The Forgotten Science
A. Blanton Godfrey
I recently prepared a conference lecture titled “Six
Sigma and Reliability: Is There Any Connection?” Because
reliability is such a key part of design for Six Sigma and
many opportunities exist to apply reliability methods in
related projects, I was sure I could put together numerous
examples of reliability’s role in most Six Sigma programs.
To my surprise, I found very few applications.
Perhaps I shouldn’t have been so surprised. Most
companies focus on cost reduction; very few have expanded
their projects into design areas or top-line-growth efforts.
It could be that companies are still in the early stages
of their Six Sigma efforts and haven’t yet considered
design or warranty problems or field failures. But why haven’t
organizations implemented reliability methods more widely?
In many companies, the reliability function (if there’s
one at all) is separate from the quality department or line
management. A few years ago, while teaching a design for
Six Sigma workshop in an aerospace company, I found that
most designers didn’t know what a failure mode and
effects analysis was. The company had a reliability department,
which was responsible for completing FMEAs on new designs,
but the results weren’t shared with the designers.
Reliability engineering was considered a separate function.
Also, very little reliability engineering, estimation
or prediction are covered in most four-week Six Sigma Black
Belt courses. Most design for Six Sigma courses merely introduce
these subjects, and most don’t cover life testing.
Six Sigma courses have become very rigid. Almost everyone
teaches the same topics, and few instructors customize their
courses to meet their clients’ real needs. They have
the packaged materials, the software, the case studies and
the exercises, and they’re intent on going through
all the material with no deviations. I suspect that very
few Six Sigma instructors are even qualified to teach reliability.
Moreover, most of the time it doesn’t matter whether
a Black Belt understands reliability because the company
doesn’t have the data it needs for making estimates
or predictions. In one company where improving throughput
was critical, the only available numbers were hours that
the production lines weren’t running. Very little
useful information existed for estimating machine failure
rates or complete production systems. At many companies,
the different functions are so separated that collecting
data from failures during or after warranty is almost impossible.
Designers receive almost no information on how the products
they’ve designed actually perform. Nor do they have
information on competitive product performance. If the company
lacks information on parts reliability, there’s little
a designer can do to predict reliability. And if the company
contracts out service and repair, it’s extremely difficult
to get reliable information about field failures.
By splitting companies’ operations across many functions,
we’ve also made it easy to blame someone else. The
service department blames designers, purchasing, manufacturing
or even shipping for problems they see every day. However,
they don’t really have the information to determine
where the problem originated. In many companies, the service
function is quite profitable and there’s little motivation
to reduce failures. When purchasing has little information
on quality or reliability, we shouldn’t be surprised
that it buys simply on lowest price. When manufacturing
has little information on field failures caused by process
changes or shortcuts, we shouldn’t be surprised to
see these changes proliferate. When designers don’t
receive reliable data about failures during tests or in
the field, we should expect to see the same design errors
It’s easy to collect initial quality data on a wide
variety of products. Organizations such as Consumers Union
or J.D. Power and Associates provide useful quality studies
and even make some attempt to portray a product’s
reliability, at least during early months or years of use.
But for many products, we have no idea of long-term reliability.
Trade magazines generally describe the features and test
results of only small samples of new product. Many companies
find it’s far easier to replace failed items rather
than repair them. I was shocked recently to learn that a
garment meant to protect against chemical agents lost all
protective ability after one washing. It seems that the
buyer never bothered to test whether it met the specs of
So how do we encourage the use of reliability methods?
One way is to expand the scope of initial Six Sigma projects.
The best way to expand thinking is to provide organized
lists of past projects and examples of results from other
companies. Clients often recognize that the same problems
exist in their companies. It’s also important to become
more customer-centered in Six Sigma training and customize
material to meet clients’ needs. For four or five
people working on reliability problems, provide special
sessions to introduce the right methods. Then offer one-on-one
support to ensure they learn how to use the right method
at the right time. Add follow-up workshops on advanced methods
to expand their knowledge and get results in their organizations.
A. Blanton Godfrey, Ph.D., is dean and Joseph D. Moore
Distinguished University Professor at North Carolina State
University’s College of Textiles. Letters to the editor
about this column can be e-mailed to email@example.com.