Scribd
Upload
25 views

Quality Through Product and Process Optimization

The document discusses quality improvement through product and process optimization using design of experiments techniques. It explains conventional factorial designs and their limitations. It then introduces Taguchi's approach which uses standardized orthogonal arrays to design robust experiments and outer array designs to make products insensitive to noise factors. The document outlines key concepts like quality definition, two-step optimization, loss function and signal-to-noise analysis in Taguchi methodology.

Uploaded by

Joshua Jasmin Grospe
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
25 views

Quality Through Product and Process Optimization

The document discusses quality improvement through product and process optimization using design of experiments techniques. It explains conventional factorial designs and their limitations. It then introduces Taguchi's approach which uses standardized orthogonal arrays to design robust experiments and outer array designs to make products insensitive to noise factors. The document outlines key concepts like quality definition, two-step optimization, loss function and signal-to-noise analysis in Taguchi methodology.

Uploaded by

Joshua Jasmin Grospe
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 49

QUALITY

through
PRODUCT & PROCESS
OPTIMIZATION
What do you are the
reasons why technology
increases?
Today’s TECHNOLOGY is testimony
to man’s incessant desire to provide
a HIGHER LEVEL OF QUALITY in
products and services to increase
market share and profits.
DESIGN OF EXPERIMENTS
THE CONVENTIONAL APPROACH
The technique of defining and
investigating all possible
conditions in an experiment
involving multiple factors is
known as the design of
experiments (DOE)
Design of experiments
concepts have been in use
since Sir Ronald A. Fisher’s
work in agricultural
experimentation during the
late 1920s.
Example (Design of Experiments)
You, as a student, are planning to improve your scores on
engineering examinations. You are thinking that study time and
part of the day are essential in order to improve your exam. Two
kinds (levels, S1 and S2) of study time, 3 hours and 5 hours, and two
parts (levels, D1 and D2) of the day, Morning and Evening. To select
the best combination of these that will make you improve your
scores, you have tried each of the combination.
• It involves two factors (time and part of the
day) at two different levels (low and high or
subscripts 1 and 2) that affect the scores
on the exam.

• Such an experiment is described as a 2 × 2


factorial experiment.

• There are four (22) possible treatments or


combinations.
Analysis of EXAMINATION SCORES shows a 10% (65 – 55) increase in preference
for EVENING (D2) at the low level (S1, 3 HOURS) OF STUDY TIME.
However, EXAMINATION SCORES shows a 15% (90 – 75) increase when
more study time is used for EVENING.
On the other hand, for the HIGHER STUDY TIME, the taste preference
increased from 55% to 75% at (D1) MORNING
And further increased to 90% on the EVENING (D2).
The mean response which is 12.5% is called the main effect of PART OF THE DAY.
Similarly, the main effect for STUDY TIME is 22.5%.
Factorial
Effects
Because the lines for the two levels, D1 and
D2, are almost parallel, Part of the Day and
Study Time are said to be independent, and
little or no interaction is assumed to exist.
NONPARALLEL LINES would indicate the presence
of some interaction. HIGHLY SKEWED LINES or a
HIGHER ANGLE BETWEEN LINES (need not be
intersecting) would indicate strong interaction
between the two factors.
Number of Possible Designs

where . . .
L = number of levels for each factor
m = number of factors.
Example:

If the qualities of a given product depend on


three factors, A, B, and C, and each factor is
to be tested at two levels, then how many
will be the possible designs?
Number of Possible Designs

m
N=L
3
N = 2 = 8 designs
Resulting Matrix
Example:

Now consider the case where the cookies


under consideration have 15 different
ingredients at two levels each. How many
will be the possible designs?
Number of Possible Designs

m
N=L
15
N=2 = 32,768 designs
A research program
of this magnitude
would be excessive in
COST and TIME.
FRACTIONAL FACTORIAL
experiments investigate only
a fraction of all possible
combinations.
So, while FACTORIAL and
FRACTIONAL FACTORIAL
designs of experiments are
widely and effectively used,
they suffer from the
following LIMITATIONS:
• The experiments become unwieldy
in cost and time when the number
of variables is large.

• Two designs for the same


experiment may yield different
results.

• The interpretation of the


experimental results with a larger
number of factors may be difficult
due to lack of clear design and
analysis guidelines.
DESIGN OF EXPERIMENTS
THE TAGUCHI APPROACH
Genichi Taguchi
developed a methodology
for applying STATISTICS to
improve the QUALITY of
manufactured goods.
He simplified and standardized
fractional factorial designs in such
a manner that two engineers
conducting tests thousands of
miles apart would use designs of
similar size and expect to obtain
consistent results.
Definition of Quality
Taguchi defined QUALITY in
terms of MINIMUM LOSS TO
SOCIETY which in measurable
engineering terms translates
into CONSISTENCY OF
PERFORMANCE.
CONSISTENCY is achieved
when performance is close
to the target with LEAST
VARIATION.
To improve quality,
Taguchi proposed a
two-step optimization
approach:
Two Step Optimization

1. Find the factor-level combination that reduces


performance variability.

2. Adjust the factor levels that bring performance


closer to the target.
Standardized DOE
For designing experiments,
Taguchi utilized a special set of
tables, called orthogonal arrays
which represent the smallest
fractional factorials and are used
for most common experiment
designs.
Robust Design Strategy
CONTROL FACTORS are typically
those that the engineer is able to set
under real conditions.

NOISE FACTORS vary uncontrollably


in practice (though within a
predictable range).
To make products and processes
insensitive to the influence of
uncontrollable (noise) factors,
Taguchi incorporates a formal
way to include noise factors in
the experiment layout and this is
called OUTER ARRAY DESIGN.
Loss Function
The mathematical formula
associated with the concept of
the loss function proposed by
Taguchi allows a simple way to
quantify the improvements in
monetary units.
Signal-to-Noise (S/N)
Analysis
For analysis of results from
multiple-sample tests, use of
SIGNAL-TO-NOISE RATIOS
instead of the results makes
the analysis of DOE results
much easier.
Reference:
A primer on the Taguchi Method by Ranjit K. Roy

You might also like