Abstract
For a part withdrawn from a supplier in a JIT system, its withdrawal cycle should be determined with consideration for a social cost. In this paper, we first model a supplier Kanban system with a stochastic demand and a variable withdrawal cycle. Next, we derive a stable condition of the system from an analysis of a backlogged demand and obtain a generating function of its limiting (equilibrium) distribution. These results lead to an algorithm for finding an optimal number of supplier Kanbans and an optimal withdrawal cycle that minimize a total expected cost including the social cost.
