This article discusses a practical lot-sizing and scheduling problem in a foundry, involving two interrelated decision levels: furnace scheduling - in which a type of alloy is selected for production in each period, and molding machine planning - which specifies the quantity of each item to be manufactured in each period, i.e., lot-sizing. A mixed integer programming (MIP) model is proposed for this problem and a solution method is developed based on local search. The production is planned using the rolling horizon technique, whereby only immediate-period schedules are implemented. The results of the computational tests are analyzed and compared with those obtained using a commercial package.
mixed integer programming; lot-sizing and scheduling; foundries
