Design of a multi-product Kanban system based on bottleneck analysis

Kanban system is a novel philosophy, which plays a significant role in many production systems. The design of a Kanban system addresses the selection of two important parameters, i.e. the number of Kanbans and the lot size. In this paper, a simulation model of a single-card Kanban system capable of handling different types of products with different demand requirements has been developed and used for the performance analysis. The number of Kanbans and lot size can be different for different product types at each work station. A bi-criteria objective function comprising weighted tardiness and weighted work-in-process has been used for evaluation. The bottleneck analysis has been first used as a sensitivity analysis of the parameters in this field, which is very useful to guide the design procedure of the Kanban system to achieve the best performance. A modified simulated annealing algorithm is also given in which the neighborhood-generation is according to a probability based on the bottleneck analysis. It can find same or better solutions with obviously fewer iterations compared to a classical simulated annealing algorithm. Several case studies in different sizes have been tried out and showed relatively positive results.