The market requires successive shipments of iron ore to be uniform in composition, for several elements (particularly iron, silicon, aluminium and calcium). Variations in mine production cause individual train loads arriving at the shipping yard to vary in composition. Appropriate stacking of incoming ore helps smooth the variability. This can be done by simple random filtering through an array of stockpiles. Intelligent stacking, guided by composition of the ore and of candidate stockpiles, gives better smoothing. The objective function is a composite of the multiple element criteria and the benefits of improved composition uniformity trade against costs of handling and storage. The problem is not soluble analytically. This paper describes the design and application of a simulation model used to plan iron ore handling procedures and improve quality performance. The model, written in Extend(TM) , has a dynamic graphical interface which has proved helpful both for developing the model and for constructive interaction with managers and operating staff.