EVOLUTIONARY OPERATION

Most process-control techniques measure one or more output quality characteristics, and if these quality characteristics are satisfactory, no modification of the process is made. However, in some situations in which there is a strong relationship between one or more controllable independent process variables and the observed quality characteristic or dependent variable, other process-control methods can sometimes be employed.

A method is needed for continuous operation and monitoring of a process with the goal of moving the operating conditions towards the optimum or following a " drift ." The method should not require large or sudden changes in operating conditions that might disrupt production. Evolutionary operation (EVOP) was proposed by Box (1957) as such an operating procedure. It is designed as a method of routine plant operation that is carried out by operating personnel with minimum assistance from the quality or manufacturing engineering staff. EVOP makes use of principles of experimental design, which is usually an "off-line" quality-control method. Thus, it is an on-line application of designed experiments. For a detailed discussion, see Montgomery (1984, 1986).

EVOP consists of systematically introducing small changes in the levels of the process-operating variables. The procedure requires that each independent process variable be assigned a high and a low level. The changes in the variables are assumed to be small enough so that serious disturbances in product quality will not occur, yet large enough so that potential improvements in process performance will eventually be discovered .

In theory, EVOP can be applied to k variables; however, in practice, only two or three variables are used at the time. An excellent application of three variables is discussed by Box and Draper (1969).