Decreasing Changeover Time in a Sewing Process with Multi-dimensional Sequence Dependent Changeovers

What is it about?

Oxford Industries, Inc., a leading U.S. apparel manufacturer, implemented a short-term scheduling algorithm in their in-house embroidery facility which significantly reduced equipment changeovers. The company was experiencing difficulties meeting due dates in one of their more profitable new markets, branded custom embroidered golfwear. Too much time was spent on equipment changeovers leaving insufficient processing time to meet a surge in demand. We identified an opportunity for increased production capacity through more efficient scheduling of sequence dependent changeovers among garments. The type of changeovers encountered in this scheduling problem was uniquely multi-dimensional in nature. Each changeover did not depend only on the two garments involved in the direct serial changeover, but also on all other garments sewn in parallel on a common run of an embroidery machine as well. We designed an integer program (IP) to perform optimal garment to machine assignments to minimize machine changeovers and increase effective capacity. Problem size, however, prevented actual implementation of the IP on a daily basis. Instead, we designed an effective heuristic which offers good solutions for use in scheduling orders. During the historical period studied, the scheduling algorithm reduced the number of changeovers required on a daily basis by half.

Featured Image

Why is it important?

To my knowledge, the article introduces the concept of multi-dimensional sequence dependent changeover times and provides an integer programming model that minimizes such changeovers in the apparel industry.