Abstract
POSCO $\#2$ Stainless steel making plant produces more than 600 thousand ton per year with a variety of products consisting of austenite and ferrite stainless steel to meet custrmers' needs since 1996. The plant has four different major processes, that are, EAF-AOD-VOD-CC to finally produce semi-product called as slab. In this study, we importantly took AOD process into consideration due to its roles such as to check and verify the final qualities through sampling inspection. But the lead-time from sampling to its verification takes five to ten minutes causing produrtivity loss as muck as the lead-time as a result. Of all indices for quality and process control the plant has, carbon ingredient in liquid type of steel is the most important since it affects in a great way to the characteristics of steel, if any problem. customers not satisfied with quality could issue a claim; therefore there is no way hut to guarantee it before delivery. in this study, to reasonably reduce lead-time ran save a cycle time and finally improve our productivity from a state-or-art alternative just such as applying statistical model based on multi-regression analysis into the A.O.D line by analyzing the statistical and technical relationship between carbon and the relevant some vital independent variables. In consequence, the model with R-square $87\%$ allowed the plant to predict, abbreviating the process in relations to sampling to verification. approximately the value of [C] so that operators could run the process line with reliability on data automatically calculated instead of actual inspection. In the future, we are going to do the best to share this type of methodology with other processes, if possible, to apply into them.