Abstract

Designers play an important and critical role in developing innovative product design, which is the key for a company to survive in the highly competitive market with ever-increasing demands from customers. However, how designers think, reason, judge, and make decisions has not yet been studied well. This thesis aims to develop experimental approaches to qualitatively and quantitatively understand designers' cognitive activities in order to explore product innovation in the early design stages. To study designers' cognitive activities, a new protocol analysis method is developed. In the protocol analysis experiment, designers' activities were recorded when solving a design problem. Designers were interviewed to recall their design process immediately after finishing the design. Then, the recorded verbal data were analyzed by transcribing, segmentation and encoding for further analysis. In general, protocol analysis is used to transform the unstructured data collected from designers into structured data. Existing protocol analysis methods used to study designers' cognitive activities heavily depend on specific design problems, specific domains and the persons who analyze the protocol data. The new protocol analysis method presented in this thesis is based on the concept of the state of design and recursive object model. It can be easily applied to any design problem and any domain. Using this method, the changes of designers' cognitive activities during the design process can be quantified. Some guidelines and recommendations for assisting designers to deliver an innovative design are summarized at the end of the present thesis based on the experimental results. The protocol analysis results have also been used to evaluate Environment-Based Design (EBD) as a descriptive design model; hence, EBD can be used as the theoretical foundation to illustrate, describe, and explain how designers conduct a design task. Another part of my thesis is to simulate the design process by developing a virtual experiment. In the virtual experiment, the selected design problem: finite element mesh design, can be automatically solved using a program developed in the computer environment of Visual C++. Net 2003. Then, different settings of the parameters, different strategies, different formulations of design requirements, and different sequences for solving the design problem are simulated to compare the differences of meshes. Three routes leading designers to innovative designs are proposed and examined through this virtual experiment. Based on the three routes, an ANN-based element extraction method for finite element mesh design is developed to illustrate the feasibility of the three routes for changing design solutions. Therefore, the two experimental approaches proposed in this thesis can be used for understanding design activities in a more systematic manner