THE EFFECT OF SYSTEM DECOMPOSITION ON PRODUCT PERFORMANCE AND PROCESS CONVERGENCE TIME

Abstract

The theory of complex systems, which has been applied successfully in evolutionary biology, is gaining popularity for the modeling and analysis of complex product development (PD) systems. Modeling complex PD systems is essential to understand how system elements and their dependencies impact system properties in several aspects such as performance, convergence, and evolution. In this thesis we use the NK and NKC models to simulate and analyze complex PD systems, which are represented by the design structure matrix (DSM). The main objective is to assess whether these models can be useful in analyzing DSMs; particularly, assessing the effect of architecture and system decomposition on product performance and evolution. All in all, this thesis mainly focuses on the effect of system decomposition when dealing with complex systems and how it helps in reducing the convergence time. However, we show that this comes at reduced system performance, represented by the fitness values. In addition, we include situated learning and design rules in this thesis to examine their impact on system convergence time and process performance and how they contribute to reaching system resolution. Once systems are decomposed, we use design rules through the process of standardization in which components agree in advance, upon each other, on some of the interfaces between the various decomposed subsystems Finally, we present our case study of a gas turbine aero engine decomposed into seven subsystems, each with a different number of components, with external interdependencies among them, to demonstrate our model and present the corresponding results. The case study shows how design rules play an important role in reaching system resolution faster and how standardization of the components’ state and fitness saves the system from running any additional needed iterations. This has shown to be a major reason of system convergence with less cost.