ABSTRACT
Designs that have factors at three levels can be used to estimated a quadratic relationship between the quality characteristics of a product and the process variables of the manufacturing process used to make that product. Designs that have each factor at two levels are capable of identifying and modelling linear relationships only. This chapter explains the concept of a non-linear effect, and provides the structure of a second order response surface model. The second order model is an extension of the first order response surface model and can be used to analyse the data from non-linear experimental designs. The chapter describes how least squares can be used to estimate the parameters of this second order model and how the least squares procedure can be implemented within Excel. It also describes how least squares can be used to estimate models that are more general than the second order model. The chapter analyzes the friction welding experiments.
