| "It
took America a long time to figure out what Taguchi was talking
about (regarding robust engineering)," said Neil Ressler,
Ford vice president, Research & Vehicle Technology. "For
many years, it was thought that he was talking about design
experiments-a tool that was in his portfolio-but in recent years
we at Ford have figured out that there is a lot more to it than
that."
Robust engineering refers to the elegance
of a product’s design, Ressler said. Part of the problem,
he said, is that is not an intuitive process as American engineers
originally thought. The concept basically involves four elements-the
designed experiments process, the "voice of the client,"
controlling the variation in the quality of parts and "Key
Life Tests."
Ford used the design experiment process
to improve sound quality of the current Taurus’ power
train.
There are many variables that go into
making engine sound pleasant, such as the stiffness of the
crankshaft, the clearance of the rod bearings, the length
of the exhaust system on V6 engine or flexible couplings for
the exhaust system.
"The question is, which (items)
are important and should be put in, and which can be left
out because they just cost money and don’t add anything,"
Ressler said.
Most auto companies will build prototypes
of a vehicle; those prototypes can then be used to verify
the vehicles’ subsystems. But this is a less-than efficient
verification process because a vehicle never has all of its
parts proven, said Roscoe Nash Jr. dean of the Ford Design
Institute.
Instead of going through an enormous
number of experiments and altering only one item at a time,
as normal intuition recommends, Taguchi’s designed experiment
process is much simpler, he added.
"You can actually define a set
of experiments that doesn’t involve many changes by changing
the combinations of independent variables and figuring out
the separate contributions of each (variable)," Ressler
said.
Ford engineers using the designed experiment
process discovered that some components of the conventional
logic to reduce noise-tolerance design specifications and
materials used-were" simply ineffective" and added
to the cost of the vehicle, Ressler said.
A second part of robust engineering
is to understand "the voice of the client"-how they
perceive the result of your design, Nash said.
The third element to take into account
involves the quality of parts and materials. For example,
a part that barely meets the manufacturers specifications
is not the equal of a part that is better than average. The
best result comes when parts meet the desired target level
of quality because the variability has then been minimized,
Nash said.
Another component of robust engineering
is testing vehicle systems using a technique called "Key
Life Tests" to see how a system performs in long-duration,
normal use, Nash said.
In the past, Ford might have tried
testing the strength of a door handle by hanging a 150-pound
weight off it, for example. These types of strength tests,
however, don’t give a true picture of how that door handle
will perform over the life of a vehicle, which is 150,000
miles or 10 years.
" If a normal client’s load
on a door handle is 25 pounds, that’s what you want to
test for," Nash said.
The results of a survey of the actual
use of remote keyless entry systems gave Ford engineers a
surprise and illustrates why it is important to test a system
based on normal use, said Ford spokesman Jon Harmon.
"When talking to people in the
real world about key fobs, we found that they would hit the
buttons over and over and over to see how far it worked,"
Harmon said. "We found out that it was used eight or
nine times (more times) than what they had been tested for."
|