The impact of aggregation level on lumpy demand management

Spare parts management is a rather complex issue. One of the reasons of its complexity is the lumpy pattern of the demand that spare parts frequently present. Several methods have been proposed to cope with this particular kind of problem and improvements have been proved compared to classical forecasting techniques. Literature has however devoted minor attention to the choice of aggregation level when demand is lumpy. This paper aims at studying whether aggregating data when demand is lumpy may be beneficial in terms of impact on inventory performances. An installation stock inventory model is considered and aggregation over time is taken into account; in particular for a single item different time buckets are considered and performances are evaluated in terms of service and inventory level. Based on simulation experiments on real demand data coming from the spare parts unit of a multinational white goods manufacturer, we identify that aggregation of data can significantly impact on inventory management performances. A contingency analysis based on demand characteristics allows us to draw some guidelines on when aggregation over time can be beneficial.