AI computation time cut from up to 28 seconds to about 2 seconds

Automated AI control improves quality stability and operational convenience

Company expects annual LNG cost savings of around 700 million won

A view of the No. 3 CGL annealing furnace at KG Steel's Dangjin factory [KG Steel]
A view of the No. 3 CGL annealing furnace at KG Steel's Dangjin factory [KG Steel]

KG Steel announced Wednesday that it plans to upgrade its AI-based predictive control technology at the annealing furnace of its Dangjin factory, raising the AI usage rate from roughly 60 percent to more than 80 percent to improve quality stability and energy efficiency.

KG Steel partnered with Inergy, an industrial AI solutions company, and in 2023 became the first in South Korea to develop AI predictive control technology for the No. 3 CGL annealing furnace at the Dangjin factory. The company has since continuously refined the technology by incorporating on-site operational data and workers' process expertise.

Since introducing the AI predictive control technology, the company has reduced LNG consumption by approximately 400,000 normal cubic meters over the past year, cutting costs by about 400 million won ($298,000). It expects annual LNG cost savings of around 700 million won once the AI usage rate exceeds 80 percent.

An annealing furnace heats steel sheets by burning fossil fuels such as LNG. Precisely controlling the temperature required for each product and reducing unnecessary heating directly lowers energy consumption and carbon emissions.

KG Steel's AI predictive control technology analyzes current operating conditions and the characteristics of the next product to be produced, calculating in advance the target steel sheet temperature and optimal control values. This allows operating conditions to be adjusted ahead of product transitions, reducing the need for manual intervention by workers.

In the latest upgrade, the predictive model was enhanced by incorporating a product's thickness, width, steel grade, production speed and expected quality characteristics to improve accuracy. Computation time was cut from a maximum of 28 seconds to about 2 seconds, enabling rapid response in actual operating environments.

KG Steel also incorporated shop-floor workers' operational experience into the AI automatic control system, designing it to adjust temperatures in stages during product transitions and minimize unnecessary temperature fluctuations.

A comparison of AI-operated and worker-operated performance for major coated products — hot-dip galvanized steel (GI), galvannealed steel (GA) and aluminum-zinc alloy coated steel (GL) — showed that LNG consumption per ton of output was lower under AI operation by about 9.9 percent for GI, about 2.9 percent for GA and about 4.9 percent for GL.

KG Steel plans to raise the annealing furnace AI usage rate from roughly 60 percent to more than 80 percent and to extend the technology's application to similar processes where AI can be deployed.

"We are combining AI technology with the operational experience and process expertise of our shop-floor workers to improve quality stability and energy efficiency at the same time," a KG Steel official said. "We will continue to expand AI utilization and the range of processes it covers to strengthen our production competitiveness, while steadily reducing LNG consumption and carbon emissions."

Meanwhile, KG Steel recently obtained the national standard certification KS D 3034 for painted hot-dip zinc-aluminum-magnesium alloy coated steel sheets and strips produced at its Dangjin and Incheon factories. The certification covers not only the ternary alloy-coated base material but also finished color steel sheets coated with it, establishing a quality management system in line with national standards.


likehyo85@heraldcorp.com