A simulation framework for forecasting uncertain lumpy demand

Abstract

Managing uncertainty when the demand is irregular and sporadic (i.e. “lumpy”) is a complex issue as well as an unexplored one. This study analyses the behaviour of forecasting techniques when dealing with lumpy demand at the master production scheduling level. In particular, three forecasting techniques (exponentially weighted moving average, early sales and order overplanning) are considered. It is argued that demand lumpiness is a multidimensional phenomenon, generated by different market characteristics, namely, numerousness of customers, heterogeneity of customers, frequency of order placing, variety of each customer's request and correlation between customers behaviour. Hence, the performances of the three forecasting methods are compared, through simulation experiments, with different values of the sources of lumpiness. A framework is then proposed that indicates the domain of applicability of each forecasting technique according to the prevailing source of lumpiness in the market.