Speaker
Description
In the modern steel industry, using large process datasets to improve quality control and production efficiency is an essential engineering task. This study focuses on predicting the main mechanical properties—yield strength (YS), ultimate tensile strength (UTS), and elongation—of hot-rolled steel coils for automotive parts. For these components, consistent material quality is critical for manufacturing stability. To achieve accurate results, we developed a data-linkage model that integrates variables from the entire production line, from the initial chemical composition to the final cooling process in the yard.
A key part of this research is the estimation of the cooling history during yard storage. Since the steel is stored in an as-coiled state, the cooling rate differs significantly depending on the radial and longitudinal positions within the coil. To reflect this, we estimated specific thermal profiles for the inner and outer laps, considering the heat transfer characteristics of the as-coiled configuration. This thermal data was then linked to the model through precise data mapping, which connects parameters from the steelmaking and rolling stages to their exact locations within the finished coil.
The model uses over 30 process parameters, including alloy elements, cooling rates on the run-out table (ROT), and coil dimensions. By matching these variables to their specific positions along the strip, the model can estimate material properties across the entire length and width of the coil. This "full-body" approach helps engineers evaluate internal property deviations that are difficult to measure through standard destructive tests. The results show that this method, combining localized cooling history with systematic data mapping, is a practical tool for managing material consistency and optimizing hot rolling parameters in real production environments.