Page 263 - Six Sigma Demystified

Part 3 s i x s i g m a to o l s 243

George Box proposed EVOP in 1957 as a method of routine plant operation
to move toward an optimal. The techniques were further defined in Box and
Draper (1969).

When to Use

Analyze Stage
• To determine the significance of process factors

Improve Stage
• To define new process factor settings resulting in an improved response

EVOP has a number of disadvantages. A large number of repeat runs often
are needed for each phase because the factor level changes are small, with an
effect that is weak relative to process noise. The repeat runs provide a reduction
in the statistical error of estimating the effects.
The experiments generally are much longer term than traditional designed
experiments because only two to three factors are changed at a time. In other
words, it may take weeks or months to determine the significant effects com-
pared with days for a designed experiment on the same process.
Despite these shortcomings, EVOPs provide a number of advantages. Most
important, during the course of the experimentation, the process continues to
make usable product, unlike a designed experiment that requires the process
be shut down to manipulate factors beyond their normal ranges.
Unlike happenstance data that are collected without specific purpose, EVOP
data are collected at predefined operating conditions. In addition, the data may

be collected as part of revised operating procedures by operational personnel.

Methodology

Definitions

• A cycle refers to a collection of data at a given experimental condition. A
cycle is complete when one observation has been collected at each point
in the design (the replicates for that condition). A cycle is a blocking fac-
tor for the EVOP.

• A phase refers to a new iteration of cycles centered about a previously
defined optimal condition. After several cycles, a phase is completed when
the operating conditions are changed to improve the response.