16-04-2024

SIX SIGMA

 Focus:
 To establish world-class business-performance benchmarks
 to providing an organizational structure and road-map to achieve

SIX SIGMA them.

 Team based approach
 Six Sigma as BPM methodology
 at improvement of the processes in an organization
 quality of the product depends on the variability at the current task
and the combined variability of all the tasks
Dr. Dhananjay A. Jolhe

Department of Mechanical Engineering,

Visvesvaraya National Institute of Technology, Nagpur
© Dr. Dhananjay Jolhe 2

SIX SIGMA: HISTORY SIX SIGMA: DEFINITIONS

 1736: Abraham de Moivre: concept of Normal Curve  “Business process that allows companies to drastically
 1 81 8: Carl Fredrick Gauss: use of Normal Curve improve their bottom line by designing and monitoring
 1 896: Vilfredo Alfredo Pareto: Pareto principle everyday business activities in ways that minimize waste and
resources while increasing cu stomer satisfaction”
 1924: Walter Shewhart: Control Charts
(Harry and Schroeder, 2000).
 1941: Alex Osborn: Brainstorming
 1949: U.S. Department of Defence: FMEA
 “Six Sigma is many things, and it would perhaps be easier to
 1960: Kaoru Ishikawa: Cause and Effect Diagram list all the things that Six Sigma quality is not. Six Sigma can
 1970: Dr. Noriaki Kano: Kano Model be seen as: a vision; a philosophy; a symbol; a metric; a goal;
 1980s: TQM (Mass production of electronics items/ Japanese a methodology.”
Competition) (Tennant, 2001)
 1986: Motorola: “Six Sigma” methodology
 1987: Malcolm Baldrige National Quality Award
 1994: Allied Signal: adopted “Six Sigma”
 1995: General Electric: adopted “Six Sigma”
© Dr. Dhananjay Jolhe 3 © Dr. Dhananjay Jolhe 4

1
 16-04-2024

S IX -S IGMA : M ETHODOLOGY
SIX SIGMA: METHODOLOGY

 SIPOC model  De fine cu stomers &  Presentation skills

DEFINE CTQs  Project management
 Draft p roject charter skills
 Form team: Identify  Leadership skills
MEASURE Champion & process  Gantt Chart
owner: Role &  QFD
SUPPLIER INPUT PROCESS OUTPUT CUSTOMER
Responsibility
ANALYSE  SIPOC Process Map
 Map the p rocess
 Project Charter
 Affinity Diagram/
IMPROVE Interrelation Diagram

 DMAIC: 5 Step Metho dology

 Implemented by Team: Champion, MBB, BBs, GBs & other team members CONTROL
© Dr. Dhananjay Jolhe 5 © Dr. Dhananjay Jolhe 6

SIX SIGMA: METHODOLOGY SIX SIGMA: METHODOLOGY

 Define d efect, o pp or tunity,  FME A  Analyse data  Hyp othesis testing

u nit and m etrics  MSA (Measu rement System  Find root causes  Mu lti-variate analysis
DEFINE DEFINE
 Id entify p arameter s to be Analysis)  Establish Cause-Effect  AN OVA
measured  Check/Data Sheet relationship  Regression analysis
MEASURE  Input  Pareto Char t MEASURE  Analyse process  C&E /Fishbone Diagram
 Output  Gage R&R  Identify source of  Sc atter Diagram
 Relationship: Y = f(X)  Histogram variation  Cp & Cpk
ANALYSE  Valid ate the m easurement  Process C ap ability ANALYSE  Box Plots
s ystem  Brainstorming  Marginal Plots
 D evelop d ata c ollection  C&E /Fishbone Diagram
p lan  Interaction Plots
IMPROVE  XY Matrix IMPROVE
 Calculate the p rocess  FME A
s i gma l evel  Tree d iagram  Opp ortunity Map s
 Run Charts
CONTROL CONTROL
 Control C har ts
© Dr. Dhananjay Jolhe 7 © Dr. Dhananjay Jolhe 8

2
 16-04-2024

SIX SIGMA: METHODOLOGY SIX SIGMA: METHODOLOGY

 Develop potential  DOE  Develop standards and  SP C

solutions  RSM (Response Su rface p roc edures  E WMA
DEFINE DEFINE
 Assess failure modes of Methodology)  Implement statistical  Poka-Yoke
 Tree Diagram p roc ess c ontrol
potential solutions  Reliability analysis
MEASURE  FME A MEASURE  Determine p roc ess  Process C ontrol Plan
 Validate potential cap ability
improvement by pilot  Brainstorming  Control/Precontrol Char t
 TPM  Develop transfer p lan,
studies hand off to p rocess owner  Pareto Char t
ANALYSE  Correct/re-evaluate  Benchmarking ANALYSE  Process C ap ability
 Verify b enefits, c ost
potential solution  Af finity/ID saving/ avoid ance, profit  Check Sheets
 Compare the results to  Hyp othesis Testing growth  Run Charts
IMPROVE  Process Map ping IMPROVE
the sigma baseline.  Close p rojec t, finalize  Histograms
 PE RT /CPM d oc umentation  Sc atter Diagrams
 PDPC (P roc ess Dec ision
CONTROL Program Char t)
CONTROL
© Dr. Dhananjay Jolhe  Gantt Charts 9 © Dr. Dhananjay Jolhe 10

SIX SIGMA: METRICS SIX SIGMA: METRICS

µ=T µ = T  1.5
 Indicates quality level Sigma Level Defect Rate Cp Cpk Defect Rate Cp Cpk

SIGMA  Measured in terms of: SIGMA 2 4.56% 0.67 0.67 30.87% 0.67 0.17

LEVEL  Cp =
– LEVEL 3 0.26% 1 1 6.68% 1 0.50
 4 (0.63*10-2)% 1.33 1.33 0.62% 1.33 0.83
–
 C pk = min (C pu , C pl ) [ C pu = 
C pl =

] 6 (0.02*10-2)% 2 2 (0.034*10-2)% 2 1.5
DPU DPU
 Two types:
 Short term, Z ST (µ = T)
DPO  Long term, Z LT (µ = T  1.5)
DPO
 Z ST = 3 * C PK & Z ST = Z LT + 1.5

1σ = 691,462 DPMO Sigma Level (Z) is the number of standard

PPM 2σ = 308,538 DPMO deviations that can fit between the mean and
PPM
3σ = 66,807 DPMO the specification limit, e.g. for a process
DPMO 4σ = 6,210 DPMO operating at three sigma level, three standard DPMO
5σ = 233 DPMO deviations can be fitted between the process
6σ = 3.4 DPMO mean and the specification limit. 11 12

3
 16-04-2024

D E F E C T R AT E S O F [ A] C E N T E R E D P R OC E SS E S ( Μ = T ) D E F E C T R AT E S O F [ A] C E N T E R E D P R OC E SS E S ( Μ = T )
[ B ] O F F S E T P RO C E S SE S ( Μ = T  1 .5 ) [ B ] O F F S E T P RO C E S SE S ( Μ = T  1 .5 )

© Dr. Dhananjay Jolhe 13 © Dr. Dhananjay Jolhe 14

D E F E C T R AT E S O F [ A] C E N T E R E D P R OC E SS E S ( Μ = T ) D E F E C T R AT E S O F [ A] C E N T E R E D P R OC E SS E S ( Μ = T )
[ B ] O F F S E T P RO C E S SE S ( Μ = T  1 .5 ) [ B ] O F F S E T P RO C E S SE S ( Μ = T  1 .5 )

© Dr. Dhananjay Jolhe 15 © Dr. Dhananjay Jolhe 16

4
 16-04-2024

SIX SIGMA: METRICS SIX SIGMA: METRICS

 Six sigma process:  Six sigma process:

SIGMA ≤
𝟏
(Specification Spread) and SIGMA ≤
𝟏
(Specification Spread) and
𝟏𝟐 𝟏𝟐
LEVEL  µ should be within ±1.5 of T.
LEVEL  µ should be within ±1.5 of T.

DPU DPU  DPU = =

DPO DPO

PPM PPM

DPMO DPMO
© Dr. Dhananjay Jolhe 17 © Dr. Dhananjay Jolhe 18

SIX SIGMA: METRICS SIX SIGMA: METRICS

 Six sigma process:  Six sigma process:

SIGMA ≤
𝟏
(Specification Spread) and SIGMA ≤
𝟏
(Specification Spread) and
𝟏𝟐 𝟏𝟐
LEVEL  µ should be within ±1.5 of T.
LEVEL  µ should be within ±1.5 of T.

DPU  DPU = = DPU  DPU = =

DPO  DPO = = = DPO  DPO = = =

∗ ∗ ∗ ∗

PPM PPM  PPM = ∗ 10

DPMO DPMO
© Dr. Dhananjay Jolhe 19 © Dr. Dhananjay Jolhe 20

5
 16-04-2024

SIX SIGMA: METRICS SIX SIGMA: METRICS

 Six sigma process:

SIGMA ≤
𝟏
(Specification Spread) and
𝟏𝟐
LEVEL  µ should be within ±1.5 of T. SIGMA
LEVEL,
DPU  DPU = =
PPM &
 DPO = Cpk
DPO ∗
=
∗
=

PPM  PPM = ∗ 10

DPMO  DPMO = ∗
∗ 10 = DPO ∗ 10

© Dr. Dhananjay Jolhe 21 © Dr. Dhananjay Jolhe 22

SIX SIGMA: METRICS SIX SIGMA: METRICS

CLASSICAL  Does not consider effect of rework CLASSICAL  Does not consider effect of rework
YIELD YIELD
( ) ( )
 CY =  CY =
( ) ( )

FIRST TIME FIRST TIME ( )

YIELD YIELD  FTY =
 FTY = 𝑒
ROLLED ROLLED
THROUGHPUT THROUGHPUT
YIELD YIELD

NORMALIZED NORMALIZED
YIELD YIELD
© Dr. Dhananjay Jolhe 23 © Dr. Dhananjay Jolhe 24

6
 16-04-2024

SIX SIGMA: METRICS SIX SIGMA: METRICS

CLASSICAL  Does not consider effect of rework CLASSICAL  Does not consider effect of rework
YIELD YIELD
( ) ( )
 CY =  CY =
( ) ( )

FIRST TIME ( ) FIRST TIME ( )

YIELD  FTY = YIELD  FTY =
 FTY = 𝑒  FTY = 𝑒
ROLLED
THROUGHPUT
 RTY = ∏ 𝐹𝑇𝑌 ROLLED
THROUGHPUT
 RTY = ∏ 𝐹𝑇𝑌
YIELD YIELD

NORMALIZED NORMALIZED  NY = 1 − 𝐷𝑃𝑂 ∗ 1 − 𝐷𝑃𝑂 ∗ ⋯ … ∗ (1 − 𝐷𝑃𝑂 )

YIELD YIELD
© Dr. Dhananjay Jolhe 25 © Dr. Dhananjay Jolhe 26
(Pyzdek, 2003)

SIX SIGMA: EXAMPLE SIX SIGMA: EXAMPLE

P ROCESS 1 P ROCESS 2 P ROCESS 3
IN 1000 994 983
O UT 994 983 979
D EFECTIVE 188 181 206
1 DEFECT/ UNIT 164 152 178
2 DEFECTS/UNIT 16 13 16
3 DEFECTS/UNIT 2 5 8
4 DEFECTS/UNIT 6 11 4

P ROCESS 1 P ROCESS 2 P ROCESS 3

DPU (164*1+16*2+2*3+6*4)/1000 (152*1+13*2+5*3+11*4)/9 (178*1+16*2+8*3+4*4)/983
= 226/1000 = 0.226 94 = 250/983 = 0.254
= 237/994 = 0.238
DPO 226/1000*4 = 0.0565 237/994*4 = 0.057 250/983*4 = 0.0508

PPM (182+6)*106/1000 = 188000 (170+11)*106/994 = (202+4)*106/983 = 209563

182093
DPMO = 56500 57000 50800
CY 994/1000 = 99.4% 983/994 = 98.89% 979/983 = 99.59%
FTY (1000-188)/1000 = 81.2% (994-181)/994 = 81.79% (983-206)/983 = 79.04%

FTY# E-DPU = 79.77% 78.82% 77.57%

RTY 0.812 * 0.8179 * 0.7904 = 0.5249 = 52.49%
RTY# 0.7977 * 0.7882 * 0.7757 = 48.77 %
NY = 94.52%
© Dr. Dhananjay Jolhe 27 Z LT 0.885 (∵ FTY= 81.2%) 0.905 (∵ FTY= 81.79%) 0.805 (∵ FTY= 79.04%) 28
Z ST 2.385 2.405 2.305

7
 16-04-2024

SIX SIGMA: TOOLS AND TECHNIQUES

XY MATRIX

 Many tools (quantitative + qualitative)

 Sleeper (2006) has explained 59 tools (SS + DFSS)
 Also called as Cause-Effect Matrix or
 Seven basic tools Correlation Matrix or CTQ Matrix.
 (i) Flow chart,
 (ii) Histograms,
 Correlates customers’ CTQ requirements (Y) to
 (iii) Pareto charts, the process inputs (X).
 (iv) Check sheets,
 (v) Scatter diagram,  Steps:
 (vi) Control charts and  List out CTQs (i.e., Y’s).
 (vii) Cause and effect diagram
 Seven new tools (Brassard) (Ryan, 2000)  Decide the weight of each CTQ.
 (i) Affinity diagram,  Find out input process parameters (i.e., X’s).
 (ii) Interrelationship diagraph,
 (iii) Tree diagram,  Form matrix. Correlate X with Y using suitable scale.
 (iv) Prioritization matrix,
 (v) Matrix diagram,
 Find out weighted sum for each X.
 (vi) Process decision program chart and
 (vii) Activity network diagram© Dr. Dhananjay Jolhe 29 © Dr. Dhananjay Jolhe 30

XY MATRIX: EXAMPLE XY MATRIX: EXAMPLE

To improve the quality of a newspaper, readers’ survey was carried Output parameters (CTQs) (Y’s)
out and following C TQs were found out (weightage shown in Clearly Good Harmless
bracket): Weighted
readable quality to health
print photo Sum of X’s
 Y 1 . Clearly readable print (15) Weightage 15 10 10
 Y 2 . Go od quality photo (10) of Y’s
 Y 3 . Harmless to health (10) Input parameters Good 9 9 1 235
(X’s) quality ink
After brainstorming, the input process parameters (X’s) were found Less 9 3 1 175
as below: vibration
Paper 1 1 9 115
quality
 X 1 . Go od quality ink
 X 2 . Less vibration during operation of printing press Scale: 1 to 9
 X 3 . Paper quality 9 = strong relationship, 3 = weak relationship, 1 = very weak relationship.

© Dr. Dhananjay Jolhe 31 © Dr. Dhananjay Jolhe 32

8
 16-04-2024

XY MATRIX: EXAMPLE XY MATRIX: EXAMPLE

 Correlation coefficient (r) between two data sets: Vibration in Press (X2) versus Readability (Y1)
12
   Experiment Vibration Readability Experimen Vibration Readability
run in press index* t run in press index*
𝑟= (cps) (cps)
10
 

Readability index
8
1 2 10 16 5 8
 significance o f the correlation: 2 4 8 17 4 9
6

 t = r* 𝑁 , -3 < t < +3, then correlation is not significant 3 5 7 18 2 8 4

4 6 4 19 7 3
5 8 6 20 9 1 2
r Correlation 6 9 3 21 6 5
0
+1.0 Strong positive 7 3 8 22 5 9 0 5 10
8 5 8 23 3 7 Vibration in press (cps)
+0.5 Weak positive 9 4 7 24 4 9
10 9 2 25 2 8
0 No correlation 11 7 4 26 7 2 r = -0.843 => highly negative
12 4 8 27 8 1 correlation
-0.5 Weak negative 13 2 9 28 6 4 t = -0.834* 30 = -5.029 =>
14 6 3 29 4 7 correlation is significant.
-1.0 Strong negative 15 3 7 30
© Dr. Dhananjay Jolhe 33 © Dr. Dhananjay Jolhe 8
5 34

SIX SIGMA: CONCLUSION

 Well-structured methodology and well defined

metrics
 Defect reduction leads to yield improvement
 Yield improvement results in improved
customer satisfaction

© Dr. Dhananjay Jolhe 35 © Dr. Dhananjay Jolhe 36