University of North Carolina Wilmington

Rebecca A Scott

University of North Carolina Wilmington

Many factors contribute to a successful implementation of the Japanese Just-In-Time (JIT) with kanban manufacturing philosophy. Much of the JIT research points out success with JIT is not automatic. Fairly stable distributions are necessary for both internal and external sources of production variation. A critical question for successful JIT implementation is how smooth timing between workstations will be achieved when highly stochastic distributions are present. This research looks at three sources of variation: product demand, machine processing times, and vendor supply. This research attempts to build a computer simulation model of a JIT system with probabilistic distributions for these three sources of uncertainty. However our model will have constant central tendencies across all distributions and the sole source of uncertainty will be a highly variable stochastic dispersion for all distributions. The simulation model can then be used to experiment on the system to find ways to improve it based on these conditions. A simulation model will be developed using the SLAM II discrete event modeling software package. The shop design chosen is an extension of the shop used by Rakes et al. (1994). The shop utilized in this study is expanded from the previous shops to allow for multiple workcenters, multiple products, and collection of shop statistics over two time periods; it consists of six workcenters with two final products. Data will be generated by the simulation model as opposed to collected from a process, database, or by other means. The data generation process is a multi-period design to look at how stochastic distributions for product demand, machine processing time, and vendor supply impact shop-control decisions in a dynamic production environment. A total of 560 different scenarios will be evaluated in terms of cost by varying the number of kanbans from one to 10. A total cost function will be developed based on three components: overtime cost, work-in-process inventory cost, and finished goods inventory cost. Parameters will be chosen for inventory costs and overtime costs so the optimal number of kanbans for each of the 560 scenarios will be in the range of one to eight. The analysis of results will look at optimal numbers of kanbans and cost sensitivity to changes in the number of kanbans.

Proceedings Home | AITP | EDSIG | ISCAP | EDSIGCON | CONISAR | ISEDJ | JISAR | JISE

© 2008 - 2014 EDSIG (Education Special Interest Group of AITP)

© 2015 - Present ISCAP (Information Systems & Computing Academic Professionals)