*2.6.1.6 Box–Behnken design*

This is a specially made design, which needs only three levels for each factor and it is widely employed in response surface methodology for fitting second-order models for all responses. In this method, 15 experiments are run three levels for each factor. Box–Behnken designs are a combination of incomplete block designs with two-level factorial designs and these are almost rotatable. This design has the benefits that there are no runs where three levels for each factor and that there are no corner points run. Runs at the corner points may be expensive or inconvenient [52].
