aim macro system model micro system model taguchi methods at aerojet rocketdyne future state

The New EconomicsThe New Economicsof Variation: Learning from of Variation: Learning from Genichi TaguchiGenichi TaguchiPresented by Bill BellowsPresented by Bill Bellows

Associate FellowAssociate FellowInThinking NetworkInThinking NetworkAerojet RocketdyneAerojet RocketdyneEmail: Email: [email protected]: 818-519-8209Cell: 818-519-8209

The Boeing CompanyThe Boeing CompanyOctober 7, 2013October 7, 2013

mailto:[email protected]

AimMacro System ModelMicro System ModelTaguchi Methods at Aerojet

RocketdyneFuture State

AgendaAgenda

The aim of this session is to offer insights on how W. Edwards Deming’s “new economics” has integrated the thinking of two pioneers in the field of variation management – Walter Shewhart and Genichi Taguchi. In doing so, this presentation will contrast these systemic views of variation management with those of the prevailing style of management, in both cases offering reflections on the decisions made in quality improvement efforts under guidance from the “old” and “new “ economics.

W. Edwards Deming Genichi TaguchiWalter Shewhart

Macro System ModelMacro System Model

D E F

P

GI H

Process FlowProcess Flow

Handoff Requirements?

D E F

P

GI H

Process FlowProcess Flow

Part Production

Part AStep 1

Step 2Step N

GOOD


“Zero defects is another way of saying ‘do it right the first time’”

Quality is defined as conformance to requirements

Source: Let’s Talk Quality, P. Crosby, 1989

PartPart Quality Quality

Part Production

Part AStep 1

Step 2Step N

Part B

Part O

Part P

GOOD

GOOD

GOOD

GOOD

Step 1Step 2Step N

Step 1Step 2Step N

Step 1Step 2Step N


Part Production

Part AStep 1

Step 2Step N

Part B

Part O

Part P

AssemblyGOOD

GOOD

GOOD

GOOD

Step 1Step 2Step N

Step 1Step 2Step N

Step 1Step 2Step N


Part Production

Sub-Assembly 1

Part AStep 1

Step 2Step N

Part B

Part O

Part P

Assembly

FIT

GOOD

GOOD

GOOD

GOOD

Step 1Step 2Step N

Step 1Step 2Step N

Step 1Step 2Step N


Part Production

Sub-Assembly 1

Part AStep 1

Step 2Step N

Part B

Part O

Part P

Sub-Assembly 2

Assembly

FIT

FIT

GOOD

GOOD

GOOD

GOOD

Step 1Step 2Step N

Step 1Step 2Step N

Step 1Step 2Step N


Part Production

Sub-Assembly 1

Part AStep 1

Step 2Step N

Part B

Part O

Part P

Sub-Assembly 2

Assembly Final Assembly

FIT

FIT

GOOD

GOOD

GOOD

GOOD

Step 1Step 2Step N

Step 1Step 2Step N

Step 1Step 2Step N


Part Production

Sub-Assembly 1

Part AStep 1

Step 2Step N

Part B

Part O

Part P

Sub-Assembly 2

ProductAssembly


FIT

FIT

FIT

GOOD

GOOD

GOOD

GOOD

Step 1Step 2Step N

Step 1Step 2Step N

Step 1Step 2Step N


Part Production

Sub-Assembly 1

Part AStep 1

Step 2Step N

Part B

Part O

Part P

Sub-Assembly 2

ProductAssembly


FIT

FIT

FIT

GOOD

WORKS

GOOD

GOOD

GOOD

Step 1Step 2Step N

Step 1Step 2Step N

Step 1Step 2Step N


Quality is defined as conformance to requirements, not as 'goodness' nor 'elegance'.

The system for causing quality is prevention, not appraisal.

The performance standard must be Zero Defects, not 'that's close enough'.

The measurement of quality is the Price of Non-conformance, not indices.

The “Absolutes” of The “Absolutes” of PartPart QualityQuality

Source: Quality is Free, Philip Crosby, 1979

Quality Focus: Conformance to Requirements

Goal: Defect-Free Parts Activities: Assess Non-

Conformances, Scrap and Rework

Mindset: Reactive / Victim Skills: Fire-Fighting and

Problem Solving Impact: No Improvement in

Quality After Zero Defects, Temporary Solutions

Attributes: Physical and Mental Handoffs (separation, blame)

Macro System Model(Part Work)

Micro System ModelMicro System Model

Given a piece of wood that will be cut into 2 pieces....

how many lines will be drawn acrossthe top face before the cut is made ?

Wood ManagementWood Management

target

Wood ManagementWood Management

0 A B C diameter

Circle ManagementCircle Management

Which 2 of these 3 circles are closest to having the same diameter?

0 A BC diameter

Decisions DecisionsDecisions Decisions

MIN MAX

Which 2 of these 3 circles are closest to having the same quality?

UpperSpecification

Limit

Lower Specification

Limit

TARGET(desired value of

parameter)

“Loss to Society”

Taguchi’s Quality Loss Taguchi’s Quality Loss FunctionFunction

A BC

Part Production

Sub-Assembly 1

Part AStep 1

Step 2Step N

Part B

Part O

Part P

Sub-Assembly 2

ProductAssembly


Step 1Step 2Step N

Step 1Step 2Step N

Step 1Step 2Step N

Degrees of GOOD

Degrees of WORKS

Degrees of FIT

Degrees of FIT

Degrees of FIT

Micro System ModelMicro System Model

MAXMIN

HOLE DIAMETER

MAXMIN

2520

8:00 PM6:00 PM

ARRIVAL TIME

Examples of Variation Examples of Variation ManagementManagement

PAGE COUNT

OUTER DIAMETER

Taguchi Methods at Taguchi Methods at Aerojet RocketdyneAerojet Rocketdyne

Dr. Taguchi’s first visit to Canoga Park – May 1994. While here he met with 6 application teams, including manufacturing engineers on the Space Shuttle Main Engine.

Taguchi Methods – Taguchi Methods – History in Canoga ParkHistory in Canoga Park

Management briefings provided by American Supplier Institute (ASI) in 1989

"Introductory" training (36-42 hours) by ASI began in 1989

Internal expert hired in 1990 In-house "Introductory" seminars through the

TQM Office began in 1991 800+ graduates to date

Taguchi Methods – Taguchi Methods – History in Canoga ParkHistory in Canoga Park

First visit by Dr. Taguchi in May 1994 Dr. Taguchi reviewed applications with 6 teams

Second visit by Dr. Taguchi in November 1994 Dr. Taguchi reviewed applications with 4 teams Meetings arranged with Dr. Taguchi and executives

to review application strategies for R&D Third visit by Dr. Taguchi in May 1996

Dr. Taguchi and Professor Yuin Wu reviewed RS-68 combustion design activities

Taguchi Methods – Taguchi Methods – Deployment (1900 – 1993)Deployment (1900 – 1993)

First application (1990) – National Launch System - turbo-machinery design

First process application (1990) – etching process problem solving study

First “advanced" application (1993) – silver plating problem solving study

Taguchi Methods – Taguchi Methods – Directions (1994 – present)Directions (1994 – present)

Focus of implementation on problems, repairs, and rework was examined for clues of how to expand applications to “proactive use”

Links established between Taguchi Methods and the management theory of Dr. Deming

Using Deming’s ideas, broader implications of the thinking of Dr. Taguchi have been re-examined

Transformational thinking efforts began in 1993

Taguchi Methods – Taguchi Methods – Directions (1994 – present)Directions (1994 – present)

Curriculum of “thinking” courses established in 1995 – “A Thinking Roadmap”

Taguchi Methods seminar expanded into 2 40-hour seminars as part of “A Thinking Roadmap”

Thinking transformation efforts focus on a need to “manage variation”, not “reduce variation” “manage costs”, not “reduce costs” “manage resources”, not “reduce resources” “continuous investment”, not “continuous

improvement”

Taguchi Methods – Lessons Taguchi Methods – Lessons LearnedLearned

1. Variation Management2. Quality Loss Function3. Parameter (Robust) Design4. Tertiary Design

Background: Consider the following two processes and the specification limits and target provided. What is the motive for selecting process 2 over process 1?

USLLSL TARGET

1

2

Variation ReductionVariation Reduction

Background: Consider the following two processes and the specification limits and target provided.

USLLSL

TARGET

1

2

Variation ManagementVariation Management

Given the following 5 criteria, which is the preferred process – 1 or 2?

1 – Equal purchase prices2 – Equal delivery schedules3 – Zero defects are guarantee from both processes4 - Distributions will never change shape and/or location5 – The “team” selected the “target” value, which is preferred, plus the tolerances

USLLSL

target

1

2

Decisions DecisionsDecisions Decisions

UpperSpecification

Limit

Lower Specification

Limit

target(desired value of

parameter)

“Loss to Society”

(a greater system)

Taguchi’s Quality Loss Taguchi’s Quality Loss FunctionFunction

12

“Quality is the minimum of lossimparted to the Society by a product after its shipment to a customer”

Source: Introduction to Quality Engineering , G. Taguchi, 1983

Relationship QualityRelationship Quality““Mind the Gap”Mind the Gap”

Genichi TaguchiGenichi Taguchi Born in Japan in 1924 Joined Electrical Communication Laboratories in 1949 Received Deming Prize in 1951, 1953, 1960, and 1984 Received Ph.D. (Science) in 1962 Visited U.S. (Princeton and Bell Labs) in 1962 Employed as Professor (Japan) until 1982 Introduced Fuji-Xerox to methodology in 1972, AT&T in

1980, Xerox in 1982, and Ford in 1982 Honored by Emperor of Japan in 1989 Died in Japan in 2012

END VIEW OF OVEN

PROCESS FLOW

APPORTION,

PULVERIZE, AND

MIX MATERIALSMOLD PREFIRE

INSPECT,

PACKAGE, AND

SHIPGLAZEFIRE

BATH ROOM TILE MANUFACTURING, APPLICATION BY INA SEITO, CIRCA 1952

Early Example of Applying Early Example of Applying “Taguchi Methods”“Taguchi Methods”

TILE

SIZE

UPPER LIMIT

LOWER LIMIT

OUTSIDE INSIDETILES TILES

BEFORE STUDY

BEFORE STUDY

END VIEW OF OVEN


PROBLEM: OUTSIDE TILES OUT OF SPEC (30 % REJECTED)

CAUSE: NON-UNIFORM TEMPERATURE VARIATION WITHIN TILES

SOLUTION OPTIONS:




SOLUTION OPTIONS:

COMMON FEATURE:




ATTRIBUTE OF "BEST" SOLUTION:

SOLUTION OPTIONS:

COMMON FEATURE:


AFTER STUDY

TILE

SIZE

UPPER LIMIT

LOWER LIMIT

AFTER STUDY

OUTSIDE INSIDETILES TILES

BEFORE STUDY

BEFORE STUDY

END VIEW OF OVEN


PROCESS FACTORS

(EXISTING)

A) CONTENT OF LIME 5 % 1 %

B) FINENESS OF ADDITIVE COARSE FINE

C) CONTENT OF AGALMATOLITE 43 % 53 %

D) KIND OF AGALMATOLITE EXISTING NEW

E) RAW MATERIAL CHARGING QUANTITY 1300 KG 1200 KG

F) CONTENT OF WASTE RETURN 0 % 4 %

G) CONTENT OF FELDSPAR 0 % 5 %

CONTROL FACTORS

LEVELS

1 2

TOTAL (POSSIBLE) COMBINATIONS

OF 7 FACTORS AT 2 LEVELS EACH


TRIAL

1

2

3

4

5

6

7

8

LIME

5

5

5

5

1

1

1

1

(A)

% FINE-

CSR

CSR

FNR

FNR

CSR

CSR

FNR

FNR

(B)

NESS AGAL.

43

43

53

53

53

53

43

43

%

(C)

AGAL.

TYPE

EXSTG

NEW

EXSTG

NEW

EXSTG

NEW

EXSTG

NEW

(D)

QTY

CHRG

1300

1200

1300

1200

1200

1300

1200

1300

(E)

RTN

WSTE

0

4

4

0

0

4

4

0

(F)

FDSPR

0

5

5

0

5

0

0

5

(G)

1 2 3 4 5 6 7

1

1

1

1

2

1 1 1 1 1 1

1 1 2 2 2 2

2 2 1 1 2 2

2 2 2 2 1 1

1 2 1 2 1 2

2 1 2 2 1 2 1

2 2 1 1 2 2 1

2 2 1 2 1 1 2

A B C D E F G

L (2 )7

8

ORTHOGONAL ARRAY

TILE SIZE

(L/W/T)


Taguchi Methods – Taguchi Methods – Three Stages of DesignThree Stages of Design

1. Concept Selection (System Design) – define the product or process technology

2. Parameter Design – given levels of input (process or product) variation and average, define the “best” value for each input average (the “target”)

3. Tertiary Design – given levels of input (process or product) variation and a preferred average for each, define the “best” variation level for each input

Stages of DesignStages of Design

Inputs

AB

1. Concept Selection (System Design) – identify the product or process inputs and outputs

OutputsX

Y

ControlFactors(can be controlled)

AB

1. Concept Selection (System Design) – sort the inputs into Control Factors and Noise Factors

NoiseFactors(cannot be controlled) P Q

OutputsX

Y



AB



OutputsX

Y


Types of Noise Factors


AB



OutputsX

Y


Types of Noise FactorsInner Noise - wear


AB



OutputsX

Y


Types of Noise FactorsInner Noise - wearOuter Noise - use


AB



OutputsX

Y


Types of Noise FactorsInner Noise - wearOuter Noise - usePiece-to-Piece Variation

The Language of Taguchi The Language of Taguchi Methods (S/N = Robustness)Methods (S/N = Robustness)

Performance Measure


Performance Measure

"Signal" (average)


Performance Measure

"Signal" (average)

"Noise" (variation)


Performance Measure

"Signal" (average)

"Noise" (variation)

m, Target


Performance Measure

"Signal" (average)

"Noise" (variation)

m, Target(m = 0 ? , m = ? )oo


Performance Measure

"Signal" (average)

"Noise" (variation)

m, Target(m = 0 ? , m = ? )oo

Quality Loss


Performance Measure

"Signal" (average)

"Noise" (variation)

m, Target(m = 0 ? , m = ? )oo

Quality Loss

Simultaneous Objectives:


Performance Measurem, Target(m = 0 ? , m = ? )oo

Quality Loss

"Signal" (average)

"Noise" (variation)

Simultaneous Objectives:•Adjust average to target (lower “loss”)


Performance Measurem, Target(m = 0 ? , m = ? )oo

Quality Loss

"Signal" (average)

"Noise" (variation)

Simultaneous Objectives:•Adjust average to target (lower “loss”)•Increase S/N (lower “loss”)


Performance Measure

"Signal" (average)

"Noise" (variation)

m, Target(m = 0 ? , m = ? )oo

Simultaneous Objectives:•Adjust average to target (lower “loss”)•Increase S/N (lower “loss”)•Lower UMC (unit manufacturing cost)

Quality Loss

2. Parameter Design – exploit non-linearity to define the “target” value for Control Factor

1 2

Output - X

Input - A


3. Tertiary Design – fine-tune process variation for each Control Factor

2

Output - X

Input - A


Resource Management Resource Management (US vs. Japan)(US vs. Japan)

1. Concept Selection – 70 % vs. 40%

2. Parameter Design – 2% vs. 30%

3. Tertiary Design – 28% vs. 30%

Source: Introduction to Quality Engineering , G. Taguchi, 1983

Taguchi Methods – Taguchi Methods – Deployment (1900 – 1993)Deployment (1900 – 1993)

First application (1990) – National Launch System - turbo-machinery design

First process application (1990) – etching process problem solving study

First “advanced" application (1993) – silver plating problem solving study

The Language of Taguchi The Language of Taguchi Methods – Technology Methods – Technology DevelopmentDevelopment

Performance Measure

"Signal" (average)

"Noise" (variation)

Static Target

Variable Target

Future State Future State (The (The NewNew Economics of Economics of RelationshipRelationship Quality) Quality)

0.22980.2299

0.23000.2301

0.23020.2303

0.23040.2305

0.23060.2307

0.23080.2309

0.23100.2311

0.23120.2313

0

50

100

150

200

250

300

350

Hole Diameter, inches

Frequency

LOWER SPEC LIMIT 295 UPPER SPEC LIMIT

101

42

121

206

166

80

3112 10 1 5 7 2

Good Good PartsParts vs. Bad vs. Bad PartsParts

.2305 + .0005 Hole

Tube

Consider a brazing process for SSME nozzle joining

Better ValueBetter Value

Manifold

Liquid hydrogen

Flames

Traditional Approach

Better Value – Better Value – Tube in Hole - Tube in Hole - NextNext Assembly Assembly

Tube

Manifold

Traditional Approach Ream/ rework holes


Tube

Manifold

Traditional Approach Ream/ rework holes Braze flow thru holes


Tube

Excess braze

Manifold

Traditional Approach Ream/ rework holes Braze flow thru holes Crack welds


Tube

Next assembly part

Excess braze

Manifold

Traditional Approach Ream/ rework holes Braze flow thru holes Crack welds Add grind operation


Tube

Next assembly part

Excess braze

Manifold

Traditional Approach Ream/ rework holes Braze flow thru holes Crack welds Add grind operation Cracking continues


Tube

Next assembly part

Excess braze

Manifold

Traditional Approach Ream/ rework holes Braze flow thru holes Crack welds Add grind operation Cracking continues Add acid etch


Tube

Next assembly part

Excess braze

Manifold

Traditional Approach Ream/ rework holes Braze flow thru holes Crack welds Add grind operation Cracking continues Add acid etch Cracking continues


Tube

Next assembly part

Excess braze

Manifold

Traditional Approach Ream/ rework holes Braze flow thru holes Crack welds Add grind operation Cracking continues Add acid etch Cracking continues Add “better” etch


Tube

Next assembly part

Excess braze

Manifold

MACHINING RESULTS FOR 1080 HOLES

HOLE DIAMETER, INCHES

FREQUENCY

LOWER SPEC LIMIT UPPER SPEC LIMITTARGET

0

100

200

300

400

500

101

42

121

206

166

295

80

31

12 101 5 7

1 2 2 1 1 1 1 10.2300

0.2301

0.2302

0.2303

0.2304

0.2305

0.2306

0.2307

0.2308

0.2309

0.2310

0.2311

0.2312

0.2313

0.2314

0.2315

0.2316

0.2317

0.2318

0.2319

Better Value – Better Value – Drilled Hole DataDrilled Hole Data

MACHINING RESULTS FOR 1080 HOLES - "BEFORE" & "AFTER"

"BEFORE"

HOLE DIAMETER, INCHES

FREQUENCY

LOWER SPEC LIMIT UPPER SPEC LIMITTARGET

0

100

200

300

400

500

101

42

1

121

4

206

166

295

80

31

12 10

53

179

246

109

34

20

"AFTER"

13

74

112 2 4

209

450

315

16 5 3 7

1 2 1 2 1 1 1 1 1

0.2300

0.2301

0.2302

0.2303

0.2304

0.2305

0.2306

0.2307

0.2308

0.2309

0.2310

0.2311

0.2312

0.2313

0.2314

0.2315

0.2316

0.2317

0.2318

0.2319

Better Value – Better Value – Drilled Hole Data Drilled Hole Data Post Taguchi ExperimentPost Taguchi Experiment

Better Approach Improve hole drilling

To target Better distribution

Successful first-cycle braze No excess braze, No

etching required


Tube

Next assembly part

No excess braze

Manifold

Traditional Approach Ream/ rework holes Braze flow thru holes Crack welds Add grind operation Cracking continues Add acid etch Cracking continues Add “better” etch

MAXMIN

HOLE DIAMETER

MAXMIN

RelationshipRelationship Management Management

OUTER DIAMETER

Tube

Next assembly part

No excess braze

Manifold

RelationshipRelationship Quality – Quality – The The NewNew Economics Economics

“The Taguchi Loss Function is a better view of the world.”

W. Edwards Deming

Source: Out of the Crisis, W. Edwards Deming, 1986

Dr. Taguchi at 80 in 2004Dr. Taguchi at 80 in 2004








Quality Focus: Relationships Between Parts (Target Thinking)

Goal: Profit Beyond Measure Activity: Seeking Opportunities to

Invest in Better Relationships between Parts

Mindset: Proactive / Leader Skills: Process Control and Systemic

Solutions Impact: Continuous Investment in

Quality of Relationships, Long-Lasting Solutions

Attributes: Physical Handoffs, without Mental Handoffs (no separation nor blame)

Micro System Model(Team Work)


















MIND THE PARTMIND THE PART

















MIND THE PARTMIND THE PART MIND THE GAPMIND THE GAP

Opportunities to ThinkOpportunities to Think

An InThinking RoadmapAn InThinking RoadmapAKA The Hotel CaliforniaAKA The Hotel California

InThinking Together (9 hrs)(Formerly known as “ET” and “Understanding Variation”)

Six Thinking Hats (8 hrs)

Kepner-Tregoe(24 hrs)

(Problem Solving and Decision Making)

Workshop”)

ManagingVariation as a

System (9 hrs)

The NewEconomics

Study Session

(14 hrs)

Understanding Taguchi Methods – Part 2 (40 hrs)

UnderstandingTaguchi Methods – Part 1

(40 hrs)

ools

DATT(16 hrs)irectppliedhinking

-

(Problem Solving and

DATT(16 hrs)DATT(16 hrs)

Design of Experiments & Taguchi Methods – An Overview (16 hrs)

Leading Systems(12 hrs)

(AKA the “Organization

Resource Leadership(8 hrs)

OD

iscu

ssio

n

ngoi

ng

(4th week, Th/Fri, 12-2pm PT)

Prerequisites

BTA…webinar

hink

ing

ette

r

(2nd week, Th/11:30-1pm PT)

bout

Lateral Thinking(16 hrs)






Workshop”)


System (9 hrs)

The NewEconomics

Study Session

(14 hrs)



(40 hrs)

ools


-







OD

iscu

ssio

n

ngoi

ng


Prerequisites

Distance LearningOpportunities

BTA…webinar

hink

ing

ette

r


bout

Lateral Thinking(16 hrs)






Workshop”)


System (9 hrs)

The NewEconomics

Study Session

(14 hrs)



(40 hrs)Lateral Thinking

(16 hrs)

ools


-







OD

iscu

ssio

n

ngoi

ng


Prerequisites

BTA…webinar

hink

ing

ette

r


bout

Beth Thompson, myrna.e.thompson, from

BCAG, is a certified instructor

Distance LearningOpportunities with Rocketdyne

Monthly AnnouncementMonthly Announcement

TARGET AUDIENCES: Members of management, individual contributors, suppliers, and customers who are providing leadership in "enterprise thinking" activities. Family members, "members of the community" and students are welcome to attend. "Members of the community" are citizens who are involved full or part time, or in a volunteer capacity, in community related work. Examples include hospital employees, teachers, religious leaders, scouting leaders, and youth sports volunteers.

An InThinking RoadmapAn InThinking Roadmap

The New EconomicsThe New Economicsof Variation: Learning from of Variation: Learning from Genichi TaguchiGenichi TaguchiPresented by Bill BellowsPresented by Bill Bellows

Associate FellowAssociate FellowInThinking NetworkInThinking NetworkAerojet RocketdyneAerojet RocketdyneEmail: Email: [email protected]: 818-519-8209Cell: 818-519-8209

The Boeing CompanyThe Boeing CompanySeptember 24, 2013September 24, 2013

mailto:[email protected]

aim macro system model micro system model taguchi methods at aerojet rocketdyne future state

Documents

step nstep

quality focus

measurement of quality

quality improvement

new economicsof variation

edwards demings new

prevailing style of

genichi taguchipresented