UNIT III

Statistical Process Control

Introduction
Best technical tool for improving products and service
quality is Statistical Process control.
There are seven basic tools
1. Histogram and stem-and-leaf plot
2. Check sheet
3. Pareto chart
4. Cause-and-effect diagram
5. Defect concentration diagram
6. Scatter diagram
7. Control chart

This four does not
involve Statistical, so
word statistical is
somewhat Misnomer
Descriptive statistics are concerned with
describing the characteristics of frequency
distributions
Where is the center?
What is the range?
What is the shape [of the distribution]?

Frequency Distributions
Simple depiction of all the data
Graphic easy to understand
Problems
Not always precisely measured
Not summarized in one number or datum
Frequency Table
Test Scores
Observation Frequency
65 1
70 2
75 3
80 4
85 3
90 2
95 1
Frequency Distributions
4

3

2

1
65 70 75 80 85 90 95
65 70 75 80 85 90 95
Frequency
Test Score
Normally Distributed Curve
Characteristics of the Normal
Distribution
It is symmetrical -Half the cases are to one side of the
center; the other half is on the other side.
The distribution is single peaked, not bimodal or multi-
modal
Most of the cases will fall in the center portion of the curve
and as values of the variable become more extreme they
become less frequent, with outliers at each of the
tails of the distribution few in number.
It is only one of many frequency distributions but the one
we will focus on .
The Mean, Median, and Mode are the same.
Percentage of cases in any range of the curve can be
calculated.
Measures of Central Tendency
What is a measure of central tendency?
Measures of Central Tendency
Mode
Median
Mean
Shape of the Distribution
Considerations for Choosing an Appropriate
Measure of Central Tendency

What is a measure of Central
Tendency?
Numbers that describe what is average or typical of
the distribution

You can think of this value as where the middle of a
distribution lies.

The Mode
The category or score with the largest frequency (or
percentage) in the distribution.

The mode can be calculated for variables with levels
of measurement that are: nominal, ordinal, or
interval-ratio.

The Mode: An Example
Example: Number of Votes for Candidates for
Mayor. The mode, in this case, gives you the
central response of the voters: the most popular
candidate.

Candidate A 11,769 votes The Mode:
Candidate B 39,443 votes Candidate C
Candidate C 78,331 votes

The Median
The score that divides the distribution into two
equal parts, so that half the cases are above it
and half below it.

The median is the middle score, or average of
middle scores in a distribution.

Median
Middle-most Value
50% of observations are above the Median,
50% are below it
The difference in magnitude between the
observations does not matter
Therefore, it is not sensitive to outliers
Formula Median = n + 1 / 2

To compute the median
- first you rank order the values of X from low to high:
85, 94, 94, 96, 96, 96, 96, 97, 97, 98
then count number of observations = 10.
add 1 = 11.
divide by 2 to get the middle score the 5

score
here 96 is the middle score


Median
Find the Median
4 5 6 6 7 8 9 10 12
Find the Median
5 6 6 7 8 9 10 12
Find the Median
5 6 6 7 8 9 10 100,000

The Mean
The arithmetic average obtained by adding up
all the scores and dividing by the total number
of scores.
Formula for the Mean
N
Y
Y

=
Y bar equals the sum of all the scores, Y, divided by the
number of scores, N.
Calculating the mean with grouped
scores
where: f Y

= a score multiplied by its frequency

N
Y f
Y

=
Mean: Grouped Scores
Mean: Grouped Scores
Shape of the Distribution
Symmetrical (mean is about equal to median)
Skewed
Negatively (example: years of education)
mean < median
Positively (example: income)
mean > median
Bimodal (two distinct modes)
Multi-modal (more than 2 distinct modes)
Distribution Shape
Dispersion
Once you determine that the variable of interest is
normally distributed, ideally by producing a
histogram of the scores,
the next question to be asked about the NDC is its
dispersion: how spread out are the scores
around the mean.
Dispersion is a key concept in statistical thinking.
The basic question being asked is how much do
the scores deviate around the Mean?
The more bunched up around the mean the
better your ability to make accurate predictions.

Means
Consider these means for weekly candy bar
consumption.


X = {7, 8, 6, 7, 7, 6, 8, 7}
X = (7+8+6+7+7+6+8+7)/8
X = 7
X = {12, 2, 0, 14, 10, 9, 5, 4}
X = (12+2+0+14+10+9+5+4)/8
X = 7
What is the difference?


How well does the mean represent the
scores in a distribution?

The logic here is to determine how much spread is in the
scores. How much do the scores "deviate" from the
mean? Think of the mean as the true score or as your
best guess. If every X were very close to the Mean, the
mean would be a very good predictor.

If the distribution is very sharply peaked then the mean is
a good measure of central tendency and if you were to use
the mean to make predictions you would be right or close
much of the time.


What if scores are widely distributed?
The mean is still your best measure and your best
predictor, but your predictive power would be
less.

How do we describe this?
Measures of variability
Mean Deviation
Variance
Standard Deviation


Mean Deviation
The key concept for describing normal distributions
and making predictions from them is called
deviation from the mean.
We could just calculate the average distance between each
observation and the mean.
We must take the absolute value of the distance, otherwise
they would just cancel out to zero!
Formula:
| |
i
X X
n

Mean Deviation: An Example

1. Compute X (Average)
2. Compute X X and take the
Absolute Value to get
Absolute Deviations
3. Sum the Absolute
Deviations
4. Divide the sum of the
absolute deviations by N
X Xi Abs. Dev.
7 6 1
7 10 3
7 5 2
7 4 3
7 9 2
7 8 1
Data: X = {6, 10, 5, 4, 9, 8} X = 42 / 6 = 7
Total: 12
12 / 6 = 2
What Does it Mean?
On Average, each observation is two units away
from the mean.
Is it Really that Easy?
No!
Absolute values are difficult to manipulate
algebraically
Absolute values cause enormous problems for calculus
(Discontinuity)
We need something else
Variance and Standard Deviation
Instead of taking the absolute value, we square the
deviations from the mean. This yields a positive
value.
This will result in measures we call the Variance and
the Standard Deviation
Sample- Population-
s: Standard Deviation : Standard Deviation
s
2
: Variance
2
: Variance
Calculating the Variance and/or Standard
Deviation
Formulae:

Variance:







2
( )
i
X X
s
N

2
2
( )
i
X X
s
N

Standard Deviation:
QUALITY TOOLS
35
Histograms
What is it?
A Histogram is a bar graph
usually used to present frequency data
LSL
USL
How does it Work?
Define Categories for Data
Collect Data, sort them into the categories
Count the Data for each category
Draw the Diagram. Each category finds its place on the x-Axis.
The bars will be as high as the value for the category
What is its use?
Histograms provide an easy way to evaluate the distribution of
Data over different categories
36
Interpretations


The histogram is an effective, practical working tool in the early
stages of data analysis. A histogram may be interpreted by
asking three questions:
How well is the histogram centered? The centering of the data
provides information on the process aim about some mean or
nominal value.
How wide is the histogram? Looking at histogram width defines
the variability of the process about the aim.
What is the shape of the histogram? Remember that the data is
expected to form a normal or bell-shaped curve. Any significant
change or anomaly usually indicates that there is something
going on in the process, which is causing the quality problem.
37
Typical Distributions


Depicted by a bell-shaped curve
Most frequent measurement appears as center of
distribution
Less frequent measurements taper gradually at
both ends of distribution
Indicates that a process is running normally (only
common causes are present).
Normal
38
Typical Distributions


Distribution appears to have two peaks
May indicate that data from more than one process
are mixed together
o Materials may come from two separate
vendors
o Samples may have come from two separate
machines.
BIMODAL
39
Typical Distributions


Appears to end sharply or abruptly at one end
Indicates possible sorting or inspection of non-
conforming parts.
CLIFF-LIKE
40
Limitations of Histograms


Histograms are limited in their use due to the random
order in which samples are taken and lack of
information about the state of control of the process.
Because samples are gathered without regard to
order, the time-dependent or time-related trends in
the process are not captured.
This lack of information on process control may lead
to incorrect conclusions being drawn and, hence,
inappropriate decisions being made.
Still, with these considerations in mind, the histogram's simplicity of
construction and ease of use make it an invaluable tool in the
elementary stages of data analysis.
Mode and Median
Stem and Leaf Diagram
43
Run Charts
What is it?
Run Charts are representing change
in measurement over a sequence or time
Time
M
e
a
s
u
r
e
m
e
n
t

How does it Work?
Gather Data
Organize Data
Measurements (y) must be confronted with time or
sequence of the events.
Chart Data
Interpreting Data
What is its use?
Determining Cyclic Events and there average character
44
Run Chart
Example
45
Run Charts
Two ways to misinterpret run charts:
You conclude that some trend or cycle exists, when in fact you
are just seeing normal process variation (and every process will
show some variation).
You do not recognize a trend or cycle when it does exist.
Both of these mistakes are common, but people are generally less aware
that they are making the first type, and are tampering with a process,
which is really behaving normally.

To avoid mistakes, use the following rules of thumb for run chart
interpretation:

46
Run Charts

Look at data for a long enough period of time, so
that a "usual" range of variation is evident.

Is the recent data within the usual range of variation?

Is there a daily pattern? Weekly? Monthly? Yearly?

47
Pareto Charts
What is its use?
Pareto Charts are used to apply the 80/20
rule of Joseph Juran which states that 80%
of the problems are the result of 20% of
the problems. A Pareto Chart can be used
to identify that 20% root causes of
problem.
Type
X
48
Pareto Charts
Decide the collection period
Identify the main problem causes or
categories
Collect data on check sheet or tally sheet
Tabulate the frequency of each category
List them in descending order
Arrange the data as in a bar chart
Determine cumulative totals and %
49
Pareto Charts
In most cases, two or three categories will tower
above the others. These few categories, which
account for the bulk of the problem, will be the high-
impact points on which to focus. If in doubt, follow
these guidelines:

Look for a break point in the cumulative
percentage line. This point occurs where the slope
of the line begins to flatten out. The factors under
the steepest part of the curve are the most
important.
Type
X
50
Pareto Charts

If there is not a fairly clear change in the
slope of the line, look for the factors that
make up at least 80% of the problem.

If the bars are all similar sizes or more than
half of the categories are needed to make
up the needed 80%, try a different
breakdown of categories that might be
more appropriate.
Type
X
Pareto
Chart
Box Plot

A box plot summarizes data using the median,
upper and lower quartiles, and the extreme
(least and greatest) values. It allows you to see
important characteristics of the data at a glance
53
How does it Work?
Determine what Process or Procedure you want
to represent.
Start at a certain point and go then step by
step using flow chart symbols
Document the elements with titles. Let it close
with an ending point.
What is its use?
A Flow chart lets a process or procedure be
understood easily. It also demonstrate the
relationships between the elements.
Flow Charts
Input
Within
Spec?
Process Output
adjust
Yes
No
54
The most common symbols
Diamond - A choice
between two or more
alternatives
Arrow - Connects two or
more symbols. The
diamond is the only symbol
that has more than two
arrows connected to it
Start - stop
Where the process
starts and ends
Box
A symbol for
action steps.
The action is
spelled out in
the box
Flow Charts
55
Output Input Processing
Flow Charts
56
Flowchart - next level down
Input
Output
Processing
57
Flowchart - one more step down
Dept. 1 Dept. 3
Operation 1
Storage
Inspection
Rework
Operation 2
No
Yes
Processing in Dept. 2
Dept. 1
Dept. 3
58
How does it Work?
Determine what Process or Procedure you want
to represent.
Start at a certain point and go then step by
step using flow chart symbols
Document the elements with titles. Let it close
with an ending point.
What is its use?
A Flow chart lets a process or procedure be
understood easily. It also demonstrate the
relationships between the elements.
Flow Charts
Input
Within
Spec?
Process Output
adjust
Yes
No
59
Example:
You intend to repair a certain
machine.
First you perform the repair
thought to be necessary
Then You check it
If it does not work you continue
with repairs
If it works you finish
Flow Charts
start
Repair machine
OK?
end
Yes
No
Check machine
60
Scatter Diagrams
What is it?
Statistical tool showing a trend in a series of values.
How does it Work?
Draw graph with value points
Draw trend line: m*x+a
Calculate m value
Calculate a value
Calculate points for trend line.
What is its use?
Demonstrating correlations between values and showing
trends for value changes.
Y
X
61


Y
X
Scatter Diagrams
62


Scatter Diagrams
63
Cause and Effect Diagrams
What is it?
Its a diagram that demonstrates the relationship
between Effects and the categories of their causes
The Arrangement of the Diagram lets it look like a
fishbone it is therefore also called fish-bone
diagram
How does it Work?
Determine the Effect or Problem you would like to
examine
Categorize the possible causes and find
subcategories
Describe the possible causes
effec
t
Cause b
Cause a
Cause d Cause c
64
What is its use?

Enables a team to focus on the content of a
problem

Focuses the team on causes, not symptoms

It is an effective tool that allows people to easily see
the relationship between factors to study processes,
situations, and for planning.

Cause and Effect Diagrams
65
Causes in a cause & effect diagram are frequently arranged
into four major categories. While these categories can be
anything:

Manpower, methods, materials, and machinery
(recommended for manufacturing)

Equipment, policies, procedures, and people
(recommended for administration and service).
Cause and Effect Diagrams
66
To successfully build a cause and effect diagram:

1. Be sure everyone agrees on the effect or problem statement before
beginning

2. Be concise.

3. For each node, think what could be its causes. Add them to the tree.

4. Pursue each line of cause back to its root cause.

5. Consider grafting relatively empty branches onto others.

6. Consider splitting up overcrowded branches.

7. Consider which root causes are most likely to merit further
investigation.
Cause and Effect Diagrams
67
EFFECT
Main Cause -1 Main Cause -2
Main Cause -3 Main Cause -4
EFFECT
1. Draw an arrow horizontally pointing to an effect
2. Draw four or more branches off the large arrow to represent main categories of potential causes.
Typical categories are man, machinery, methods, and materials.
Cause and Effect Diagrams
68
3. Secondary causes can be listed on branches off the main category branches.
EFFECT
Main Cause -1 Main Cause -2
Main Cause -3 Main Cause -4
Secondary Cause
Cause and Effect Diagrams
Main Cause
70
What is it?
Statistical tool, showing whether a process is in
control or not
How does it Work?
Define Upper limit, lower limit and Center line
Draw Chart.
Gather values and draw them into chart
What is its use?
Taking samples of a process and detect possibility of process
being out of control
Y
X
Upper limit
Lower limit
Average/Spec
Control Charts
71
Control Charts
Run charts turn into control charts
One of the single most effective quality control
devices for managers and employees
Y
X
Upper limit
Lower limit
Average/Spec
Control Charts
72
Periodic tracking of a process
Common types
X bar, R or range, p or percent nonconforming
Elements of a control chart
upper control limit (UCL), the highest value a
process should produce
central line (x bar), the average value of
consecutive samples
lower control limit (LCL), the lowest value a
process should produce

Y
X
Upper limit
Lower limit
Average/Spec
Control Charts
73
Constructing a control Chart
Decide what to measure or count
Collect the sample data
Plot the samples on a control chart
Calculate and plot the control limits on the control
chart
Determine if the data is in-control
If non-random variation is present, discard the data
(fix the problem) and recalculate the control limits
Y
X
Upper limit
Lower limit
Average/Spec
Control Charts
74
Y
X
Upper limit
Lower limit
Average/Spec
Y
X
Upper limit
Lower limit
Average/Spec
Control Charts
75
A Process is In Control if:
1. No sample points are outside control limits

2. Most points are near the process average

3. About an equal # points are above & below the
centerline

4. Points appear randomly distributed
Y
X
Upper limit
Lower limit
Average/Spec
Control Charts
76
Upper
control limits
Lower
control limits
Centre Line
Stable process
Special cause
Unstable process
Common cause
Upper
control limits
Lower
control limits
Centre Line
Stable process
Special cause
Unstable process
Common cause
Control Charts
Normal bell shaped curve
78
Potential Reasons of Variations
The Operator
Training, supervision, technique
The Method
Procedure, Set-up, temperature, etc.
The Material
Moisture content, blending, contamination
The Machine
Machine condition, inherent precision
Management
Poor Process management, Poor System
79
Causes of Variations
Common Causes : Improvement if system
Variation inherent in process
Can be eliminated only through improvement in system
No pattern

Assignable Causes : Control of Process
Variation due to identifiable factors
Can be modified through operator or management action
May exhibit a pattern
Examples of special causes include: wrong tool, wrong production method,
improper raw material, operators skill, wrong die etc.
SPC uses samples to identify that Assignable causes have occurred
80
Control Charts:
Recognizing Sources of Variation
Variables
Things we measure
Attributes
Things we count
Length Blood Pressure Diameter
Weight Volume Tensile strength
Time Temperature
Number or percent defective items
Number of defects per item
Types of defects etc.
81
Control Charts: Types
Control Charts for Variables Data
_
X and R charts : for sample averages and ranges
_
X and s charts : for sample averages and standard deviations

Md and R charts : for sample medians and ranges

X and R
m
charts : for individual measures and moving ranges
Control Charts for Attributes Data
p charts : proportions of units nonconforming

np charts : number of units nonconforming

c charts : number of defects/nonconformities

u charts : number of defects/nonconformities per unit
82
Control Charts for Variable Type of Data
and R charts
In the x bar chart the sample means are plotted in order to
control the mean value of a variable.

In R chart, the sample ranges are plotted in order to control
the variability of a variable
83
Control Charts for Variable Type of Data
X X i i
i
R
) min( ) ( max
=
g
R
R
g
i
i
=
=
1
R
i
= range of i
th
sample
X
max(i)
= maximum value of the data in i
th
sample
X
min (i)
= minimum value of the data in i
th
sample
= mean of g samples
Centre line, upper, & lower control limit for R charts are calculated.
The formulae used are as following:
R D UCL
R 4
=
R D LCL
R 3
=
(Upper control limit for R chart)
(Lower control limit for R chart)
Where D
3
and D
4
are
constants and their values
can be obtained from
standard tables. These
values depend upon the
size of sample.
R
84
Control Charts for Variable Type of Data
n A
2
D
3
D
4

2 1.880 0 3.267
3 1.023 0 2.575
4 0.729 0 2.282
5 0.577 0 2.115
6 0.483 0 2.004
7 0.419 0.076 1.924
8 0.373 0.136 1.864
9 0.337 0.184 1.816
10 0.308 0.223 1.777
Factors for Calculating Limits for and R Charts
85
Control Charts for Variable Type of Data
= mean of g samples
Centre line, upper, & lower control limit for x bar chart are calculated.
The formulae used are as following:
(Upper control limit for X bar chart)
(Lower control limit for X bar chart)
Where A
2
is
constants and its
value can be
obtained from
standard table. Its
value depend upon
the size of sample.
i
X
n
n
i
i X

=
=
1


= mean of the ith sample
n = sample size,
X
i
= ith data
i
X
CL
X
x
g
i
i
g
= =

=1
X
X
R A X UCL
x
2
+ =
R A X LCL
x
2
=
86
Control Charts for Variable Type of Data
Example
Mean values and ranges of data from 20 samples (sample size = 4) are shown in the
table below:


S.N
Mean
of
Sample
Range S.N
Mean
of
Sample
Range S.N
Mean
of
Sample
Range S.N
Mean
of
Sample
Range S.N
Mean
of
Sample
Range
1 10 4 5 9 5 9 10 4 13 12 4 17 12 4
2 15 4 6 11 6 10 11 6 14 12 3 18 15 3
3 12 5 7 11 4 11 12 5 15 11 3 19 11 3
4 11 4 8 9 4 12 13 4 16 15 4 20 10 4
Average of Ranges of 20 samples =
20
20
1

= i
R
Upper Control Limit of R chart = D
4
*4.15 (D
4
= 2.282 for sample size 4)
= 9.47 ~ 9.5
Lower Control Limit of R chart = D
3
*4.15 (D
3
= 0 for sample size 4) = 0
= 4.15 (Center Line of R Chart)
87
Control Charts for Variable Type of Data
Example
Mean values and ranges of data from 20 samples (sample size = 4) are shown in the
table below:


S.N
Mean
of
Sample
Range S.N
Mean
of
Sample
Range S.N
Mean
of
Sample
Range S.N
Mean
of
Sample
Range S.N
Mean
of
Sample
Range
1 10 4 5 9 5 9 10 4 13 12 4 17 12 4
2 15 4 6 11 6 10 11 6 14 12 3 18 15 3
3 12 5 7 11 4 11 12 5 15 11 3 19 11 3
4 11 4 8 9 4 12 13 4 16 15 4 20 10 4
Upper Control Limit of x bar chart = 11.6 + A
2
*4.15 = 14.63 (A
2
= 0.729 for sample size 4)

Lower Control Limit of x bar chart = 11.6 - A
2
*4.15 = 08.57 (A2 = 0.729 for sample size 4)
Average of mean values of 20 samples =
= 11.6 (Center Line of x bar Chart)
20
20
1

= i
X
88
Control Charts for Variable Type of Data


0
2
4
6
8
10
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
4
6
8
10
12
14
16
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
X-Bar Chart
R Chart
All the data are within
the LCL and UCL in R
Chart.
Hence variability of the
process data is not an
issue to worry.
Sample data at S.N 2,
16, and 18 are slightly
above the UCL. Efforts
must be made to find
the special causes and
revised limits are
advised to calculate
after deleting these
data.
89
Control Charts for Variable Type of Data
Suppose we have g preliminary samples at our disposition,
each of size n, and let s
i
be the standard deviation of the i
th
sample.
Then the average of the g standard deviations is

S B UCL
R 4
=
(Upper control limit for s chart)
S B LCL
R 3
=
(Lower control limit for s chart)

=
g
i
i
s
g
S
1
1
(Center Line for s chart)
and s charts
Centre line, upper, & lower control limit for s charts are calculated.
The formulae used are as following:
Where B
3
and B
4

are constants and
their values can be
obtained from
standard tables.
These values
depend upon the
size of sample.
90
Control Charts for Variable Type of Data
n A
3
B
3
B
4

2 1.880 0 3.267
3 1.023 0 2.575
4 0.729 0 2.282
5 0.577 0 2.115
6 0.483 0 2.004
7 0.419 0.076 1.924
8 0.373 0.136 1.864
9 0.337 0.184 1.816
10 0.308 0.223 1.777
Factors for Calculating Limits for and S Charts
91
Control Charts for Variable Type of Data
= mean of g samples
Centre line, upper, & lower control limit for x bar charts are calculated.
The formulae used are as following:
(Upper control limit for X bar chart)
(Lower control limit for X bar chart)
Where A
3
is
constants and its
value can be
obtained from
standard table. Its
value depend upon
the size of sample.
i
X
n
n
i
i X

=
=
1


= mean of the i
th
sample
n = sample size,
X
i
= i
th
data
i
X
CL
X
x
g
i
i
g
= =

=1
X
X
S A X UCL
x
3
+ =
S A X LCL
x
3
=
92
Control Charts for Variable Type of Data
Example
SN Mean SD SN Mean SD SN Mean SD SN Mean SD SN Mean SD
1 55.6 9.63 7 46.8 6.72 13 44 14.35 19 50.2 7.6 25 44.6 8.96
2 61 8.63 8 44.2 8.53 14 51.6 5.18 20 44 8.46 26 46.8 6.5
3 45.2 7.4 9 50.8 11.95 15 53.2 5.36 21 50 5.15 27 49.2 3.19
4 46.2 4.09 10 48.4 6.19 16 52.4 9.48 22 47 5.15 28 45.6 7.96
5 46.8 7.22 11 51.2 6.83 17 50.6 3.44 23 50.6 5.55 29 57.6 14.38
6 49.8 8.76 12 49.4 5.46 18 56 7 24 48.8 6.5 30 51.4 6.8
Mean values and SD of data from 30 samples (sample size = 5) are shown in the table
below:
Average of Ranges of 30 samples =
Upper Control Limit of s chart = B
4
*7.41 (B
3
= 2.115 for sample size 5)
= 15.48
Lower Control Limit of R chart = B
3
*7.41 (B
4
= 0 for sample size 5) = 0
= 7.41 (Center Line of s Chart)

=
g
i
i
s
g
S
1
1
93
Control Charts for Variable Type of Data
Example
SN Mean SD SN Mean SD SN Mean SD SN Mean SD SN Mean SD
1 55.6 9.63 7 46.8 6.72 13 44 14.35 19 50.2 7.6 25 44.6 8.96
2 61 8.63 8 44.2 8.53 14 51.6 5.18 20 44 8.46 26 46.8 6.5
3 45.2 7.4 9 50.8 11.95 15 53.2 5.36 21 50 5.15 27 49.2 3.19
4 46.2 4.09 10 48.4 6.19 16 52.4 9.48 22 47 5.15 28 45.6 7.96
5 46.8 7.22 11 51.2 6.83 17 50.6 3.44 23 50.6 5.55 29 57.6 14.38
6 49.8 8.76 12 49.4 5.46 18 56 7 24 48.8 6.5 30 51.4 6.8
Mean values and SD of data from 30 samples (sample size = 5) are shown in the table
below:
Upper Control Limit of x bar chart = 49.63 + A
2
*7.41 = 60.21 (A
2
= 1.427 for sample size 5)

Lower Control Limit of x bar chart = 49.63 - A
2
*7.41 = 39.05 (A2 = 1.427 for sample size 5)
= 49.63 (Center Line of x bar Chart)
Average of mean values of 30 samples =
30
30
1

= i
X
94
40
45
50
55
60
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
X Bar chart
S
a
m
p
l
e

M
e
a
n
-2
2
6
10
14
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
S chart
S
a
m
p
l
e

S
D
Sample data at S.N 2, is
slightly above the UCL.
Efforts must be made to find
the special causes and
revised limits are advised to
calculate after deleting these
data.
All the data are within the
LCL and UCL in S Chart.
Hence variability of the
process data is not an issue
to worry.
Control Charts for Variable Type of Data
95


DIFFERENT PATTERNS OF CHARTS AND THEIR
INFERENCES
96
Process In Control
When a process is in control, there occurs a
natural pattern of variation. (only chance
causes will occur)
Natural pattern has:
About 34% of the plotted point in an imaginary
band between 1 o on both side CL.

About 13.5% in an imaginary band between 1 o
and 2 o on both side CL.

About 2.5% of the plotted point in an imaginary
band between 2 o and 3 o on both side CL.
97
Process Out Of Control
The term out of control is a change in the process due to
an assignable cause.

When a point (subgroup value) falls outside its control
limits, the process is out of control.
98
Extreme Variation

Fig:4
99
Shifts(Runs)
When 7 consecutive points lie above or below centre line
shifts occur

Occur due to change in material, operator, inspector,
machine, inspection instrument or carelessness of new
operator
100
Shifts
Fig:5
101
Trends
They may occur due to:

1.Fatigue of worker

2.Accumulation of dirt and waste material

3.Deterioration of environmental conditions
102
Trends
Fig:6
103
Recurring Cycles
Recurring cycles may occur due to:

1.Operators fatigue

2.Recurring effects of temperature and humidity

3.Periodic rotation of operators or machines

4.Gradual change in homogeneity of incoming material
quality
104
Recurring Cycles
Fig:7
105
Hugging (Too Close)
Occur due to

1.Improper selection of the samples

2.Errors in testing equipment or errors in readings

3.Wrong calculation of control limits
106
Hugging

Fig:8
107
Erratic Fluctuations
These may occur due to:

1.Frequent adjustment of machines and their settings

2.Machines, materials, operators, test equipment, etc.
108
Erratic Fluctuations
Fig:9
109
Control Limits And Errors
Type I error:
Probability of searching for
a cause when none exists
(a) Three-sigma limits
UCL
Process
average
LCL
Fig:10
110
Control Limits And Errors
Type I error:
Probability of searching for
a cause when none exists
(b) Two-sigma limits
UCL
Process
average
LCL
Fig:11
111
Control Limits And Errors
(a) Three-sigma limits
Type II error:
Probability of concluding
that nothing has changed
Shift in process
average
UCL
Process
average
LCL
Fig:12
112
Control Limits And Errors
(b) Two-sigma limits
Type II error:
Probability of concluding
that nothing has changed
Shift in process
average
UCL
Process
average
LCL
Fig:13
113
Achieve The Purpose
1. Our goal is to decrease the variation inherent in a
process over time.

2. As we improve the process, the spread of the data will
continue to decrease.

3. Quality improves!!
114
Examine The Process
A process is considered to be stable and in a state of
control, or under control, when the performance of the
process falls within the statistically calculated control
limits and exhibits only chance, or common causes.
115
Process Variation
When a system is subject to only chance
causes of variation, 99.74% of the
measurements will fall within 6 standard
deviations

If 1000 subgroups are measured, 997 will fall
within the six sigma limits.

(according to normal distribution)
116
Chart Zones
Based on our knowledge of the normal curve, a
control chart exhibits a state of control
when:

1. Two thirds of all points are near the center value.
2. The points appear to float back and forth across
the centerline.
3. The points are balanced on both sides of the
centerline.
4. No points beyond the control limits.
5. No patterns or trends.
117
Control Charts for Attribute Type of Data
p charts
p-charts calculates the percent defective in sample. p-charts
are used when observations can be placed in two categories
such as yes or no, good or bad, pass or fail etc.
118
Control Charts for Attribute Type of Data
Centre line, upper, & lower control limit for p charts are calculated.
The formulae used are as following:
(Upper control limit for p chart)
(Lower control limit for p chart)
samples all in items of number total
samples all in defectives of sum
p =
n
p p
p UCL
) 1 (
3

+ =
n
p p
p LCL
) 1 (
3

=
= centre line of p chart

Where n is the sample size. Sample size in p chart must be > 50
Sometimes LCL in p chart becomes negative, in such cases LCL should be taken as 0
119
Control Charts for Attribute Type of Data
Example
Data for defective CDs from 20 samples (sample size = 100) are shown in the table
below:


Sample No. No. of
Defective
CDs = x
Proportion
Defective =
x/sample size
Sample No. No. of
Defective
CDs = x
Proportion
Defective =
x/sample size
1 4 .04 11 6 .06
2 3 .03 12 5 .05
3 3 .03 13 4 .04
4 5 .05 14 5 .05
5 6 .06 15 4 .04
6 5 .05 16 7 .07
7 2 .02 17 6 .06
8 3 .03 18 8 .08
9 5 .05 19 6 .06
10 6 .06 20 8 .08
12 .
100
) 051 . 1 ( 051 .
3 051 . 0
) 1 (
3 =

+ =

+ =
n
p p
p UCL
0 02 .
100
) 051 . 1 ( 051 .
3 051 . 0
) 1 (
3 ~ =

=
n
p p
p LCL
051 . 0
2000
101
= =
allsamples of sum
defectives of sum
CL =
120
Control Charts for Attribute Type of Data


There is important observation that is clearly visible from the data points that there
is an increasing trend in the average proportion defectives beyond sample number
15. Also, data show cyclic pattern. Process appears to be out of control and also
there is strong evidence that data are not from independent source.
P CHART
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
MEANP
UCLP
LCLP
P
121
Control Charts for Attribute Type of Data
np charts
np-charts calculates the number of defective in sample. np-
charts are used when observations can be placed in two
categories such as yes or no, good or bad, pass or fail etc.

It is easier to relate the number of defectives than to the
proportion defective. However, the np charts should not be
used when the sample size varies as the Centre line and the
control limits will vary.
122
Control Charts for Attribute Type of Data
Centre line, upper, & lower control limit for np charts are calculated.
The formulae used are as following:
(Upper control limit for np chart)
(Lower control limit for np chart)
samples number total
defectives of sum
p n =
|
|
.
|

\
|
+ =

p p n p n UCL 1 3
= centre line of np chart
Where n is the sample size. Sample size in np chart must be > 50
Sometimes LCL in np chart becomes negative, in such cases LCL should be taken as 0
|
|
.
|

\
|
=

p p n p n LCL 1 3
123
Control Charts for Attribute Type of Data
Example
Data for the number of defective products from 20 samples (sample size = 300) are
shown in the table below:


Sample No. No. of
Defective
Sample No. No. of
Defective
1 10 11 6
2 12 12 19
3 8 13 10
4 9 14 8
5 6 15 7
6 11 16 4
7 13 17 11
8 10 18 10
9 8 19 6
10 9 20 7
16 . 18 )
300
2 . 9
1 ( 2 . 9 3 2 . 9 ) 1 ( 3 = + = + =

p p n p n UCL
2 . 9
20
184
= =
samples of Number
defectives of sum
CL =
24 . 0 )
300
2 . 9
1 ( 2 . 9 3 2 . 9 ) 1 ( 3 = = =

p p n p n LCL
124
Control Charts for Attribute Type of Data


Sample data at S.N 12, is above the UCL. Efforts must be made to find the special
causes and revised limits are advised to calculate after deleting these data.
-2
0
2
4
6
8
10
12
14
16
18
20
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
D
e
f
e
c
t
i
v
e
s
125
Control Charts for Attribute Type of Data
c charts
c-charts counts the number of defects in an item. c-charts are used only when
the number of occurrence per unit of measure can be counted such as number
of scratches, cracks etc.
126
Control Charts for Attribute Type of Data
Centre line, upper, & lower control limit for c charts are calculated.
The formulae used are as following:
(Upper control limit for c chart)
(Lower control limit for c chart)
= centre line of c chart

samples all in items of number total
samples all in defects of sum
c =
c c UCL 3 + =
c c LCL 3 =
127
Control Charts for Attribute Type of Data
Example
Data for defects on TV set from 20 samples (sample size = 10) are shown in the table
below:


Sample
No.
No. of
Defects
Sample
No.
No. of
Defects
Sample
No.
No. of
Defects
Sample
No.
No. of
Defects
1 5 6 4 11 6 16 5
2 4 7 5 12 5 17 4
3 5 8 6 13 4 18 6
4 6 9 8 14 7 19 6
5 4 10 7 15 6 20 6
45 . 12 45 . 5 3 45 . 5 3 = + = + = c c UCL
0 55 . 1 45 . 5 3 45 . 5 3 = = = = c c LCL
45 . 5
20
109
= =
samples of number
defects of sum
CL =
128
Control Charts for Attribute Type of Data


None of the sample is out of the LCL and UCL. But the chart shows cyclic trend.
0
2
4
6
8
10
12
14
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
D
e
f
e
c
t
s
129
Control Charts for Attribute Type of Data
u charts
u-chart counts the number of defect per sample. The u chart is used when it is
not possible to have a sample size of a fixed size.
130
Control Charts for Attribute Type of Data
Centre line, upper, & lower control limit for u charts are calculated.
The formulae used are as following:
(Upper control limit for u chart)
(Lower control limit for u chart)
= centre line of u chart

i
n
u
u UCL 3 + =
i
n
u
u LCL 3 =
samples all in items of number total
samples all in defects of sum
u =

=
=
k
i
i
k
i
i
n
c
1
1
=

c
i
=number of defects in i
th
sample
k = number of samples
n
i
= size of i
th
samples
131
Example
Data for defects in a product from 20 samples are shown in the table below. The number
of products varies from sample to sample. Construct a control for number of defects per
100 products.


Sample
No.
No. of
products
inspected
No. of
Defects
Sample
No.
No. of
products
inspected
No. of
Defects
Sample
No.
No. of
products
inspected
No. of
Defects
1 200 5 8 150 10 15 100 6
2 300 14 9 150 6 16 200 8
3 250 8 10 250 10 17 200 5
4 150 8 11 300 9 18 100 5
5 250 12 12 250 16 19 300 14
6 100 6 13 200 12 20 200 8
7 200 20 14 250 10
68 . 4
41
192
1
1
= = =

=
=

k
i
i
k
i
i
n
c
u
27 . 9
2
68 . 4
3 68 . 4 = + = UCL
092 . 0
2
68 . 4
3 68 . 4 = = LCL
132


S No. Sample size
n
i

No. of
Defects per
100 product
UCL LCL S No. Sample size
n
i

No. of
Defects per
100 product
UCL LCL
1 2.0 2.50 9.3 0.1 11 3.0 3.00 8.4 0.9
2 3.0 4.67 8.4 0.9 12 2.5 6.40 8.8 0.6
3 2.5 3.20 8.8 0.6 13 2.0 6.00 9.3 0.1
4 1.5 5.33 10.0 0.0 14 2.5 4.00 8.8 0.6
5 2.5 4.80 8.8 0.6 15 1.0 6.00 11.2 0.0
6 1.0 6.00 11.2 0.0 16 2.0 4.00 9.3 0.1
7 2.0 10.00 9.3 0.1 17 2.0 2.50 9.3 0.1
8 1.5 6.67 10.0 0.0 18 1.0 5.00 11.2 0.0
9 1.5 4.00 10.0 0.0 19 3.0 4.67 8.4 0.9
10 2.5 4.00 8.8 0.6 20 2.0 4.00 9.3 0.1
-2
0
2
4
6
8
10
12
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
D
e
f
e
c
t
s
X
X
X
X
X
X
X
X
X X
X
Variation
Which Pilot Would You Choose?
Airplane landing example
X
X
X
X
X
X
X
X
X
X


Six Sigma Process


A Six Sigma process is one that has at least six standard deviations
between the process mean and the nearest specification limit when
centred.
+/- 6 sigma 99.99996% *
Mean Value
Lower Design Specification
Upper Design Specification
*Distribution shifted +/- 1.5 o
1.5 o 1.5 o
Six Sigma as a Metric
1
) (
2

n
x xi

o
Sigma = o = Deviation
( Square root of variance
)
-
7

-
6

-
5

-
4

-
3

-
2

-
1

0

1

2

3

4

5

6

7

Axis graduated in Sigma
68.27 %
95.45 %
99.73 %
99.9937 %
99.9943 %
99.99966 %
317300 ppm outside
(deviation)
45500 ppm
2700 ppm
63 ppm
57 ppm
3.4 ppm
between + / - 1o
between + / - 2o
between + / - 3o
between + / - 4o
between + / - 5o
between + / - 6o
o
Decrease

1

2

3

4

5

6
317,300

45,500

2,700

63

57

3.4
o
Process
Capability
Defects per
Million
Opportunities
Industry
Average
Examples of Sigma
Electric companies if operated at
4 Sigma = no power for 1 hour/week
6 Sigma = no power for 2 seconds/week
Telephone Companies
4 Sigma = No phone service: 4 hours/month
6 Sigma = No phone service: 9 seconds/month
History of Six Sigma





Six Sigma had its birth at Motorola in 1979.

Six Sigma was developed by Mikel Harry.

Within four years, Six Sigma had saved the
company $2.2 billion.



What is Six Sigma ?


Measure of how well a process is performing
- Six Sigma process produces 3.4 defects per
million opportunities
- Most companies are at 3-4 sigma level
Philosophy of reducing defects so we can
improve customer satisfaction and reduce
costs
Business Strategy that increases process
performance resulting in enhanced customer
satisfaction and improved bottom line.

CONCEPT OF SIX SIGMA


Six-sigma is an approach to continually improve and reduce
variations in the process that contribute to waste and poor
quality.
It is a measure of how well a process is performing.
Six sigma is a well structured, data driven methodology for
eliminating defects, waste, or quality control problems of all
kinds in every area.
It Focus on not making mistakes (Mistake proofing) and
reducing the variability in the processes.
Six Sigma properly implemented will
Improve customer satisfaction,
significantly improve quality,
improve the way we launch new products, and
create a continuous improvement mindset.


Fundamental Themes of
Six Sigma
Six Sigma is based around a number of central
themes
A genuine focus on the customer
Data and fact driven Management
- Process focus, management & improvement
Proactive management
Team based problem solving projects
Drive for perfection, tolerance for failure
Contd.
Fundamental Themes of
Six Sigma
A genuine focus on the customer
Six Sigma measures start with customer satisfaction
Emphasis on understanding customer expectations and
requirements

Data and fact driven Management
Speaking with data (not opinions!)
Management decisions based on fact
Develop understanding of internal processes
Contd.
Fundamental Themes of
Six Sigma
Process focus, management & improvement
Understanding the process is the key aspect
Control the inputs, and improve process

Proactive management
Understanding of Six Sigma principles
Defining the root cause of problems
Setting and maintaining aggressive improvement targets
Challenging why things are done this way
Contd.
Fundamental Themes of
Six Sigma
Team based collaboration
Six Sigma is teamwork focused
Project lead by process expert
Facilitated by Six Sigma Black Belt
Multi-skill experts within project team

Drive for perfection, tolerance for failure
Being prepared to fail during improvement
Understanding why something failed
Improving it next time
Six Sigma
Improvement Strategy
Phase 0: Process Definition
Phase I: Process Measurement
Phase II: Process Analysis
Phase III: Process Improvement
Phase IV: Process Control


DMAIC
Six Sigma Improvement Model
Control
Improve
Analyze
Define
Measure
Six Sigma Structure
Management
Team
Project
Champion
Master
Black Belt
Green
Belt
Black Belt
and
Team

Functional Groups
HR, Finance, IT
Engg, QA, etc.


A typical six sigma structure in manufacturing organization
Quality Leader/ Manager (QL/QM) - The quality leaders
responsibility is to represent the needs of the customer and to
improve the operational effectiveness of the organization.
Master Black Belt (MBB) - Master Black Belts are typically
assigned to specific area or function of a business or organization.
Black Belt (BB) - Black Belts lead quality projects and work full
time until they are complete.
Green Belt (GB) - Green Belts are employees trained in six
sigma who spend a portion of their time in completing projects, but
maintain their regular work role and responsibilities.

SIX SIGMA
ORGANIZATIONAL STRUCTURE
Case Study
Bombay Tiffenwallas



28
Case Study
Bombay Tiffenwallas
(Source: Forbes Magazine)
Bombay Tiffanwallas (Raghunath Medge president)
- They make
one Error on every 16 millions transactions
5000 Tiffanwallas deliver 175,000 lunches
everyday and take the empty tiffin back
- They make
one mistake in 2 month
175,000x2(transaction per day) x30 (days)=10,500,000
10,500,000x2 months= 21,000,000 (21 millions)

Case Study
Bombay Tiffenwallas
(Source: Forbes Magazine)
The charge for this service = 150 Rupees per month
Tiffenwallah monthly income= Rs 3,250
Tiffenmens Association gets Rs 10 per worker
Send to Charity = Rs. 50, 000 per month
Superb!!!
Service and Charity Too
Can anyone ask for more????