Allocation of buffer spaces in a flexible manufacturing system: a model and a case example

A flexible manufacturing system (FMS) comprises many stations such as a load/unload station, a set of workstations, and a common buffer, that are linked together with a material handling system (MHS). Each workstation consists of a limited input buffer, several identical machines and a limited output buffer. The MHS includes several unit-load transfer tools (UTTs) moving parts in the system according to the process routings required by the parts. The function of the common buffer is to temporarily store blocked parts that are treated following an FMS blocking mechanism. We model the FMS by an open queueing network with the FMS blocking mechanism and a block dependent static Markov (BDSM) routing policy. A multiple objective optimization problem is discussed for allocating spaces to the input and output buffers so as to simultaneously maximize the throughput of the workstations, and minimize the mean work-in-process (WIP) in the system. Two heuristic algorithms, based on marginal and balancing mechanisms respectively, are proposed for solving the optimization problem. The efficiency of the algorithms is evaluated through numerical experiments. A case example is provided for solving an FMS design problem in a reconfigurable transforming planning by applying the algorithms.