Data-driven iron and steel inventory control policies

Abstract

In this study, we investigated the optimal material inventory policy with regard to the iron and steel industry's effort to reduce massive overstocking issues in the face of increased corporate competitiveness. We gathered actual data, including sales and inventory numbers, from a steel and iron company over a period of 216 weeks between January 2010 and February 2014. We then utilized the Markov decision process (MDP) to analyze this data for inventory problems, such as relevant reorder points and reorder quantity issues as they relate to lead time, stock on hand and the limitations of having stock in-transit. The purpose of the study was to determine the most effective method for minimizing costs by using the optimal inventory policy to calculate and verify the effectiveness of the results. The final 52 weeks of data were put aside, while the initial 164 weeks were used to create an inbound material receipt system to ultimately establish a yearly (52-week) policy based on the inventory and sales data for weeks 113-164. Finally, we verified the effectiveness of the policy using the data from the final 52 weeks. The results showed that our proposed categorization method was effective for reducing the quantity of inventory while still meeting quarterly demands.