Functional Decomposition of the Clustering Approach for Matrix-Based Structuring

In engineering design, matrices have been commonly used to capture dependency relationships for structure-related problems (e.g., product architecture, process workflow, and team organization). In this context, structuring is considered a group formation process that clusters the design entities and identifies the interactions among the formed groups. To support matrix-based design structuring, this paper proposes a clustering approach that has three phases in the working procedure. Firstly, the coupling analysis is used to assess the coupling strength of any two entities according to the application context. Secondly, the sorting analysis is used to organize the matrix’s rows and columns by bringing the highly coupled entities close to each other, thus yielding a sorted matrix. Thirdly, the partitioning analysis is applied to form a structured matrix that identifies the groups of entities and their interactions based on some structural criteria (e.g., number of groups, limits on group sizes, etc). The proposed clustering method has been applied to four engineering examples to demonstrate its flexibility and adaptability in tackling different design structuring problems.