Estimation of production preparation time by using data envelopment analysis

Abstract

Recently, in the most of small and medium-sized industries in Japan, they are facing with the problems that occur from small amounts of orders with too much variety. In Maruzen Industry, about 70% of products are orders that have never been produced. Therefore, it is very difficult to estimate required production time. These miss estimation lead to overtime working, deadline breaking and so on. So it is very important to estimate it as accurate as possible. We have proposed prediction of production preparation time for each set-up for each machine by using Data Envelopment Analysis (DEA). Once we grasp production preparation time, we can grasp daily production ability for each machine. Moreover, once we grasp product abilities of machines in the factory, we can estimate total amount of products, and we can improve the scheduling that we do not need to overtime work and/or dead line break. In the previous study, some data did not match with the preparation time that part managers have had. In this study, we are going to propose reasonable preprocess method to derive accurate production preparation time. © 2013 The Japan Society of Mechanical Engineers.