This is the fifth in a series on design of experiments. The earlier columns discussed two-level factorial designs, useful in the preliminary stages of an investigation when the effects of many factors on some system need to be studied in order to identify the important ones. Once a small number of factors have been identified it is sometimes of interest to investigate their effects in more detail such as establishing whether the response of the system to changes in the factors is linear or not, or to identify optimal settings. For this purpose response surface designs are used.

Keywords: