This paper discusses lot-sizing in time-phased order point systems under three different conditions: with no forecast errors present, with forecast errors present but no safety stocks, and finally, with forecast errors present but with safety stocks introduced to counter the effects of the demand uncertainty. Fourteen different single stage lot-sizing procedures have been observed during simulation experiments where each of these operating environments have been modeled. The existence of forecast errors radically affects the behavior of the lot-sizing procedures compared to situations without forecast errors. For example, forecast errors not only lead to stockouts, they also induce larger inventories. Introduction of safety stocks, in turn, generates even larger inventories and also more orders. These environments, therefore, are so different, as are the associated lot-sizing performances, that the relevance of previous research which has not considered stochastic environments must be questioned.
One must conclude that, despite the enormous interest in lot-sizing research over the years, the "scientific" body of knowledge, as it can be applied in practice, remains relatively undeveloped. Few guidelines with respect to lot-sizing can, therefore, be offered to the practitioner by the research community. More relevant, i.e., more realistic, lot-sizing research should alleviate such a situation.
Introduction
Numerous studies devoted to comparative lot-sizing research have been carried out in the last decade or so (see, for example,