Some optimization procedures used in response surface methodology

This thesis is a literature survey into selected optimization topics of response surface methodology. In a typical response surface problem one of the main problems to be solved is to find those levels of the controllable variables to provide an optimum response such as highest yield or lowest cost. Several methods for attaining this optimum response are discussed, such as: the method of steepest ascent, sequential one-factor at a time procedure, a random balance method also known as random evolutionary operation (REVOP), rotating square evolutionary operation (ROVOP), and some one-factor optimization techniques. In addition two methods for optimizing an existing production process are discussed: evolutionary operation (EVOP) and simplex EVOP.

A technique useful in the interpretation of a complex surface is also discussed. This method is called “Ridge Analysis” and the development of this technique is presented and a worked example is given.