In this paper, a discrete-event simulation model is
coupled with a genetic algorithm to treat highly combinatorial
scheduling problems encountered in a production campaign of a fine
chemistry plant. The main constraints and features of fine chemistry
have been taken into account in the development of the model, thus
allowing a realistic evaluation of the objective function used in the
stochastic optimization procedure. After a presentation of problem
combinatorics, the coupling strategy is then proposed and illustrated by
an example of industrial size (24 equipment items, 140 products, 12
different production recipes and 40 products to be recycled during the
campaign). This example serves as an incentive to show how the approach
can improve production performance. Three technical criteria have been
studied: campaign completion time, average product cycle time, respect
of due-dates. Two kinds of optimization variables have been considered:
product input order and/or allocation of heuristics for conflit
treatment. The results obtained are then analysed and some perspectives
of this work are presented.