Statistical Methods for Quality Improvement, Third Edition

CHAPTER 15

Evolutionary Operation

Evolutionary operation (EVOP) was introduced by G. E. P. Box (1957) as a technique that can be used to facilitate continuous process improvement. It is based on experimental design concepts but is used much differently than the experimental designs presented in Chapter 13. With those designs the objective is to determine the relationship between a response variable and a number of process variables, and this determination is made by varying the process variables over a reasonable range. Doing so, however, will generally disrupt the normal production process. With EVOP only very small changes are made in the settings of the process variables so that the process is not disrupted, and, in particular, there is (hopefully) no increase in the percentage of nonconforming units.

Evolutionary operation has been employed successfully in the chemical industry, as well as in other industries. It is appropriate for continuous processes. As discussed by Box and Draper (1969), the evolution of an industrial process from the laboratory through a pilot plant and finally to a full-scale plant does not guarantee that the optimum operating conditions have been identified. Rather, the values of process variables that are used initially in production are probably only reasonable approximations, at best, to the optimum values. It should be noted that EVOP is not a true optimization procedure, as in mathematical optimization. Rather, the user of EVOP is attempting to ...