This paper describes a production scheduling application for a petrochemical plant. The main techniques used are heuristic reasoning, simulation and object oriented (OO) models. Each product manufactured by the plant has its' own scheduling object, which contains a set of rules for producing a heuristic estimate of the need to manufacture the product in question. Simultaneous simulation with an OO model is used to verify that the schedule is good as well as to correct errors during the planning stage.