1 Introduction to Quality Engineering

Quality Engineering is designed to generate

quality processes. Based on the purposes of the
Quality engineering, TAGUCHI developed a
approach to experimental design with the objective
to reduce the costs arising from experimentation,
this approach is more practical than theoretical and is
more interested in productivity and costs of
production that according to statistical rules.
The concepts of these techniques are based on the
relationships of costs y savings.
There are some noise factors that affect the
processes, and they are those that cause a
functional characteristic deviates from a target value,
these are causes of variability and loss of
quality.
According to TAGUCHI, this loss of quality
constitutes a long-term waste of time and money
both for the consumer and for the manufacturer.
Within the quality control activities, the
Quality engineering consists of the activities
aimed at reducing variability and
losses.
1.2 General factorial experiment

The results of the ANOVA for two factors can be

extended to a general case where a are the levels
of factor A, b are the levels of factor B, c are the
factors of level C, and so on, which
they can be arranged in a factorial experiment, in the
which the number of replicas is n.
It is designed to generate quality processes.
TAGUCHI developed an approach to design
experiments aimed at reducing costs
emanating from experimentation, this approach is
more practical than theoretical and is more interested in the
productivity and the production costs that in the
statistical rules. The concepts of these techniques
they are based on cost and savings relationships.
Design a manufacturing system to produce a
the product requires technical knowledge as well
of a great experience in the area to which it belongs
product.
Factorial designs are widely used in
experiments involving several factors
to study the combined effect of these on a
response. There are several special cases of design
factorial general that are important because they
they are widely used in research work, and
because they form the basis for other larger designs
value practical.
In recent years, there has been a growing
interest in some of Professor Genechi's ideas
Taguchi about experimental design and its application
the improvement of the quality
The fractional factorial design 2 k-p is used in
scrutiny experiments to quickly identify
and efficiently the subset of factors that
are assets, and to obtain some information about the
interaction. The projection property of these
designs make it possible in many cases to examine the
active factors in more detail. The combination
sequence of these designs through folding is
a very effective way to obtain extra information
about the interactions, which can be identified
in an initial experiment as potentially
important.
1.2.1 General Factorial Design 2k

Factorial designs are widely used in

experiments involving multiple factors
to study the combined effect of these on a
response. There are several special cases of the design
factorials are important because . they
they are widely used in research work, and
because they form the basis for other large designs
value practical.
The most important of these special cases occurs
when there are k factors, each with two
levels. These levels can be quantitative such as
it would be the case of two values of temperature pressure or
time. They can also be qualitative as would be the
case of two machines, two operators, the levels
superior and inferior of a factor, or perhaps, the absence
o presence from a factor.
A complete replica of such a design requires that we
collect 2 x 2 x .... x 2 = 2observations
k and it
know how design general 2k.
The second special case is that of k factors with three.
levels each one, known as factorial design 3k.
If supposes what:
a) the factors son fixed
b) the designs are completely random
c) the usual assumption of normality is satisfied
The design 2kit is particularly useful in the early phases
of the experimental work, when it is likely that there will be
many factors for investigate.
It involves the least number of runs with which
k factors can be studied in a factorial design
complete. Because there are only two levels for each
factor, it should be assumed that the response is
approximately linear in the range of levels
chosen from the factors.
DESIGN 22

The first design of the series 2kit is that which has only
two factors, A and B, each with two
levels. Arbitrarily, the levels of the factor can
to be called 'inferior' and 'superior'.
DESIGN 23

Assume that three factors A are under study,

B and C, each with two levels. This design is known
as factorial design, 23and the eight combinations of
treatments can be graphically represented
through a cube.
There are actually three distinct notations that...
they are widely used for the runs or executions in the
design 2k:
a. The first is the notation "+,-"called "geometric".
b. The second consists of the use oflowercase lettersto identify thecombinations of
treatments.
c. In the third one, thedigits 1 and 0to denote thehigh and low levels of the factor,
respectively.
 1.3 General Factorial Design 3k

This design is a variation of design 2.kand they are very

tools like those used when everyone
factors they act a three levels.
In recent years, there has been a growing
interest in some of Professor Genechi's ideas
Taguchi about experimental design and its application
to the improvement of the quality.
This is a design consisting of k factors with three
levels each one. The factors and the interactions are
represented by capital letters. The three
levels of factors can be referred to as level
inferior, intermediate, and superior. These levels
represented by the digits 0 (lower level), 1
(intermediate) y 2 (superior).
Each treatment combination of a 3 design if k

present through k digits, where the first one affects the

level A, the second pointed to level B, ..... and the k-
th digit the level del factor k.
For example, it is a design 32the 00 represents the
combination of treatments, in which both factor A
as the B is at the lower level, and the 01 represents the
treatment combination corresponding to the factor
A at the lower level and B at the intermediate level.
In this, the notation system that is preferred to use is
the de + - by virtue of facilitating interpretation
geometric design and what it directly is
applicable to regression modeling, the formation of
blocks and the construction of fractional factorials.
The addition of a third level allows for modeling with a
quadratic relation the relationship between the response and each
factor.
DESIGN 32

The simplest design is 32which consists of two factors

with three levels each one.
As there are 32There are 9 combinations of treatments.
8 degrees of freedom between them, the main effects
A and B have two degrees of freedom each, and the
Interaction AB has four degrees of freedom. If there are n
there will be a total of n3 replicas21 degree of freedom,
corresponding to error 32(n-1) degrees of
freedom.
DESIGN 33

If it is assumed that three factors (A,

B, C) and that each factor has three arranged levels
In a factorial experiment. This is a design 33. The
27 combinations have 26 degrees of freedom.
FACTOR A
Under Medium High
0 1 2
Under 0 0 10 20
FACTOR B Medium 1
1 11 21
High 2 2 12 22

1.4 Process Design

Design a manufacturing system to produce a

the product requires technical knowledge as well
from extensive experience in the area to which it belongs
product.
Mr. TAGUCHI defines quality as follows.
THE QUALITY OF A PRODUCT IS THE MINIMUM LOSS IMPARTED BY THE
PRODUCT TO THE SOCIETY, FROM THE MOMENT IT IS LOADED.

The manufacturer is the one who feels the losses the most because
to the negative reaction of the consumer to a product
of poor quality.
FACTORIAL EXPERIMENTS

Many experiments require the study of the

effects of 2 or more factors. In general, the
factorial experiments are the most efficient for
this type of analysis. In a factorial experiment, it
measure at each complete stage or replica of the experiment,
all possible combinations of the levels of the
factors.
When the factors are arranged in an experiment
factorial, it is often said that they are crossed. The
effect of a factor is defined as the change in the
response produced by a change in the level of
factor. This is often called a main effect
why it refers to the primary factors of interest in
the experiment.
For example, consider the data that is represented in
the following table:

Factor A

The main effect of factor A could be calculated.

like the difference between the average response of the
first level of A and the average of the response in the
second level of A; this
A = 40 + 52/2 - 20 + 30/2 = 21

This value is interpreted as the increase of factor A.

from level 1 to level 2 causes an average response
increase by 21 units. In the same way, the
main effect of B. It is calculated as follows:

B = 30 + 52/2 - 20 + 40/2 = 11

In some experiments, it is found that the difference

in the response, between the levels of a factor is not the
same at all levels of the other factors;
when this happens it is because there is an interaction
between the factors.
For example, using the data from the following table,
calculate the effect of A at the 1st level of factor B which
if carry out of the next way
A = 50-20 =30
And the effect of A on the second level of factor B is:
A = 12-40 = -28
Since the effect of A depends on the selected level
of factor B, there is an interaction between A and B

Factorial Experiment with Interaction

FACTOR B
B1 B2
FACTOR
A1 20 40
A
A2 50 12
Graphically, the interaction can be represented.
like your absence; using the data from the tables
Previous ones, two graphs will be constructed to analyze the
analyzed concepts.

Factorial Experiment without interaction

Factorial experiment with interaction

Hypothesis testing is contributed to and approved in the following way:

For the effect of factor A:

Ho: (i = 0
Hi: (i = 0 at least for one i

Fo = MSA
MSAB F(, (a-1), (a-1) (b-1)

For the effect of factor B:

Ho: (2( = 0
Hi (2( ( 0 at least for one (
Fo = MSB
MSE F(, (b- l), ab (n-1)

For the purpose of the interaction:

Ho: (2(( = 0
Hi: (2(( = 0
Fo = MSAB
MSE F( , (a-1) (b-1), ab (n-1)

1.5 Design of fractional factorial experiments

The number of factors in a factorial design 2k ó

3k increase the number of runs required for
make all possible combinations of the levels
of the factors, which quickly consumes the
resources available for researchers to
carry out the experiments.
A complete replica of a design 26requires 64
runs to analyze all possible combinations
of the treatments; in this design only 6 of the 63
degrees of freedom correspond to the effects
mainly, only 15 degrees of freedom correspond to
the interactions of two factors. The remaining 42
degrees of freedom are associated with the
interactions of 3 or more factors. In the series 3kthe
the situation is worse, for example, factorial 36requires
243 runs, and only 12 of the 242 degrees of freedom
correspond a the effects main.
These designs factorials fractionated son
widely used in industrial research. A
important use of the factorial fractionated is in the
research experiments, which are
generally carried out in the initial stages of a
project, and most of the factors initially
often considered have little or no
effect on the response of the analyzed variable. The
factors that are identified as important are
then investigated more deeply into
subsequent experiments.
1.5.1 Fractional Factorial Design 2k-p

A fractional factorial design 2kwhat contains 2k-

pessays are known as fraction 1/2pof the design

factorial 2kor simply fractional factorial design

2k-pThese designs require that p be chosen.
generators independents.
The relationship that defines these designs consists of the p
originally chosen generators and their 2P - P - 1
generalized interactions. The structure of the aliases
it can be determined by multiplying each effect, modulus
2, for the relationship that defines design. It must be taken into account.
be careful when choosing the generators so that the effects
of potential interest are not aliases of each other. Each effect
has 2P-1 alias.
Typically, they are considered negligible.
higher-order interactions (those of third, fourth or
higher order) when the values of k are
moderately large. This simplifies a lot the
structure of the aliases.
1.6 Fractional Factorial Design 3k-p

A greater subdivision is often desired of the

design 3k for moderate to large values of k. In
general, a fraction (1/3) P of a
design 3kfor p < k, where the fraction 3k-2 it is the fraction
(1/9), the 3k-3 it consists of selecting p components of
interaction, and use these effects to decompose the
3 k combinations of treatments of 3pblocks. Each
block now constitutes a fractional factorial design
3k-p.
The relation 1 defines any fraction, consists of the p
initially chosen effects and their (3p-2p-1) / 2
generalized interactions. The aliases of any
main effect or interaction component
determine by multiplying the effect by 1 and 12module 3.
Example:

THE MAXIMUM VOLTAGE OUTPUT OF A TYPE

SPECIFIC TO BATTERY, IT IS THOUGHT THAT IT MAY
BE INFLUENCED BY THE MATERIAL USED IN
The dishes and the temperature in the
LOCATION WHERE THE BATTERY IS
PLACED. FOUR REPLICAS ARE MADE IN THE
EXPERIMENT IN A FACTORIAL EXPERIMENT,
FOR THREE TYPES OF TEMPERATURE AND THREE
MATERIALS. THE RESULTS ARE:
FACTOR (B)
TEMPERATURE
1 130 155 30 40 20 70 998
74 180 80 75 82 58
2 150 188 136 122 25 70 1300
FACTOR (A) 159 126 106 115 58 45
MATERIAL
3 138 110 174 120 96 104 1501
168 160 150 139 82 60
1738 1291 770 3799

FIRST, THE SUM OF THE SQUARES OF IS DONE

ALL SAMPLES, EXCEPT THE TOTAL SUM
OF ROWS AND COLUMNS, BETWEEN THE
MULTIPLICATION OF ROWS OF COLUMNS * #
SAMPLES
n = 4A=3B=3

THEN THE SUM OF THE TOTALS IS MADE TO THE

SQUARE OF THE FACTOR, A:

NOW THE SUM OF THE TOTALS IS MADE TO THE

SQUARE OF THE FACTOR, B:

NOW WE SUM UP THE FOUR SAMPLES OF EACH

COMBINATION AND WE EVALUATE THEM THE SAME. THAT IS THE
SUM OF SQUARES OF THE INTERSECTION AB
EXAMPLE: 130+155+74+180 =539

Finally, the sum of the squares of

THE ERRORS

THE MEAN SQUARES MUST BE TAKEN OUT OF THE

TWO FACTORS AS WELL AS THE INTERACTION AND
ERRORS IN THE FOLLOWING MANNER:
RESULTING IN THE ANOVA TABLE OF THE FOLLOWING
WAY:

SOURCES OF SUM OF DEGREES OF

CUAD.MEDIA FO
VARIATION SQUARES FREEDOM

TYPE OF MAT. 10683.72 2 5341.86 7.91 is greater than 3.35

28.97 >
TEMPERATURE 39118.72 2 19559.36
3.35

INTERACTION 9613.77 4 2403.44 3.56 is greater than 2.73

ERROR 18230.75 27 675.21

TOTAL 77646.96 35

The GRAPHIC DEL MATERIAL TYPE -

THE TEMPERATURE IS THE FOLLOWING:
TYPE 2 MATERIAL BEHAVES MORE
STABLE. PROVIDES BETTER VOLTAGE TO THE DIFFERENT
TEMPERATURES.

ANOVA TABLE FOR 2 FACTORS

FIXED EFFECTS MODEL
SOURCE OF SUM OF DEGREES OF SQUARES
FO
VARIATION SQUARES FREEDOM MEDIA

A SSA a-1 MSA MSA/MSE

B SSB b-1 MSB MSB/MSE

INTERACTION SSAB (a-1) 8b-1) MSAB MSAB/MSE

ERROR SSE ab (n-1) MSE

TOTAL SST abn-1

CONCLUSIONS:

a. THE DATA REFLECTS SUFFICIENT EVIDENCE THAT THE TYPE OF MATERIAL

IT AFFECTS THE OUTPUT VOLTAGE, CONSIDERED AT THE LEVEL OF SIGNIFICANCE
5% off
b. THE DATA PRESENTS SUFFICIENT EVIDENCE THAT THE TEMPERATURES
USADA -5 IN THE EXPERIMENTS IF THEY AFFECT THE OUTPUT VOLTAGE,
CONSIDERING A SIGNIFICANCE LEVEL OF 5%
c. THE DATA SHOWS SUFFICIENT EVIDENCE OF THE TYPE OF
MATERIAL USED AND THE CONSIDERED TEMPERATURES IN THE
EXPERIMENT HAS A JOINT EFFECT ON THE OUTPUT VOLTAGE
CONSIDERED k A SIGNIFICANCE LEVEL OF 5%