Simultaneous lot sizing and scheduling : a heuristic algorithm for dynamic, discrete demands on a single facility

Abstract

One of major decisions which firms must face is to determine when and how much to produce in what sequence. The existing practice is to go through stages of lot sizing, capacity planning and scheduling sequentially to arrive at a production schedule. This sequential type of decision making is heavily dependent upon past experience, judgment and trial and error methods. Using a procedure based on what has been described in this study, the lot sizing, capacity planning and scheduling problems can be solved in one step. The problem considered in this study is arising from a production system in which a group of different products are produced on a single machine which requires major setups between processing different products. The problem has been further restricted by imposing a constraint which all the requirements be met no later than their due dates within a given level of production capacity. The research described in this dissertation has developed a simultaneous approach for making decisions on lot sizing and sequencing for the problem. In order to attack the problem, there have been made two different kinds of effort. The first effort made was to reduce the computational burden required to find a solution by systematically curtailing the search-state space on which the problem solution is searched. The second was to design an efficient schedule generation procedure which searches a solution in the predefined search space. The simultaneous procedure developed in this study, which incorporates the capacity limitation, due dates, and non-zero setup times, shows potential for an alternative approach to manufacturing logistics systems to better realize the Just-In-Time production philosophy.