An iterative approach for system setup problems of flexible manufacturing systems.

One of several phases of decision issues in operating flexible manufacturing systems (FMSs) is system setup (also called FMS planning), for which the following subproblems must be solved: (1) selecting a subset of part types for immediate and simultaneous production; (2) partitioning machines into groups; (3) allocating operations and tools to machine groups; and (4) determining the numbers of pallets and fixtures and allocating them to the selected part types. In this research, we consider the first three subproblems, which are called the part type selection problem, the grouping problem, and the loading problem, respectively. Maximizing system throughput is used as the objective. Since it is very difficult to maximize throughput directly in the loading problem, we use balancing or unbalancing workloads of machines, which is known to maximize system throughput., as a surrogate objective in the loading problem. Since the system setup problem is computationally intractable, a sequential method has been suggested in which the solution of a subproblem is used as given conditions for the subsequent subproblems. However, the sequential approach may fail to provide a feasible solution for each of the subproblems. As an alternative, an iterative approach is suggested in this research. In the iterative approach, the subproblems are solved many times until a satisfactory solution for the entire system set up problem is obtained. Since even the individual subproblems cannot be solved optimally in a reasonable amount of time, heuristic algorithms are employed. In the suggested approach, for given priorities of orders, successively larger sets of orders, which are selected in the order of their priorities, are considered for immediate production. For each set, the grouping and loading problems are solved, and the set of orders that gives the highest throughput is selected. We present methods to generate all possible machine groupings, to find the ideal workloads of machine groups for each grouping, and to assign operations to the machine groups. We also suggest a method to determine priorities of orders. These methods are combined in an iterative procedure to solve the system setup problem. Computational experience is reported.PhDIndustrial engineeringOperations researchUniversity of Michiganhttp://deepblue.lib.umich.edu/bitstream/2027.42/162073/1/8907074.pd