Page 157 - The Six Sigma Project Planner

established procedures, as we did in the Analyze phase. There are many reasons for
first determining the process capability:
• Experiments may not be necessary if procedures are rigorously followed, thus
eliminating the need for process optimization experiments.
• Experiments are costly.
• Experiments disrupt operations.

• Experiments are inherently risky and may lead to additional problems.
• Experiments may produce misleading results if process variation isn’t reduced
beforehand.
• Variables to experiment on are often discovered while auditing a process,
performing capability studies, or investigating special causes.
• Levels at which to set (or not set) experimental variables may be determined
during the initial process investigation.
• Key personnel may be identified during auditing or SPC, thus helping us design
a better experiment.
• “Noise” variables that need to be monitored during experiments may be
identified during SPC investigations.

• The scope of the experiment is easier to determine if the process is well
understood. By definition, we don’t understand an unstable process as well as
one we can control.
At this point in the project, we know the actual performance of the critical
characteristics for this process and what the process is capable of doing if it is operated
according to established procedures. However, it is possible that the process can do
much better if we changed established procedures.

It is common practice, when new products or processes are introduced, to start out with
very poor yields, often in the single-digit range. Process designers work diligently to
improve things until they are called away by more urgent matters. At that point things
are “carved in stone,” i.e., the process is documented in standard operating procedure
manuals and further changes to it are forbidden unless special permission is obtained.
Although the standardized process may be much better than when it was introduced as
a pilot, there is often a great deal of room for additional improvement. This is especially
true if the SOP was written in the pre-Six Sigma era, when failures were still measured
in percentages (parts per hundred) rather than DPMO (parts per million).
How much better might a process do? Listen to this true story.
A major computer manufacturer sold a wave solder machine because it believed that
the best solder joints the machine could do were about 5 defects per thousand (DPMO =
5,000 or a yield of 99.9%). Its new wave solder machine was 10 times better, i.e., the
defect rate improved to 500 PPM. Naturally, the company felt its investment was

140