Identifying Architectural Modularity in the Smart Grid: an Application of Design Structure Matrix Methodology

This paper demonstrates a vetted methodology for identifying areas of architectural modularity using two detailed architecture references: NISTIR Logical Reference Model [1] – a work product of the NIST smart grid standards effort that establishes actors and interfaces in the smart grid NRECA’s Demonstration Architecture [2] – a planned architecture for a federally funded smart grid demonstration project. The design structure matrix methodology [3] is applied with the intent to demonstrate how this approach can apply to defining smart grid architectures and to help identify architectural groupings that can lead to better modularization of smart grid systems and standards efforts. This paper is intended to inform current and future smart grid architecture efforts and to help improve the organization by which smart grid systems and standards can be established. The paper concludes that initial smart grid architectural efforts (as documented in [1] and [2], actual architectures may have other constraints such as backward compatibility) can be improved upon by identifying areas of modularity and organizing around them. The tools demonstrated can be best applied when the full dimensionality and scope of the problem is made explicit, but the demonstrations in this paper that use only publically available information also yield interesting findings. 1. WHY MODULARITY? One significant interoperability challenge today is to integrate with legacy systems while driving toward elegant solutions for future integrations. This challenge is compounded by a phenomenon called “accidental architecture.” [4] An accidental architecture is the de facto structure of a system resulting from numerous point-to-point integrations between various applications to achieve near-term objectives. Point-to-point integrations are not scalable and often create unintended ripple effects on downstream applications. The result of this haphazard evolution is a unique and customized system that becomes increasingly difficult to maintain, update and integrate with. Minimizing the accidental architecture phenomenon requires both backward-looking and forward-looking efforts – how do we integrate with the existing architectures of today while ensuring robust architectures tomorrow. Part of the mechanism that leads to accidental architectures is the highly integrated nature of these grid communication systems. The characteristics of each dependency and interface of the system are so nuanced that custom approaches are required. The original architectural principals of these systems can be difficult and costly to maintain, and therefore they are not maintained. Modular systems [5], on the other hand, are less tightly integrated and tend to integrate easily, evolve flexibly, and operate simply and reliably [6]. Modularity can often be difficult to achieve in large complex systems. This paper demonstrates a vetted approach to identifying modularity in complex systems that can lead to improved system structures and ultimately reduce the impetus for point-topoint integrations that lead to accidental architectures.