Design for Six Sigma

An Overview
Damon R. Stoddard
Master Black Belt

© 2003 All Rights Reserved eNPI Process. Do not reproduce

 Conceptual Y=f(x)
Coffee Strength By Coffee Amt

The function (f)

Y is What Really quantifies the
Matters to the 20 relationship
Customer! between x and Y
Customer
Requirement
Y=Coffee Strength
15
for Taste >6
Tighten
Stronger

Tolerances on
Manufacturin
10 g by
Inspection
X’s are the factors
(knobs) that change the Y
5

0
.0 .2 .4 .6 .8 1.0 1.2

x=CoffeeCoffee
AmountAmt
(Scoops)
Fitting More

Companies Cannot Achieve Six Sigma Quality Through Process Improvements….

© 2003 All Rights Reserved eNPI Process. Do not reproduce
 1000:1
100:1
PRODUCT
PRODUCT
DEFINITION 10:1
DESIGN
PROCESS
DESIGN 1:1
PRODUCTION
PRODUCT
IMPROVEMENT
QFD
PCS
DFSS Process
Capability

Taste Critical To Customer! Profitable Approach Exists!

Coffee Simulation Model

#Scoops Taste
Taste LSL
Water Volume

Water Temperature
I/O
1.5 Score 9.5
AGE SYSTEM
Defect Rate 65%
TIME

Coffee Maker
Taste
Coffee Type

6 Score 9.5
LSL Defect Rate: 32 dpm
Design Exceeds Customer Needs!
© 2003 All Rights Reserved eNPI Process. Do not reproduce Profitable @ Launch!
 Why Does it Work?
“It’s the difference between getting a tune-up and a brand
new engine; between patching your pants and getting a new
pair. Instead of constantly debugging products and processes
that already exist-an effort that never ends, of course-DFSS
starts from scratch to design the product to be virtually error
free.”
Design for Six Sigma by Subir Chowdhury

© 2003 All Rights Reserved eNPI Process. Do not reproduce

 Elements of Successful DFSS
Deployments
Customer
Focussed

Professional
Skills Team Based
Development

Leadership
Development
Baked Design to
Profitability
Into Your
NPI
Statistical
Program Design for
Management Six Sigma

Phased
Development

© 2003 All Rights Reserved eNPI Process. Do not reproduce

 Phase Gate
Reviews

A Set of Tools and Methodologies applied at various stages of

the New Product Introduction Process which enable
development of products, processes, or services that meet or
exceed customer requirements while maximizing quality and
profitability and minimizing design cycle time
© 2003 All Rights Reserved eNPI Process. Do not reproduce
 Team Based
High
Performance
Team
Performance Impact

Real Team

Working
Group
Potential
Team
Pseudo
Team

Team Effectiveness
Distance Effects on Communication

60% Thomas J. Allen , “We have been in contact with thousands of

Least Once Per Week
Communicating at

50%
Managing the Source of people who have worked on development
Probability of

Technology teams. Without exception, those that have

40%
worked on colocated teams insist that this is
30%
the most effective way to do development.”
20%
Donald G. Reinertsen, Managing the Design
10%
Factory
0%
0 5 10 15 20 25 30 35 40 45 50
Separation Distance (Feet)

© 2003 All Rights Reserved eNPI Process. Do not reproduce

 Gemba
"We must go the "real place" where are customers are, and observe
first-hand their situation, their problems, their opportunities. Methods
such as in-context market research and contextual inquiry are used.
From this time in the gemba we gather the "voices of our customers"--
their verbatims, or actual statements.“ Unknown

© 2003 All Rights Reserved eNPI Process. Do not reproduce

 Satisfied

Kano “That’s Cool”

Customer

•Web page editing on IE Tool Bar ds

e e
•Free Internet Access @ Starbucks •Web page editing ”N l
on IE Tool Bar
ona
•Retractable light under hood s i
en
m
•Send digital pix with your cell phone
1 Di
“ “More Is Better”
r
nea
s Li •Speed of Internet Connection
eed
eN •Flavor of coffee
c tiv
t tra •Gas Mileage
A
•Battery life on digital camera

Indifferent Needs
Poor “I don’t care” Excellent
Functionality •MapPoint on standard tool bar Functionality
s
•Color of coffee cup a v e Need
H
Must
•Size of tires “If it’s not there, I’ll never
•Voltage of battery be satisfied”
•Ability to Print
•Coffee is hot
•Car has brakes
•Ability to transfer digital
pictures to computer

Dissatisfied
© 2003 All Rights Reserved eNPI Process. Do not reproduce Customer
 I want to setup
CTQ Flow Down
wireless in my
home…! How Measured

Kano Need Type

Range of Wireless
Technical Support
% Successful w/o
Importance

Cost of Wireless
Priority
Ability

Time to Setup

Time to Setup
Base Station

Download
Speed of
Network
Customer Outcomes
Easy Initial Setup 5.00 1.00 45.00 One-Dimensional 9 9 3 1
Lightning Fast Downloads 5.00 1.00 45.00 One-Dimensional 1 9
Works Everywhere in my Home 4.00 1.00 36.00 Must-Have 3 1 9
Less Expensive Than Hardwiring 3.00 1.00 27.00 Attractive 1 3 3 1 3 9
Fast to Set Up Home Network 3.00 1.00 27.00 One-Dimensional 9 9 1 1
Overall Ability Score 10% 48% 33% 31% 27% 27% 19%
P re vie w

S ingle La be l
Easy Initial
Setup
Measurement Units %
Target Value
Upper Specification
100
Min
10
15
min
5
15
Mb/s
54
ft
Time to Setup - (T)
300
$
79
125
Lower Specification 85 10 200
Ma ke La be ls Statistical Tolerance
Smaller Is Smaller Is Bigger is Bigger is Smaller Is
R e a d Ma trix Parameter Type Target Better Better Better Better Better
T USL

0
-2 22
© 2003 All Rights Reserved eNPI Process. Do not reproduce
 Design to Profitability
Economic Analysis Tool
LBR : Labor DC : Development Cost

LBRR : Labor Rate PPU : Profit Per Unit

ODC : Other Direct Costs PPY : Profit Per Year

I/O
TTD : Time to Develop SYSTEM BE : Break Even Point

SP : Sales Price PM : Profit Margin

QTY : Quanity of Units/Year

COGS : Cost of Goods Sold/Unit

Profit Margin - (PM)

Labor - (LBR) Statistical Tolerance
Statistical Tolerance LSL USL
T

LSL T USL
-113.67 % 98.395

Characteristic Value
Average: -7.6351
Standard Deviation: 35.343
Cp: 0.165
-22.484 362.48 Cc: 1.71
Cpk: -0.166
Def. Rate (normal): 75.9 %
Z-Score (short-term): -0.704
Sigma Level: -0.704

Interval for Values = (-113.67, 98.395) +/-3SD

© 2003 All Rights Reserved eNPI Process. Do not reproduce
 Visualize Robustness
Spring Force F=K*X
100
K X F
90

80

70

60
Force (lb)

K=200
50
K=50

40 Design Nominals

30

20

10

Process Variation
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2
Displa cement (Inches)

© 2003 All Rights Reserved eNPI Process. Do not reproduce

 Statistical Tolerancing
Dimensions Target Tolerance Stdev
A 4.500 +/- 0.010 0.006
B 0.125 +/- 0.005 0.015
C 0.375 +/- 0.005 0.006
D D 4.000 +/- 0.010
A
B
Gap - (D)
Statistical Tolerance
C
LSL T USL

D  A B C
3.9483 inches 4.0517

D   2
A  2
B  2
C Characteristic Value

Average: 4
Standard Deviation: 0.017

Def. Rate : 56.2 %

Sigma Level:
-0.155
© 2003 All Rights Reserved eNPI Process. Do not reproduce
 Scorecard
Parameter Units Discrete or Continuous USL Target LSL Mean Stdev Zl Zu pl pu PPM Defective Zt (Sigma Score)
A cm Continuous 12 10 8 9.5 2.3 0.65 1.09 26% 14% 395,673 0.26
B m Continuous 3 2 1 2 0.3 3.33 3.33 0% 0% 858 3.14
C mm Continuous 5 3 1 3 1 2.00 2.00 2% 2% 45,500 1.69
D kg Continuous
Parameter 20
Units Discrete or Continuous 15
USL10Target
17.2LSL2.1 3.43Stdev
Mean 1.33 0% Zl 9% Zu pl pu 91,515 PPM Defective Zt 1.33
(Sigma Score)
E Xmg Continuous
cm Continuous 125 100 5575 96
35 15 20 1.05
35 1.45 15% 7%
5 4.00 4.00 0% 0% 220,388 63 0.77 3.83
Part A YYield Discrete
m Continuous 14 12 10 12 0.8 2.50 2.50 1% 1%230,000 12,419 0.74 2.24
Part B ZYield Parameter
Discrete
mm Continuous 5 USL 3Target
Units Discrete or Continuous 1 LSL3 Mean1 Stdev
2.00 2.00Zl 2%Zu 2%pl 5,000
pu PPM45,500
Defective2.58 Zt (Sigma
1.69Score)
Part C Yield GG
BB Discrete
kg Continuous
cm Continuous 100 4590 8040 35 90 2.1
38 4.76 1.14.76
2.73 0%
6.36 0%680,000
0% 0% -0.47
2 3,193 4.62 2.73
Part D Yield MM
CC Discrete
mg Continuous
m Continuous 125 100 75 9
10 96
8 95 4.200.35.80
3.33 0%
3.33 0%150,000
0% 0% 13 858 1.04 4.20 3.14
Part E Yield LL
Process Discrete
A Yield Discrete
mm Continuous 5 3 1 3 0.5 4.00 4.00 0% 0% 50 4,000 3.89
63 2.65 3.83
ProcessBBB Yield Discrete
kg Continuous 20 15 10 16 1.5 4.00 2.67 0% 0% 5,000 3,862 2.58 2.66
ProcessCCC Yield Discrete
mg Continuous 125 100 75 101 4 6.50 6.00 0% 0% 68 0 3.82 5.99
Process D Yield
Mode A Discrete
Yield Discrete 10 23 4.26 4.08
Module KLOC Defects Defects/KLOC Zt (Sigma Score)
Process E Yield
Mode B Discrete
Yield Discrete 50 5,000 3.89 2.58
A 20 200 10 2.33
Mode C Yield Discrete B 10 75 120 8 3.67 2.43
Mode D Yield Discrete C 5 55 80,000 11 1.41 2.29
Mode E Yield Discrete D 30 120 5 4 4.42 2.65
E 80 450 6 2.53

Parts Process Performance Total

PPM 1,818,984 67,126 93,125 1,979,235
Opportunities 150 223 450 823
DPMO 12,127 301 207 2,405

Sigma 2.25 3.43 3.53 2.82

© 2003 All Rights Reserved eNPI Process. Do not reproduce

 Design of Experiments
Factors Response 1 Controllable Factors
Name Post RB Arm Stop 3 Factor Name Units Low Level (-) High Level (+)
2 Runorder A B C D Y A Post Pin Pos 1 4
-1 -1 -1 -1 85 B RB # 2 3
1 -1 -1 -1 0 C Arm Pin Pos 1 5
-1 1 -1 -1 140 D Stop Pin Pos 1 3
1 1 -1 -1 14
-1 -1 1 -1 150 Y=f(x) Optimization Settings 4
1 -1 1 -1 57 Effect Coeff Coded Levels Uncoded Levels
-1 1 1 -1 201 Constant 93.375 93.375
1 1 1 -1 100 A -101.25 -50.63 -0.58 1.64
-1 -1 -1 1 85 B 40.75 20.38 -1.00 2.00
1 -1 -1 1 0 C 67.25 33.63 0.83 4.67
-1 1 -1 1 140 D 0.00 0.00 1.00 3.00
1 1 -1 1 14 AB -12.25 -6.13 0.58
-1 -1 1 1 150 AC 4.25 2.13 -0.48
1 -1 1 1 57 AD 0.00 0.00 -0.58
-1 1 1 1 201 BC 6.25 3.13 -0.83
1 1 1 1 100 BD 0.00 0.00 -1.00
CD 0.00 0.00 0.83
Average + 42.75 113.75 127.00 93.38 ABC 8.25 4.13 0.48
Average - 144.00 Effect
73.00 Pareto
59.75 93.38 ABD 0.00 0.00 0.58
Delta (Effect) -101.25 40.75 67.25 0.00 ACD 0.00 0.00 -0.48
80.0
Delta/2 (Coefficient) -50.63 20.38 33.63 0.00 BCD 0.00 0.00 -0.83
ABCD 0.00 0.00 0.48
60.0 5
Prediction 125.00
40.0

Main Effects Plot

20.0

0.0 160
ABCD
A

AB

AC

AD

BC

BD

CD

ABC

ABD

ACD

BCD

A 1 144 140
Effect

-20.0 4 42.75
120
Response

B 2 73 100
-40.0 3 113.75 80
60
-60.0 C 1 59.75
5 127 40
20
-80.0 D 1 93.375 0
3 93.375
1 4 2 3 1 5 1 3
-100.0
A B C D

-120.0 Factor & Level

© 2003 All Rights Reserved eNPI Process. Do not reproduce

 Risk Mitigation
LO IMPACT HI
1 2 3 4 5
HI

5 5

4 4
LIKELIHOOD

3 3

2 2

1 1
LO

1 2 3 4 5

© 2003 All Rights Reserved eNPI Process. Do not reproduce

 Phased Development
“If you don’t bake DFSS into your New
Product Introduction Process, and hold
teams accountable to the use of DFSS at
Phase Gates, your deployment will fail.”
David Cote, CEO Honeywell

Phase Requirements
Gate Met

Requirements
Not Met

© 2003 All Rights Reserved eNPI Process. Do not reproduce

 Compare and Contrast DFSS
With Current Practices
•
•
•
•
•

© 2003 All Rights Reserved eNPI Process. Do not reproduce

 Deployment Lessons Learned
• Cultural Change will be your biggest barrier
– Get a few wins
– Don’t do “Parallel Processing”!
– Failure means you’re learning
• Leadership critical
• Start Early in the design phase!
– Trust the process to define the product.,,,
• This will be very tough!
• Master the basics-don’t complicate it
– If it’s easy to do, people will embrace and try it
• If you can’t measure it, you can’t improve it!
– Put the measurement system in place-ASAP!
• Reward success
– “Not rewarding a team for quality work is analogous to a practicing to be a concert
pianist in a deaf world!”
• Have Fun

© 2003 All Rights Reserved eNPI Process. Do not reproduce

 Resources
Organizational Reference Materials and Articles:
Managing the Design Factory, Donald G. Reinertsen Managing the Design Factory, Donald G. Reinertsen
Setting the PACE in Product Development, Michael Setting the PACE in Product Development, Michael
E. McGrath E. McGrath
Lean Thinking, James P. Womack and Daniel T. Total Quality Development, Don Clausing
Jones
Optimization & Variation Reduction in Quality,
The 21 Irrefutable Laws of Leadership, John C. Wayne Taylor
Maxwell
Improving Quality Through Planned
Hot Groups, Jean Lipman-Blumen & Harold J. Leavitt Experimentation, Ronald D. Moen, Thomas W. Nolan,
and Lloyd P. Provost
The Wisdom of Teams, Katzenbach
Statistics for Experimenters, George E.P. Box,
The 7 Habits of Highly Effective People, Stephen
William G. Hunter, and J. Stuart Hunter
Covey
Design and Analysis of Experiments, Douglas C.
Who Moved my Cheese, Spencer Johnson
Montgomery
Leading Change, John P. Kotter Robust Design,
The Fifth Discipline, Peter M. Senge
http://www.variation.com/techlib/ta-1.htm
The 17 Indisputable Laws of Teamwork, John C.
Maxwell New Product Development,
http://www.newproductdynamics.com/
What Every Manager Should Know about Training,
Robert F. Mager New Product Introduction Process,
www.enpiprocess.com
Design for Six Sigma, Subir Chowdhury

Customer Centric Product Definition, Sheila Mello

© 2003 All Rights Reserved eNPI Process. Do not reproduce

 x f() Y
R1 R1 R 2
T=50 ±10% Impedance  Impedance
R1  R 2 LSL T USL

35 65
25.713 Ohms 40.657

R2 Impedance R1 R2 Characteristic Value

T=100 ±10% Average: 33.2
Standard Deviation: 2.5
Def. Rate (%): 42.3
Sigma Level: .19
70 130

© 2003 All Rights Reserved eNPI Process. Do not reproduce

© 2003 All Rights Reserved eNPI Process. Do not reproduce
 Solving The Problem
Variation Contributors Manufacturing Solution
to Coffee Taste Taste
Statistical Tolerance
Contribution

VOL
LSL

5.5187 Goodness 7.3749

Characteristic Value
VOL AMT All Average: 6.4468
Design Solution Standard Deviation: 0.30937
Cp: ----
Cpk:
Taste 0.4814
Root MSE: Tolerance
Statistical ----
Defect Rate: 7.4342 %
LSL
3 Std. Dev. Tolerance for Values = (5.5187,

6 Goodness 8.8452

Characteristic Value
© 2003 All Rights Reserved eNPI Process. Do not reproduce