ABSTRACT
In the previous chapters, it was implicitly assumed that the experimental trials performed in a response surface investigation were carried out under homogeneous conditions. This assumption, however, may not be valid in some experimental situations. For example, batches of raw material used in a production process may be different with regard to source, composition, or other characteristics. In another situation, changes in the experimental conditions may be difficult to avoid or control, particularly if a long period of time is needed to perform the experiment. In general, heterogeneity of conditions can be caused by one or several extraneous sources of variation. In such circumstances, the experimental trials should be carried out in groups, or blocks, within each of which a greater degree of homogeneity of conditions can be maintained. This amounts to a division of the corresponding response surface design into several blocks. As a result, a block effect must be added to the fitted model. This effect is discrete, that is, does not have values on a continuous scale, as is the case with the model’s input variables.
