Six Sigma Methodology
Overview
Y Dependent Output Effect Symptom Monitor
X1 . . . XN
Independent Input-Process Cause Problem Control
To get results, should we focus our behavior on the Y or X?
Y = f (X)
Six Sigma Basis
Practical Problem Statistical Problem
Statistical SolutionPractical Solution
y f x x xk= ( , ,..., )1 2
Measure Analyze
ImproveControl
Overall Approach
D M A I CMethodology
• Data-driven Quality Methodology
• A Five-step approach to process improvement
• Can be used for both incremental and exponential improvement projects
• Iterative process
Define Define : Customer expectations of the process ?
MeasureMeasure : What is the frequency of Defects ?
AnalyzeAnalyze : Why, When and Where do defects occur ?
ImproveImprove : How can we fix the Process ?
ControlControl : How can we make the process stay fixed ?
DMAIC
DefineProject Charter, CTQs, High Level Process Map
MeasureProject Y and Current Performance,
Determine Process Capabilities,Validate MSA, Set Project goals
AnalyzeAnalyze Root Causes,
Identify sources of variation & Screen potential causes
ImproveDiscover variable relationships,
Establish tolerances & Suggest improvements
ControlDetermine Process Capabilities,
Implement process control & Institutionalize improvements
The 5 step
DMAIC
METHODOLOGY
Questions to be answered
1. What is the Problem or Opportunity?
2. Why should be this a Priority?
3. What is the Effect of this Problem?
4. What Processes are affected?
5. Who should be on the Team?
1. What is the Problem or Opportunity?
2. Why should be this a Priority?
3. What is the Effect of this Problem?
4. What Processes are affected?
5. Who should be on the Team?
Deliverables - Phase level
1. Project CTQs
2. Approved Project Charter
3. High Level Process Map
1. Project CTQs
2. Approved Project Charter
3. High Level Process Map
Tools
Pareto Analysis Voice of Customer Affinity DiagramInterviews/Questionnaire QFD COQ Analysis Project CharterBrainstorming Process Mapping Stratification
Pareto Analysis Voice of Customer Affinity DiagramInterviews/Questionnaire QFD COQ Analysis Project CharterBrainstorming Process Mapping Stratification
Define
1. What is the Primary CTQ and how isit measured?
2. How good is the measurementsystem?
3. How bad is the problem?
4. Are there any clues about its behavior?
1. What is the Primary CTQ and how isit measured?
2. How good is the measurementsystem?
3. How bad is the problem?
4. Are there any clues about its behavior?
1. Project response variable (Y)
2. Data collection on Y
3. Assess the performance standards on Y (Specs)
4. Analysis of the measurementsystem capability
5. Determine performance capability
6. Establish improvement goals
1. Project response variable (Y)
2. Data collection on Y
3. Assess the performance standards on Y (Specs)
4. Analysis of the measurementsystem capability
5. Determine performance capability
6. Establish improvement goals
Process Map Graphical Tools Control charts Z ScoreProcess Capability Descriptive statistics Project CharterCause & Effect Matrix FMEA Gage R & R Brainstorming
Process Map Graphical Tools Control charts Z ScoreProcess Capability Descriptive statistics Project CharterCause & Effect Matrix FMEA Gage R & R Brainstorming
Measure
Questions to be answered Deliverables - Phase level
Tools
1. What could cause the symptoms?
2. How much impact will we have?
3. What returns will be there?
1. What could cause the symptoms?
2. How much impact will we have?
3. What returns will be there?
1. Prioritize list of all X’s
2. Identify the likely families ofvariability (Vital few)
3. Perform Graphical Analysis
4. Perform ANOVA
5. Use of basic statistical tools
6. Quantified financialopportunity
1. Prioritize list of all X’s
2. Identify the likely families ofvariability (Vital few)
3. Perform Graphical Analysis
4. Perform ANOVA
5. Use of basic statistical tools
6. Quantified financialopportunity
Brainstorming Cause & Effect Correlation/RegressionTest of Hypothesis/ANOVA COQ Analysis Statistical tool-kitGraphical Analysis FMEA Project Charter
Brainstorming Cause & Effect Correlation/RegressionTest of Hypothesis/ANOVA COQ Analysis Statistical tool-kitGraphical Analysis FMEA Project Charter
Questions to be answered Deliverables - Phase level
Tools
Analyze
1. What is the root cause?
2. How was this verified?
3. What could be done about each cause?
4. What will be done to address these?
1. What is the root cause?
2. How was this verified?
3. What could be done about each cause?
4. What will be done to address these?
1. Confirmed Vital X’s
2. Optimized solutions
3. Confirmed solutions
4. Risk Assessment plan
1. Confirmed Vital X’s
2. Optimized solutions
3. Confirmed solutions
4. Risk Assessment plan
DOE Z scores Brainstorming/Benchmarking FMEA CapabilityCost benefit analysis Sample size Project Charter Process MapStatistical Tolerancing Regression Hypothesis/ANOVA
DOE Z scores Brainstorming/Benchmarking FMEA CapabilityCost benefit analysis Sample size Project Charter Process MapStatistical Tolerancing Regression Hypothesis/ANOVA
Improve
Questions to be answered Deliverables - Phase level
Tools
1. How will the solution be implemented?
2. Who will be affected?
3. How will the change be sustained?
4. What are the verified impacts?
1. How will the solution be implemented?
2. Who will be affected?
3. How will the change be sustained?
4. What are the verified impacts?
1. Sustained Process- Mistake proof- Tolerance the process- Monitor (SPC)
2. Project Documentation
3. Translated Learning's
4. Project Closure
1. Sustained Process- Mistake proof- Tolerance the process- Monitor (SPC)
2. Project Documentation
3. Translated Learning's
4. Project Closure
SPC Process Capability Process MapCommunication plan Error proofing Project CharterControl plan Project documentation
SPC Process Capability Process MapCommunication plan Error proofing Project CharterControl plan Project documentation
Questions to be answered Deliverables - Phase level
Tools
Control
D M A D VMethodology
Define Determine objectives & deployment
MeasureUnderstand customer needs & specify
measurable design requirements
Analyze Develop concept design
Design Developed details & optimized design
Verify Verify Performance
The 5 step
DMADV
METHODOLOGY
The 5 step
DMADV
METHODOLOGY
Measure Define Analyze Design Verify
Measure Define Analyze Improve Control
Does a process
exist?
Is the improvement
a new or redesigned
product/service?
IsIncremental improvement
enough?
Y Y N
N YN
DMADV-DMAIC Relationship and Transition Points
Six Sigma Metrics
Identify the CTQs
Critical to Quality Characteristics of the Customer requirements for a Product (or) Service
Define Defect Opportunities
Any step in the process where a defect could occur in a CTQ
Look for defects in
products (or) services
Count defects or failures to meet CTQ requirements in all process steps
Arrive at DPMO
Defects per million opportunities
Convert DPMO to Sigma level
Use the Sigma Conversion table (or) Normsinv Function in Ms-Excel
How do we arrive at Sigma Levels
Report as Z-Short term
Value
Z-LT = Z-ST - 1.5
CTQs and Technical Requirements are the foundation for all Six Sigma Projects. They must be defined UP FRONT!
Before we can begin, the “Voice of the Consumer” (Consumer Cue)must be translated into the “Voice of the Engineer” (Technical Requirement).
schedules should be adhered to. A call-taker must answer all incoming calls
(Telephone promptness)
Translate the customer/consumer cue into “Critical to Quality” requirements (dimensions/ parameters that need to bemeasured and/or controlled)
Schedule Adherence - 99% Answer rate (% of incoming calls answered) - 100%
Technical Requirement
CTQ/System of CTQs
Translation process
(“Cues” to “CTQs” or Critical to Quality
requirements)
Renovated Code must reach at specified time Call-takers must be available to answer calls
Consumer Cue
Translation of Customer “Cues into
CTQs
1. Anything that results in customer dissatisfaction
2. Anything that results in non-conformance
3. Any variation of a required characteristic of a product or service or its parts which prevents the product or service from fulfilling the functional / physical requirements of the customer
Defect Definition...
1. A measure of volume of output from your area
2. It is observable and countable
3. It should have a definite starting and stopping point
- Where work is normally reviewed (e.g., inspection, test)
4. That “thing” for which you are counting defects
Unit Definition...
Financial Report
Defect : Entry not correct
Unit : Each entry
# of Defects = 50
# of Units = 50000
DPU = .001
Telephone response
Defect : not answered within 5 rings
Unit : every call
# of defects = 45# of calls = 2000DPU = .0225
DPU Examples...
Can we say that processes of Supplier no 1 are better than supplier no 2? WHY?
SUPPLIER No 1
•Blank A4 size paper supplier’s DPU = 0.002
(1 defect / 500 papers)
SUPPLIER No 2
•PC manufacturer (Compaq, Dell) ‘s DPU = 0.2
(1 defect / 5 PC units)
DPU Examples...
Can we say that processes of Supplier no 1 and supplier no 2 are equal in performance? WHY?
SUPPLIER No 1
•Software Product DPU = 0.01
(5 defect / 500 Products)
SUPPLIER No 2
•Software Product DPU = 0.01
(5 defect / 500 Products)
DPU Examples...
Defect Opportunity is the number of chances in a process where a defect could occur in a Product/Process standard (CTQs)
•Indicates complexity of process
•Need to be independent
•Measurable
Defect Opportunity...
How many No. of Defect Opportunities are there
for a Product if there are 5 CTQs on the Product
and 1 Opportunities for each CTQ across the
Process steps...
Total No. of Opportunities = 5 x 1 = 5
Defect Opportunity...
Calculate Defects per Unit (DPU):
DPU=Total # of Defects
Total # of Units Produced
Calculate Defects per Million Opportunities (DPMO):
Go to a Sigma Chart and Estimate the Sigma Level
DPMO=DPU x 1,000,000
(# of Opportunities for Error in One Unit)
Calculate Defects per Opportunity (DPO):
DPO=DPU
(# of Opportunities for Error in One Unit)
Calculating DPMO...
Finance Report
CTQs : Amount and Date of TransactionDefect : Amount & Date not accurate
# Defects = 500Unit : Each entry
# of units = 50000# of opportunities = 2 steps in the Entry process
DPU = 500/50000 = 0.01DPMO = (0.01 x 1000000)/ (2)
= 5000Sigma Value : 4.075
DPMO Example...
Assumptions:Assumptions:1Unit = 1 Invoice1Unit = 1 Invoice
Invoicing to the customer has 3 process steps that are CTQ’s for the customerInvoicing to the customer has 3 process steps that are CTQ’s for the customer1) Correct Quantity1) Correct Quantity2) Correct Price2) Correct Price 3) Invoice Received Timely3) Invoice Received Timely
•• 3 opportunities for a defect exists for each invoice.3 opportunities for a defect exists for each invoice.•• There is one opportunity for a defect in each processThere is one opportunity for a defect in each process
To evaluate actual invoicing performance:To evaluate actual invoicing performance:
Process Process # of # of ## # # Opp Opp Yield Yield ImpactingImpacting UnitsUnits ofof of of for for for for
CTQCTQ ProducedProduced DefectsDefects SuccessesSuccesses ProcessProcess ProcessProcess
Correct QuantityCorrect Quantity 500500 465465 11 0.930.93Correct PriceCorrect Price 500500 350350 11 0.70.7Invoice Received TimelyInvoice Received Timely 500500 425425 11 0.850.85
260260
35351501507575
DPMO Example...
Sigma Value = 2.4
Less than 3-Sigma: Process is not meeting the
customer requirements
3-Sigma: Process barely meets requirements; no
latitude for upsets
4-Sigma: Typical corporate performance
5-Sigma: Excellent performance
6-Sigma: Robust to nearly all upsets
Interpreting your Sigma levels
Rolled Throughput
Yield...
66% 90%
... why not?
ScrapScrap
90% Customer Quality
ReworkRework
Hidden FactoryNOTOK
Yield After Inspection or Test
OperationOperationInputsInputs InspectInspect First TimeFirst TimeYieldYield =
OK
ProcessA B C
90%Yield
90%Yield
81 % 73 %
Final TestD
Rolled-Throughput Yield Classical First-Time Yield
90%Yield
66 %
90%Yield
Rolled Yield
RTY Vs Final Yield
• Calculate Yield
Yield = e-DPU
(Using Poisson Approximation)
Rolled Throughput Yield = Y1*Y2*Y3*Y…
Calculating Yield...
Rolled yield of a multi step process = Yield (process 1) * Yield (process 2) * etc.
For ONE process, rolled yield = 93%
For TEN processes, rolled yield = 50%
For ONE HUNDRED processes, rolled yield = 0%
For ONE processes, rolled yield = 100.00%
For TEN processes, rolled yield = 100.00%
For ONE HUNDRED processes, rolled yield = 99.97%
Rolled Throughput Yield...
Six Sigma and COPQ...
• What is the – hidden cost of failing to meet customer expectations
the first time – hidden opportunity for increased efficiency– hidden potential for higher profits– hidden loss in market share– hidden increase in production cycle time– hidden labor associated with ordering replacement
material– hidden costs associated with disposing of defects
• For most companies today, the cost of poor quality is likely to be 25-40 % of sales.
What is Cost of Poor Quality?
40%
35%
30%
25%
20%
15%
10%
5%
3.4 233
6210
66,807
308,537
691,462
Cost
of
Failu
re
(%
Sale
s)
DPMO
6 5 4 3 2 1
Sigma
Cost of Poor Quality Vs Sigma
Cost of Poor Quality PPM Sigma
30-40% revenue 308,537 2
20-30% revenue 66,807 3
15-20% revenue 6,210 4
10-15% revenue 233 5
less than 10% revenue 3.4 6
Non-competitive
Industry average
World class
Correlation to Sigma Level...
Engineering change orders
Traditional Quality Costs
Lost Opportunity
Hidden Factory
Lost sales
Late delivery
Long cycle times
Expediting costs
Excess inventory
(intangible)
(tangible)
(Difficult or impossible to measure)
Lost Customer Loyalty
More Setups
Scrap
Rework Inspection
Warranty
Rejects
Administration / Disposition
Concessions
Average COPQ approximately 15% of Sales
Improvement program costs
Quality audits
Additional Cost of Poor Quality
COPQ Iceberg
Defect rate Cost of control
Point of diminishingeconomic returns
fromquality investments
Failures Cost
Quality improvement
The highest quality
producer is not
necessarily the lowest
cost producer
Classical COPQ Model
Failures Cost
Defect rate
Cost of control
3
4 5
6
The highest quality producer
is also the lowest cost producer
Six Sigma COPQ Model