2004 Volume 54 Issue 6 Pages 399-409
This paper deals with a sequencing problem of a mixed-model assembly line in the Toyota production system. It is important to level work loads on each station of the assembly line. The sequencing problems of leveling the work loads have been studied by many researchers and are known as NP-hard problems. Most researchers assume that the work overload on each station is always completed by utility workers and the assembly line does not stop. That is, the number of utility workers is sufficiently large and workers are helped by the utility workers at any time. Few researchers have studied models of the assembly line with line stoppages. In fact, line stoppages occur in the assembly line of the Toyota production system. Whenever a worker in the assembly line judges a possible failure to complete operations within the work zone, a push of an assembly line button summons an auxiliary worker, different from the utility worker, who will assist in the operation. A group, which is a formal organization of the workplace of the assembly line, consists of about 20 workers and is manned by only one auxiliary worker. Therefore, the worker calls the specific auxiliary worker. If the worker can complete operations with the auxiliary worker within the work zone, the line does not stop. Otherwise, the line stops at the downstream limit of the worker's station until they complete the operations. Moreover, when the auxiliary worker helps the other station, the worker can not call the auxiliary worker. That is, the worker can not be helped by the auxiliary worker and the line may stop. To our knowledge, no one has studied the sequencing problem of minimizing total line stoppage time of the mixed-model assembly line with the auxiliary workers. This paper investigates this sequencing problem. Since this problem is NP-hard and the size of practical problems at Toyota Motor Corporation is huge, we propose an approximation algorithm that adapts the goal chasing method, which was developed by the first author of this paper, and show its efficiency through numerical examples.
