Professional Documents
Culture Documents
Topics (Session 1)
Understanding Six Sigma
Sigma is a measure
of variation
(the data spread)
What does variation mean?
20
every time. 0
-5
Some variation will exist in
-10
all processes.
Variation directly affects customer experiences.
the variation!
s
0 10 20 x 30 40 50
0 10 20 x 30 40 50
The police make 7 false arrests There are fewer than 4 false
every 4 minutes. arrests per month.
1818: Gauss uses the normal curve 1924: Walter A. Shewhart introduces
to explore the mathematics of error the control chart and the distinction of
analysis for measurement, probability special vs. common cause variation as
analysis, and hypothesis testing. contributors to process problems.
Key Define Tools: Key Measure Key Analyze Key Improve Key Control
Cost of Poor Tools: Tools: Tools: Tools:
Quality (COPQ) Critical to Quality Histograms, Solution Control Charts
Voice of the Requirements Boxplots, Multi- Selection Matrix Contingency
Stakeholder (CTQs) Vari Charts, etc. To-Be Process and/or Action
(VOS) Sample Plan Hypothesis Tests Map(s) Plan(s)
Project Charter Capability Regression
As-Is Process Analysis Analysis
Map(s) Failure Modes
Primary Metric and Effect
(Y) Analysis (FMEA)
Define DMAIC Project
What is the project?
$
Project Cost of
Poor Voice of
Charter Quality S ta k e h o ld e r s the
Stakeholde
r
Six Sigma
Observ e package
weight (1 or 2) on
back of package
Look up
appropriate
Weight Fee and
write in top middle
box on package
back
process works?
Outbox to appropriate
Accounts Sender column on
Superv isorInbox. Accts. Supv .s log
Take packages
Draw 5-point Star
f rom Accounts
in upper right
Superv isor
corner of package
Outbox to Out-
f ront
Sort Clerk Inbox.
Add steps
placing in outbox
Deliv erPackages
Delivery
to customers
according to N, S,
E, W route
Submit log to
Submit log to Submit log to
General Manager
General Manager General Manager
at conclusion of
at end of round at end of round
round.
Define Customer Requirements
What are the CTQs? What motivates the customer?
Voice of the Customer Key Customer Issue Critical to Quality
SECONDARY RESEARCH
Industry
Intel
Market
Data Industry
Benchmarking
Listening
Posts
Customer
Correspondence
Customer
Service
PRIMARY RESEARCH
Survey
s Survey
s
OTM
Obser-
Focus Groups vations
Measure Baselines and
Capability
What is our current level of performance?
Descriptive Statistics
Sample some data / not all data Variable: 2003 Output
Mean 23.1692
StDev 10.2152
60 Percent
Count
50
40
20
0 0
t
un er s
Defect La
te
A mo Oth
Count 79 17 4
Percent 79.0 17.0 4.0
Cum % 79.0 96.0 100.0
Measure Failures and Risks
Where does our process fail and why?
Subjective opinion mapped into an objective risk profile number
X1
X2
X3
X4
etc
Analyze Potential Root Causes
What affects our process?
Ishikawa Diagram
(Fishbone)
Six Sigma
100 100
80
Percent
60
Count
50
40
20
0 0
E x p e r im e n ta l D e s ig n
Defect
Count 79 17 4
Percent 79.0 17.0 4.0
Cum % 79.0 96.0 100.0
Data Regression
Stratification Analysis
Pareto Chart for Amt Defects
15
100
80
Process
Simulatio
Percent
10 60
Count
40
5
20 n
0 0
Defect
Count 12 3 2
Percent 70.6 17.6 11.8
Cum % 70.6 88.2 100.0
Six Sigma
Generat
Clarify
Evaluat
Decision
e
e
y = f (x1, x2, x3 . . . xn)
Critical Xs
Divergent | Convergent
Improve Solution Selection
How do we choose the best solution?
Solution Selection Matrix
Qualit
Solution Sigma Time CBA Other Score
y
Time Cost
Six Sigma
Solution
Right Wrong
Implementation
Nice
Good
Solution
Try
Implementatio
Bad
Nice n Plan
Idea X
Control Sustainable Benefits
How do we hold the gains of our new process?
Some variation is normal and OK
How High and Low can an X go yet not materially impact the Y
Pre-plan approach for control exceptions
35
UCL=33.48
Individual Value
25
Mean=24.35
15 LCL=15.21
0 10 20 30
Observation Number
DFSS The Design Methodology
Design for Six Sigma
Uses
Design new processes, products, and/or services from scratch
Replace old processes where improvement will not suffice
Differences between DFSS and DMAIC
Projects typically longer than 4-6 months
Extensive definition of Customer Requirements (CTQs)
Heavy emphasis on benchmarking and simulation; less emphasis
on baselining
Key Tools
Multi-Generational Planning (MGP)
Quality Function Deployment (QFD)
Topics
Understanding Six Sigma
Financial Controllers
Ensure validity and reliability of financial figures used
by Six Sigma project teams
Assist in development of financial components of
initial business case and final cost-benefit analysis
Topics
Understanding Six Sigma
90
87
At first glance, the yield would seem to be When in fact the FPY is (1 x .90 x .96 x .97 = .
85% (85/100 but.) 838)
85
Problem Identification
Rolled
100 Units Outputs / Inputs
Throughput
Yield (RTY): Step 1 90 / 100 = .90
The yield of
individual Re-Work
process steps 10 Units 100 Units
multiplied 97 / 100 = .97
Step 2
together.
Reflects the
Re-Work
hidden factory 100 Units
3 Units
rework issues
associated with Step 3 98 / 100 = .98
a process.
Re-Work
2 Units 100 Units
Step 4 .90 x .97 x .98 = .855
100 Units
Problem Identification
RTY Examples - Widgets
50
50
50
Function 4
5
Put another way, this process is operating
50 a 65% efficiency
Problem Identification
RTY Example - Loan Underwriting
50
Complete Full
Paperwork
1 x .82 x .86 x .93 = .66
2 1
43
Decide not to
Close
borrow
40
30
Frequency
20
10
Descriptive Statistics
Variable: CT
Mean 80.1824
StDev 67.6003
Variance 4569.81
Skewness 2.31712
Kurtosis 8.26356
N 170
25 100 175 250 325 400
Minimum 1.000
1st Quartile 31.000
Median 66.000
3rd Quartile 105.000
95% Confidence Interval for Mu Maximum 444.000
95% Confidence Interval for Mu
69.947 90.417
54 64 74 84 94 95% Confidence Interval for Sigma
61.098 75.664
95% Confidence Interval for Median
95% Confidence Interval for Median
55.753 84.494
Problem Identification
Pareto The Pareto principle states that 80% of the impact of the
problem will show up in 20% of the causes. A bar chart that displays by
frequency, in descending order, the most important defects.
Pareto Chart for WEB
100
100
80
Percent
60
Count
50
40
20
0 0
Defect
Count 96 15
Percent 86.5 13.5
Cum % 86.5 100.0
Topics (Session 2)
Problem Identification
Cost of Poor Quality
Problem Refinement
Process Understanding
Potential X to Critical X
Improvement
Cost of Poor Quality
COPQ - The cost involved in fulfilling the gap between the desired and
actual product/service quality. It also includes the cost of lost opportunity
due to the loss of resources used in rectifying the defect.
Hard Savings - Six Sigma project benefits that allow you to do the same
amount of business with less employees (cost savings) or handle more
business without adding people (cost avoidance).
Soft Savings - Six Sigma project benefits such as reduced time to market,
cost avoidance, lost profit avoidance, improved employee morale,
enhanced image for the organization and other intangibles may result in
additional savings to your organization, but are harder to quantify.
Examples / Buckets
Roll Throughput Yield Inefficiencies (GAP between desired result and
current result multiplied by direct costs AND indirect costs in the process).
Cycle Time GAP (stated as a percentage between current results and
desired results) multiplied by direct and indirect costs in the process.
Square Footage opportunity cost, advertising costs, overhead costs, etc
Topics (Session 2)
Problem Identification
Cost of Poor Quality
Problem Refinement
Process Understanding
Potential X to Critical X
Improvement
Problem Refinement
Multi Level Pareto Logically Break down initial Pareto data into sub-
sets (to help refine area of focus)
100
100
80
Percent
60
Count
50
40
20
Percent
60
Count
50
40
20
0 0
Defect
Count 45 35 13 16
Percent 41.3 32.1 11.9 14.7
Cum % 41.3 73.4 85.3 100.0
Problem Refinement
Problem Statement A crisp description of what we are trying to solve.
Primary Metric An objective measurement of what we are attempting
to solve (the y in the y = f(x1, x2, x3.) calculation).
Secondary Metric An objective measurement that ensures that a Six
Sigma Project does not create a new problem as it fixes the primary
problem. For example, a quality metric would be a good secondary
metric for an improve cycle time primary metric.
Problem Refinement
Fish Bone Diagram - A tool used to solve quality problems by
brainstorming causes and logically organizing them by branches. Also
called the Cause & Effect diagram and Ishikawa diagram
Full Form
Control Open
Start Size Sorts Pull & Sort
Receipt / Docs
Extract
Ck / Vouch
Verify
Perfection
Requal Group
No
Vouchers
Key from
Balance
Data Cap image
No
Vouch
OK
Inventory Yes
Prep
Folders / Full Form Ship to
Box QCReview Cust
Process Understanding
Yes Yes
New &
Other Take off Set 14
Take off Set 14
People IPS month
IPS month
call in system flag (on
system flag (on
IPS?)
IPS?)
Yes
Training Gen rpt for
Ops Kronos
Recruit
Train
No Yes Update
Pass?
IPS
Topics (Session 2)
Problem Identification
Cost of Poor Quality
Problem Refinement
Process Understanding
Potential X to Critical X
Improvement
Potential X to Critical X
Y is the dependent output of a variable process. In other
words, output is a function of input variables (Y=f(x1, x2,
x3).
Through hypothesis testing, Six Sigma allows one to
determine which attributes (basic descriptor (generally
limited or binary in nature) for data we gather ie. day of
the week, shift, supervisor, site location, machine type,
work type, affect the output. For example, statistically,
does one shift make more errors or have a longer cycle
time than another? Do we make more errors on Fridays
than on Mondays? Is one site faster than another? Once we
determine which attributes affect our output, we determine
the degree of impact using Design of Experiment (DOE).
Potential X to Critical X
A Design of Experiment (DOE) is a structured, organized
method for determining the relationship between factors
(Xs) affecting a process and the output of that process (Y).
Not only is the direct affect of an X1 gauged against Y but
also the affect of X1 on X2 against Y is also gauged. In
other words, DOE allows us to determine - does one input
(x1) affect another input (x2) as well as Output (Y).
Potential X to Critical X
DOE Example
Main Effects Plot (data means) for Elapsed
Main Effects Plot
1.4
Lo
w
Hig
h
Lo
w
Hig
h
Lo
w
Hig
h
Lo
w
Hig
h
Direct impact to Y
1.3
Elapsed
1.2
1.1
1.0
Jams DCDEL SK P2Jam
1.50
Jams
1 1.25
3 1.00
1.50
DCDEL
3 1.25
Interaction Plot
1 1.00
1.50
SK
Impacts of Xs on 3
1
1.25
1.00
3 1.25
1 1.00
Potential X to Critical X
DOE Optimizer
Allows us to
statistically predict the
Output (Y) based on
optimizing the inputs
(X) from the Design of
experiment data.
Topics (Session 2)
Problem Identification
Cost of Poor Quality
Problem Refinement
Process Understanding
Potential X to Critical X
Improvement
Improvement
Once we know the degree to which inputs (X) affect our
output (Y), we can explore improvement ideas, focusing
on the cost benefit of a given improvement as it relates
to the degree it will affect the output. In other words, we
generally will not attempt to fix every X, only those that
give us the greatest impact and are financially or
customer justified.
Control
Once improvements are made, the question becomes, are the
improvement consistent with predicted Design of Experiment
results (ie are they what we expected) and, are they statistically
different than pre-improvement results.
LSL USL
Process Data
USL 0.23000
Within
Target *
LSL -1.00000 Overall
Mean -0.02391
Sample N 23
StDev (Within) 0.166425
StDev (Overall) 0.221880
Overall Capability Observed Performance Exp. "Within" Performance Exp. "Overall" Performance
Z.Bench 1.14 % < LSL 0.00 % < LSL 0.00 % < LSL 0.00
Z.USL 1.14 % > USL 13.04 % > USL 6.35 % > USL 12.62
Z.LSL 4.40 % Total 13.04 % Total 6.35 % Total 12.62
Ppk 0.38
Control
Control Chart - A graphical tool for monitoring changes that occur
within a process, by distinguishing variation that is inherent in the
process(common cause) from variation that yields a change to the
process(special cause). This change may be a single point or a series
of points in time - each is a signal that something is different from
what was previously observed and measured.
I and MR Chart for Sept
1
Individual Value
0.5 UCL=0.5293
0.0 Mean=0.03
2
-0.5 LCL=-0.4693
0.5
0.4
0.3
0.2 R=0.1877
0.1
0.0 LCL=0
Wrap Up