Digitalization to drive quality: the synergy between ladle tracking system and prediction models

Abstract

ABSTRACT Tata Steel caters to the needs of its customers in two segments – long products and flat products. In the long products division, LD1 is the plant that manufactures steel and semi-finished products in the form of billets. At LD1, the Quality rejections are mainly due to the off chemistry in heats, which happens when different elements such as carbon, manganese, and silicon go beyond the specified chemitry range of the steel grade. A detailed analysis of the previously downgraded heats revealed that the major reason for the deviation was the erratic recovery of ferro alloys (FAs) added at the ladle furnace stage which in turn was due to the high variations in the steel weight in every ladle. Due to the variations in the tonnage of steel weight tapped and the actual heat weight, which were a crucial input to an exisitng static FA model, model the predictions of this model were inaccurate. In order to improve the accuracy of the prediction model, an Internet of things based ladle tracking system was implemented. This RFID-based system combined with the in-house developed real-time prediction model, for hot metal (liquid iron) + scrap to be charged in converters, resulted in improving the accuracy of steel weight being tapped and hence improving the FA predictions which eventually resulted in achieving the lowest ever quality rejections at LD Shop#1 in FY18. There have been multi-fold savings as a result of this: not only the loss due to rejections of heat was minimized but also the FA consumptions were reduced and due to tapping weight optimization, so there was improvement in the overall heat weight.