Complex system analysis through discrete event simulation

Abstract

E-commerce is generally thought of as a world without walls. Although a computer monitor may replace a storefront window, the products that are purchased online have to be distributed from a brick and mortar warehouse. Amazon.com now makes it possible to instantly order and quickly receive everything from CDs and books to large toys and home furnishings. Amazon's success not only depends on their e-business capabilities, but also on their distribution and warehouse management systems that support them. Fulfillment center management therefore has become an important component of Amazon's unique set of system activities that serve as a corporate strategy. In an attempt to improve current and future warehouse management practices, the engineering group at Amazon has recently invested in Discrete Event Simulation technology. This thesis focuses on the creation of a discrete event simulation of the Fernley, NV semi-automated distribution center's outbound flow process. More specifically, the business subjects investigated include: picker variability, tote diversion, item-per-tote reduction, and conveyor merge logic. The model presented is Amazon's first attempt at simulating this environment and serves as an initial step towards a more detailed simulation of this facility. Preliminary findings from the simulation are presented and the report is concluded with a cultural evaluation of the present engineering directive. It should be noted that this published version of the thesis has been approved by Amazon.com and does not contain any proprietary data. A fully detailed version of this thesis was only submitted to Amazon.com and to the advisors listed below.