SIX SIGMA

HOW GOOD IS GOOD ENOUGH?

99.9% is already VERY GOOD

But what could happen at a quality level of 99.9% (i.e., 1000 ppm),
in our everyday lives (about 4.6)?

• 4000 wrong medical prescriptions each year

• More than 3000 newborns accidentally falling

from the hands of nurses or doctors each year
• Two long or short landings at American airports each day

• 400 letters per hour which never arrive at their destination

HOW CAN WE GET THESE RESULTS

 13 wrong drug prescriptions per year

 10 newborn babies dropped by doctors/nurses
per year
 Two short or long landings per year in all the
airports in the U.S.
 One lost article of mail per hour
THE ANSWER IS:

Six Sigma
WHAT IS SIX SIGMA
 A Vision and Philosophical commitment
to offer the highest quality, lowest cost
products

 A Metric that demonstrates quality levels at

99.9997% performance for products and
processs

 A Benchmark of product and process

capability for comparison to ‘best in class’

 A practical application of statistical Tools

and Methods to help us measure, analyze,
improve, and control a process
 SIX SIGMA AS A PHILOSOPHY
 is a measure of how much
Internal &
External
Prevention & variation exists in a process
Appraisal
Failure
Costs
Costs
Old Belief
Costs

Old Belief
4 High Quality = High Cost

Quality
Internal & Prevention &
External Appraisal
Costs Failure Costs Costs

New Belief 4
New Belief
5
High Quality = Low Cost
6

Quality
 3 SIGMA VS. 6 SIGMA
The 3 sigma Company The 6 sigma Company
• Spends 15~25% of sales dollars • Spends 5% of sales dollars on
on cost of failure cost of failure
• Relies on inspection to find • Relies on capable process that
defects don’t produce defects
• Does not have a disciplined • Use Measure, Analyze, Improve,
approach to gather and analyze Control and Measure, Analyze,
data Design
• Benchmarks themselves • Benchmarks themselves
against their competition against the best in the world
• Believes 99% is good enough • Believes 99% is unacceptable

• Define CTQs internally • Defines CTQs externally

 FOCUS: THE END USER
 Customer: Internal or External
 Consumer: The End User

the “Voice of the Consumer” (Consumer Cue)

must be translated into
the “Voice of the Engineer” (Technical Requirement)
 SIX SIGMA AS A METRIC

2
   ( xi  x )
=
Sigma =  = Deviation n 1
( Square root of variance )

Axis graduated in Sigma

-3
-7
-6
-5
-4

-2
-1

7
1
2
3

6
0

4
between + / - 1 68.27 % result: 317300 ppm outside
(deviation)
between + / - 2 95.45 % 45500 ppm

between + / - 3 99.73 % 2700 ppm

between + / - 4 99.9937 % 63 ppm

between + / - 5 99.999943 % 0.57 ppm

between + / - 6 99.9999998 % 0.002 ppm

3-sigma Process Spec Limits
(centered)
Cp = 1.0
Cpk = 1.0
2,700 ppm
 3-sigma Process Spec Limits
(shifted 0.5 std. dev.)
Cp = 1.0
Cpk = 0.833
ppm = 6,442
(about 2.72-sigma)
 3-sigma Process Spec Limits
(shifted 1.0 std. dev.)
Cp = 1.0
Cpk = 0.667
ppm = 22,782
(about 2.28-sigma)
 3-sigma Process Spec Limits
(shifted 1.5 std. dev.)
Cp = 1.0
Cpk = 0.5
ppm = 66,811
(about 1.83-sigma)
NON-LINER DECREASE
 PPM

2 308,537 80 800000

70 From 5 to 6 700000

3 66,811 60 600000

50 500000

% Change
PPM % Change

PPM
40 400000
4 6,210 30 From 4 to 5 300000

20 200000
From 3 to 4

5 233 10
From 1 to 2
100000

0 0

1 2 3 4 5 6
6 3.4
Process Sigma
Process Defects per Million
Capability Opportunities

* Includes 1.5 shift

Focusing on  requires thorough process

understanding and breakthrough thinking
SIX SIGMA AS A TOOL
Process Mapping Tolerance Analysis

Structure Tree Components Search

Pareto Analysis Hypothesis Testing

Gauge R & R Regression

Rational Subgrouping DOE

Baselining SPC

Many familiar quality tools applied in a

structured methodology
 SIX SIGMA AS A METHOD

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

•Y •X1…Xn
•Dependent •Independent
•Output •Input-Process
•Effect •Cause
•Symptom •Problem
•Monitor •Control
A TRADITIONAL VIEW

Market Share

Sales Growth
• Output Variables
Profitability

Manage the outputs.

 A NON-TRADITIONAL VIEW

Product Quality
COQ Service

• Input Variables On-Time Delivery

Relationships
Credit Terms
Customer
Training

Customer Satisfaction

Market Share

Sales Growth
• Output Variables
Profitability

Manage the inputs; respond to the outputs.

 DISTINGUISH “VITAL FEW”
FROM “TRIVIAL MANY”

Material Environment
Measurements
Methods Process
Output
(Parameters)
Machine
People

Define the Problem / Defect Statement

Y = f ( x1*, x2, x3, x4*, x5. . . Xn)

Y= Dependent Variable Output, Defect

x= Independent Variables Potential Cause
x* = Independent Variable Critical Cause
 Measure

STRATEGY BY PHASE -
Improvement
Phase Step Focus

Process Characterization
Measure What is the frequency of Defects? Measure

• Define the defect Y

Control
(What)

Analyze
• Define performance standards Y
• Validate measurement system Y Improve

• Establish capability metric Y

Measure

Analyze Where, when and why do Defects occur?

Control
• Identify sources of variation

Analyze
(Where, When, Why) X
• Determine the critical process parameters Vital X
Improve

Process Optimization
Improve How can we improve the process? Measure

(How) • Screen potential causes X

Control
Analyze
• Discover relationships Vital X
• Establish operating tolerances Vital X Improve

Were the improvements effective?

• Re-establish capability metric Y, Vital X Measure

Control
Analyze
Control How can we maintain the improvements? Y, Vital X
(Sustain, Leverage) • Implement process control mechanisms
• Leverage project learning's Improve

• Document & Proceduralize

 Measure
Characterize Process

Evaluate Control
Understand Process Maintain New Process

Improve
Improve and Verify Process
 Measure Phase

Define Understand Collect Process

Problem Process Data Performance

 Defect  Define Process-  Data Types  Process Capability

Statement Process Mapping - Defectives - Cp/Cpk
 Project  Historical - Defects - Run Charts
Goals Performance - Continuous  Understand Problem
 Brainstorm  Measurement (Control or
Potential Defect Systems Evaluation Capability)
Causes (MSE)

Understand the Process and Potential Impact

Baselining:
Quantifying the goodness (or badness!) of the current
process, before ANY improvements are made, using
sample data. The key to baselining is collecting
representative sample data

Sampling Plan
- Size of Subgroups
- Number of Subgroups
- Take as many “X” as possible into consideration
 How do we know our process?

Process Map

Fishbone

Historical Data
 RATIONAL SUBGROUPS
Minimize variation within subroups
BLACK NOISE Maximize variation between subrgoups
(Signal)
WHITE NOISE
(Common
Cause Variation)
RESPONSE
PROCESS

TIME
RATIONAL SUBROUPING Allows samples to be taken that
include only white noise, within the samples. Black noise
occurs between the samples.
 Visualizing the Causes
Within Group

Time 1

Time 2

Time 3

Time 4

•Called  short term (st)

 st + shift = •Our potential – the best
we can be
total
•The s reported by all 6
sigma companies
•The trivial many
 Visualizing the Causes

Time 1

Time 2

Time 3

Time 4
•Called shift (truly a
measurement in sigmas of how
far the mean has shifted)
•Indicates our process control
 st + shift = total •The vital few
Between Groups
 Assignable Cause
 Outside influences
 Black noise

 Potentially controllable

 How the process is actually performing

over time

Fishbone
 Common Cause Variation
 Variation present in every process
 Not controllable

 The best the process can be within the

present technology

Data within subgroups (Z.st) will contain only Common Cause

Variation
 Gauge R&R

2Total = 2Part-Part + 2R&R

R&R
Part-Part
• Repeatability (Equipment variation)
Variation observed with one measurement device when used several times by one operator
while measuring the identical characteristic on the same part.

• Reproducibility (Appraised variation)

Variation Obtained from different operators using the same device when measuring the
identical characteristic on the same part.

•Stability or Drift
Total variation in the measurement obtained with a measurement obtained on the same
master or reference value when measuring the same characteristic, over an extending time
period.
Map the Process

Identify the variables - ‘x’

Measure the Process

Understand the Problem -

’Y’ = function of variables -’x’
Y=f(x)

To understand where you want to be,

you need to know how to get there.v
 Measure
Characterize Process

Evaluate Control
Understand Process Maintain New Process

Improve
Improve and Verify Process
In many cases, the data sample can be transformed so that it is approximately normal.
For example, square roots, logarithms, and reciprocals often take a positively skewed
distribution and convert it to something close to a bell-shaped curve
 What do we Need?
LSL USL LSL USL

Off-Target, Low Variation On Target

High Potential Defects High Variation
Good Cp but Bad Cpk High Potential Defects
No so good Cp and Cpk
LSL USL

Variation reduction and process


centering create processes with
less potential for defects.
 The concept of defect reduction
applies to ALL processes (not just
On-Target, Low Variation manufacturing)
Low Potential Defects
Good Cp and Cpk
Eliminate “Trivial Many”
 Qualitative Evaluation
 Technical Expertise
 Graphical Methods
 Screening Design of Experiments Identify “Vital Few”
 Pareto Analysis
 Hypothesis Testing
 Regression
Quantify  Design of Experiments
Opportunity
 % Reduction in Variation
 Cost/ Benefit
Our Goal:
Identify the Key Factors (x’s)
 Graph>Box plot Graph>Box plot
DBP
Without X values 10

75% 10
DBP 4

109 Day DBP

99
10
104
50%
94
9
99
10
94
4
25%
DBP
10 99 Operator

9 94

10

99 Shift

Box plots help to see the

94
data distribution
 Statistical Analysis
Apply statistics to validate actions & improvements
Hypothesis Testing
7 30

5
20
Frequency

Frequency
4

3
10
2

0 0

0.000 0.005 0.010 0.015 0.020 0.025 0.000 0.005 0.010 0.015 0.020 0.025
New Machine Machine 6 mths

Regression Analysis
Regression Plot
Y = 2.19469 + 0.918549X  Is the factor really important?
R-Sq = 86.0 %

60

50  Do we understand the impact for

40

the factor?
Y

30

20

10 Regression
95% PI
 Has our improvement made an
0

5 15 25 35 45 55
impact
X

 What is the true impact?

 poor 2.5
A B
CONTROL
2.0
1.5

Zshift
1.0
0.5
C D
1 2 3 4 5 6
good poor good
TECHNOLOGY
ZSt

A- Poor Control, Poor Process

B- Must control the Process better, Technology is fine
C- Process control is good, bad Process or technology
D- World Class
 Measure
Characterize Process

Evaluate Control
Understand Process Maintain New Process

Improve
Improve and Verify Process
 What Do I need to do to improve my Game?

6
GUTTER!
 DESIGN OF EXPERIMENTS (DOE)
 To estimate the effects of independent Variables on Responses.

X Y
PROCESS

 Terminology
 Factor – An independent variable
 Level – A value for the factor.
 Response - Outcome
 THE COFFEE EXAMPLE

Level
Factor
Low High
Coffee Brand Maxwell House Chock Full o Nuts
Water Spring Tap
Coffee Amount 1 2
Main Effects: Effect of each individual factor on response

3.7

ME
2.2

Bean ‘A’ Bean ‘B’

 Concept of Interaction

Bean ‘A’ Bean ‘B’

Temp ‘X’ Temp ‘Y’
WHY USE DOE ?
 Shift the average of a process.
x1 x2

 Reduce the variation.

 Shift average and reduce variation

DOE TECHNIQUES
 Full Factorial.
4
 2 = 16 trials
 2 is number of levels
 4 is number of factors

 All combinations are tested.

 Fractional factorial can reduce number of
trials from 16 to 8.
 Factor Effect Plot

6.5
5.58 5.65 5.58

5.5 5.43 5
4.6

＋ - ＋ - ＋ - ＋ -

Adhesion Thk of Foam Thk of logo

Area Type of Glue
Styrene
 STEPS IN PLANNING AN EXPERIMENT
1. Define Objective.
2. Select the Response (Y)
3. Select the factors (Xs)
4. Choose the factor levels
5. Select the Experimental Design
6. Run Experiment and Collect the Data
7. Analyze the data
8. Conclusions
9. Perform a confirmation run.
 Measure
Characterize Process

Evaluate Control
Understand Process Maintain New Process

Improve
Improve and Verify Process
CONTROL PHASE - SIX SIGMA

Control Phase Activities:

-Confirmation of Improvement
-Confirmation you solved the practical problem
-Benefit validation
-Buy into the Control plan
-Quality plan implementation
-Procedural changes
-System changes
-Statistical process control implementation
-“Mistake-proofing” the process
-Closure documentation
-Audit process
-Scoping next project
CONTROL PHASE - SIX SIGMA

How to create a Control Plan:

1. Select Causal Variable(s). Proven vital few X(s)

2. Define Control Plan
- 5Ws for optimal ranges of X(s)
3. Validate Control Plan
- Observe Y
4. Implement/Document Control Plan
5. Audit Control Plan
6. Monitor Performance Metrics
CONTROL PHASE - SIX SIGMA
Control Plan Tools:

Statistical Process Control (SPC)

- Used with various types of distributions
- Control Charts
•Attribute based (np, p, c, u). Variable based (X-R, X)
•Additional Variable based tools
-PRE-Control
-Common Cause Chart (Exponentially Balanced
Moving Average (EWMA))
THREE LEVELS OF TRANSFORMATION IN SHINGO
MODEL
FAIL SAFE
 A fail-safe in engineering is a design feature or practice that in the event of a specific
type of failure, inherently responds in a way that will cause no or minimal harm to
other equipment, the environment or to people.
 Unlike inherent safety to a particular hazard, a system being "fail-safe" does not
mean that failure is impossible or improbable, but rather that the system's design
prevents or mitigates unsafe consequences of the system's failure. That is, if and
when a "fail-safe" system "fails", it is "safe" or at least no less safe than when it was
operating correctly.
 Since many types of failure are possible, failure mode and effects analysis is used to
examine failure situations and recommend safety design and procedures.
 Some systems can never be made fail safe, as continuous availability is needed.
Redundancy, fault tolerance, or recovery procedures are used for these situations
(e.g. multiple independent controlled and fuel fed engines). This also makes the
system less sensitive for the reliability prediction errors or quality induced
uncertainty for the separate items. On the other hand, failure detection & correction
and avoidance of common cause failures becomes here increasingly important to
ensure system level reliability.
ISO 9000
 ISO 9000 is defined as a set of international standards on quality management and
quality assurance developed to help companies effectively document the quality
system elements needed to maintain an efficient quality system. They are not
specific to any one industry and can be applied to organizations of any size.
 ISO 9000 can help a company satisfy its customers, meet regulatory requirements,
and achieve continual improvement. It should be considered to be a first step or the
base level of a quality system.

 ISO 9001 Certification

 ISO 9001 is the only standard within the ISO 9000 family to which organizations can
certify.
ISO 14000
 This standard specifies the requirements for an environmental management system
that an organization can use to enhance its environmental performance.

 This standard helps an organization achieve the intended outcomes of its

environmental management system, which provide value for the environment, the
organization itself and interested parties. The intended outcomes of an
environmental management system include:

 enhancement of environmental performance;

 fulfilment of compliance obligations;
 achievement of environmental objectives.

 This standard is applicable to any organization, regardless of size, type and nature,
and applies to the environmental aspects of its activities, products and services that
the organization determines it can either control or influence considering a life cycle
perspective. This International Standard does not state specific environmental
performance criteria