Identification of Clusters and Interfaces for Supporting the Implementation of Change Requests

Abstract

Due to the presence of dependence linkages, changing one element of a product (e.g., functions and components) can trigger changes to other related elements and lead to numerous possible propagation paths (i.e., the “snowball effect”). To address this issue, this paper proposes the matrix-based clustering method. Two matrix models are considered in this research: design structure matrix for the linkages of components and domain mapping matrix for the linkages of functions and parameters. After denoting some product elements as “target” representing the initial changes, the clustering method is used to form and classify the clusters according to the change impacts from target elements. The interfaces between the clusters are also identified to manage the propagation process. The purpose of this method is to provide the cluster and interface information for implementing change requests. In view of the methodical advancement, the clustering method can tailor a clustered matrix for specific change requests and handle two types of matrices. Two examples have been used to demonstrate and support the utility and flexibility of the proposed method to manage matrix-based change propagation.