Simulation-based performance measurement and analysis: an industrial application

Scheduling problems occur in a lot of application fields such as production, timetabling and computer systems. These problems have already been widely studied and modeled (Carlier and Chrétienne, 1988; Pinedo, 1995; Pinedo and Chao, 1999; Vernadat, 1999). Each problem has its own characteristics; it is much more complicated when tackling real industrial problems. Starting from object-oriented techniques we are developing a “generic” Framework dedicated to simulation and optimization of real discrete-continuous planning and scheduling problems. In this paper, a real problem, inspired by the Fontainunion plant (at FontaineLévêque, Belgium), is considered. The production consists in high quality steel wires and strands. The process is composed of continuous and discrete operations. Planning is done by an ERP (Enterprise Requirement Planning), so we focus our description on the scheduling part of the problem. Our objective is to simultaneously minimize late orders (both in terms of number of late orders and average lateness) and makespan (i.e. the maximum completion time of jobs).