Design for Six Sigma

Opportunities for Medical Device Companies

Product Genesis www.productgenesis.com

245 Bent Street info@productgenesis.com
Cambridge, MA 02141
617-234-0070

© 2005 Kevin Otto

 Kevin Otto

Kevin_N_Otto@yahoo.com
www.kevinotto.com
Experience
– Lead Consultant, Product Genesis Inc.
– Lead Consultant, PDSS Inc.
– Former Associate Professor, MIT
– Six Sigma Master Black Belt
– Many past clients

Winner

© 2005 Kevin Otto – March 2005 –Slide # 2

 What is DFSS?

What if you were asked not to develop your product

instantly, but to “do it right”?

A “shift from deterministic to a probabilistic design culture”

(from DFSS: 15 Lessons Learned; Quality Progress; Jan. 2002)

© 2005 Kevin Otto – March 2005 –Slide # 3

 Historical Development Process
Development Process: Do it quick!

Define Design Alpha Beta PPU

Overall
Requirements

BUILD, TEST, FIX CYCLES!

BUILD, TEST, FIX CYCLES!

Product Evolution

© 2005 Kevin Otto – March 2005 –Slide # 4

 Quality: What it Means

Actual Margin $ Delivered

“Say / Do Ratio” =
Margin Promised in the Business Case

Goal

Today

50% 75% 100% 200%

2002 Say/Do Ratio Shortfall = $1.0 Billion.

Less than ½ of their product development projects
earned their entitlement business case margin.
PDMA Visions, 2004
© 2005 Kevin Otto – March 2005 –Slide # 5
 Quality: What it Means

The cost of fixing a single defect:

$35 during the design phase
$177 before procurement
$368 before production
$17,000 before shipment
$690,000 on customer site

Mr. Hiroshi Hamada, President of Ricoh

Source: European Community Quarterly Review, Third Quarter 1996

© 2005 Kevin Otto – March 2005 –Slide # 6

 Capability Defects / Cost as
Sigma
Million % Sales
6 3.4 < 1%
5 233 5 – 15%

1,000,000 4 6,210 15 – 25%

IRS Tax Advice 3 66,807 25 – 40%
100,000
Restaurant Bill Airline Baggage
10,000
DPMO

Handling
Prescription Writing
1,000

100
World Class Quality

10

Airline Fatality Rate

1
2 3 4 5 6
Sigma
Most companies operate at ≤ 4 sigma
What are your warrantee and service costs?
© 2005 Kevin Otto – March 2005 –Slide # 7
 Operations & Production Six Sigma

Define Measure Analyze Improve Control

What’s the problem How to measure it What changes it What to change How to keep it there
Clarify it - Not defect counts - Controllable - Controllable - Control plans
- Defect data - Output variation - Noise
- What defects - Measurement system - Signals
- What productn steps

Tools: Tools: Tools: Tools: Tools:

- Fishbone diagrams - P-diagram - DoE - Robust Design - SPC
- Observation - MSA - Robust Design - Signal to Noise - MSA
- Discussion - Data collection - CPM
- Stat. Tolerances

Generates “bottom line” financial value by eliminating Cost-

Of-Poor-Quality (COPQ) in production & business transactions

© 2005 Kevin Otto – March 2005 –Slide # 8

 It works

GE published a net benefit of

$2 billion in its 1999 annual report
Jack Welch, has said Six Sigma will
save his company $12 billion over
five years and will add $1 to its
earnings per share

Allied Signal has saved $1.2 billion in direct costs since

1994

Asean Brown Boveri (ABB) saved $898 million each year for
two years

© 2005 Kevin Otto – March 2005 –Slide # 9

 Six Sigma Across the Enterprise

Product Strategy & Planning Product Development

PSP
P DFSS
I D E A C D O V

TDFSS OSS
I2 D O V D M A D V
Research & Development Support Engineering

PSS
D M A I C
The six-sigma data-driven approach is Manufacturing
expanding out of manufacturing and
into every aspect of business. MFSS
L M A D
Sales and Marketing
© 2005 Kevin Otto – March 2005 –Slide # 10
 Improve Existing or New Products First?

Bob Galvin, CEO of Motorola stated that…

If he would start six sigma again,

he would focus on product development
rather than manufacturing.

Galvin’s view is that mfg. process improvement is often the

result of poor product development.

With any corporate Six Sigma implementation, there

occurs a natural evolution out into the organization

R & D offers the highest leverage against the cost of poor

quality. DFSS.

© 2005 Kevin Otto – March 2005 –Slide # 11

 Design for Six Sigma
Generates “top line” financial value by providing new products
with no problems and thereby generate new revenue.
VOC QFD Module Concept FMEA
CPM
Gen & Sel
0

Response (Y1)

sodne tfowoH

?MF roesuaccteetd
oH

uoynacl lewwoH
veS w

F ro
e v
al uat o
i n? par t ?

eht ot tcefe
eht sie re

?re mtosu
c

?Wrucco M
P art under char act eri st c
i or f uncti onali ty ? E xter nal )? bef ore desi gnr ele ase?

c e
C har acter s
i t i cor pur pose of t he char act eri sti cl ose ti s C usto mer ( I nter nal or f nct o
u i n? t echniqu est od s i cover t he cause

esua
Wha t si h t e What s i hte I nw hat ways does th s i What s i h t e i mpact to t he W h at causes the l ossof hat ar e the t ests, met hods or

V C T N
Par t Pa rt Funct o i n C ont r ol
Pot ent i al Fai l ur eM od e Pot ent i al Fai l ur eE f ect s E Po e
t n ti al C auses C E P
Char act er i sti co r C har act er ist i cor C ur
r en t Desi gn Eval uat i on or
S O D R

Functional Architecture Variable (n1)

Va riable (x 1)
Su b- Resp onse (y 3)

Variable (x3)
Variable (n3) Va riable (x 4)
Sub- Response (y5)

Variable (x6) Variab le (x 7)

Reliability Prediction
Design of
99

Capability
95
90

80

Experiments
70

Equations
60

Robust Design
50
40
30
20

10
5

Response (Y 1) 1

System Stress
10 1 00
Ti me to Fai lu re

Variable (n1 )
Variable (x1 )
Sub- Response (y3 )

Tests Capability
Variable (n3 )
Variable (x3 )
Sub- Response (y 5)
Variable (x 4)
D

C
Verification
Signifies the OPTIMUM S/N Set point…
B
1

A
Variable (x6 ) Variable (x 7) 55

0:00 0:10 0:20 0:30

Time (hour:minute)
50

45
0 10 20 30

© 2005 Kevin Otto – March 2005 –Slide # 12

 What’s Different

Statistical tools for design

Eliminate or accommodate variability
Functionally Parametric Designs – Data and Equations
Shared Focus – Critical Parameter Management
Process Scorecards
Test Planning
Subsystems first
Experimentation over regions of the design space

Result:
1. No surprises, scrambles, ECOs at manufacturing launch.
2. Confidence against any surprise at manufacturing launch.

© 2005 Kevin Otto – March 2005 –Slide # 13

 Key Elements in Improving Product Development

Product Development
Processes

WHAT to do…
& WHEN to do it!

Best Practices Project Management

Schedule your project activities with risk and backup

planning project management tools
Map specific project activities to standard work with
Phases & Gates of an End-to-End Development Process
Insert appropriate Tools & Best Practices with your
detailed Project Activities

© 2005 Kevin Otto – March 2005 –Slide # 14

 Project Risk Management
Gather the Voice of the Generate Product
Customer (VOC) RequirementsDocument Develop Reliability Phase IGate Review
Ralph 10 days George 1 day Requirements,Initial
Wed 8/22/01 Tue 9/4/01 Thu 9/20/01 Thu 9/20/01 George 48 days VP of Div. 1 day
Fri 9/21/01 Tue 11/27/01 Mon 2/25/02 Mon 2/25/02

Refine & Rank the VOC

using KJ Analysis Define theFunctions that

High
Ralph 3 days Fulfill the System
Wed 9/5/01 Fri 9/7/01 Ed 2 days Analyze, Characterize,
Model& Predict Nominal
Fri 9/21/01 Mon 9/24/01
Ed 72 days Marketing resources
Risk
Create Product or System Thu 11/15/01Fri 2/22/02
Level House of Quality
Ralph 5 days Generate System
not available
Thu 9/13/01 Wed 9/19/01 ConceptEvaluation
Ralph 2 days Select Superior System
Wed 9/26/01 Thu 9/27/01 Concept
Segment X requirements unknown

Consequence
Ralph 3 days
Mon 11/12/0 Wed 11/14/0
Competitive Position/
Benchmarking Analysis Generate System
Ed 8 days Concepts that Fulfill the
Mon 9/3/01 Wed 9/12/01 Ralph 25 days
Fri 9/28/01 Thu 11/1/01
1. Market Segment rankings not complete
EvaluateSystem
2. Competitor ranking surveys not complete

Moderate
Concepts
Ralph 6 days
Fri 11/2/01 Fri 11/9/01

Risk
Microsoft Project

Phase
Task:
% Task
Fulfillment
Task
Result vs.
Red Yellow Green Deliverable Requirements
Low 3rd World Developing market
Risk
D&R Reqts
requirements unknown

Likelihood
Risk Scorecard for each Task Gate Review: Risk Summary
and backup plans

Manage not only time, but risk.

© 2005 Kevin Otto – March 2005 –Slide # 15

 Standard Work – The CDOV Process

Value Proposition
Concept Voice of Customer
Platform and Portfolio Architecture
Development Plan and Risks Superior Concept Gate

Minimize Complexity – System Arch

Design Minimize Risk – FMEA, Scorecards
Requirements Flowdown - CPM
Identify Noise – DOE / RSM
Characterized Design Gate

Optimize – Taguchi DOE

Optimize Stress the Design – HALT
Statistical Tolerancing
Robust Design Gate

Reliability Engineering – ALT

Verify Product Launch – CPM, SPC

Reliable Launch Gate

© 2005 Kevin Otto – March 2005 –Slide # 16

 Critical Parameter Management

Critical: focus on only the critical 10% of all requirements

Parameter: we will be quantitative, measurable and

testable

Management: we will improve, control, tradeoff with a total

systems perspective

© 2005 Kevin Otto – March 2005 –Slide # 17

 Criticality Scorecards
CFR Output Variability
CFR Output (Y) Y = f ( x1 ,..., xn )
Design For Six Sigma Scorecard
Estimates Based on Mean Condition of x's and n's Listed Below
Performance Transfer Function Specification Predicted Performance Capability 6σ Score
Characteristic Units Y/N Formula (enter below) Target USL LSL mean: µ s.d.: σ Short/Long Confidence z σ-shift DPM
Voltage V Y 2 2 2.1 1.9 2 0.008654 11.55 0.00 0.0

x's, Input Control Factors

Variables Range Contribution to Variability Specification Sample/Database Statistics 6σ Score
No. Characteristic Units Min Max Sensitivity % USL LSL mean: µ s.d.: σ Short/Long Confidence z σ-shift DPM
1 X1 ohms 20 500 0 0.00% 20 0.04899 -408.25 1000000.0
2 X2 ohms 2 50 -0.3108194 32.03% 6.433029 0.015758 -408.25 1000000.0
3 X3 ohms 2 50 0.4176437 32.04% 4.788771 0.01173 -408.25 1000000.0
4 X4 volts 1.2 30 0 0.00% 30 0.03873 -774.60 1000000.0
5 X5 ohms 2 50 0 0.00% 2 0.004899 -408.25 1000000.0
6 X6 ohms 0.7444038 32.04% 2.686714 0.006581 -408.25 1000000.0
7 I amp -10.449776 3.89% 0 0.000163 0.00 933192.8
8
9
10
11
12
13
14

CTS Control factors (x) CTS Control factor Variability

© 2005 Kevin Otto – March 2005 –Slide # 18

 Tracing Critical Relationships
Requirement allocation
flows down…
Y: System Level CFRs

Rqmt S, Cp Y = f ( x1 ,..., xn )

y: Subsystem Level CFRs

Y = f ( x1 ,..., xn )
Rqmt S, Cp

x: Component Level
CTFspecs.
Y = f ( x1 ,..., xn )
Rqmt S, Cp

p: Mfg. Process Level CFRs

Effects of variation flow
Reqmt USL-LSL up…
Cp = =
Capability 6s

© 2005 Kevin Otto – March 2005 –Slide # 19

 Design of Experiments and Robust Design
y

Generate equations to Less

Variation

Different Output Variation

relate responses to factors.
Y = f ( x1 ,..., xn )
Noise Factors

More
Variation

Signal Function Response Same Input Variation

Less More x
Robust Robust

Control Factors L12 NA NB NC ND NE NF NG NH NI NJ NK Y1 Y2 … Yn

1 1 1 1 1 1 1 1 1 1 1 1
2 1 1 1 1 1 2 2 2 2 2 2
3 1 2 2 2 2 1 1 1 2 2 2
4 1 2 1 2 2 1 2 2 1 1 2
5 1 1 2 1 2 2 1 2 1 2 1
6 1 1 2 2 1 2 2 1 2 1 1
7 2 2 2 2 1 1 2 2 1 2 1
8 2 2 2 1 2 2 2 1 1 1 2
9 2 1 1 2 2 2 1 2 2 1 1
10 2 1 2 1 1 1 1 2 2 1 2
11 2 2 1 2 1 2 1 1 1 2 2
12 2 2 1 1 2 1 2 1 2 2 1

© 2005 Kevin Otto – March 2005 –Slide # 20

 Effect of Critical Parameter Modeling at Toyota
$
1977

Launch
$
1984
Production launches become
non-events for engineering

Hauser and Clausing – “The House of Quality”

Launch Harvard Business Review, May-June 1988

What will happen when product starts, and there is a

problem with a component, assembly step, …?
– Off target, Too much variation
– Acting differently than the development prototypes
With DFSS, you know what to do Y = f ( x1 ,..., xn )
– You have pre-defined factors x to shift every response (Y or y)
– Factors x that production and design agree to use

© 2005 Kevin Otto – March 2005 –Slide # 21

 Problems You Will Encounter

Lack of executive support

Yet another program of the month

We already do this

We don’t need to do this

Hired external design firms are incapable

© 2005 Kevin Otto – March 2005 –Slide # 22

 Success Factors

Executive champion
– One who believes and is staking their job on DFSS success
– Clears obstacles faced by project champion
Project Manager
– One who believes and is staking their job on DFSS success
– Person skills: can attain buy-in from support functions
DFSS Master Black Belt
– Training and consulting, reports to Executive and PM
– Responsible for launch quality and lifetime reliability
– Ensure effective tool use
Defined Development Process
Test facilities, Field data collection, Supply chain support

© 2005 Kevin Otto – March 2005 –Slide # 23

Product Genesis www.productgenesis.com
245 Bent Street info@productgenesis.com
Cambridge, MA 02141
617-234-0070

© 2005 Kevin Otto