Incorporating Order Crossover Information into Service-Oriented Base Stock Policy Decisions

Abstract

We investigate the impact of order crossover on a periodic, order-up-to (R,S) inventory system where attaining a target service level is the focus. Order crossover, when orders arrive in a sequence different from that in which they were originally placed, is increasingly likely to occur in modern supply chains. One reason for this is the increasing pressure to focus on service performance, which we refer to as the “Amazon Service Effect”. Many service-oriented changes to order placement and fulfillment strategies will increase the chance of order crossover. While the majority of inventory control studies, including crossover-focused studies, tend to be cost-oriented, in practice many inventory systems are driven by service goals. To explore the use of crossover information in inventory policy decision-making we adopt the “effective lead-time” (ELT) approach (Hayya et al., 2008) to representing order lead-times where order crossover is present. We develop a hybrid discrete-event/continuous simulation model that utilizes a unique structure to provide a detailed representation of a stochastic inventory system and collect precise information about order crossover, effective lead-time, service performance, and related operating characteristics. We first address two crossover related assumptions that have not been directly addressed in the literature. We find that order crossover clearly impacts service-oriented inventory system performance (much as it does cost-oriented systems) and that a protection period of R ELT time units is appropriate for an R,S inventory policy under crossover. We then explore five methods of including ELT information into the process of determining the order-up-to level (S) for a periodic, order-up-to (R,S) inventory policy. We perform a full-factorial simulation study under various conditions of target service level, demand variance, and lead-time variance, for a total of 375 scenarios. We find the method used to include effective lead-time information into the decision significantly impacts service performance. We compare and contrast the different methods, discuss issues such as data requirements and the sources of error, and address the research and managerial implications of our findings.