Skip to main content
eScholarship
Open Access Publications from the University of California

Process Flexibility Design in Unbalanced and Asymmetric Networks

  • Author(s): Deng, Tianhu
  • Advisor(s): Shen, Zuojun
  • et al.
Abstract

In a multi-product multi-plant manufacturing system, process flexibility is both the abil- ity of each plant to produce a variety of products and the ability to produce products at multiple plants. A balanced network has the same number of plants and products. In a symmetric network, all the plants have the same capacity and all the products have the same demand. Several design guidelines and flexibility indices have been developed in the literature to inform the design of flexible production networks. Yet little insights have been gained in unbalanced networks and asymmetric networks. In this thesis, we aim to provide new insights and results of unbalanced and asymmetric networks.

First, we propose additional flexibility design guidelines by refining the well-known Chain- ing Guidelines (Jordan and Graves, 1995). We study unbalanced and symmetric networks where each product is built at two plants. We also briefly discuss cases where (1) each product is built at three plants and (2) some products are built at only one plant. An ex- tensive computational study suggests that our refinements work very well for finding flexible configurations with minimum shortfall.

Then, instead of assuming a joint demand distribution for all products, we optimize the worst case over all the demand joint distributions that have the given marginal mean and variance. We present two results. First, we show that most known results for the effects of plant capacity and product demand in symmetric networks can be extended to asymmetric networks. Second, the analysis sheds light on how to incorporate demand standard deviation into the existing flexibility design guidelines and indices. We show that the proposed method significantly enhances the performance of the Node Expansion Guideline.

Main Content
Current View