Browsing by Author "Brochtrup, Brad M."
Now showing 1 - 2 of 2
Results Per Page
Sort Options
Item Classifying and Comparing Design Optimization Problems(2006) Brochtrup, Brad M.; Herrmann, Jeffrey W.; ISRResearch in product design optimization has developed and demonstrated a variety of modeling techniques and solution methods, including multidisciplinary design optimization. As new techniques migrate to the industrial world engineers are faced with much more complex problems often extending beyond their realm of knowledge. A novel classification scheme is proposed and demonstrated to offer engineers a method of organizing and searching for relevant example problems to assist in the production of their own optimization problem. To explore the tradeoff between information requirements and solution quality, computational experiments are conducted on two design problems, a bathroom scale and a universal electric motor. In particular, the results of these experiments identify the additional information required to solve a profit maximization problem, demonstrate the role of rules of thumb in formulating design optimization problems, show how decomposition affects solution quality and computational effort, and uncover the impact of using target matching in the objective function instead of as constraints. In addition, the results show how the values of targets and objective function weights impact solution quality. In general, these results show the extent to which correct information is critical to finding a high quality solution, perhaps more critical than the optimization model selected. That is, the quality of the information used is more important than the amount of information used.Item Selecting an Optimization Model for Product Development(2006) Brochtrup, Brad M.; Herrmann, Jeffrey W.; Herrmann, Jeffrey W.; ISRDesign optimization is an important engineering design activity. When used during the development of a new product, the overall profitability of that product depends upon the quality of the solution that the optimization model returns as well as the time and cost of using optimization. There exist many different ways to use optimization. The design engineer wants to select the most appropriate optimization model to create a profitable design. This paper discusses this meta-design (or meta-reasoning) problem and presents a method for selecting the best (most profitable) optimization model from a set of candidate optimization models. The approach allows multiple ways to handle uncertainty about the optimization models. We demonstrate the approach by considering the optimization of a universal electric motor.