Sensitivity Analyses Using
Design of Experiments Techniques

Terry Bahill
Systems and Industrial Engineering
University of Arizona
Tucson, AZ 85721-0020, USA
terry@sie.arizona.edu
© 1998-2004 Bahill

We have used Design of Experiments (DoE) techniques to perform sensitivity analyses and to design robust control systems. In particular, the use of uncontrollable factors and second-order effect estimates is suggested to ensure the robustness of the analyzed system to parametric uncertainty. These second order effects are often called interactions of the parameters. An augmentation to Yates' algorithm for 3-level factorial designs is described to speed the process of obtaining estimates. This method can be used to optimize the system design.

Reference [58]. This lecture is suitable for engineers. This talk requires an overhead projector. This talk takes one hour.