Abstract
There has been considerable interest recently in the application of parameter design methodology to make a system's performance robust over a wide range of input conditions. This has been referred to as robust design with dynamic characteristics. In this paper, we first review this overall strategy and its usefulness and then demonstrate the limitations of the data analysis methods recommended by Taguchi. We then propose two graphical methods, the sensitivity-standard deviation plot and the gamma-plot, for identifying suitable measures of dispersion that are valid more generally. These graphical methods are related to the mean-variance plot and the lambda-plot that have been found useful in analyzing data from static robust design studies. The implementation and the usefulness of the proposed graphical methods are illustrated through two examples including a real application on engine idling performance.
Additional information
Notes on contributors
Mahesh Lunani
Mr. Lunani is an Engineer in the Powertrain Operations Quality Office.
Vijayan N. Nair
Dr. Nair is a Professor of Statistics and of Industrial and Operations Engineering. He is a Senior Member of ASQC.
Gary S. Wasserman
Dr. Wasserman is a Professor of Industrial and Manufacturing Engineering. He is a Senior Member of ASQC.