Configuration of Multiproduct Kanban Systems Using a Distributed Evolutionary Algorithm

Abstract

Performances of multiproduct Kanban systems are strongly dependent on a set of parameters, such as the number of kanbans between machines, the transport lot sizes, the safety storage sizes and the sequencing rules. Configuring such systems consist of determining a value for each parameter in order to optimise a production performance criterion. Recent works have shown that simulation optimisation of Kanban systems can be efficientlY addressed using evolutionary algorithms. We propose a distributed evolutionary optimisation approach, which can manage several searches for solutions simultaneously. This new approach speeds up the search of the optimal value and enlarges the explored search space, giving therefore a better chance to find the optimal solution. The benefits of this approach are illustrated through the example of a multiproduct Kanban system.