Planning and scheduling in assembly systems with parallel-machine workstations

Abstract

This dissertation focuses on production planning and scheduling problems in an assembly system operating on a make-to-order basis. Due dates are considered as constraints in the problems, that is, tardiness is not allowed. The assembly system considered in this study consists of several workstations each of which consists of one or more parallel machines or workers. The planning problem is to determine starting and completion periods of items, i.e., final products, subassemblies and parts, and their production quantities in each period under the constraints of due dates, precedence relationships among items, and resource limits. The scheduling problem is to determine schedules of items on parallel machines. An item can be released after a given ready time and must be completed before its due date. An item is split into multiple jobs (batches) and a job is processed on one of the parallel machines. Since the scheduling problem is a lower-level decision making than the planning problem, the scheduling problem is solved using a production plan obtained from the planning problem. The objective of the planning problem considered here is to minimize holding costs for final product inventory as well as work-in-process inventory. A mixed integer linear programming model is presented and a heuristic algorithm is developed using a solution property and a network model for defining solutions of the problem. The suggested algorithm is shown to outperformed a commercial planning/scheduling software and backward finite-loading methods that employ various priority rules. For the scheduling problem, two local search heuristics, simulated annealing and taboo search algorithms, are suggested for the objective of minimizing the holding costs of items including work-in-process as well as finished job inventories. Results of computational experiments showed that the simulated annealing algorithm outperformed the taboo search algorithm. Since aggregated information is used when t...