Kanban scheduling system

Abstract

Nowadays many manufacturing plants have adopted a demand-driven production control system also known as just-in-time production (JIT). This greatly reduces inventory costs because the buffers between the processes act as a blocking mechanism to indicate when production should stop and when eventually it can continue. One implementation of a JIT system is the Kanban system. Manufacturing in a JIT fashion, we face the challenge of flexibility to respond to changes in the customers' demands, while at the same time remaining cost-effective. This work attempts to introduce automation to the process of kanban scheduling in a manufacturing plant environment with a multi-level part type. The proposed solution is a Memetic Algorithm (MA) which tries to optimize schedules such that they meet the deadlines of customer orders without causing buffer overflows, while at the same time keeping setup time to a minimum and the placing free periods sensibly if these occur. The evaluation shows that this was achieved and the MA gave better results than the manual system currently in use.