|
DESIGN OF EXPERIMENTS:
Who is Using it? And Why?
As both product quality
and product competition rise throughout the world, a growing number
of companies that have relied on conventional research and testing
methods, and have accepted a certain percentage of product failures
as normal and unavoidable, are realizing their trial-and-error methods
are costing them more in dollars, and perhaps in customers, than they
can logically afford. In response, many firms are replacing or enhancing
traditional experimental approaches with Design of Experiments, the
advanced statistical tool that can generate research breakthroughs
and far higher manufacturing yields.
After years of consulting, we've learned the reasons companies
avoid upgrading their statistical methods typically fall into a
few categories which can loosely be defined as "It's tradition,"
"We haven't got the time," or "Three batches and we're done (we
hope!)." Companies with long histories and a fair amount of success
often use tradition as a defense against change, partly because
of fear the waste accepted as part of the old ways will become glaringly
evident. Other firms complain their research and manufacturing departments
are already sorely overworked, and the energy required to train
problem solvers in new methodologies is lacking."I really understand
where these companies are coming from," says Tunnell Principal Larry
Meyers, who has taught Design of Experiments to companies for some
25 years. "It is true that DOE can represent a major shift for research
and manufacturing departments and that learning DOE requires an
initial investment of energy. But the payback is so ultimately satisfying
that people are often shocked they've ignored DOE for so long. The
fact is, for a relatively small investment in time and resources,
DOE can save an extraordinary amount of money."
As an example, Larry sites a popular success story involving Motorola,
detailed in World Class Quality by Keki Bhote; Motorola reported
a 500-fold increase in quality and a $1 billion savings during a
three-year period following the adoption of advanced statistical
tools. In this newsletter, we recently reported on the success of
Trans-Tech, a major technical ceramics firm that "exploded" capacity
in one operation by more than 100% -- and satisfied its client,
the U.S. Navy -- by using experimental design to unravel a process
problem.
What other types of companies are using this tool, and what kinds
of problems are they trying to solve? We posed that question to
students in one of our recent 3-day Design of Experiments courses:
HOW
ARE YOU USING DOE?
Elliot Lipin
Advanced Components Engineer
General Signal Networks-Tautron
PRODUCT:Telecommunications
Products
In the past I've done things by the seat of the pants -- common sense,
logic -- all those things you need to have anyway, but DOE brings
some rigor to my analysis of data. It gives me some experimental evidence
to make sure if we invest time and capital that, in fact, we know
what the variables are, we know what impacts what, what the trade-offs
are, so we can make an educated recommendation in terms of what the
processes should be and also understand what the possible risks and
problems are at the same time. That's where Design of Experiments
comes in perfectly; it puts a handle on these things and quantifies
the variables.
George Gillespie
Mechanical Engineer
Naval Air Warfare Center
PRODUCT:
Vehicle Restraint Systems
It's so expensive to do big systems -- large seats and big restraint
systems for various kinds of vehicles and aircraft -- that it really
is cost-effective to cut down the number of tests you have to do.
One issue we have involves a horizontal accelerator -- it's a sled
with a mannequin on it that we use to test a seat belt and seat by
firing it up at x-number of G-forces. We're trying to arrange a business
venture with a university that is 1500 miles away, but we haven't
been able to make sure the sleds are exactly alike. DOE gives us this
possibility.
Peter Giancana
Packaging Analyst
Estee Lauder Inc.
PRODUCT:
Cosmetics and Perfumes
We test all the packaging that the cosmetics go into, testing for
compatibility with products and putting the packaging materials through
stress and environmental tests because products can really be exposed
to extremes once they leave the plant; something can get put on a
truck and wind up sitting in the Arizona desert overnight. In the
packaging field, some of the things that are handed down are decades
old and sometimes you stop and scratch your head and wonder where
the criteria you're using has come from. More often than not you wind
up only 80% sure what's wrong in a situation. With DOE, we can redesign
our experiments and be confidently able to predict why something is
happening, and what we can expect if we do things a certain way.
George Smith
Advanced Technologist
Owens Corning
PRODUCT:
Replacement Windows
People buy a replacement window and they want it to look nice but
they also don't want to pay any more than they can afford. We've been
competing against the higher priced windows, against some of the big
manufacturers like Andersen and Pella, but for some people the cost
is too high. We need to find out exactly what variables can help us
reduce the cost of production so we can get an even better quality
product and a better process, but charge less so our volume will go
up. We need to look at a number of variables, such as temperature
of the die used, how fast the product comes out of the die, and so
on. Reducing cost is the bottom line.
John Soboslai
Project Engineer
Wyeth Nutritionals, Inc.
PRODUCT:
Infant Formula
We've had some problems attaining certain physical properties in a
new powdered infant formula we've introduced, and DOE could help us
get better control of the variables involved, specifically with regards
to bulk volume and the reconstitutability of the product. For example,
if our bulk volume is off, let's say on the high side, the formula
is very fluffy and the volume of a scoop would be less than the required
30 grams. In such a case, the infant would be fed less nutrients than
they need, and certainly we would never be comfortable with that.
Nasser El-Awar
Project Engineer
Keystone Thermometrics
PRODUCT:
Thermistors
Although most of our products are for the automotive industry, they
have a wide variety of applications. In all our applications, the
degree of stability is very important. Stability depends on the manufacturing
process and this process is controlled by many factors such as material,
temperature and time at every stage. With a DOE study, I can quickly
determine the most critical and effective temperatures and durations
for a particular process, thus I am able to save a great amount of
time on testing all the possible combinations. And I am able to further
improve my process for a better and more stable product.
Tunnell Consulting's Design of Experiments course is taught by
Larry Meyers. Please call him at (610) 337-0820 for more information
on courses or DOE consulting services.
Back to Performance Excellence Practice
Back to Courses
|