Some thoughts on the evolution of quality engineering
| Pages | 153-157 |
| DOI | https://doi.org/10.1108/02635579710173239 |
| Published date | 01 June 1997 |
| Date | 01 June 1997 |
| Author | Chang Chung Li,Tsong Shin Sheu,Yuh Ren Wang |
| Subject Matter | Economics,Information & knowledge management,Management science & operations |
[ 153 ]
Industrial Management &
Data Systems
97/4 [1997] 153–157
© MCB University Press
[ISSN 0263-5577]
Some thoughts on the evolution of quality
engineering
Chang Chung Li, Tsong Shin Sheu, Yuh Ren Wang
Department of Industrial Engineering, National Tsing Hua University, Hsinchu,
Taiwan, ROC
Recently, Taguchi methods
have become very popular
among industries worldwide.
Begins by reviewing the past
discussion on quality engi-
neering from viewpoints of
Taguchi himself, practitioners
and academic scholars. The
weakness of Taguchi’s
approach in quality engineer-
ing is not in the statistical
analysis aspect. Posits that,
on the contrary, it provides a
way of thinking that empha-
sizes a philosophy of freely
using the methods of design
of experiments to solve engi-
neering problems. States
that, in this regard, quality
engineering will be further
empowered via the aid of
various engineering tech-
niques. Proposes, based on
these kind of thoughts, two
directions for evolution of
quality engineering.
An introduction to quality
engineering
Recently, Taguchi methods have become very
popular among industries worldwide. Many
academic research papers and industrial
application reports have been published
under the title “Taguchi methods”. In fact,
the so-called “Taguchi methods” is a collec-
tive term. It consists of system of experiment
design, quality engineering (QE), business
data analysis, divisional evaluation system
and pattern recognition. Furthermore, QE
can be divided into online QE and off-line QE.
Online QE is in contrast with traditional
statistical process control (SPC) and off-line
QE is in contrast with classical design and
analysis of experiments. The difference
between online and off-line QE is that activi-
ties of the former are done along mass pro-
duction lines while that of the latter are done
prior to mass production. There are three
steps in off-line QE, i.e. system design, para-
meter design and tolerance design. What
most people in North America call robust
design refers to parameter design and toler-
ance design together. This is also the focus of
discussions among statisticians and quality
specialists. In this paper, QE refers to off-line
QE in Taguchi methods.
Taguchi defines quality of a product/
process as the loss to society. Losses are
incurred because performances of products
deviate from their targets (ideal function).
And deviation is caused by noise. Noises are
uncontrollable variables, such as environ-
mental factors (temperature, humidity, dust,
magnetic, etc.), variation within components
and deterioration (wear-out). Therefore,
noises are always there and we can never
eliminate them. So are losses. The best we can
do is to make our products/processes
“strong” enough to be less sensitive to the
effects of noises. That is, through minimiza-
tion of loss, we obtain a robust product/
process.
In system design, design engineers examine
a variety of architectures and technologies
for achieving the desired function of the prod-
uct and select the most suitable ones for the
product/process. Engineering expertise plays
a big part here. Next, in parameter design,
designers start with the lowest possible cost
and try to determine the best settings for each
control factor to minimize the sensitivity to
noises. Taguchi suggests using two orthogo-
nal arrays (control array and noise array) to
study the effects of noises combinations on
the various combinations of control factors.
If, at the end of parameter design the quality
loss is acceptable, we have a design with the
lowest cost. Otherwise, we have to go to the
third step – tolerance design. In tolerance
design, a trade-off is made between reduction
in the quality 1oss and increase in manufac-
turing cost.
Review on discussions of quality
engineering
During the panel discussion of the “Third
Symposium on Taguchi Methods”, held by
the American Supplier Institute, Inc., in Octo-
ber 1985, the late Prof William G. Hunter
wrote the following concerning professional
attitudes with regard to the functions of a
manufactured product:
Engineers: Make it work
Scientists: Understand why it works
Mathematicians: Don’t care
However, Taguchi[1] pointed out that the task
of an engineer is not merely to “make it
work”. As a matter of fact, when he worked
for Morinaga Pharmaceutical company in
1948, Taguchi was interested in determining
whether it was possible to produce caramels
whose plasticity would not change when
room temperature changes. Therefore,
Taguchi had concepts like ideal function and
loss to society some 48 years ago.
Taguchi[1] continued to point out some
different aspects between Taguchi methods
and classical design and analysis of experi-
ments as follow:
What characteristics does Taguchi Methods
have that are distinct from the design of
experiments methods in general?
The answer lies in an efficient research
method for designing, not in order to find
responses (functional relations), but to reduce
To continue reading
Request your trial