A distributed evolutionary simulation optimization approach for the configuration of multiproduct kanban systems

The performance of multiproduct kanban systems is strongly dependent on a s et of numerical and non-numerical parameters, such as the number of kanbans between machines, the transport lot sizes, the safety storage sizes and th e sequencing rules. Configuring such systems consists of determining a valu e for each parameter in order to optimize a performance criterion. The auth ors propose a distributed simulation optimization method based on evolution ary principles, which is capable of simultaneously taking into account vari ous parameters involved in the configuration of multiproduct kanban systems . In order to improve the quality of the results and to reduce computing ti me, concurrent searches are managed through a network of workstations. This approach is illustrated by the configuration of a kanban system comprising four machines in line, which process three types of product.