process?from lime injection to slag removal?can be completed with the push of a single button.

It is expected that the data obtained through this automation process will enable the advancement of steelmaking process management.

Additionally, by using thermal cameras and video recognition systems to accurately measure the amount and location of slag, the loss rate of molten iron during impurity removal has been reduced, and the amount of impurity removal can be precisely adjusted according to the finished products produced from the molten iron.

Lee Seung-heon, head of the Steelmaking Department at Pohang Steelworks, said, “With AI that has learned the intuition and know-how of skilled operators being deployed on-site, we were able to reduce variation between operators and quantify data such as slag removal rates in real time. We will continue to improve the model according to field conditions and operator needs and conduct continuous monitoring to ensure AI technology is more usefully applied in the field.”

This content was produced with the assistance of AI translation services.