g GE Global Research

Green Belt DMAIC Workshop

ANALYZE
Part 1

Define
Measure
Analyze
Improve
Control

i 1
GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.1
 g GE Global Research

The 12-Step Process

Step Description Focus Tools SSQC Deliverables
Define
A Identify Project CTQs Project CTQs (1)
B Develop Team Charter Approved Charter (2)
C Define Process Map High Level Process Map
(3)
Measure
1 Select CTQ Y Fishbone, FMEA, Pareto Project Y (4)
Characteristic Customer, QFD
2 Define Performance Y, X Customer, Blueprints Performance Standard
Standards for Project Y (5)
3 Establish Data Y, X Gage Study Data Collection Plan &
Collection Plan, MSA (6), Data for Project
Validate Measurement Y (7)
System*, & Collect Data
Analyze
4 Establish Process Y Capability Indices Process Capability
Capability for Project Y (8)
5 Define Performance Y Team, Benchmarking Improvement Goal
Objective for Project Y (9)
6 Identify Variation Sources X Hypothesis Tests Prioritized List of all Xs
(10)
Improve
7 Screen Potential Causes X DOE-Fractional, Process List of Vital Few Xs (11)
Map, Fishbone, FMEA
8 Discover Variable X DOE-Full, Prediction Proposed Solution (13)
Relationships Eqns., DFSS,
& Propose Solution Statistical Tolerancing
9 Establish Operating Y, X Piloted Solution (14)
Tolerances & Pilot Solution
Control
10 Validate Measurement* Y, X Gage Study MSA*
System
11 Determine Process Y, X Capability Indices Process Capability Y, X
Capability (15)
12 Implement Process X Risk Analysis, Sustained Solution (15),
Control System Mistake Proof, SPC Documentation (16),
& Project Closure QPT Leveraged Solution (17),
* You may want to validate the Measurement System throughout the DMAIC steps. Financial Audit (12)

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.2

 g GE Global Research

Analyze: Learning Objectives

 Review 3 steps of Analyze

 Review the deliverables of each step in
Analyze
 Review the concept of measuring process
capability in Six Sigma terms.
 Recognize Analyze is the phase we start
to work on x’s.

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.3

 g GE Global Research

Objectives of Analyze

 Establish an improvement goal—

the performance objective
 Study the stability, shape, center,
and spread of their process
 Determine the vital Xs that impact
the project Y
 Make recommendations for the
Improve phase

4
GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.4
 g GE Global Research

Analyze Phase

Define Measure Analyze Improve Control

4. Establish 5. Define Perfor- 6. Identify Vari-

Process Capability mance Objectives ation Sources

Deliverable: Baseline Deliverable: Deliverable: List of

Current Process Statistically Define Statistically
the Goal of the Significant X’s,
Tools: Project Chosen Based on
 Basic statistics
Analysis of
Tools: Historical Data
 Graphical Analysis
 Benchmarking
 Sampling
Tools:
 Continuous Zst, Zlt
 Process Analysis
 Normality
 Graphical Analysis
 Discrete Zst, Zlt
 Hypothesis Testing
 Regression
Analysis
 GLM

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.5

 g GE Global Research

Step 4: Establish Process Capability

x 
Z
 USL
USL

Y = f (X1 . . . X n )

The variation inherent to any dependent variable (Y) is determined by

the variations inherent to each of the independent variables. (X)

Excellent
Poor Process Process
Capability Capability
Very High Very Low
Probability Very High Very Low
Probability
of Defects Probability of Defects Probability
of Defects of Defects

LSL USL LSL USL

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.6

 g GE Global Research

Step 5: Define Performance Objective

The Driving Need for

Breakthrough
The
TheBasic
BasicObjective
Objective
PPM
1000000

100000

10000

1000
Your
YourProcess
Process
100
Performance
Performance
10
Objective
Objective
1
22 33 44 55 66 77

Sigma Scale of Measure

How do you determine the Performance Objective

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.7

 g GE Global Research

Defining Performance Objectives

p(x) Benchmark
Entitlement
Baseline

Defects

Benchmark:
Benchmark: World-Class
World-Classperformance
performance

Entitlement:
Entitlement: The
Thelevel
levelofofperformance
performanceaabusiness
business
should
shouldbebeable
ableto
toachieve
achieve
given
giventhe
the investments
investmentsalready
alreadymade
made

Baseline:
Baseline:The
Thecurrent
currentlevel
levelofof
performance
performance

Benchmarking
BenchmarkingSets
Setsthe
theUltimate
UltimateGoal,
Goal,While
WhileBaselining
Baselining
Takes
TakesCurrent
CurrentMeasurements
MeasurementstotoMonitor
MonitoraaProcess
Process

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.8

 g GE Global Research

Step 6: Identify Sources of Variation

Y= f (X)

To get results, should we focus our behavior on the Y or X ?

 Y  X1 . . . Xn
 Dependent  Independent
 Output  Input-
 Effect Process
 Symptom
 Cause
 Monitor
 Problem
 Control
Historically the Y, … with Six Sigma the Xs

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.9

 g GE Global Research

Process Capability
 Continuous Data
• Process Centering

 Discrete Data
• Discrete Distributions
• Yield

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.10

 g GE Global Research

Process Capability: Learning Objectives

 Review definition and calculation of Zbench

 Describe the concept of process centering
— Describe ration sub grouping and its relationship to
determining short and long term process capability
— Describe special and common causes and how they
relate to Zst and Zlt.

 Determine process capability using Six

Sigma Process Report for continuous
data
— Understand the proper use of subgroup size
in Process Report.
 Describe the calculation of DPMO for
discrete data.
 Determine process capability using Six
Sigma Product Report for discrete data

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.11

 g GE Global Research

Process Capability
 Continuous Data
• Process Centering

 Discrete Data
• Discrete Distributions
• Yield

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.12

 g GE Global Research

Calculating Z

You can calculate a Z value for any given value of

x. Z is the number of standard deviations which
will fit between the mean and the value of x.

x 
Z


    



GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.13

 g GE Global Research

Calculating Capability

x = 8.5
s = 0.1
xx

Probability
Probability Probability
Probability
ofofaadefect
defect LSL
LSL USL
USL ofofaadefect
defect
less than
less than greater
ZUSL greater
LSL
LSL ZLSL than
thanUSL
USL

Standard Deviations -4 -3 -2 -1 0 1 2 3 4
Units of Measure 8.1 8.2 8.3 8.4 8.5 8.6 8.7 8.8 8.9

ZUSL = USL - x = 8.7 - 8.5 = 0.2 = 2

s 0.1 0.1

ZLSL = x - LSL = 8.5 - 8.2 = 0.3 = 3

s 0.1 0.1

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.14

 g GE Global Research

Z-Bench

P(d)LSL P(d)USL
_
LSL T x USL

x  LSL USL  x
Z LSL  Z USL 
S S

P ( d ) L S L  fro m Z tab le P ( d ) U S L  fro m Z tab le

P (d ) Total  P (d ) LSL  P (d )USL

Z B en ch  fro m Z tab le

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.15

 g GE Global Research

Probability of a Defect
Let us suppose that we
calculate the standard normal
deviate for a given
performance limit and discover
that “Sigma” = 1.91. The
question becomes, “What
Probability
portion of the total area under Of A Defect
the normal curve lies beyond a Example = .028066
“Sigma” value of 1.91?”
Answering this question will Specification
give us the probability of
producing a defect.
Limit

“Sigma”

Units of Measure

0.20 .420740315 1.71 .043632958 3.22 .000640954 4.73 .000001153

0.25 .401293634 1.76 .039203955 3.27 .000537758 4.78 .000000903

“Sigma” = Z .382088486 1.81

1.86
1.91
.035147973
.031442864
.028066724
3.32
3.37
3.42
.000450127
.000375899
.000313179
4.83
4.88
4.93
.000000705
.000000550
.000000428
1.96 .024998022 3.47 .000260317 4.98 .000000332
2.01
2.06 .022215724 3.52 .000215873 5.03 .000000258
2.11 .019699396 3.57 .000178601 5.08 .000000199
2.16 .017429293 3.62 .000147419 5.13 .000000154
2.21
2.26 .015386434 3.67 .000121399 5.18 .000000118
2.31 .013552660 3.72 .000099739 5.23 .000000091
2.36 .011910681 3.77 .000081753 5.28 .000000070

To the right 2.41

.010444106
.009137469
.007976235
3.82
3.87
3.92
.000066855
.000054545
.000044399
5.33
5.38
5.43
.000000053
.000000041
.000000031

of the 2.46
2.51
2.56
.006946800
.006036485
.005233515
3.97
4.02
4.07
.000036057
.000029215
.000023617
5.48
5.53
5.58
.000000024
.000000018
.000000014

“Sigma” “Sigma” = 1.91 2.61

2.66
.004527002
.003906912
4.12
4.17
.000019047 5.63 .000000010

2.71 .003364033 4.22

value, you 1.35 .088507862

2.76
2.81
2.86
.002889938
.002476947
.002118083
4.27
4.32
4.37
Table Of
will note the 1.40
1.45
1.50
.080756531
.073529141
.066807100
2.91
2.96
3.01
.001807032
.001538097
.001306156
4.42
4.47
4.52
Area Under
tail area The Normal
Curve

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.16

 g GE Global Research

z Single-Tail Z Table
(Values of Z from 0.00 to 4.99)

Z 0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09

0.00 5.00e-001 4.96e-001 4.92e-001 4.88e-001 4.84e-001 4.80e-001 4.76e-001 4.72e-001 4.68e-001 4.64e-001
0.10 4.60e-001 4.56e-001 4.52e-001 4.48e-001 4.44e-001 4.40e-001 4.36e-001 4.33e-001 4.29e-001 4.25e-001
0.20 4.21e-001 4.17e-001 4.13e-001 4.09e-001 4.05e-001 4.01e-001 3.97e-001 3.94e-001 3.90e-001 3.86e-001
0.30 3.82e-001 3.78e-001 3.74e-001 3.71e-001 3.67e-001 3.63e-001 3.59e-001 3.56e-001 3.52e-001 3.48e-001
0.40 3.45e-001 3.41e-001 3.37e-001 3.34e-001 3.30e-001 3.26e-001 3.23e-001 3.19e-001 3.16e-001 3.12e-001
0.50 3.09e-001 3.05e-001 3.02e-001 2.98e-001 2.95e-001 2.91e-001 2.88e-001 2.84e-001 2.81e-001 2.78e-001
0.60 2.74e-001 2.71e-001 2.68e-001 2.64e-001 2.61e-001 2.58e-001 2.55e-001 2.51e-001 2.48e-001 2.45e-001
0.70 2.42e-001 2.39e-001 2.36e-001 2.33e-001 2.30e-001 2.27e-001 2.24e-001 2.21e-001 2.18e-001 2.15e-001
0.80 2.12e-001 2.09e-001 2.06e-001 2.03e-001 2.00e-001 1.98e-001 1.95e-001 1.92e-001 1.89e-001 1.87e-001
0.90 1.84e-001 1.81e-001 1.79e-001 1.76e-001 1.74e-001 1.71e-001 1.69e-001 1.66e-001 1.64e-001 1.61e-001

1.00 1.59e-001 1.56e-001 1.54e-001 1.52e-001 1.49e-001 1.47e-001 1.45e-001 1.42e-001 1.40e-001 1.38e-001
1.10 1.36e-001 1.33e-001 1.31e-001 1.29e-001 1.27e-001 1.25e-001 1.23e-001 1.21e-001 1.19e-001 1.17e-001
1.20 1.15e-001 1.13e-001 1.11e-001 1.09e-001 1.07e-001 1.06e-001 1.04e-001 1.02e-001 1.00e-001 9.85e-002
1.30 9.68e-002 9.51e-002 9.34e-002 9.18e-002 9.01e-002 8.85e-002 8.69e-002 8.53e-002 8.38e-002 8.23e-002
1.40 8.08e-002 7.93e-002 7.78e-002 7.64e-002 7.49e-002 7.35e-002 7.21e-002 7.08e-002 6.94e-002 6.81e-002
1.50 6.68e-002 6.55e-002 6.43e-002 6.30e-002 6.18e-002 6.06e-002 5.94e-002 5.82e-002 5.71e-002 5.59e-002
1.60 5.48e-002 5.37e-002 5.26e-002 5.16e-002 5.05e-002 4.95e-002 4.85e-002 4.75e-002 4.65e-002 4.55e-002
1.70 4.46e-002 4.36e-002 4.27e-002 4.18e-002 4.09e-002 4.01e-002 3.92e-002 3.84e-002 3.75e-002 3.67e-002
1.80 3.59e-002 3.51e-002 3.44e-002 3.36e-002 3.29e-002 3.22e-002 3.14e-002 3.07e-002 3.01e-002 2.94e-002
1.90 2.87e-002 2.81e-002 2.74e-002 2.68e-002 2.62e-002 2.56e-002 2.50e-002 2.44e-002 2.39e-002 2.33e-002

2.00 2.28e-002 2.22e-002 2.17e-002 2.12e-002 2.07e-002 2.02e-002 1.97e-002 1.92e-002 1.88e-002 1.83e-002
2.10 1.79e-002 1.74e-002 1.70e-002 1.66e-002 1.62e-002 1.58e-002 1.54e-002 1.50e-002 1.46e-002 1.43e-002
2.20 1.39e-002 1.36e-002 1.32e-002 1.29e-002 1.25e-002 1.22e-002 1.19e-002 1.16e-002 1.13e-002 1.10e-002
2.30 1.07e-002 1.04e-002 1.02e-002 9.90e-003 9.64e-003 9.39e-003 9.14e-003 8.89e-003 8.66e-003 8.42e-003
2.40 8.20e-003 7.98e-003 7.76e-003 7.55e-003 7.34e-003 7.14e-003 6.95e-003 6.76e-003 6.57e-003 6.39e-003
2.50 6.21e-003 6.04e-003 5.87e-003 5.70e-003 5.54e-003 5.39e-003 5.23e-003 5.08e-003 4.94e-003 4.80e-003
2.60 4.66e-003 4.53e-003 4.40e-003 4.27e-003 4.15e-003 4.02e-003 3.91e-003 3.79e-003 3.68e-003 3.57e-003
2.70 3.47e-003 3.36e-003 3.26e-003 3.17e-003 3.07e-003 2.98e-003 2.89e-003 2.80e-003 2.72e-003 2.64e-003
2.80 2.56e-003 2.48e-003 2.40e-003 2.33e-003 2.26e-003 2.19e-003 2.12e-003 2.05e-003 1.99e-003 1.93e-003
2.90 1.87e-003 1.81e-003 1.75e-003 1.69e-003 1.64e-003 1.59e-003 1.54e-003 1.49e-003 1.44e-003 1.39e-003

3.00 1.35e-003 1.31e-003 1.26e-003 1.22e-003 1.18e-003 1.14e-003 1.11e-003 1.07e-003 1.04e-003 1.00e-003
3.10 9.68e-004 9.35e-004 9.04e-004 8.74e-004 8.45e-004 8.16e-004 7.89e-004 7.62e-004 7.36e-004 7.11e-004
3.20 6.87e-004 6.64e-004 6.41e-004 6.19e-004 5.98e-004 5.77e-004 5.57e-004 5.38e-004 5.19e-004 5.01e-004
3.30 4.83e-004 4.66e-004 4.50e-004 4.34e-004 4.19e-004 4.04e-004 3.90e-004 3.76e-004 3.62e-004 3.49e-004
3.40 3.37e-004 3.25e-004 3.13e-004 3.02e-004 2.91e-004 2.80e-004 2.70e-004 2.60e-004 2.51e-004 2.42e-004
3.50 2.33e-004 2.24e-004 2.16e-004 2.08e-004 2.00e-004 1.93e-004 1.85e-004 1.78e-004 1.72e-004 1.65e-004
3.60 1.59e-004 1.53e-004 1.47e-004 1.42e-004 1.36e-004 1.31e-004 1.26e-004 1.21e-004 1.17e-004 1.12e-004
3.70 1.08e-004 1.04e-004 9.96e-005 9.57e-005 9.20e-005 8.84e-005 8.50e-005 8.16e-005 7.84e-005 7.53e-005
3.80 7.23e-005 6.95e-005 6.67e-005 6.41e-005 6.15e-005 5.91e-005 5.67e-005 5.44e-005 5.22e-005 5.01e-005
3.90 4.81e-005 4.61e-005 4.43e-005 4.25e-005 4.07e-005 3.91e-005 3.75e-005 3.59e-005 3.45e-005 3.30e-005

4.00 3.17e-005 3.04e-005 2.91e-005 2.79e-005 2.67e-005 2.56e-005 2.45e-005 2.35e-005 2.25e-005 2.16e-005
4.10 2.07e-005 1.98e-005 1.89e-005 1.81e-005 1.74e-005 1.66e-005 1.59e-005 1.52e-005 1.46e-005 1.39e-005
4.20 1.33e-005 1.28e-005 1.22e-005 1.17e-005 1.12e-005 1.07e-005 1.02e-005 9.77e-006 9.34e-006 8.93e-006
4.30 8.54e-006 8.16e-006 7.80e-006 7.46e-006 7.12e-006 6.81e-006 6.50e-006 6.21e-006 5.93e-006 5.67e-006
4.40 5.41e-006 5.17e-006 4.94e-006 4.71e-006 4.50e-006 4.29e-006 4.10e-006 3.91e-006 3.73e-006 3.56e-006
4.50 3.40e-006 3.24e-006 3.09e-006 2.95e-006 2.81e-006 2.68e-006 2.56e-006 2.44e-006 2.32e-006 2.22e-006
4.60 2.11e-006 2.01e-006 1.92e-006 1.83e-006 1.74e-006 1.66e-006 1.58e-006 1.51e-006 1.43e-006 1.37e-006
4.70 1.30e-006 1.24e-006 1.18e-006 1.12e-006 1.07e-006 1.02e-006 9.68e-007 9.21e-007 8.76e-007 8.34e-007
4.80 7.93e-007 7.55e-007 7.18e-007 6.83e-007 6.49e-007 6.17e-007 5.87e-007 5.58e-007 5.30e-007 5.04e-007
4.90 4.79e-007 4.55e-007 4.33e-007 4.11e-007 3.91e-007 3.71e-007 3.52e-007 3.35e-007 3.18e-007 3.02e-007

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.17

 g GE Global Research

z
Single-Tail Z Table
(Values of Z from 5.00 to 9.99)
Z 0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09

5.00 2.87e-007 2.72e-007 2.58e-007 2.45e-007 2.33e-007 2.21e-007 2.10e-007 1.99e-007 1.89e-007 1.79e-007
5.10 1.70e-007 1.61e-007 1.53e-007 1.45e-007 1.37e-007 1.30e-007 1.23e-007 1.17e-007 1.11e-007 1.05e-007
5.20 9.96e-008 9.44e-008 8.95e-008 8.48e-008 8.03e-008 7.60e-008 7.20e-008 6.82e-008 6.46e-008 6.12e-008
5.30 5.79e-008 5.48e-008 5.19e-008 4.91e-008 4.65e-008 4.40e-008 4.16e-008 3.94e-008 3.72e-008 3.52e-008
5.40 3.33e-008 3.15e-008 2.98e-008 2.82e-008 2.66e-008 2.52e-008 2.38e-008 2.25e-008 2.13e-008 2.01e-008
5.50 1.90e-008 1.79e-008 1.69e-008 1.60e-008 1.51e-008 1.43e-008 1.35e-008 1.27e-008 1.20e-008 1.14e-008
5.60 1.07e-008 1.01e-008 9.55e-009 9.01e-009 8.50e-009 8.02e-009 7.57e-009 7.14e-009 6.73e-009 6.35e-009
5.70 5.99e-009 5.65e-009 5.33e-009 5.02e-009 4.73e-009 4.46e-009 4.21e-009 3.96e-009 3.74e-009 3.52e-009
5.80 3.32e-009 3.12e-009 2.94e-009 2.77e-009 2.61e-009 2.46e-009 2.31e-009 2.18e-009 2.05e-009 1.93e-009
5.90 1.82e-009 1.71e-009 1.61e-009 1.51e-009 1.43e-009 1.34e-009 1.26e-009 1.19e-009 1.12e-009 1.05e-009

6.00 9.87e-010 9.28e-010 8.72e-010 8.20e-010 7.71e-010 7.24e-010 6.81e-010 6.40e-010 6.01e-010 5.65e-010
6.10 5.30e-010 4.98e-010 4.68e-010 4.39e-010 4.13e-010 3.87e-010 3.64e-010 3.41e-010 3.21e-010 3.01e-010
6.20 2.82e-010 2.65e-010 2.49e-010 2.33e-010 2.19e-010 2.05e-010 1.92e-010 1.81e-010 1.69e-010 1.59e-010
6.30 1.49e-010 1.40e-010 1.31e-010 1.23e-010 1.15e-010 1.08e-010 1.01e-010 9.45e-011 8.85e-011 8.29e-011
6.40 7.77e-011 7.28e-011 6.81e-011 6.38e-011 5.97e-011 5.59e-011 5.24e-011 4.90e-011 4.59e-011 4.29e-011
6.50 4.02e-011 3.76e-011 3.52e-011 3.29e-011 3.08e-011 2.88e-011 2.69e-011 2.52e-011 2.35e-011 2.20e-011
6.60 2.06e-011 1.92e-011 1.80e-011 1.68e-011 1.57e-011 1.47e-011 1.37e-011 1.28e-011 1.19e-011 1.12e-011
6.70 1.04e-011 9.73e-012 9.09e-012 8.48e-012 7.92e-012 7.39e-012 6.90e-012 6.44e-012 6.01e-012 5.61e-012
6.80 5.23e-012 4.88e-012 4.55e-012 4.25e-012 3.96e-012 3.69e-012 3.44e-012 3.21e-012 2.99e-012 2.79e-012
6.90 2.60e-012 2.42e-012 2.26e-012 2.10e-012 1.96e-012 1.83e-012 1.70e-012 1.58e-012 1.48e-012 1.37e-012

7.00 1.28e-012 1.19e-012 1.11e-012 1.03e-012 9.61e-013 8.95e-013 8.33e-013 7.75e-013 7.21e-013 6.71e-013
7.10 6.24e-013 5.80e-013 5.40e-013 5.02e-013 4.67e-013 4.34e-013 4.03e-013 3.75e-013 3.49e-013 3.24e-013
7.20 3.01e-013 2.80e-013 2.60e-013 2.41e-013 2.24e-013 2.08e-013 1.94e-013 1.80e-013 1.67e-013 1.55e-013
7.30 1.44e-013 1.34e-013 1.24e-013 1.15e-013 1.07e-013 9.91e-014 9.20e-014 8.53e-014 7.91e-014 7.34e-014
7.40 6.81e-014 6.31e-014 5.86e-014 5.43e-014 5.03e-014 4.67e-014 4.33e-014 4.01e-014 3.72e-014 3.44e-014
7.50 3.19e-014 2.96e-014 2.74e-014 2.54e-014 2.35e-014 2.18e-014 2.02e-014 1.87e-014 1.73e-014 1.60e-014
7.60 1.48e-014 1.37e-014 1.27e-014 1.17e-014 1.09e-014 1.00e-014 9.30e-015 8.60e-015 7.95e-015 7.36e-015
7.70 6.80e-015 6.29e-015 5.82e-015 5.38e-015 4.97e-015 4.59e-015 4.25e-015 3.92e-015 3.63e-015 3.35e-015
7.80 3.10e-015 2.86e-015 2.64e-015 2.44e-015 2.25e-015 2.08e-015 1.92e-015 1.77e-015 1.64e-015 1.51e-015
7.90 1.39e-015 1.29e-015 1.19e-015 1.10e-015 1.01e-015 9.33e-016 8.60e-016 7.93e-016 7.32e-016 6.75e-016

8.00 6.22e-016 5.74e-016 5.29e-016 4.87e-016 4.49e-016 4.14e-016 3.81e-016 3.51e-016 3.24e-016 2.98e-016
8.10 2.75e-016 2.53e-016 2.33e-016 2.15e-016 1.98e-016 1.82e-016 1.68e-016 1.54e-016 1.42e-016 1.31e-016
8.20 1.20e-016 1.11e-016 1.02e-016 9.36e-017 8.61e-017 7.92e-017 7.28e-017 6.70e-017 6.16e-017 5.66e-017
8.30 5.21e-017 4.79e-017 4.40e-017 4.04e-017 3.71e-017 3.41e-017 3.14e-017 2.88e-017 2.65e-017 2.43e-017
8.40 2.23e-017 2.05e-017 1.88e-017 1.73e-017 1.59e-017 1.46e-017 1.34e-017 1.23e-017 1.13e-017 1.03e-017
8.50 9.48e-018 8.70e-018 7.98e-018 7.32e-018 6.71e-018 6.15e-018 5.64e-018 5.17e-018 4.74e-018 4.35e-018
8.60 3.99e-018 3.65e-018 3.35e-018 3.07e-018 2.81e-018 2.57e-018 2.36e-018 2.16e-018 1.98e-018 1.81e-018
8.70 1.66e-018 1.52e-018 1.39e-018 1.27e-018 1.17e-018 1.07e-018 9.76e-019 8.93e-019 8.17e-019 7.48e-019
8.80 6.84e-019 6.26e-019 5.72e-019 5.23e-019 4.79e-019 4.38e-019 4.00e-019 3.66e-019 3.34e-019 3.06e-019
8.90 2.79e-019 2.55e-019 2.33e-019 2.13e-019 1.95e-019 1.78e-019 1.62e-019 1.48e-019 1.35e-019 1.24e-019

9.00 1.13e-019 1.03e-019 9.40e-020 8.58e-020 7.83e-020 7.15e-020 6.52e-020 5.95e-020 5.43e-020 4.95e-020
9.10 4.52e-020 4.12e-020 3.76e-020 3.42e-020 3.12e-020 2.85e-020 2.59e-020 2.37e-020 2.16e-020 1.96e-020
9.20 1.79e-020 1.63e-020 1.49e-020 1.35e-020 1.23e-020 1.12e-020 1.02e-020 9.31e-021 8.47e-021 7.71e-021
9.30 7.02e-021 6.39e-021 5.82e-021 5.29e-021 4.82e-021 4.38e-021 3.99e-021 3.63e-021 3.30e-021 3.00e-021
9.40 2.73e-021 2.48e-021 2.26e-021 2.05e-021 1.86e-021 1.69e-021 1.54e-021 1.40e-021 1.27e-021 1.16e-021
9.50 1.05e-021 9.53e-022 8.66e-022 7.86e-022 7.14e-022 6.48e-022 5.89e-022 5.35e-022 4.85e-022 4.40e-022
9.60 4.00e-022 3.63e-022 3.29e-022 2.99e-022 2.71e-022 2.46e-022 2.23e-022 2.02e-022 1.83e-022 1.66e-022
9.70 1.51e-022 1.37e-022 1.24e-022 1.12e-022 1.02e-022 9.22e-023 8.36e-023 7.57e-023 6.86e-023 6.21e-023
9.80 5.63e-023 5.10e-023 4.62e-023 4.18e-023 3.79e-023 3.43e-023 3.10e-023 2.81e-023 2.54e-023 2.30e-023
9.90 2.08e-023 1.88e-023 1.70e-023 1.54e-023 1.39e-023 1.26e-023 1.14e-023 1.03e-023 9.32e-024 8.43e-024

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.18

 g GE Global Research

The Universal Equation for Z


. so what are the possibilities? SL - 
Z=


USL
SL T (Target)
LSL (Mean)

Z =
st (short-term)
lt (long-term) st
lt

and how do we choose the right one?

Dr. Mikel J. Harry

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.19

 g GE Global Research

Selecting an Appropriate Z
 SL - T  SL - T
Eq. 8.1 Zst = Eq. 8.3 Z=
st  lt

This Z value is designated as Z.st. It describes how precise This Z value is designated as Z.lt.d. It is a measure
This Z value is designated as Z.st. It describes how precise
the process is at any given moment in time. For this ofThis Z valueprocess
long-term is designated as Z.lt.d.
capability. It is athe
It reflects measure
the process is at any given moment in time. For this of long-term process capability. It
influence of white noise, as well as the dynamic reflects the
reason, it is referred to as “instantaneous capability.” It is influencedueof white noise, as process
well as the dynamic
reason, it is referred to as “instantaneous capability.” It is variations to nonrandom centering
also called “short-term capability.” In context of the Six variations due to
also called “short-term capability.” In context of the Six error; i.e., shifts andnonrandom
drifts in theprocess
process centering
mean
Sigma Program, it is the value used when referring to the error;sampling
across i.e., shiftssubgroups.
and drifts in It the process
assumes thatmean
the
Sigma Program, it is the value used when referring to the across
“SIGMA” of process. It represents the true potential of the
“SIGMA” of process. It represents the true potential of the errors in sampling subgroups.
process centering are It assumesand
dynamic thatwill
the
process technology to meet the given performance errors in process centering are dynamic
eventually average out (over a great many cycles) to and will
process technology to meet the given performance eventually average out In (over a great many
specification(s); I.e., what the process can do if everything the target specification. context of the Sixcycles)
Sigmato
specification(s); I.e., what the process can do if everything the target specification. In context of
Program, it is not often used, except in some design the Six Sigma
is controlled to such an extent that only background noise
is controlled to such an extent that only background noise Program, itapplications.
engineering is not often used,
This except
metric in some design
assumes the
is present, common cause variation. It reflects the process
is present, common cause variation. It reflects the process engineering
data applications.
were gathered This metric
in accordance to the assumes
principalsthe
capability under the assumption of random variation and data were gathered in accordance to the principals
capability under the assumption of random variation and and spirit of a “rational sampling” plan. For a
does not give consideration to the process center. This and spirittolerance
of a “rational sampling” plan.this For a
does not give consideration to the process center. This unilateral with no target value,
metric assumes the data were gathered in accordance to the unilateral tolerance withInnosuch
target
metric assumes the data were gathered in accordance to the equation cannot be used. an value,
event, thisEq. 4
principals and spirit of a “rational sampling” plan. For a equation
should cannot be to
be employed used. In such
estimate an event,process
long-term Eq. 4
principals and spirit of a “rational sampling” plan. For a should be employed to estimate long-term process
unilateral tolerance with no target Eq. 2 should be used. capability.
unilateral tolerance with no target Eq. 2 should be used. capability.

 SL -   SL - 
Eq. 8.2 Z= Eq. 8.4 Zlt =
 st  lt

This Z value is designated as Z.lt. It is a measure of This Z value is designated as Z.lt.s. It describes the
This Z value
long-term is designated
capability and, when as Z.lt.
used Itproperly,
is a measure of This Z value
sustained is designated
reproducibility of aasprocess.
Z.lt.s. ItBecause
describes of the
long-term capability and, when used
reflects process accuracy when compared to Z.st; properly, this, it is also called “long-term capability.” In of
sustained reproducibility of a process. Because
reflects
e.g., Z.st -process accuracyExpressed
Z.lt = Z.shift. when compared to Z.st;
differently, it this, it of
context is the
alsoSix
called
Sigma“long-term
Program,capability.”
it is the value In
e.g., Z.st
reflects how - Z.lt
well=the
Z.shift.
processExpressed
remains differently,
centered over it context
used of the Six
to estimate the Sigma Program,
long-term process it is the value
“PPM.” It
reflects
time. how well
Of course, the process
it ignores any remains
nonrandom centered
processover used tothe
reflects estimate the of
influence long-term process
special cause “PPM.” It
variation,
time. Oferrors
centering course, it ignores
which any nonrandom
may occur between samplingprocess reflects nonrandom
dynamic the influence of special
process cause error,
centering variation,
and
centering errors which may occur
intervals. This metric assumes the data were between sampling dynamic nonrandom process centering
any static offset present in the process mean. error,From
and
intervals.
gathered This metricto
in accordance assumes the dataand
the principals were spirit of anyperspective,
this static offsetitpresent
considersin the process
all of mean.
the “vital few”From
gathered in accordance to the principals
a “rational sampling” plan. However, in the and spirit of this perspective, it considers all of the
sources of manufacturing error. It is a measure of “vital few”
a “rational
instance of a sampling” plan. However,
unilateral tolerance with nointarget
the sources
how well ofthemanufacturing error. It(over
process is controlled is a measure
many of
instance of athe
specification, unilateral
given Ztolerance
value willwith no target
reflect only how well
cycles) when thecompared
process istocontrolled
Z.st. This(over metric many
assumes
specification,
short-term the given
capability. Z value
In this will reflectthe
circumstance, only the data were gathered in accordance to the assumes
cycles) when compared to Z.st. This metric
short-term
mean becomes capability.
the target.In Consequently,
this circumstance,it willthe the data were
principals gathered
and spirit of a in accordance
“rational to the plan.
sampling”
mean becomes the target. Consequently,
produce the same result as Eq. 1; therefore, it it will principals and spirit of a “rational sampling”
This equation is applicable to all types of tolerances. plan.
produce
should the same result
be designated as Eq.
as Z.st and 1;
sotherefore,
interpreted. it This equation is applicable to all types of tolerances.
should be designated as Z.st and so interpreted.

Dr. Mikel J. Harry

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.20

 g GE Global Research

Z-Bench

Short Term
Long Term

P(d)LSL P(d)USL
_
LSL T x USL

Z-Long Term Z-Short Term

 SL -   SL - T
Zlt = Zst =
 lt st

Z-Bench-Long Term Z-Bench-Short Term

   - LSL  USL- T  T - LSL
USL- 
ZUSL = ZLSL= ZUSL = ZLSL=
 lt  lt  st  st

P(d)USL = from P(d)LSL = from P(d)USL = from P(d)LSL = from

Z table Z table Z table Z table

P(d)Total = P(d)USL + P(d)LSL P(d)Total = P(d)USL + P(d)LSL

ZB-lt = from Z ZB-st = from

table Z table

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.21

 g GE Global Research

Process Capability
 Continuous Data
• Process Centering

 Discrete Data
• Discrete Distributions
• Yield

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.22

 g GE Global Research

Generalizing the Correction

Six Sigma Centered

Process
Capability

.0005 ppm .0005 ppm

T
T
LSL
LSL ± 6 USL
USL

Six Sigma Shifted 1.5




3.4 ppm
T
T
LSL
LSL 4.5 USL
USL

The 1.5 shift is used as a compensatory offset in the mean to generally account
for dynamic nonrandom variations in process centering. It represents the average
amount of change in a typical process over many cycles of that process.

Dr. Mikel J. Harry

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.23

 g GE Global Research

Rational Subgrouping Principles

Process Response

Time
RATIONAL
RATIONALSUBGROUPING
SUBGROUPING
CHOOSE
CHOOSE SUBGROUPSSO
SUBGROUPS SOTHAT:
THAT:

1.1. There
Thereisismaximum
maximumchance
chancefor
forthe
themeasurements
measurementsinineach
eachsubgroup
subgroup
totobe alike. A subgroup should only contain common cause
be alike. A subgroup should only contain common cause
variation.
variation.

2.2. There
Thereisismaximum
maximumchancechancefor
forsubgroups
subgroupstotodiffer
differfrom
fromone
onetotothe
the
the
thenext.
next. The
Thedifference
differencebetween
betweenthe
thesubgroups
subgroupsisisthethespecial
special
cause variation.
cause variation.

Special Cause Common Cause

Variation Variation
Long term
Capability X X
Short term
Capability X

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.24

 g GE Global Research

The Components of Variation

Production
Unit
Production Sequence

••••••••••••••••••••••••••••••••••• Goal: Establish a sampling window

small enough to force the exclusion of
systematic nonrandom influences.

Sampling Window n = 5

SS T SS B SS W

x  x x  x
g n g g n

x 
2 2 2
 
j 1 i 1
ij n 
j 1
j 
j 1
i 1
ij  xj

Total
Total Between
Between Within
Within
Sustained Reproducibility Instantaneous Reproducibility

Capability
Capability Accuracy
Accuracy Precision
Precision
Long term “Shift” Short term
Total Variation Special Cause Common Cause
Performance Variation Variation
Control Technology
Dr. Mikel J. Harry Vital few Trivial many

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.25

 g GE Global Research

Visualizing the Components

g n
Process 2
  (x ij- x j)
Centering j=1 i =1
is Within
Dynamic Group
by Nature

Time 1

Time 2

Time 3

Time 4

g n
2 g
  (x ij- x) n  ( x j -x )
2
j=1 i =1
j=1

Total Between
Groups

LSL T USL
Dr. Mikel J. Harry

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.26

 g GE Global Research

Zst
• Short term capability (short term Z Bench).
• Process potential when process centering
principle applied.
• Containing only common cause variation.

Zlt
• Long term capability (long term Z Bench).
• True capability when process centering
principle applied.
• Containing both common cause and special
cause variations

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.27

 g GE Global Research

Exercise: Six Sigma Process Report

Exercise: Process Capability

• Customers of CHI (Cellulose Helicopters Inc.) have

indicated that the flight time is an important CTQ of CHI
helicopters.

• Customers want the flight time no higher than 30 and

no less than 20 flight time units.

• CHI has collected a set of data over a course of one

year to understand their process capability.

• A sample of the data is shown below (data file= CHI

Copter Capability.mtw)

• Use Minitab to calculate Zst, Zlt and DPMO

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.28

 g GE Global Research

Minitab Six Sigma Process Report

Important Note:
 Subgroup=1
— Using a subgroup size = 1 in the Minitab Process
report results in misleading and, in most cases,
wrong ZST, ZSHIFT, and DPMOST values.
— This discrepancy is due to the way that Minitab
estimates ST.
— Only the long term values are valid with a
subgroup size = 1.

 One-sided Specifications
— Minitab assumes the mid-point between
specifications is the target and calculate ZST,
based on the process centering principle.
— One sided spec will results in invalid ST unless a
target is entered.

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.29

 g GE Global Research

Process Capability
 Continuous Data
• Process Centering

 Discrete Data
• Discrete Distributions
• Yield

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.30

 g GE Global Research

Data Analysis Roadmap

Discrete
DiscreteData
Data Continuous
Continuous
Data
Data
Defects per Six Sigma Process
Opportunity Report:
Defects per Million Zlong term
Opportunities
Zshort term
L1 Spreadsheet - Z
Zshift
Six Sigma Product
Report

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.31

 g GE Global Research

Definitions

Unit (U)
The number of parts, sub-assemblies,
assemblies, or systems inspected or tested.
– Squares: 4 units

Opportunity (OP)
A characteristic you inspect or test.
– Circles: 5 opportunities per unit

Defect (D)
Anything that results in customer
dissatisfaction. Anything that results in a non-
conformance.
– Black circles: 9 defects

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.32

 g GE Global Research

Formulas
Defects per Unit
DPU = D/U
9/4 = 2.25

Total Opportunities
TOP = U*OP
4*5 = 20

Defects per Opportunity (Probability of a Defect)

DPO = D/TOP
9/20 = .45

Defects per Million Opportunities

DPMO = DPO*1,000,000
.45*1,000,000 =
450,000

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.33

 g GE Global Research

Linking DPO to Probability of a Defect

DPO
DPO==Probability
Probabilityof
ofaaDefect
Defect
Probability
Specification Of A Defect
Limit Example = .028066

Area of Yield

Units of Measure

0.20 .420740315 1.71 .043632958 3.22 .000640954 4.73 .000001153

0.25 .401293634 1.76 .039203955 3.27 .000537758 4.78 .000000903
.382088486 1.81 .035147973 3.32 .000450127 4.83 .000000705
1.86 .031442864 3.37 .000375899 4.88 .000000550
1.91 .028066724 3.42 .000313179 4.93 .000000428
1.96 .024998022 3.47 .000260317 4.98 .000000332
2.01
2.06 .022215724 3.52 .000215873 5.03 .000000258
2.11 .019699396 3.57 .000178601 5.08 .000000199
2.16 .017429293 3.62 .000147419 5.13 .000000154
2.21
2.26 .015386434 3.67 .000121399 5.18 .000000118
2.31 .013552660 3.72 .000099739 5.23 .000000091
2.36 .011910681 3.77 .000081753 5.28 .000000070
.010444106 3.82 .000066855 5.33 .000000053
.009137469 3.87 .000054545 5.38 .000000041
2.41 .007976235 3.92 .000044399 5.43 .000000031
2.46 .006946800 3.97 .000036057 5.48 .000000024
2.51 .006036485 4.02 .000029215 5.53 .000000018
2.56 .005233515 4.07 .000023617 5.58 .000000014
“Sigma” = 1.91 2.61
2.66
.004527002
.003906912
4.12
4.17
.000019047 5.63 .000000010

2.71 .003364033 4.22

Table Of
2.76 .002889938 4.27
2.81 .002476947 4.32
1.35 .088507862 2.86 .002118083 4.37
1.40
1.45
.080756531
.073529141
2.91
2.96
.001807032
.001538097
4.42
4.47 Area Under
1.50 .066807100 3.01 .001306156 4.52
The Normal
Curve
© 1994 Dr. Mikel J. Harry V3.0

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.34

 g GE Global Research

Converting DPMO to Z
Sigma table
Long term Actual Reported
DPMO Sigma (long term) Sigma (short term)
500,000 0 1.5
460,172 0.1 1.6
420,740 0.2 1.7
382,089 0.3 1.8
344,578 0.4 1.9
308,538 0.5 2
274,253 0.6 2.1
241,964 0.7 2.2
211,855 0.8 2.3
184,060 0.9 2.4
158,655 1 2.5
135,666 1.1 2.6
115,070 1.2 2.7
96,801 1.3 2.8
80,757
66,807
1.4
1.5
2.9
3
Z = Sigma
Capability
54,799 1.6 3.1
44,565 1.7 3.2
35,930 1.8 3.3
28,716 1.9 3.4
22,750 2 3.5
17,864 2.1 3.6
13,903 2.2 3.7
10,724 2.3 3.8
8,198 2.4 3.9
6,210
4,661
2.5
2.6
4
4.1
ZZ DPMO
DPMO
3,467 2.7 4.2
2,555 2.8 4.3
1,866
1,350
2.9
3
4.4
4.5
2 308,537
968
687
3.1
3.2
4.6
4.7
3 66,807
483
337
3.3
3.4
4.8
4.9
4 6,210
233
159
3.5
3.6
5
5.1
5 233
108
72
3.7
3.8
5.2
5.3
6 3.4
48 3.9 5.4
32 4 5.5
21 4.1 5.6
13 4.2 5.7
9 4.3 5.8
5 4.4 5.9
3.4 4.5 6

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.35

 g GE Global Research

Exercise: Six Sigma Product Report

Exercise: Process Capability

• You, as the manufacturing manager at CHI (Cellulose

Helicopters Inc.), was asked to provide the process
capability of the assembly line in the upcoming operation
meeting.

• There are three sub-assembly lines, (1) rotor system; (2)

engine; (3) navigation system. Based on the standard
operating procedure, each completed unit needs to pass
120, 140 and 50 inspections at the corresponding sub-
assembly line, respectively.

• According to the production record, CHI has assembled

248 helicopter in the past six months and accumulated the
following number of “failed inspections”.
Rotor 9
Engine 8
Navigation 2
• In preparation for your meeting, use Minitab to calculate
the process capability of the assembly line.

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.36

 g GE Global Research

Process Capability
 Continuous Data
• Process Centering

 Discrete Data
• Discrete Distributions
• Yield

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.37

 g GE Global Research

Learning Objectives

 Binomial Distribution
 Application of Binomial Distribution
 Poisson’s Distribution
 Application of Poisson’s distribution

38
GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.38
 g GE Global Research

Binomial Distribution
Consider that an activity has a success rate of 80%, i.e. 0.8
probability of success, this means there is a 20% chance of
failure. The question is, if this activity is repeated say 1000
times, what is the likelihood that 800 times out of these trials,
there is a success? Such trials are called Bernoulli trials and the
probability distribution is called Binomial ( Bernoulli) distribution
given by
n!
B(n, r , p )nCr p n (1  p ) n  r where nCr 
r!n  r !
In the above example, n=1000, r=800 and p=0.8
Bernoulli's Distribution

0.14
0.12
0.1
Probability

p=0.1
0.08
p=0.5
0.06
p=0.7
0.04
0.02
0
0 10 20 30 40 50 60 70 80 90 100
Number of Trials

p < 0.5 Skewed to the right

p = 0.5 No Skewness
p > 0.5 Skewed to the left

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.39

 g GE Global Research

Application of Binomial Distribution

When both the defects and non-defects of a product or

process are countable, this distribution is used to find the
probability of defects.

If no. of letters not received is a defect,

no. of letters received is a non-defect
And both are countable.

Example : A person send 10 letters to his friend who

resides in a remote village. What is the probability that
his friend receives 8 letters out of them if the probability
of receiving the letter is 0.2 ?

Pr(Receiving a letter)  p 0.2

 Pr( Not receiving a letter) 1  p 1  0.2 0.8
Out of 10 trials,
Pr(Receiving letter 8 times)10C8 p 8 (1  p) 2 10C8 (0.2)8 (0.8) 2
 Pr(Receiving letter 8 times) (45)(0.00000256)(0.64) 0.000073728

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.40

 g GE Global Research

Poisson’s Distribution
When there is a known average failure rate, such as no. of
accidents per annum, no. of defects per sq. meter, no. of viruses
per cc of blood etc, the probability that a failure occurs in the a
unit or over a period of units such as time, area, volume etc. is
given by Poisson’s distribution.

e  t ( t ) n Mean  t
P ( n,  t ) 
n! Variance  2 t

Poisson's Distribution

0.9

0.8

0.7
lamda=0.05
0.6
Probability

lambda=0.5
0.5 lambda=1
lambda=1.5
0.4
lambda=10
0.3

0.2

0.1

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

Mean and Variance are equal.

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.41

 g GE Global Research

Application of Poisson Distribution

When the defects per unit of a product or process are

countable, but even if non-defects per unit are not countable,
this distribution is used to find the probability of defects.

While no. of accidents per annum is a defect, and it can be counted,

no. of no-accidents per annum is a non-defect, and it cannot be
counted. (It will be always zero.)

Example : Average no. of dust-particles in a clean room

is 10 per cubic meter. What is the probability of finding
1450 particles in a clean room of 5m x 10m x 3m ?

 10 particles/m 3 ; t 5 X 10 X 3 150m 3

 t 10 X 150 1500
(1500)1450 e  1500
 Pr(1450 particles)  0.00451
1450!

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.42

 g GE Global Research

Which distribution to be used and when?

When number of opportunities is not very large, the

distribution of zero defects is BINOMIAL and is given by

DPU OP
P{0} = 1  
 
 OP 

As the number of opportunities approaches infinity(i.e. very

large), the distribution of zero defects is POISSON and is
given by

P{0} = e-DPU

Example: Given an average DPU=1;

P(0) = e-DPU = e-1 = 1/e = .367878

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.43

 g GE Global Research

Process Capability
 Continuous Data
• Process Centering

 Discrete Data
• Discrete Distributions
• Yield

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.44

 g GE Global Research

Yield

Four parts are made

1. After first inspection: 1 passed, 3 failed
Rework 3 parts

2. After second inspection: 1 passed, 2 failed

Rework 2 parts

3. After third inspection: 1 passed, 1 scrapped

What
Whatis
isthe
theyield
yieldof
ofthis
thisprocess?
process?

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.45

 g GE Global Research

Types of Yield

Classical Yield = YC = 3/4 = 75%?

Classical yield is the number of defect-free parts
for the whole process divided by the total number
of parts inspected. If we say the yield is 3/4 or
75%, we lose valuable data on the true
performance of the process. This loss of insight
becomes a barrier to process improvement.

First Time Yield = YFT = 1/4 = 25%?

First time yield is the number of defect-free parts
divided by the total number of parts inspected for
the first time. If we say the yield is 1/4 or 25%,
we are really talking about the First Time Yield
(FTY). This is a better yield estimate to drive
improvement.

Throughput Yield =
YTP = P(0) = e-DPU = e-2.25 = .1054 = 10.54%
YTP is the percentage of units that pass through
an operation without any defects. This is the
best yield estimate to drive improvement.

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.46

 g GE Global Research

Comparison of the Yield Models

For Example:
3
YC = = .75, or 75%
4
– DPU – 2.25
YTP = e =e = .1054 , or 10.54%

Process Capability 10.54%

Throughput
Throughput
Yield
Yield
YY.TP
75% Customer Quality
.TP
Yield Prior to Inspection or Test
Yield Prior to Inspection or Test

Operation Verify
Classical
ClassicalYield
Yield
Operation Verify YY.C .C or Test
Yield after Inspection
Yield after Inspection or Test

DPU = 2.25
Not
OK
Rework
Rework

10.54% = 75%
Scrap
Scrap ... why not?

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.47

 g GE Global Research

Extending the Concept

A given process has two operations.

Each operation has a throughput yield
of 99 %. The rolled yield equals:
Process Centered Process Centered
Op 1 x Op 2 = Output

99% 99% 98%

Without Inspection or Test Without Inspection or Test Without Inspection or Test

. . . There is an 98% probability that any given unit

of product could pass through both operations
defect free.

If YTP = e-DPU = Throughput Yield,

Does YRT = e-TDPU = Rolled Yield?

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.48

 g GE Global Research

Let’s try it on our earlier example:

Process Centered Process Centered

Op 1 x Op 2 = Output

99% 99% 98%

Without Inspection or Test Without Inspection or Test Without Inspection or Test

OR
= e
-TDPU
YRT
Each Operation has a 0.01 probability of a
defect. Therefore:

TDPU = .01 + .01 defect per unit

= .02 defects per unit
YRT = e-.02 = .98019
or 98%

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.49

 g GE Global Research

Rolled Throughput Yield (YRT)

Receive parts from Supplier

95.5% Yield (YTP)

Following Receiving
Inspection and Line Fall-out...
45,000 DPMO wasted

97% Yield (YTP)

From Machining Operations

94.4% Yield (YTP)

28,650 DPMO wasted At Test Stands on
first attempt

YRT = .955*.97*.944 = 87.5%

51,876 DPMO wasted

Right
First
125,526 DPMO
Time
wasted opportunities

G. Reimer 12/21/94 - Charlotte NC

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.50

 g GE Global Research

Why Z = 3 Is Not Good Enough

Rolled Yield (%) Vs

Z (distribution centered )

Capability
Complexity

Number of Z = 3.0 Z = 4.5 Z = 5.5 Z = 6.0

characteristics
1 93.32 99.865 99.997 99.99966
20 25.09 97.334 99.941 99.9932
60 1.58 92.214 99.820 99.9796
100 --- 87.363 99.700 99.9660
200 --- 76.324 99.402 99.9320
500 --- 50.892 98.511 99.8301
1000 --- 6.696 93.857 99.3223

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.51

 g GE Global Research

Run Chart

Purpose: To track process over time in order to

display trends and focus attention on
changes in the process

When: To establish a baseline of performance

for improvement
To uncover changes in your process
To brainstorm possible causes for trends
To compare the historical performance of
a process with the improved process

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.52

 g GE Global Research

Run Chart
How: Determine what you want to measure
Determine period of time to measure and
in what time increments
Create a graph (vertical axis =
occurrences, horizontal axis =
time)
Collect data and plot
Connect data points with solid line
Calculate average of measurements, draw
solid horizontal line on run chart
Analyze results
Indicate with a dashed vertical line when a
change was •introduced to the
process •
70
• •
65 •
60 • •
55 •
50 •
45 •
40
35
30
25
20
15
10
5
J F M A M J J A S O

GB DMAIC - Analyze Part 1 Version 2.0 5/2002 7.53