Operational decomposition through statistical clustering of expert knowledge

The shift from a design to requirements towards a design for capability has expanded the scope of the requirements analysis to include the family of systems´ operations. With this expansion, a complete understanding and description of the operational picture is needed. The paper includes the discussion of two frameworks for operational decomposition along with their advantages and disadvantages. As a way of capturing the advantages of both of the frameworks, this paper presents a proposed methodology for combining a hierarchical treelike structured decomposition and a networked structured decomposition. Statistical clustering enables the transformation of the decomposition´s structure. A statistical clustering algorithm is used to find groupings of common operations. These common operations can be grouped to form a set of operational classes which effectively intertwine the branches of the tree into a network. This automated process mirrors the way in which operational groupings retroactively occur in organizations after commonalities or problems are exhibited, only now these core operational elements can be identified proactively as an element within the systems engineering process. Furthermore, information about how the operations support other operations is retained allowing for analysis and intelligent decision making at the operational level.