The Impact of Solution Process Architecture on the Convergence Behavior of Distributed Design Systems
Devendorf, Erich D.
MetadataShow full item record
In distributed design processes individual subsystems have local control over design variables and seek to achieve their own individual objectives, which may also be influenced by some system level objectives. The resulting network of coupled subsystems will either converge to a stable equilibrium or diverge in an unstable manner. When analyzing these decentralized systems the fundamental system characteristics to consider are the equilibrium stability and transient response. The intellectual merit of this dissertation is based on examining the dependence of equilibrium stability and transient response on the solution process architecture. The solution process architecture describes how the design subsystems are ordered and can be either sequential, concurrent or a hybrid that incorporates both. In this dissertation it is demonstrated that system stability is indeed dependent on the process architecture and a general process architecture model is created based on linear system theory. The model allows for the equilibrium stability of any process architecture to be determined by converting any process architecture into an equivalent parallel representation. Through the analysis of a system's parallel equivalent, qualitative and quantitative tools are developed to examine the transient response of distributed design processes. The qualitative tool proposes a new model to examine process architecture by leveraging concepts from social network and graph theory. The quantitative tool directly examines parallel equivalents, using control theory to create an upper bound on the system’s overall transient response time. The robustness of process architectures are also considered through the study of design impulses, defined to be unexpected inputs, interruptions and changes in the design process. This dissertation has broader impacts in scenarios where decision makers are members of large decentralized decision networks. These networks include, but are not limited to, engineering firms, emergency responders, military personnel and medical professionals. Greater insight by network managers into the mechanisms by which process architecture impacts the stability and transient response of decision networks empowers them to realize faster solutions, set realistic deadlines and allocate project resources effectively. Even when decentralized systems are not actively managed, process architectures can be specified that passively mitigate the impact of system impulses and possess desirable transient response characteristics.