Analysis and design of flexible systems to manage demand uncertainty and supply disruptions

Abstract

The "chaining strategy" has been extremely influential in recent decades, with many large automakers and service industries already making this the cornerstone of their business strategies. The purpose of this thesis is to provide analytical justification for why partially flexible strategies, like chaining, already perform nearly as well as full flexibility at only a small fraction of the cost. We develop a random walk approach to capture the exact performance of chaining under general demand distributions. In addition, we extend this theory to environments that account for factors such as increasing system size, the response dimension, partial production postponement, and supply disruptions. Under these scenarios, the performance of the celebrated b 2-chainb may deteriorate but the benefits are still non-negligible even in the worst case. That said, we contend that a third layer of flexibility is sufficient to recover most of the losses due to these additional factors.