Presented to
the faculty of the School of Engineering and Applied Science
University of Virginia
In Partial Fulfillment
of the requirements for the Degree
Doctor of Philosophy in Systems Engineering
by
John Francis Kros
May 18, 1997
ABSTRACT The use of Taguchi methods for improving critical quality characteristics is widespread in industry today. In practice, Taguchi-style designs have been used very successfully to improve one quality characteristic of a product. However, since almost all products or processes are made up of more than one desired quality characteristic, methods of multiresponse design become very important issues for design engineers.
This research investigates the Taguchi methods of parameter design when multiple responses exist and develops an improved model for selecting design constants (system parameters). Design problems which endeavor to simultaneously achieve numerous quality characteristics, many of which are in conflict, are referred to as multiple criteria decision making problems (MCDM). A special class of MCDMs called multiple objective decision making (MODM) problems will be studied.
Taguchi recognizes the customer's desire to have products of higher quality. The quality of a product is related to the concept of "loss to society". This loss is composed of the costs incurred in the production phase as well as the costs encountered during the use by the customer. A truly high quality product will have minimal loss as it progresses through its life cycle. Taguchi's strategy is to minimize "loss to society" through the reduction of variance. A modeling approach based on Taguchi's idea of "loss to society" for solving MODM quality engineering problems is proposed.
The approach is built upon Taguchi's concepts of loss functions (Us), signal-to noise (S/N) ratios, and the three general quality characteristics: smaller-the-better (STB), larger-is-better (LTB), and nominal-the-best (NTB). A MODM technique will be proposed that allows for the modeling of multiple Us, the integration of those multiple Us into the S/N ratios, and the optimization of the multiple quality characteristics. Taguchi methods have traditionally been used in manufacturing and product improvement. This approach will be applied to master production scheduling -- a dynamic decision making environment.