Matrix-based hierarchical clustering for developing product architecture

Abstract

Product architecture can influence different aspects of product lifecycle including manufacturing, assembly, and supply chain. The purpose of this article is to employ hierarchical cluster analysis for developing product architecture to support product variety. Design structure matrix is used to visualize and analyze product architecture in view of product modules, overlapping modules, and bus components. The proposed method for design structure matrix clustering consists of three phases. The first phase is component filtering to identify components that should be classified as bus components. The second phase is approximate structure formation that preliminarily organizes similar components to form a diagonal matrix. The third phase is partitioning analysis that finalizes the modules’ boundary to yield the structured matrix as the solution of design structure matrix clustering. To examine the solution’s quality, minimum description length from literature is used. Then, the proposed method is demonstrated via two literature examples and compared with the solutions by the manual and genetic algorithm approaches. One unique advantage of the proposed method is that the user can obtain and inspect the approximate structure in view of the diagonal matrix before finalizing the structured solution (e.g. estimate the number of modules).