Design for Six Sigma

Opportunities for Medical Device Companies

Product Genesis
Innovation Genesis, LLC
One Canal Park
Cambridge, MA 02141
617-234-0070

www.productgenesis.com
info@productgenesis.com

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
Overall
Requirements

Design

Alpha

Beta

PPU

BUILD, TEST, FIX CYCLES!

BUILD, TEST, FIX CYCLES!

Product Evolution

2005 Kevin Otto March 2005 Slide # 4

Quality: What it Means

Say / Do Ratio =

Actual Margin $ Delivered

Margin Promised in the Business Case
Goal

Today

50%

75%

100%

200%

2002 Say/Do Ratio Shortfall = $1.0 Billion.

Less than 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

Sigma

Defects /
Million

Cost as
% Sales

3.4

< 1%

233

5 15%

6,210

15 25%

66,807

25 40%

1,000,000
IRS Tax Advice
100,000

DPMO

Restaurant Bill

Airline Baggage
Handling
Prescription Writing

10,000
1,000
100

World Class Quality

10
1

Airline Fatality Rate

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
How to keep it there
- Control plans

Whats the problem

How to measure it
Clarify it
- Not defect counts
- Defect data
- Output variation
- What defects
- Measurement system
- What productn steps

What changes it
- Controllable
- Noise
- Signals

What to change
- Controllable

Tools:
- Fishbone diagrams
- Observation
- Discussion

Tools:
- DoE
- Robust Design

Tools:
Tools:
- SPC
- Robust Design
- MSA
- Signal to Noise
- CPM
- Stat. Tolerances

Tools:
- P-diagram
- MSA
- Data collection

Generates bottom line financial value by eliminating CostOf-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
I D E A

DFSS
C D O V

TDFSS
I2 D O V

Research & Development

The six-sigma data-driven approach is

expanding out of manufacturing and
into every aspect of business.

OSS
D M A D V
Support Engineering
PSS
D M A I
Manufacturing
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.
Galvins 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
Gen & Sel
Functional Architecture

FMEA

CPM
Response (Y1)

Variable (x1)
Sub- Response (y3)

Variable (n1)

Variable (x3)
Sub- Response (y5)

Variable (n3)

Variable (x6)

Variable (x4)

Variable (x7)

Reliability Prediction
Design of
Experiments

99

Robust Design

Capability
Equations

95
90
80
70
60
50
40
30
20
10
5

Response (Y1)

Variable (n1)

Variable (n3)

System Stress
Tests

Variable (x1)
Sub- Response (y3)

Variable (x3)
Sub- Response (y5)

Variable (x4)

Signifies the OPTIMUM S/N Set point

10

100

T im e to Failure

Capability
Verification

Variable (x6)

Variable (x7)

A
55

0:00

0:10

0:20
Time (hour:minute)

0:30
50

45
0

2005 Kevin Otto March 2005 Slide # 12

10

20

30

Whats 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
Customer (VOC)

Generate Product
RequirementsDocument

Ralph

George

Refine & Rank the VOC

using KJ Analysis
Ralph

1 day

Thu 9/20/01 Thu 9/20/01

Develop Reliability
Requirements,Initial

Phase I Gate Review

George

48 days

VP of Div.

Fri 9/21/01

Tue 11/27/01

Mon 2/25/02 Mon 2/25/02

Define the Functions that

Fulfill the System

3 days

Wed 9/5/01 Fri 9/7/01

Ed

2 days

Fri 9/21/01

Mon 9/24/01

Analyze, Characterize,
Model& Predict Nominal
Ed

Create Product or System

Level House of Quality
Ralph

1 day

5 days

Generate System
Concept Evaluation

Thu 9/13/01 Wed 9/19/01

Ralph

2 days

Select Superior System

Concept

Wed 9/26/01 Thu 9/27/01

Ralph

3 days

Mon 11/12/0 Wed 11/14/0

CompetitivePosition/
BenchmarkingAnalysis
Ed

Generate System
Concepts that Fulfill the

8 days

Mon 9/3/01 Wed 9/12/01

Ralph

25 days

Fri 9/28/01

Thu 11/1/01

EvaluateSystem
Concepts
Ralph

6 days

Fri 11/2/01

Fri 11/9/01

Microsoft Project
Phase
Task:

% Task
Fulfillment

Task
Result vs.
D&R Reqts

Red

Yellow

Green

High
Risk

Marketing resources
not available

72 days

Thu 11/15/01Fri 2/22/02

Consequence

10 days

Wed 8/22/01 Tue 9/4/01

Deliverable Requirements

Risk Scorecard for each Task

Segment X requirements unknown

1. Market Segment rankings not complete

2. Competitor ranking surveys not complete

Moderate
Risk

Low
Risk

3rd World Developing market

requirements unknown

Likelihood

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

Concept

Value Proposition
Voice of Customer
Platform and Portfolio Architecture

Superior Concept Gate

Development Plan and Risks

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

Verify

Reliability Engineering ALT

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

Y = f ( x1 ,..., xn )

CFR Output (Y)

CFR Output Variability

Design For Six Sigma Scorecard

Performance
Characteristic
Voltage

Units
V

Y/N
Y

Transfer Function
Formula (enter below)
2

Specification
Target
USL
2
2.1

LSL
1.9

Estimates Based on Mean Condition of x's and n's Listed Below

Predicted Performance Capability
6 Score
-shift
mean:
s.d.:
DPM
z
Short/Long
Confidence
2 0.008654
11.55
0.00
0.0

x's, Input Control Factors

No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14

Variables
Characteristic
X1
X2
X3
X4
X5
X6
I

Units
ohms
ohms
ohms
volts
ohms
ohms
amp

Range
Min
Max
20
2
2
1.2
2

500
50
50
30
50

CTS Control factors (x)

Contribution to Variability
Sensitivity
%
0
0.00%
-0.3108194
32.03%
0.4176437
32.04%
0
0.00%
0
0.00%
0.7444038
32.04%
-10.449776
3.89%

Specification
USL
LSL

Sample/Database Statistics
mean:
s.d.:
Short/Long
Confidence
20
0.04899
6.433029 0.015758
4.788771
0.01173
30
0.03873
2 0.004899
2.686714 0.006581
0 0.000163

z
-408.25
-408.25
-408.25
-774.60
-408.25
-408.25
0.00

6 Score
-shift

CTS Control factor Variability

2005 Kevin Otto March 2005 Slide # 18

DPM
1000000.0
1000000.0
1000000.0
1000000.0
1000000.0
1000000.0
933192.8

Requirement allocation
flows down

Tracing Critical Relationships

Y: System Level CFRs

S, Cp

Rqmt

Y = f ( x1 ,..., xn )

y: Subsystem Level CFRs

Rqmt

S, Cp

Y = f ( x1 ,..., xn )

S, Cp

Y = f ( x1 ,..., xn )

x: Component Level
CTFspecs.
Rqmt

p: Mfg. Process Level CFRs

Reqmt
Cp =

Capability

USL-LSL
=

6s

2005 Kevin Otto March 2005 Slide # 19

Effects of variation flow

up

Design of Experiments and Robust Design

y

Generate equations to
relate responses to factors.

Different Output Variation

Noise Factors

Less
Variation

Y = f ( x1 ,..., xn )
More
Variation

Signal

Function

Response

Same Input Variation

Less
Robust

Control Factors

L12
1
2
3
4
5
6
7
8
9
10
11
12

NA
1
1
1
1
1
1
2
2
2
2
2
2

NB
1
1
2
2
1
1
2
2
1
1
2
2

NC
1
1
2
1
2
2
2
2
1
2
1
1

ND
1
1
2
2
1
2
2
1
2
1
2
1

NE
1
1
2
2
2
1
1
2
2
1
1
2

NF
1
2
1
1
2
2
1
2
2
1
2
1

2005 Kevin Otto March 2005 Slide # 20

More
Robust

NG
1
2
1
2
1
2
2
2
1
1
1
2

NH
1
2
1
2
2
1
2
1
2
2
1
1

NI
1
2
2
1
1
2
1
1
2
2
1
2

NJ
1
2
2
1
2
1
2
1
1
1
2
2

NK
1
2
2
2
1
1
1
2
1
2
2
1

Y1

Y2

Yn

Effect of Critical Parameter Modeling at Toyota

$

1977

Launch
$

1984

Production launches become

non-events for engineering
Hauser and Clausing The House of Quality
Harvard Business Review, May-June 1988

Launch

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 dont 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
Innovation Genesis, LLC
One Canal Park
Cambridge, MA 02141
617-234-0070

www.productgenesis.com
info@productgenesis.com

2005 Kevin Otto