A Design Tool Architecture for the Rapid Evaluation of Product Design Tradeoffs in an Internet-based System Modeling Environment


This thesis presents a computer-aided design tool for the rapid evaluation of design tradeoffs in an integrated product modeling environment. The goal of this work is to provide product development organizations with better means of exploring product design spaces so as to identify promising design candidates early in the concept generation phase. Ultimately, such practices would streamline the product development process. The proposed design tool is made up of two key components: an optimization engine, and the Distributed Object-based Modeling Environment. This modeling environment is part of an ongoing research initiative at the Computer-Aided Design Lab. The optimization engine consists of a multiobjective evolutionary algorithm developed at the Ecole Polytechnique F6d6rale de Lausanne. The first part of this thesis provides a comprehensive survey of all topics relevant to this work. Traditional product development is discussed along with some of the challenges inherent in this process. Integrated modeling tools are surveyed. Finally, a variety of optimization methods and algorithms are discussed, along with a review of commercially available optimization packages. The second part discusses the developed design tool and the implications of this work on traditional product development. After a detailed description of the optimization algorithm, use of the design tool is illustrated with a trivial design example. Enabled by this work, a new "target-driven" design approach is introduced. In this approach, individuals select optimal tradeoffs between competing design objectives and use them, as design targets, to configure the integrated product model so as to achieve best-overall product performance. Validation of this design approach is done through the design of a hybrid PV-diesel energy system for two different applications. It is shown that the design tool effectively evaluates design tradeoffs and allows for the rapid derivation of optimal design alternatives.