Design as a search problem : interactive visualization for system design under uncertainty

Abstract

Epoch Era Analysis (EEA) was developed to better model problems with lifecycle uncertainties and has demonstrated its usefulness in prior research studies, but it still faces significant challenges to practical application. Specifically, EEA can result in large, multivariate datasets that are difficult to generate, visualize and perform analysis on. When performing exploratory analysis on such model-generated data sets, human interaction is often necessary to identify important subsets of the data, resolve ambiguity or find inconsistencies. Although prior research towards methods for applying EEA constructs has been performed, a prescriptive framework that explicitly considers human interaction does not exist. To make informed decisions, and design successful strategies for value sustainment, effective visualization and analysis techniques are needed to derive valuable insights from this data. These challenges motivate this thesis research. The aim of this thesis is to leverage recent research in visual analytics and advanced systems engineering methods to develop a rigorous framework, with associated methods, processes, metrics and prototype applications that will result in new capabilities that better enable analysis and decision-making for long-run value sustainment. Several research contributions are outcomes of this thesis. First, the Interactive Epoch Era Analysis (IEEA) framework is introduced as a methodology for analyzing lifecycle uncertainty when designing systems to achieve sustained value delivery. IEEA provides a coherent theoretical framework to guide the development of human-usable analytic tools for early-stage system concept selection. Next, new interactive visualization applications for system concept selection are introduced to demonstrate the feasibility, usefulness and scalability of IEEA as an integrated visual analytics system. Finally, to characterize the benefits of interactive visualization applications for engineering design problems, the results of a controlled human-subjects experiment are presented.