1

CONTENTS:
ƒ Measurementofcentraltendency
ƒ Problemsolvingandgraphicaltechniques
Problem solving and graphical techniques
ƒ ConceptofSixSigma

2
 BasicsofMeasurementofcentraltendency

How often do you hear

Howoftendoyouhear…
y Whatistheaverage
What is the average pay package sizeofthecollege?
size of the college?

y Whatisthemost
Wh i h common salary
l offeredbyTCS?
ff d b TCS?

y Whatisthemiddle value of the salary ofsay100

employees?

y Yousaid“averagepaypackageis50000/,butXis
Y id “ k i / b Xi
drawing1,00,000/ whileYisdrawing20000/?”

4
•Data is a collection of any number of related observations.
observations
•A single observation is called a data point.
• A collection
ll off data
d is called
ll d a data
d set.

Yards produced by each of the thirty carpet looms :

Yardsproducedbyeachofthethirtycarpetlooms:

16 2
16.2 15 8
15.8 15 8
15.8 15 8
15.8 16 3
16.3 15 6
15.6
15.7 16.0 16.2 16.1 16.8 16.0
16.4 15.2 16.2 15.9 15.9 16.8
15.4 15.7 15.9 16.0 16.3 16.0
4
16.4 16.6 15.6
5 15.6
5 16.9
9 16.33

Testfordata
Test for data
y Whether the source is biased(it is an interest in
supplying data points leading to one conclusion rather
than another)
y Do the data support or contradict other evidence we
have?
y Is there some evidence missing?
y How many observations do we have? Do they
p
represent all the g
groups
p we wish to study?
y
y Is the conclusion logical? Have we made conclusions
that the data do not support?
6
Arrangingdatausingthedataarrayand
thefrequencydistribution
the frequency distribution
y Thedataarrayisoneofthesimplestwaystopresent
Th d i f h i l
data.Itarrangesvaluesinascendingordescending
order.
d
p p
Soourpreviousexamplebecomes:

15 2
15.2 15 7
15.7 15 9
15.9 16 0
16.0 16 2
16.2 16 4
16.4
15.4 15.7 15.9 16.0 16.3 16.6
15.6 15.8 15.9 16.0 16.3 16.8
15.6 15.8 15.9 16.1 16.3 16.8
5
15.6 15.8
5 16.0 16.2 16.4
4 16.9
9

Frequency Distribution
FrequencyDistribution
y The data array
y is not always
y helpful
p because of its cumbersome form. We need
to compress information and still be able to use it for interpretation. Frequency
distribution comes into picture then.
y The data are therefore put in different classes.
classes
y Three information are required to know the class interval:
((a)) lowest data p point
(b) Highest data point
(c) Total number of class intervals( assume it)
y Width of class interval=
next value after the highest value in the data lowest value

total number of class intervals
y Here next value after the largest g value in the data= 17.0 7
y Smallest value= 15.2
y Assuming class interval= 6,
width of class interval= (17.015.2)/6=1.8/6=0.3
8
 Class Frequency Relative Less than Cumulative
frequency ogive frequency
Frequency
Frequencydistribution(Histogram)
15.215.4 2
distribution(Histogram) (%)

7(2/30*100) <15.2 0(0%)

15 5 15 7
15.515.7 5 17(5/30*100)
( / * ) <15.5 2(7%)
( %)
15.816.0 11 36(11/30*100) <15.8 7(23%)
16 1 16 3
16.116.3 6 20(6/30*100)
(6/ * ) <16.1
6 18(60%)
(6 %)
16.416.6 3 10(3/30*100) <16.4 24(80%)
16 7 16 9
16.716.9 3 10(3/30*100)
( ) <16.7 27(90%)
( )
Total 30 100 < 17.0 30(100%)
relativefreq
l f Cumulativefrequency
l f
frequency
40 35
12 30 30
10
8 20 25
6 20
10 Cumulative
4 relativefreq
frequency 15 frequency
2 0
0 10
15.215.4
15.515.7
15.816.0
16.116.3
16.416.6
16.716.9
15.816.0
16.116.3
15.215.4
15.515.7

16.416.6
16.716.9

5
0
9
<15.2 <15.5 <15.8 <16.1 <16.4<16.7

Measure of central tendency

Measureofcentraltendency
y Mean
y Median
y Mode
M d

10
Mean
y Simplearithmeticmean
Simple arithmetic mean

 x=
 X/n(forungroupeddata)
X/ ( f dd )
 x=(f*X)/n(forgroupeddata)

11

Averagemonthlybalanceof600
g y
customers
Class(dollars) Frequency(f)
049.99 78
5099.99 123
Findthemeanvalue.
100149.99 187
150199.99 82
200249.99 51
250299.99 47
300349.99 13
350399.99 9
400449.99 6
450499.99 4
f=n=600

12
Averagemonthlybalanceof600
Average monthly balance of 600
customers
Class(dollars) Midpoint(x) Frequency(f) f *x
049.99 25 78 1950
5099.99 75 123 9225
100149.99 125 187 23375
150199.99 175 82 14350
200249.99 225 51 11475
250299.99 275 47 12925
300349.99 325 13 4225
350399.99 375 9 3375
400449.99 425 6 2550
450499.99 475 4 1900
f=n=600 (f*X)=85350

average=(f*X)/f=85350/600=142.25 13

Advantagesanddisadvantagesofmean
Advantages and disadvantages of mean
Advantages
y conceptisfamiliarandclear.
y Everydatasethasoneandonlyonemean,hence
E d h d l h
unique.
y Comparingthemeanvalueofseveraldatasetiseasier.
Disadvantages
y Getsaffectedbyextremevalues.
y Tedioustocomputeincaseoflargedatasets.
T di i fl d

14
Median
y Asinglevaluefromadatasetthatmeasuresthecentral
A single value from a data set that measures the central
iteminthedata.
If the number of items in an array is n then median (
Ifthenumberofitemsinanarrayisn,thenmedian=(
n+1/2)th iteminthearray.

5
56891056891012
9 5 9

8.5

15

AdvantagesanddisadvantagesofMedian
Advantages and disadvantages of Median
Advantages:
1. Extreme values do not affect the median as strongly as they do the
mean.
2
2. Easy to understand and calculate.
calculate
3. Median can be found out for data which are having qualitative
descriptions such as colour or sharpness.
Disadvantages:
1. Data must be p put in arrayy before median calculation. This is time
consuming for any data set having large no. of elements.
2. Some accuracy is lost by choosing a single value to represent a
distribution, e.g., the median of 2,4,5,40,100,213 and 347 is 40,
which has no apparent relationship to any other values in the
distribution.
distribution

16
MODE
y Themodeisthevaluethatisrepeatedmostoftenin
The mode is the value that is repeated most often in
thedataset.e.g.,0,1,1,2,2,4,5,6,6,7,7,7
y Here,as
Here as ‘7’
7 isrepeatedmostoften,itiscalledthe
is repeated most often it is called the
mode.Incase,twoormorevaluesarerepeatedmost
often they are called bi modal or multi modal
often,theyarecalledbimodalormultimodal.

17

AdvantagesandDisadvantagesofMode
Advantages and Disadvantages of Mode
Advantages:
1. The
h mode,d like
l k the
h median,
d can be
b used d as a centrall location
l f
for
qualitative as well as quantitative data, e.g., if a printing press turns
out five impressions, which we rate “very
sharp”,
h ” “sharp”,
“ h ” “sharp”,
“ h ” “sharp”,
“ h ” and d “blurred”,
“bl d” then
h the h modald l value
l is i
“sharp”.
2. Like the median, the mode is not unduly y affected byy extreme values.
Disadvantages:
1. Too often, there is no modal value because the data set contains no
values that occur more than once.
once
2. Other times, every value is the mode, because, every value occurs the
same no. of times.
3. When
Wh d t sett contains
data t i 2,3 or many modes, d th are difficult
they diffi lt to
t
interpret or compare.

18
DISPERSION
y The spread
p or variabilityy of data is called dispersion.
p
y It gives us additional information that enables us to judge
the reliability of out measure of the central tendency.
y If the data are widely spread, the central location is less
representative of the data as a whole.
y Financial analysis are concerned about the dispersion of a
farm’s earning. Widely dispersed earnings indicate a higher
risk
i k to stockholders
kh ld and
d creditors
di than
h d earnings
do i
remaining relatively stable.
y A drug
d th t is
that i average in
i purity
it but
b t ranges from
f very pure to
t
highly impure may endanger lives.

19

Measuresofdispersion
Measures of dispersion
1. Range valueofhighestobservation
Range= value of highest observation – valueoflowest
value of lowest
observation

2. Variance=2= ((x μ)2)/N

3. Standard deviation
3.Standard deviation=
=variance
 variance

20
S.no Observations(x) Xμ (Xμ)2
.
1. 0.04 An
An
2.
3.
0.06
0.12
p
example:
Observed
4. 0.14
percentage impurity
percentageimpurity
5. 0.14
inanexperiment.15
6. 0.15
samplesweretaken.
p
7. 0.17
8. 0.17
9. 0.18
8
10. 0.19 Variance=2=((x μ)2)/N
11
11. 0 21
0.21
StandardDeviation==Variance
12. 0.21
13
13. 0 22
0.22
14. 0.24
15. 0.25
μ=
21

S.no. Observations(x) Xμ (Xμ)2

1
1.
2.
0 04
0.04
0.06
0 126
0.126
0.106
0 016
0.016
0.011
An
An
33. 0.12 4
0.046 0.002 p
example:
4. 0.14 0.026 0.001
Observed
5. 0.14 0.026 0.001
percentage impurity
percentageimpurity
6. 0.15 0.016 0.000
inanexperiment.15
7. 0.17 0.004 0.000
8
8. 0.17 0.004 0.000
samplesweretaken.
p
9. 0.18 0.014 0.000
10
10. 0 19
0.19 0 024
0.024 0 001
0.001
11. 0.21 0.044 0.002 Variance=2= ((x
12. 0.21 0.044 0.002 μ)2)/N=0.051/15=0.0034
13. 0.22 0.054 0.003
StandardDeviation==
14. 0.24 0.074 0.005
0.0034=0.058
 34 5
15. 0.25 0.084 0.007
x=2.49,μ=2.49/15= (X
0. 66
0.166 μ)2=0.051
0.05

22
 23

Theaveragescoreofaclassis70and
standard deviation is 10 The class has 20
standarddeviationis10.Theclasshas20
students.Show thedistributionpattern.

24
Thebellcurve
The bell curve
y Thebellcurvegetsitsnamequitesimplybecauseits
g q py
shaperesemblesthatofabell.Thesecurvesappear
throughoutthestudyofstatistics.Theyarebasically
normalprobabilitydistribution.
y Itisdependantonlyontwovariables:meanand
p y
standarddeviation.68%ofthevaluesfallin±1SD,
± 95%in2SDand± 99.5%in3SD.
y Applicationsofbellcurveincludeassessment
p
patterninacompany,testresultsofapieceof
p y p
equipment,heightofstudentsofaparticular
gradeetc.

25

ProblemIdentification
Problem Identification Problem
Problem ProblemAnalysis
Problem Analysis
identification+
Problemanalysis
Flowchart Paretochart Histogram
Checksheet IshikawaDiagram Scatterdiagram
Nominalgroup
l Runchart
h Controlchart
l h
techniques
Brain storming
Brainstorming Stratification Process capability
Processcapability

26
ProblemIdentification
Problem Identification
y Flowchart
Flow chart

y Checksheet
Ch k h

y Nominalgrouptechniques

y Brainstorming

27

FlowChart
•Apictorialrepresentationshowingallthestepsofthe
p
process.

• Providesexcellentdocumentationofaprogramme andcan
beanusefultoolforexamininghowvariousstepsofaprocess
arerelatedtoeachother.

•Useseasilyrecognizablesymbolstorepresentthetypesof
processingperformed.
f d

Start/end Decision Processstep

28
 29

Checksheet
Check sheet
y Amethodbywhichcollectionofdataismadeeasier
A method by which collection of data is made easier
andsystematic.
y It helps to answer questions like “howoftenarecertain
Ithelpstoanswerquestionslike how often are certain
eventshappening?”
y S
Startstheprocessoftranslatingopinionsintofacts.
h f l i i i i f
y Needstohavedetailedstratifiedinformation.
y Showsareasofstrengthandweakness.

30
Checksheet ofatailoringshop
of a tailoring shop

31

Graphs
%rejection
j
8

3 %rejection

32
NominalGroupTechnique
Nominal Group Technique
y Each member of a group will write a problem.
y Codifying the problems. For example, A, B, C, D & E.
y Give
Gi score to the
h most important
i problem
bl as 5 and
d
least important as 1 .Each member will contribute for
this.
hi
y Addition of all the scores. Maximum is the most
important and so on.

33

AnexampleofNGT
An example of NGT
Problem Scores Total

A 2 5 2 4 1 14
4

B 3 4 4 1 4 15

C 2 5 5 1 4 17

D 5 5 4 5 4 23

E 1 1 2 1 3 7
34
Brainstorming
Brain storming
y Thistechniquebringsouttheideastappedinhuman
This technique brings out the ideas tapped in human
mindforgeneratinggreatestpossiblenumberof
solutions.
solutions
y Activeparticipationbyeachandeverymemberis
required.
required
y Recordingofallideas.
y Rejection,ifany,afterthroughexaminations.
y Implementation.
Implementation

35

ProblemAnalysis
Problem Analysis
y Histogram

y Scatterdiagram
S di

y Controlchart

y Processcapability

36
Histogram
yA form of column graph which displays the
distribution of values obtained whenever numerical
data is collected.
collected
y Objective of a histogram is to suggest probable
distribution of the population from which samples
were taken.
y General analysis through histogram are
weight, chemical analysis, voltage, length, width etc.

37

AnexampleofanalysisofsizefractionfromBIM
A l f l i f i f i f BIM
p
fineswith20 samples
Sl Samplesize(mm) Sl Samplesize(mm)
No No
1 6.1 11 9.2
2 8 12 9.5
3 10.2 13 9.7
4 88
8.8 14 9.7
5 9 15 9.8
6 10 16
6 10.1
7 10.2 17 9.2
8 10 3
10.3 18 91
9.1
9 11.4 19 9.0
10 12 5
12.5 20 89
8.9

38
FrequencydistributionofBIMfines
frequency
9
8
7
6
5
4 frequency
3
2
1
0
6.17 7.18 8.19 9.110 10.111 11.112
39

Scatterdiagram
Scatter diagram
y An analytical diagram which is obtained by plotting
the values of one variable against the other for
checking the relationship between them.
them
y Used to test for cause and effect relationship.
y It gives
i an indication
i di i whether
h h such h a relationship
l i hi
exists and the strength of such relationship.

40
Relationshipbetweencoalashandcokeashpercentage
for20samples
S.No. Coalash% Cokeash S.No. Coalash% Cokeash
% %
1. 16.2 23.5 11. 17.3 24.8
2. 17.2 24.0 12. 18 25
3. 18 25 13. 19 25
4. 19 26 14. 20 27
5. 17 24 15. 15 18
6
6. 18
8 26
6 16.
6 21 28
8
7. 19.2 20.3 17. 20.5 22.5
8
8. 16
6 20 18.
8 22.5 29.5
9. 17 24.6 19. 18 26
10
10. 17 1
17.1 24 3
24.3 20
20. 19 26

41

An example of scatter diagram

Anexampleofscatterdiagram
35

30

25

20

15

10

0
0 5 10 155 20 255

CoalAsh% 42
 Variation: common and special causes
Variation:commonandspecialcauses
y No two product or characteristics are exactly alike because any process
contains
i many sources off variability.
i bili The
Th differences
diff among the
h products
d may
be large or small, but they are always present.

y Common causes refer to many sources with a process that has a stable and
repeatable distribution over time. This is called “ in a state of statistical
control”.
t l” The
Th common causes behave
b h lik a stable
like t bl system
t off chance
h causes. E.g.
E
machine clearances, bearing wear, people performance etc. The resolution of
common cause of variation usually requires action on the system.

y Special or assignable causes refer to any factor that make the overall process
distribution change.
change If special causes of variation are there,
there the process output
is not stable over time. E.g defective shaft, wrong alignment, deliberate go slow
etc. The resolution of a special cause of variation usually requires local action.

IT IS THE JOB OF THE ORGANISATION TO ADDRESS AND ELIMINATE

SPECIAL CAUSES OF VARIATION TO BRING THE PROCESS TO A STATE
OF STATISTICAL CONTROL BEFORE CALCULATION OF PROCESS
CAPABILITY. 43

Controlcharts
Control charts
y The control charts are nothing but line graphs having 2
controll limit
li i lines
li called
ll d Upper Control
C l Limit(UCL)
i i ( C ) and d
Lower Control Limit(LCL) drawn on either side of process
average.
average
y The genesis of control chart is based on the fact that
variation in a pprocess is inherent. However, this inherent
variation takes
k place
l only
l within
h a certain limit.
l
y The values of UCL and LCL are computed by statistical
analysis.
analysis
y The control charts effectively direct attention towards
special cause of variation and reflect the extent of
common cause of variation that must be reduced by
system or process improvement.

44
Atypicalcontrolchart
A typical control chart

45

Benefitsofacontrolchart
y Effective tools to understand process variation and help
achieve statistical control. The control charts give the
people closest to the operation reliable information what
action should or should not be taken.
y A process in statistical control can be further improved
through reduction of common causes and improved
process centering. Such process improvements may reduce
cost aand
cos d improve
po ep productivity
oduc y byy dec
decreasing
eas g thee variation
a a o
around the target value.
y Control charts,
charts by distinguishing special from common
causes of variation, gives a good indication of whether any
problems are likely to be correctable locally or will require
management action.
46
 Calculation of Control limits
CalculationofControllimits

47

H l t
Hourlytemperaturedeg.C
t d C oftopgasforBF
ft f BF
Sl.No. Position1 Position2 Position3 Position4 x R(Range)
1. 200 260 180 180
2. 200 220 200 240
3. 240 190 200 220
4. 240 290 150 190
5. 210 170 170 200
6
6. 200 200 200 180
7. 210 220 200 180
8. 260 200 210 220
9. 150 200 180 120
10. 190 220 180 160
11. 170 210 180 150
12. 170 230 190 160
13. 160 220 180 150
14. 190 240 210 180
15. 100 200 210 160
16. 200 210 160 180
17
17. 210 240 200 180
18. 160 170 200 150
x= R=
 H l t
Hourlytemperaturedeg.C
t d C oftopgasforBF
ft f BF
Sl.No. Position1 Position2 Position3 Position4 x R(Range)
1. 200 260 180 180 205 80
2. 200 220 200 240 215 40
3. 240 190 200 220 212.5 50
4. 240 290 150 190 217.5 140
5. 210 170 170 200 187.5 40
6
6. 200 200 200 180 195 20
7. 210 220 200 180 202.5 40
8. 260 200 210 220 222.5 60
9. 150 200 180 120 162.5 80
10. 190 220 180 160 187.5 60
11. 170 210 180 150 177.5 60
12. 170 230 190 160 187.5 70
13. 160 220 180 150 177.5 70
14. 190 240 210 180 205 60
15. 100 200 210 160 167.5 110
16. 200 210 160 180 187.5 50
17
17. 210 240 200 180 207 5
207.5 60
18. 160 170 200 150 170 50
x=3487.5 R=1140 49

CalculationofUCLandLCL
Calculation of UCL and LCL
y GrandAverage=(x
g ( )/ )/N=

y Averagerange=(R)/N=
Average range= (R)/N=

y A2=0.729(forsamplesizeof18)
(f l i f 8)

y UCL=grandaverage+A2*averagerange
=
y LCL=grandaverageA2*averagerange
=
50
 CalculationofUCLandLCL
Calculation of UCL and LCL
y GrandAverage=(x)/N=3487.5/18=193.75

y Averagerange=(R)/N=1140/18=63.3

y A2=0.729(forsamplesizeof18)

y UCL=grandaverage+A2*averagerange
=193.75+0.729*63.3=240
*6

y LCL=grandaverageA
LCL d A2*averagerange
*
=193.750.729*63.3=147.6

51

250

UCL

200
Gasstempeeraturesindeg.C
C

150
LCL

Series1

100

50

0
0 2 4 6 8 10 12 14 16 18 20

AveragedatapointsofgastemperaturesofBF
52
ProcessCapability
Process Capability
y To say that process is in control is not enough, as it
does not give guarantee that the process will not
produce bad products.
products

y True improvement
i i a process can be
in b brought
b h by
b
balancing the repeatability and consistency with the
capability of meeting customer’s requirements. This is
known as process capability.

53

ProcessCapabilityIndex
Process Capability Index
y In order to measure the degree
g to which the p
process is
or is not meeting the requirements, capability
indices have been developed, which places the
distribution of the process in relation to specification
limits.

y Processcapabilityratio(C
p y ( p))=(USLLSL)/6
( )/
USL=UpperSpecificationLimit
LSL=Lower Specification Limit
LSL=LowerSpecificationLimit
=StandardDeviation

54
Anexampleofprocesscapability
An example of process capability
study
Acustomerwantsacoffeemachinetoproducecoffee
A customer wants a coffee machine to produce coffee
between150to170degreecentigrade.Findoutthe
process capability ratio Cp for the following st
processcapabilityratioCpforthefollowingst
deviationlevel:
y SD=5deg
SD 5 deg
y SD=3.3deg
y SD=2.5deg
y SD=2deg
SD=2 deg
y SD=1.67deg

55

Variousvaluesofprocesscapabilityindices
V i l f bilit i di

56
ProblemIdentificationand
Problem Identification and
Analysis
y Paretochart
Pareto chart

y IshikawaDiagram
I hik Di

y Runchart

y Stratification

57

Paretoanalysis
Pareto analysis
y Paretodiagramcombinestwoofthecommonformof
Pareto diagram combines two of the common form of
graphs acolumngraphandalinegraph.

y Ithelpstoidentifythe‘vitalfew’fromthe‘trivial
many’andpinpointstheareas,i.e.,vitalfew,onwhich
’ d i i h i i lf hi h
attentionshouldbefocused.

y Itisalsousedtodeterminewhichproblemtosolvein
It is also used to determine which problem to solve in
whatorder.

58
 CausesofrejectioninaCNCmachine

Sl.No. Causeof No.of %ageof CUM

rejection rejections rejections %ageof
rejections

1 Poorfinish 8 40 40

2 Sharp 7 35 75
edges

3 BadID 2 10 85
thread

4 BadOD
dO 1 5 90
thread

5 OD 1 5 95
Oversized

6 OD
OD 1 5 100
Undersize
d

Totall 40 100

59

Ishikawadiagram
Ishikawa diagram
y First developed
p byy Prof. Ishikawa of JJapan
p in 1950s.
95 Also
known as Cause and Effect diagram, or Fishbone diagram.
y It is a universal truth that there cannot be an effect without
one or more causes contributing to it.
y The Ishikawa diagram demonstrates the causes/subcauses
contributing
b to the
h effect.
ff
y The causes and subcauses are decided based on structured
b i
brainstorming.
i
y The rootcause is found out by collecting data of the
relative
l ti occurrence off theth different
diff t causes and
d arriving
i i att a
team consensus.

60
Ishikawadiagramforhighpetrolconsumptioninacar

61

RunChart
Run Chart
y A run chart, also known as a run
runsequence
sequence plot is a
graph that displays observed data in a time sequence.

y Often, the data displayed represent some aspect of the

output or performance
f off a manufacturing
f i or other
h
business process.

y It can be used to identify the meaningful trends or

shifts in the average.

62
Anexampleofarunchartbefore
An example of a run chart before
and after system intervention
andaftersystemintervention

63

Stratification
y Stratification breaks down single
g numbers into
meaningful categories or classifications to focus on
corrective actions.
y For example, data on minor injuries for a plant may be
g figure,
recorded as a single g , either rising g or falling.
g But
this number is actually the sum total of injuries like
byy department:
p Coke Ovens,, Foundry, y, SMS,,
etc.
by location of injury: head,
head eyes,
eyes hands,
hands etc.
etc
by type: burns, cuts, scratches, etc.

64
 Conceptofsixsigma

65

SixSigmaAnintroduction
Six SigmaAn introduction
y Six Sigma is a set of techniques and tools for process
improvement. Introduced by Motorola in 1986, it was widely
adopted by GE under the stewardship of Jack Welch in 1995.
y Six
Si Sigma
Si seeks
k to
t improve
i th output
the t t off process by
b identifying
id tif i
and removing the causes of defects and minimizing variability.
y Each six sigma project follows a defined sequence of
steps(DMAIC) and specific value targets( reduce process
time, costs etc).
y A six sigma process is one in which 99.99966% of opportunities
to produce some feature of a part are statistically expected to be
free of defects( 3.4
3 4 defects per million opportunities).
opportunities)

66
 ConceptofSixSigma
Concept of Six Sigma

67

Thesigmaleveland%ofdefects
The sigma level and % of defects
Sigmalevel Defects/million %Good

1 692462 31

2 308537 70

3 66807 93.4

4 6210 99 4
99.4

5 233 99.97

6 3.4 99.9996

68
SixSigmaPhilosophy
Six Sigma Philosophy

CTQ:CriticaltoQuality 69

Methodologies
Methodologies
y Sixsigmaprojectsfollowsprojectmethodologyinspiredby
g p j p j gy p y
Deming’sPDCA
1. DMAIC(definemeasureanalyse improvecontrol)
A. Define thesystem,thevoiceofthecustomerandtheir
requirementsandtheprojectgoalsetc.
B
B. Measure key aspects of the current process and collect relevant
Measurekeyaspectsofthecurrentprocessandcollectrelevant
data,measurethecurrentCp.
C. Analyse thedatatoinvestigate– seekforrootcausesofthedefect
underinvestigation.
d i i i
D. Improve thecurrentprocessthroughDOE,Poka Yokeetc.Check
theimprovedCpsubsequently.
p p q y
E. Control thefuturestateprocesstoensurethatanydeviationsfrom
thetargetarecorrectedbeforetheyresultindefects.

70
Anexampleofapossiblesixsigma
An example of a possible six sigma
project
project
y The most critical parameter of a steam boiler is its
final superheat steam temperature. Where as the
steam temperature of a typical boiler in a plant was
supposed to be 485 ± 5 deg centigrade, on a
measurement it was found that out of 100
readings, more than 20 data points are either in the
range of 460 or 500 deg centigrade.
centigrade

71

72
Implementationroles
Implementation roles
y Sixsigmaidentifiesseveralkeyrolesforitssuccessfulimplementation.
1. Executive Leadership:
d h CEO and d otherh members
b off the
h top
management. Set up vision, provide resources and empower other
role holders.
2. Champions: Take responsibility for six sigma implementation
across the organisation.
3. Master black belts: Inhouse coaches on six sigma devoting 100% of
their time. They assist champions and guide black belts and green
belts.
4
4. Black belts: They apply six sigma projects to specific projects
focusing on project execution.
5. Green belts: The employees who take up six sigma implementation
along with other responsibilities,
responsibilities operating under the guidance of
the black belts.

73

CALCULATIONOFPROCESS
CAPABILITY SOME EXAMPLES
CAPABILITY:SOMEEXAMPLES

74
S.n Observations Xμ (Xμ)2
o. ( )
(x)
1. 99.82105 Thepercentage
2. 99.69079 ofoxygenpurity
f it
3. 99.66642 forTOPIIfrom
4
4. 99 82139
99.82139
5.
day1today15for
99.67868
6
6. 99 71408
99.71408
p
aparticular
7. 99.71243 monthisas
8. 99 67065
99.67065 follows:
9. 99.6736 Variance=2=((x μ)2)/N
10. 99 67577
99.67577
StandardDeviation==Variance
11. 99.8043
12. 99.64826
13. 99.64666
14. 99.64518
15. 99.70347
μ= 75

S.no. Observations(x) Xμ (Xμ)2

1. 99.82105 0.11625 0.0135

2.
2 99 69079 0.01401
99.69079 0 01401 1.962x10
1 962x10^4
4
3. 99.66642 0.03838 1.473x10^3
4
4. 99 82139 0.11659
99.82139 0.0136
5. 99.67868 0.02612 6.822x10^4
6. 99 71408 9.28X10^3
99.71408 8.612x10^5
7. 99.71243 7.63x10^3 5.821x10^5
8. 99.67065 0.03415 1.166x10^3
9. 99.6736 0.0312 9.7344x10^4
10. 99.67577 0.02903 8.427x10^4
11. 99.8043 0.0995 9.9x10^3
12. 99.64826 0.05654 3.196x10^3
13. 99.64666 0.05814 3.3802x10^3
14. 99.64518 0.05962 3.554x10^3
15. 99.70347 1.33x10^3 1.7689x10^6
μ=99.7048 76
 •Variance=2=((x μ)2)/N=0.0271/15=1.8066x1o3

•StandardDeviation==Variance=0.0425

•DefinedUSL=100

•DefinedLSL=99.5

•Cp=(USLLSL)/6=1.96
( )

•CPk=lower((USL
((  mean)/3
) ,(meanLSL)/3
( ) ))
))
=lower(2.315,1.606)
=1.606

77

S.n Observations Xμ (Xμ)2

o. (x)
1. 11.62
11 62
2. 11.72 Thepipelength
3. 11.72
11 72 variationof
i ti f
4. 11.72
ERWPPfromday
5
5. 11 72
11.72
6. 11.71
1today15fora
7
7. 11 72
11.72 particularmonth
p
8. 11.73 isasfollows:
9. 11.71
10. 11.88 Variance=2=((x μ)2)/N
11. 11.63
StandardDeviation==Variance
12. 11.68
13.
3 11.67
14. 11.68
15. 10.88
μ=11.65
78
S.no. Observations(x) Xμ (Xμ)2

1. 11.62 0.03 0.0009

2. 11.72
11 72 0.07 0.0049
3. 11.72 0.07 0.0049
4
4. 11 72
11.72 0 07
0.07 0 0049
0.0049
5. 11.72 0.07 0.0049
6
6. 11 71
11.71 0 06
0.06 0 0036
0.0036
7. 11.72 0.07 0.0049
8. 11.73
3 0.08 0.0064
9. 11.71 0.06 0.0036
10. 11.88 0.233 0.0529
5 9
11. 11.63 0.02 0.0004
12. 11.68 0.033 0.0009
9
13. 11.67 0.02 0.0004
14. 11.68 0.03 0.0009
15. 10.88 0.23 0.0529
μ=11.65 79

•Variance=2=((x μ)2)/N=0.009827

•StandardDeviation==Variance=0.09913

•DefinedUSL=12

•DefinedLSL=11.5

•Cp=(USLLSL)/6=0.84
( )

•CPk=lower((USL
((  mean)/3
) ,(meanLSL)/3
( ) ))
))
=lower(1.177,0.504)
=0.504

80