Winkler, M. Denise (1988, ASQC) Texas Instruments, Inc., Johnson City, TN
In a positive response to today's quality challenge, many companies in the United States have become educated in statistical process control (SPC), using SPC extensively to maintain product quality. However, simply maintaining product quality may not recapture lost market share, but producing quality product at competitive prices may very well be an answer. How do we offer competitive prices when manufacturing costs seem relatively fixed? Many companies are trying various designed experiment techniques to statistically optimize processes, ultimately resulting in reduced cost through reduction of scrap and rework. These techniques range fro classical statistical methods to quality engineering approaches such as Taguchi methods.
This paper will discuss key differences between classical methods and Taguchi methods, describing certain controversial issues as statistical theory versus cost effective applications. Advantages and disadvantages of using either Taguchi or classical methods will be included. The paper will specifically compare these approaches in the areas of cost measures, interactions, non-linear effects, and data analysis. The paper will also discuss the concepts of process optimization vis both techniques and examine the means of determining best fit parameter values through both methods. To best demonstrate the similarities and differences between the classical design of experiment and Taguchi methods, a specific example is also utilized to illustrate these ideas.