A performance-based representation for engineering design
Engineering design is an unstructured but logic-based process where successive iterations of synthesis and analysis eventually converge to the desired solution. Due to the sophisticated nature of design, it is rarely feasible to deduce the final design from the initial specifications. A performance-based representation and its applications are developed in this dissertation to assist the designer in synthesizing design artifacts, gaining insight into the performance attributes of a design problem, and making rational decisions throughout an iterative design process. To achieve this goal, an Extensive Simplex Method is proposed to traverse the extreme points of linear model and build the graph-based design and performance space. Based on the established feasible space, the designer can assess the system performance capability, evaluate the feasibility of a design alternative, achieve multiple feasible solutions, negotiate the trade-off of system performance attributes, and eventually converge to a desired solution.
The methodology is applied to a beam design problem and a molding process control problem. While the first example demonstrates the utility of the representation in concept selection, parametric design refinement, and inverse reasoning, the second example focuses on the process tuning and flexibility evaluation of a digital video disc molding process. Both case studies indicate that the performance-based representation can help the designer not only to obtain a flexible design solution with better overall system performance, but also to gain insight to the design problem and its specifications. Such an insight to problem leads to increased competence in product design.