| The design process for complex systems must be substantially improved in order to create more efficient and reliable products in less time and reduced cost. However, increasing efficiency and reliability while decreasing time and cost is a difficult task to accomplish. To create a better design, more detail about the problem must be taken into account thereby yielding an invariably longer design process. Research has focused on new solution methods, development of software infrastructures, and even creation of new hardware to specifically handle large amounts of data. However, these approaches tend to be largely autonomous, with little potential for human involvement. The thinking has been that the larger the amount of data, the less involved the designer can be and the more involved the computer must be. Hence, new ways to extract meaning out of these larger amounts of data are required to involve and fully engage the designer in all aspects of a design process.; In this dissertation, the concept of Computational Steering is applied to optimal design and Multidisciplinary Design Optimization (MDO) to improve efficiency and reliability of solutions. Computational Steering calls for real-time visual interaction with the data of a problem. In this dissertation, a modified paradigm termed Visual Design Steering (VDS) is proposed and implemented. VDS allows a designer to make choices before, during, or after an analysis via a visual environment. He can then institute these changes immediately, thus steering the analysis and optimization to a better solution in less time. This proposed paradigm contains two key components. First, the visualization method to represent the data, and second, the way these representations are used to improve an optimal or MDO design process.; The visual methods developed and described in this dissertation allow a designer to gain insight into the complex relationships that exist between design variables and the objectives and constraints of a design. The methods also provide a mechanism to foster concurrent design by allowing the exchange of these insights between designers located at different geographic locations. Test cases are presented to demonstrate the capabilities of the developed methods. |
