Vous êtes sur la page 1sur 53

The Six Sigma Process

Compare business problem


solving to picking fruit off a tree

Sweet Fruit at the top


Design for Six Sigma

Bulk of Fruit
Use data Analysis with Structured Problem
Solving; I.e. Six Sigma

Low Hanging Fruit


Basic Structured Problem Solving

Ground Fruit
Use logic and intuition

Copyright Ro 2
Six Sigma: Application

For each Product, Service


Identify
Identify or Process CTQ,
Key Processes
Customer Driven use the
that cause
Critical to Quality (CTQ) Six Sigma Process
defects
Characteristics to achieve
in the CTQ
significant
Characteristic
quality improvement

Six Sigma Quality requires a rigorous analytical approach


for anticipating and solving problems

Copyright Ro 3
Identify Customer Driven
Critical to Quality (CTQ) Characteristics

 Who are the Customers of your business or process?


 External to business
 Internal to business
 What do you provide your customers?
 Products
 Services
 Are all customers equally important?
 Whose needs must be met for this product / service
to be successful?
 What is critical to the quality needs of your
customers for the products and services you provide?

Copyright Ro 4
Identifying CTQ’s
 Take the CUE from what the Customer talks about.
 QFD (Quality Function Deployment) is often used here.
 Customer Surveys
 Customer Complaints
 Use the Customer’s words, not the idea of the business
team
 Translate the Customer Cues into CTQ’s
 A Structure tree or Customer Needs Map is useful if not
doing QFD
 Screen and Prioritize the CTQs
 Failure Modes and Effects Analysis can help
 Benchmark the competition or best practice companies

AA CTQ
CTQ must
must be
be MEASUREABLE
MEASUREABLE and
and MEASURED
MEASURED
Copyright Ro 5
Specify CTQs

Product/
Product/ Customer Response
 CTQs Are The Translation Service
Service Time
Of Customer Needs Into Characteristic
Characteristic
Quantifiable Requirements
For Our Product/Service Time to provide
Measure
Measure requested service
(hours)
Customer
Customer
Need
Need Target/
Target/
CTQ
CTQ Nominal
Nominal 8 Hours
Value
Value
Prompt
Response
Specification/
Specification/
Tolerance
Tolerance 10 Hours
Limit(s)
Limit(s)
Business
Business
Goal
Goal
Allowable
Allowable
Defect
Defect < 3.4 DPMO
Six Sigma Rate
Rate
Performance
Copyright Ro 6
Specify CTQs – Let’s try it
Sigma
Customer Need Characteristic Measure Target LSL USL
Target
Quick Check-In Check-In Time Time From 2 Minutes — 4 Min. 6
At Hotel Customer Entry (<3.4 DPMO)
To Lobby To
Receipt Of Room
Key

Easy Bank
Deposits

Quick Trip on
the Ferry (Mass
Transit)

Refrigerator
keeps the milk
cold

Copyright Ro 7
What About a Company CTQ?
Sigma
Customer Need Characteristic Measure Target LSL USL
Target
Quick Check-In Check-In Time Time From 2 Minutes — 4 Min. 6
At Hotel Customer Entry (<3.4 DPMO)
To Lobby To
Receipt Of Room
Key

Copyright Ro 8
Identify Key Processes That Cause Defects

in the CTQ Characteristic

 What are the internal processes for providing the


product or service to the customer?
 Where do defects occur in these processes?

Because
Because of
of the
the nature
nature of
of Opportunities,
Opportunities, the the critical
critical process
process
defect
defect may
may be
be buried
buried several
several levels
levels into
into the
the business
business process
process

Copyright Ro 9
The Six Sigma Process - DMAIC
 Define: Frame the Problem.

 Measure: Measure CTQ Performance.

 Analyze: Determine Root Cause.

 Improve: Improve the output by


manipulation of Inputs.

 Control: Systematize the control of the


improved process to sustain the
improvement.
Copyright Ro 10
Define – Frame the Problem
 Understand the Voice of the Customer
 Through direct communication, surveying or negotiation
 Through Quality Function Deployment
 Measurable output characteristics (CTQ)
 Define the High Level Process
 SIPOC Map
 Define the Project Charter
 Problem Statement
 Project Scope (Boundaries, Resources)
 Business Case

Copyright Ro 11
Define Phase
• QFD (Quality Function Deployment)
• Process Mapping
• Rolled Throughput Yield (probability of 0
defects)
• Pareto Analysis
• Project Charter

Six Sigma project boundaries are clarified


in the ‘Define’ Phase . . .
Copyright Ro 12 11
Ability to Flow Customer Needs
down to Process CTQ’s
 Using the “Voice Of The Customer” requires a process
 It may begin with one on one communication
 It requires active listening
 The VOC cues often need to be translated to a process CTQ
 Quality Function Deployment (QFD) provides a method
for really listening
 The customer knows what they want
 They generally won’t directly verbalize
 Listen for the “OTHER” stuff that surfaces
 Watch for things they say they don’ want
 Understand The Types Of Needs Of Customers
 How does the need relate to “Customer Satisfaction”?

Copyright Ro 13
Quality Function Deployment

QFD
Customer
Customer
Customer Needs
Critical-To-Quality Refined
Refined product
product //
High-Level Research
Research  (CTQs) service
service operating
operating
Customer  ID
ID Customers
Customers Requirements processes
Organized processes or

 Determine
Determine Data
or
Needs  Data  Measures
Collection
Collection Plan
Plan

 Targets functional
functional redesign
redesign

 Collect
Collect Data
Data Prioritized  Specifications

Competition Research
 Comparative Perception
 Technical Evaluation

Copyright Ro 14
A QFD Approach
Structured Process
 Identifies CTQs At One Level
 Relates Those CTQs To The Next Level
System Features
(HOW)
Functional Req’ments
(HOW)
Customer Needs

House of Internal Processes


(WHAT)

(HOW)
System Features

Quality House of Critical to Process


(WHAT)

Functional Req’ments
Variables (HOW)

#1 Quality House of

(WHAT)

Internal Processes
#2 Quality House of

(WHAT)
#3 Quality
Customer Y
House #4
Business
Customer House X
Requirements Process
House Process
Product
Requirements Operating System
Control
Requirements
Operating
Process
Copyright Ro 15
Requirements
Customer Needs Change Over
Time
Kano Model of Customer Needs
Understand how
Satisfied the QFD defined
Customer
needs fit in the
s
eed
t eme
nt N model and from
c i
Don’t Have
Ex
Need
s
Included that determine the
a nce
Don’t Do
Perfo
rm
eeds
Do Well
ultimate effect of
ic N
Bas addressing specific
Dissatisfied outputs.
Customer

Quality
Quality Improvement
Improvement must
must focus
focus on
on high
high impact
impact needs
needs
and
and simultaneously
simultaneously improve
improve business
business profitability
profitability
Copyright Ro 16
Process Mapping
 Process mapping is an innovative
extension of process flow charting
that provides structure for defining
and communicating a process in a
simplified, graphic manner
 In Six Sigma projects, process
mapping is used to enhance the
understanding of the process, set
boundaries for the project, and
identify areas for focus or
improvement.
 Process Mapping can also be used as
a tool for documenting the process
and associated quality systems.

Copyright Ro 17
Use Process Mapping to aid in defining the
Critical Process Parameters

Supplier Expectations Critical Output Characteristics


What must we have from Customer Needs Related
our process suppliers to Measurable
assure our process can give
the required output?
Measurable

Critical Input Characteristics


What do we control to assure the output
characteristic is achieved
What could cause the output characteristic
to be unsuitable.
Measurable

Copyright Ro 18
Uses for the Map In Process
Improvement
 Define the Improvement Project (High Level )
 Understand what the process looks like
 Assess and manage the scope of the project
 Measure (Detail)
 Understand what the process looks like
 Assess and manage the scope of the project
 Identify potential X’s and Y’s
 Scope performance standards for process steps
 Analyze (Detail)
 Identify process flow anomalies
 Identify potential X’s and Y’s
 Detect non value-added steps
 Improve
 Identify alternative Process flows
 Facilitate resource planning

Copyright Ro 19
Value Stream Mapping
Questions
Customer Value-Added Business Value-Added Questions
Questions  Is this task required by law or
 Does the task add a form or feature regulatory?
to the product or device?  Does the task reduce the business’
 Does the task enable a competitive financial risk
advantage, I.e., faster delivery,  Does this task support financial
lower price, fewer defects. reporting requirements?
 Would the customer be willing to
 Would the process breakdown if this
pay extra or prefer us to the task were removed?
competitor if he or she knew we did
this task?
Non-Value-Added Questions
 Does this task include handling, transporting, moving, counting, inspecting,
delaying, storing expediting multiple signatures or network loops?

Copyright Ro 20
What Is Rolled Throughput Yield
(Why Do I Care?)
 Rolled throughput yield is the calculated probability
that a product or service provided can be processed
through all the opportunities in a process without a
defect.
In a process which has 25 stations, each with four
opportunities for a defect to occur, and with a historical
defect per opportunity of 0.001, the YRT is 90.48%
# Stations
Y 
   Yield 
# Opportunities 
station 
RT  

Y RT

 .999 
4 25
  .996  25  .9048
Process
Process Complexity
Complexity is
is an
an important
important consideration
consideration
when
when working
working towards
towards aa defect
defect free
free product
product
Copyright Ro 21
Pareto Analysis
 Focus on the vital few issues.
 80/20 Rule
 Look at Yield Issues; Quality Cost Issues; Cycle time
contributors.

Copyright Ro 22
Questions to be answered in
Define
 What is the problem addressed by the project?
 What is the intended scope of work?
 Will the system be replaced or improved?
 What is the planned completion state?
 What is the project plan and schedule?
 How will this project provide financial benefit to the
business?
 What resources will be needed to complete the
project?

These
These questions
questions should
should all
all be
be answered
answered by
by the
the project
project charter
charter

Copyright Ro 23
Measure- Describe the Output
quality
 Define the project CTQ
 What will be the focal output for improvement
 What type measurement will be used?
 Define the CTQ performance standard
 What are the acceptable limits of the CTQ?
 Plan the capability data collection
 Who will measure?
 What methods will be used for measurement?
 How long will data be collected?
 Measurement system analysis
 How do we know we can believe the data collected?
 Measure and determine process capability
 Report long term and short term capability
 Determine improvement focus for project

Copyright Ro 24
Measurements focus our
attention
 If you can’t measure something, you can’t understand it.
 If you can’t understand it, you can’t control it
 If you can’t control it, you can’t manage it
 If you can’t manage it, you can’t improve it.

 What do you need to know about your business process to


understand it? How do you measure that?
 What do you need to control within your business process? How do
you measure that?

 When we focus on the measurement of process or product


standards, new questions will continuously arise from the results.

Can
Can you
you imagine
imagine the
the World
World Cup
Cup without
without aa scoreboard?
scoreboard?
Copyright Ro 25
Measure Phase Methods
 QFD flow down of Customer Needs
 Process Mapping (drill down)
 Structure trees
 Measurement System Analysis
 Process capability
 Rational subgrouping
 Histograms
 Tolerance analysis

The ‘Measure’ phase describes


the process output (Y) …

Copyright Ro 26 12
Measure Phase is really about
describing the process output

Copyright Ro 27
Rational Subgroups give
Direction

50

40

30

20

10

-10

200 400 600 800 1000

What can we learn from this process?


Copyright Ro 28
Measurement System
Analysis Essentials
 Non Destructive Continuous Data is a small part of
the data available to us for decision making.

 What about Yield Data?

 What about database information?

Copyright Ro 29
Measure the Floppy
 Let’s look at measurement system limitations
 Measure this floppy disk width.
 Using the steel rule
 Using the caliper
 Record your measurements as
 Yellow
 White
 Blue
 Red
 Purple
 Pass the floppies to the next one measuring

Copyright Ro 30
Questions to be answered in
Measure
 What is the Project Y?
 How does the Project Y link to the customer needs?
 What kind of data is used to describe it?
 What is the target, and/or specification limits?
 Can we measure it reliably?
 What are the potential Factors (X’s) or causes for
performance limitations?
 Can we measure them reliably?
 What is the process capability for the Project Y?
 Instantaneous?
 Long Term?
 Compared to Customer needs or Specifications

The
Themeasure
measurechallenge
challengewill
willbe
bethe
theability
abilityto
tomeasure
measurereliably.
reliably.
Copyright Ro 31
Analyze – Find the Root Causes

Y  f (x)  Analyze Cause - Effect


Relationships
 Identify Potential Factors (X)
Inputs (X) Outputs (Y)  Analyze Measurement System for
X’s
 Identify Vital Factors (X’s) with data
Process  Reconfirm Improvement
Objectives
 Start with Project Charter
 Confirm validity of improvement
Move the focus upstream plan with new data
 Establish adjusted financial goals

Copyright Ro 32
Analyze Phase Methods
 Structured investigation of potential root causes.
 Graphing - scatterplots, boxplots, time series plots,
etc.
 Hypothesis tests (t-test, F-test, Chi-square test)
 Confidence intervals
 ANOVA
 Residual Analysis
 Cost / Benefit Analysis

The ‘Analyze’ phase builds understanding about


cause and effect relationships between the
output, Y, and the potentially ‘Vital’ X’s . . .
Copyright Ro 33 13
Graphs
Which shows comparison better?
These Histogram of 26880rr, with Normal Curve Histogram of 26880ft, with Normal Curve

7 10

6
Frequency

Frequency
5

4
5
3

0
0
26.863026.863526.864026.864526.865026.865526.866026.866526.867026.867526.8680
26.854 26.855 26.856 26.857 26.858 26.859 26.860 26.861 26.862 26.863
26880rr
26880ft

OR
26.869

Use
Usegraphs
graphs that
that focus
focusthe
theproblem
problem
and
andto
toprovide
providevisual
26.864

visual
26880ft

26.859
understanding
understanding
26.854

26880ft 26880rr

Copyright Ro 34
Nature of Hypotheses
 Null Hypothesis (Ho):  Alternative Hypothesis (Ha ):
 Usually describes a status  Usually describes a difference
quo  The one you accept or reject
 The one you assume unless based upon evidence
otherwise shown  Signs used in Minitab: not =or <
 Signs used in Minitab: = or >

Or

Its
Its either
either Null
Null (same)
(same) or
or Alternative
Alternative (Different)
(Different)
Copyright Ro 35
Data helps confirm hypotheses
A study was performed of cardiovascular readiness of 92 people. One of
the tests was starting heart rate prior to exercise. The results are shown
below. Compare the stating heart rate of participants by sex. Is there a
difference in starting heart rate between the sexes in those people
participating in the study?

The
Thegraphic
graphiccomparison
comparisonpoints points to
to possibility
possibility of
of
“difference”
“difference”but
but isis not
not conclusive.
conclusive.The The Hypothesis
Hypothesistest
test
confirms
confirmsthe
theanswer
answerfor forthis
thissmall
smallsample.
sample.
Copyright Ro 36
What does an ANOVA say?
Analysis of Variance (Balanced Designs)

Factor Type Levels Values


Suppleme fixed 3 1 2 3
Lake fixed 2 Dennison Rose

Analysis of Variance for Zooplank

Source DF SS MS F P
Suppleme 2 1918.5 959.2 9.25 0.015
Lake 1 21.3 21.3 0.21 0.666
Suppleme*Lake 2 561.2 280.6 2.71 0.145
Error 6 622.0 103.7
Total 11 3123.0

Not
Notonly
onlydo
dowe
weknow
knowthat
that“Supplement”
“Supplement”isisimportant
importantand
and“Lake”
“Lake”isisnot,
not,
we
weknow
knowthat
thatwe
wehave
havecaptured
captured80%
80%ofofthe
thevariation
variationand
andthe
theremaining
remaining
uncontrolled
uncontrolledvariation
variationhas
hasaastandard
standarddeviation
deviationofof10.2.
10.2.
Copyright Ro 37
What does the Regression
Provide?
Regression Analysis

The regression equation is


HeatFlux = 325 + 0.0675 Insolation + 2.55 East + 3.80 South -
22.9 North + 2.42 Time

Predictor Coef StDev T P


Constant 325.44 96.13 3.39 0.003
Insolati 0.06753 0.02899 2.33 0.029
East 2.552 1.248 2.04 0.053
South 3.800 1.461 2.60 0.016
North -22.949 2.704 -8.49 0.000
Time 2.417 1.808 1.34 0.194

S = 8.039 R-Sq = 89.9% R-Sq(adj) = 87.7%

Analysis of Variance
Source DF SS MS F P
Regression 5 13195.5 2639.1 40.84 0.000
Residual Error 23 1486.4 64.6
Total 28 14681.9
We
Wegain
gainthe
thetransfer
transferfunction,
function,
Source DF Seq SS identification
identification of the criticalfactors,
of the critical factors,
Insolati 1 5783.8
East 1 811.7 an
anestimate
estimateof ofthe
themodel
modelandandan
an
South 1 1181.5 estimate
estimateofofthe
thesigma
sigmaofofthe
the
North 1 5303.0
Time 1 115.5 remaining
remainingvariation.
variation.
Copyright Ro 38
The importance of comparing cost
and benefits of improvement
 Most every process can be improved at some cost.
 In the Analyze phase, we should gain sufficient knowledge
of improvement and defect reduction possibilities to
calculate the benefits related to that improvement.
 Compare the benefits to the related cost.
 Based on the business case, there may be an economic
limit on how much improvement is justified.
 We may find that we can improve the process to 4.5 sigma levels
by changing control methods, but achieving 6 sigma requires
redesign and automation.
 Based on the business benefits gained, 4.5 sigma may be a logical
limit.

Copyright Ro 39
Questions to be answered in
Analyze
 Which of the potential factors (X’s) are Critical?
 What is the root cause?
 How were the critical factors (X’s) determined?
 What evidence was used to “prove” the case?
 Is the improvement to be gained by control of the
identified factors enough?
 What are the expected project benefits?
 Do they justify continuing with this project as conceived?

Copyright Ro 40
Improve
 Identify Project Improvement Plan
 Confirm factor significance with systematic techniques (DOE)
 Confirm cause and effect relationships to improve process.
 DOE
 Simulation
 Define optimized configuration Y=f(x)
 Confirm improvement methodology
 Define the transfer function
 Use simulation or confirmation runs to confirm significance.

Copyright Ro 41
Improve Phase Methods
 Factorial DOEs
 Fractional factorial DOEs

 Response Surface designs (Central Composite)

 Developing the Transfer function

 Process Simulation

NOTE: regardless of the design, the DOE focus is on


both the mean and the standard deviation:

 (Y) = f(X);  (Y) = f(X)


The
The ‘Improve’
‘Improve’ phase
phase confirms
confirms which
which inputs
inputs are
are critical
critical
(Vital
(Vital X’s),
X’s), and
and enables
enables optimization
optimization of
of the
the process
process .. ....
Copyright Ro 42 14
The Focus of Experiments in Six
Sigma
 In Six Sigma projects, experimentation is used to help define
the mathematical model for the process, called the transfer
function.
 A mathematical model can be defined for the mean () of the
output and the the variation() of the output.
 Once the model has been defined, it is possible to use it to:
 Shift the mean of a process
 For example, using a longer hold time on an injection molded part may
reduce shrinkage and move the mean to the center of the spec.
 Reduce the variation of a process
 If we increase dealer technician training on awning installation, we may
reduce the variation in awning install time.
 Shift the mean and reduce the variation of the process
 If we increase the part wash cleaning solution concentration, we may
gain improved and more consistent cleanliness of parts prior to paint.

Copyright Ro 43
Different DOE’s for Different
Needs
 Screening Experiments – When you are looking to
identify the critical factors
 Characterization Experiments – When you are
looking to quantify the factor effects and interactions
 Optimization Experiments – When you are
looking to define the optimal settings of factors for
the required process output.
 Confirming Experiments - When you have
identified changes using a transfer function or other
experimentation method.
DOE
DOE types
types may
may be
be used
used individually
individually or
or may
may be
be part
part
Copyright Rooverall process of finding the solution.44
of an
Factorial DOE
A manufacturing process team was seeking to improve injection mold part
strength. From the analyze phase they had found a possible relationship
between temperature, pressure, thickness and cure time. They ran a DOE
based on these findings to confirm best conditions and settings. Their
analysis of output follows.

Interaction Plot (data means) for mean


-1 1 -1 1 -1 1

Temprtr 60

1 50
-1
40

Pressr 60

1 50
-1
40

Thikness 60

1 50
-1
40

DwelTime

Copyright Ro 45
The Transfer Function
What is the mathematical relationship between the
“Y” and the “X’s”?

Y  f ( x) .

Effect Root Causes

Output Inputs

Y  ax1  bx 2  cx3  ...  error(common cause)


Once
Oncewe
wehave
haveestablished
establishedthe
thetransfer
transferfunction
functionfor
foraaprocess
processwe
wecan
canuse
usethat
that
knowledge
knowledgeto
tomake
makechange
changeandandimprovement
improvementquicker
quickerand
andmore predictably..
morepredictably
Copyright Ro 46
Questions to be answered in
Improve
 What is the mathematical relationship between the
determined critical “X’s” and the Project Y?
 What is the most best way for our business to
improve this process?
 What settings of “X” provide for optimized
performance of the Project Y?
 Based on using the best way for process improvement
 How have these settings changed from the original settings?
 What was the process capability of the process based
on the confirmation run?
 Is the achieved improvement sufficient?

Copyright Ro 47
Control
 Implement process controls at new process settings
 Draft new quality plan
 Engage Operational Team to use new plan
 Rebaseline process
 Put new quality plan in place
 Reconfirm Measurement System Capability
 Measure as with original baseline
 Compare to Project Charter goals
 Document Process and close project
 Finish the quality plan
 Transition improved process to operational team
 Negotiate close with owner

Copyright Ro 48
Control Phase Methods
 Process Documentation
 ISO documentation
 Setup Procedures
 Operating control methods
 SPC charts - on the Vital X’s
 Training
 Information technology (IT) solutions - ‘systems
fixes’
 Process monitoring/alarming
 Transfer ownership to Operational Team
The
The ‘Control’
‘Control’ phase
phase provides
provides sustainable
sustainable process
process
modifications
modifications toto keep
keep the
the Vital
Vital X’s
X’s under
under control
control
Copyright Ro 49 15
DMAIC or DMADV
DMAIC/DMADV Transition Points:

Define Measure Analyze Improve Control

Is Is The
Yes Yes Improvement No
Does A Incremental
A New Or
Process Exist? Improvement Redesigned
Enough? Product/
Service?

No No Yes

Define Measure Analyze Design Verify

Copyright Ro 50
The Six Sigma Redesign Process
– DMADV
 Define: Frame the Problem
 Measure: Measure CTQ Performance of similar
technology
 Analyze: Define the factors needing change in the
new process and frame the high level functional
design.
 Design: Confirm factors of detail functional design
and find appropriate factor ranges necessary for
desired output.
 Verify: Use pilot runs to confirm adequacy of new
design and document new process.

Copyright Ro 51
Comparing Six Sigma to Other
Structured Problem Solving
 Project framing and root cause studies are similar.
 Traditional problem solving does not require data.
 Most structured problem solving does suggest use of data
but does not integrate advanced statistical methods
 Data collection needs may extend investigation time.
 Data used in six sigma approach generally leads to
more precision in selecting improved process
parameters.
 The ability to define a transfer function from the
accumulated data allows broader application of the
proposed change.

Copyright Ro 52
Section Summary
 Six sigma is a structured problem solving discipline that
integrates data into the analytical approach.
 Six sigma represents a measurement of a process performance
relative to the customer needs. So six sigma measurements
convert directly to defect rates.
 Six sigma analysis focuses on understanding the process output
as a function of process inputs
(Y = f(x)).
 DMAIC is focused on improving existing processes.
 DMADV is a process focused on using data based learning to
design a new or different process for meeting a specific
customer need.

Copyright Ro 53

Vous aimerez peut-être aussi