Design customization concerns the re-computation of the models of existing products in order to address the pressing customer needs and the dynamic economical environment. Due to the intrinsic couplings in the models, even minor modifications to only some parts of a model may be propagated to other parts indefinitely, causing the recomputation process to be longer than necessary. In light of this challenge, the purpose of this dissertation is to develop a redesign methodology to achieve rapid design customization.

In the methodology development, the design dependency matrix (DDM) is used to represent a redesign problem. Based on this type of matrix, decomposition is purposely applied to decompose a redesign problem to yield redesign pattern solutions. These pattern solutions can be utilized to diagnose an existing design by identifying which portion(s) of the design model should be re-computed, and the relevant interaction to control redesign propagation. Following this, an efficient redesign plan is generated to effectively manage and schedule the re-computation required to solve the redesign problem. This dissertation focuses on this planning process, namely, Pattern-based Redesign Planning (PRP).

The workflow of the PRP has three stages. The first stage (namely, Optimal Pattern Selection) is intended to estimate the potential redesign effort entailed by each redesign pattern solution such that an optimal pattern solution can be selected for redesign planning. The second stage (namely, Re-computation Strategy Selection) is intended to utilize the optimal pattern’s information in order to select an appropriate recomputation strategy to handle the decomposed redesign problem at hand. In this context, a library of re-computation strategies has been built to systematically characterize the solution roadmaps to solve different redesign problems. Finally, based on the selected strategy, the third stage (namely, Redesign Plan Generation) is invoked to synthesize a redesign plan and model it using Petri net.

For illustration and validation, the methodology has been applied to two problems:​ powertrain system redesign and relief valve system redesign. The results demonstrate the applicability and validity of the proposed methodology.