Vous êtes sur la page 1sur 49

Chapter 12 Design for Six Sigma (DFSS)

QFD, Reliability analysis, Taguchi loss function, Process capability

02/25/06

SJSU Bus. 142 - David Bentley

DFSS Activity Categories

Concept development Design development Design optimization Design verification


Well look at each of these in detail
02/25/06 SJSU Bus. 142 - David Bentley 2

Concept Development

Based on:

Customer requirements Technological capabilities Economic considerations

Tools

Quality Function Deployment (QFD) Concept engineering

02/25/06

SJSU Bus. 142 - David Bentley

Quality Function Deployment (QFD)


Structured approach for design Developed at Mitsubishis Kobe shipyards House of quality built on relationships

4 layers: product, part, process, production (quality plans)


Rev. 11/25/02 SJSU Bus. 142 - David Bentley 4

Customer requirements Design requirements Competitive assessment Technical assessment

Operations Management, Seventh Edition, by William J. Stevenson Copyright 2002 by The McGraw-Hill Companies, Inc. All rights reserved.

The House of Quality


Correlation matrix Design requirements

Customer requirements

Relationship matrix

Competitive assessment

Specifications or target values


11/21/02 SJSU Bus 142 - David Bentley 5

THE MANAGEMENT AND CONTROL OF QUALITY, 5e, 2002 South-Western/Thomson LearningTM

House of Quality
Interrelationships
Technical requirements Voice of the customer Relationship matrix Technical requirement priorities
SJSU Bus. 142 - David Bentley

Customer requirement priorities

Competitive evaluation
6

QFD Example
X

Operations Management, Seventh Edition, by William J. Stevenson Copyright 2002 by The McGraw-Hill Companies, Inc. All rights reserved.

Correlation:
X X

Water resistance

Accoust. Trans. Window

Energy needed to close door

Door seal resistance

Engineering Characteristics

Check force on level ground Energy needed to open door

Strong positive Positive Negative Strong negative


X = Us A = Comp. A B = Comp. B (5 is best) 1 2 3 4

Competitive evaluation

Customer Requirements Easy to close Stays open on a hill Easy to open

7 5 3 3 2
Reduce energy level to 7.5 ft/lb

X
X AB

AB

XAB A XB

Doesnt leak in rain


No road noise Importance weighting Target values

X A

10

6
Maintain current level

6
Reduce force to 9 lb.

9
Reduce energy to 7.5 ft/lb.

2
Maintain current level

3
Maintain current level

Relationships:
Strong = 9 Medium = 3 Small = 1

Technical evaluation (5 is best)

5 4 3 2 1

B A X

BA X

B A X

B X A

BXA

BA X

11/21/02

SJSU Bus 142 - David Bentley

QFD Steps - 1
1.

2. 3.

4.

Identify/ prioritize customer requirements Determine technical requirements Relate customer requirements to technical requirements Compare ability to meet requirements against competitive products
02/25/06 SJSU Bus. 142 - David Bentley 8

QFD Steps - 2
5.

6.

7.

Set targets for technical requirements and determine capability Look for high opportunity requirements to satisfy customer Continue QFD process to the next level.

02/25/06

SJSU Bus. 142 - David Bentley

THE MANAGEMENT AND CONTROL OF QUALITY, 5e, 2002 South-Western/Thomson LearningTM

QFD Levels

technical requirements
component characteristics process operations
SJSU Bus. 142 - David Bentley

quality plan
10

Concept Engineering

Understand customer environment Convert into requirements Deploy learning into operations Generate concepts Select appropriate concept

02/25/06

SJSU Bus. 142 - David Bentley

11

Design Development

Product and process performance issues Focus on ability to meet requirements in operations Tools

Tolerance design and process capability Design failure mode and effects analysis (DFEA) Reliability prediction
02/25/06 SJSU Bus. 142 - David Bentley 12

Tolerance Design - 1

Specification

Translation of customer requirements into design requirements Consists of nominal value and tolerances Ideal dimension or target value for meeting customer requirement

Nominal value

Tolerance

Allowable variation above and/or below nominal value Recognizes natural variation (common causes)
02/25/06 SJSU Bus. 142 - David Bentley 13

Tolerance Design - 2

Consider tradeoff between costs and performance Too tight tolerances = unnecessary cost Too loose tolerances = not meeting customer requirements End result: too loose or too tight is going to cost you money!
02/26/05 SJSU Bus. 142 - David Bentley 14

DFMEA

Design failure and effects analysis (DFMEA) Identify all the ways failures can occur Estimate effects of the failures Recommend changes in design

02/26/06

SJSU Bus. 142 - David Bentley

15

THE MANAGEMENT AND CONTROL OF QUALITY, 5e, 2002 South-Western/Thomson LearningTM

SJSU Bus. 142 - David Bentley

16

THE MANAGEMENT AND CONTROL OF QUALITY, 5e, 2002 South-Western/Thomson LearningTM

Reliability Prediction

Generally defined as the ability of a product to perform as expected over time Formally defined as the probability that a product, piece of equipment, or system performs its intended function for a stated period of time under specified operating conditions
Rev. 02/25/06 SJSU Bus. 142 - David Bentley 17

THE MANAGEMENT AND CONTROL OF QUALITY, 5e, 2002 South-Western/Thomson LearningTM (Mod 11/11/02 DAB)

Types of Failures

Functional failure

Failure that occurs at the start of product life due to manufacturing or material detects

DOA or infant mortality

Reliability failure

Failure after some period of use

11/11/02

SJSU Bus. 142 - David Bentley

18

THE MANAGEMENT AND CONTROL OF QUALITY, 5e, 2002 South-Western/Thomson LearningTM (Mod 11/11/02 DAB)

Types of Reliability

Inherent reliability predicted by product design (robust design) Achieved reliability observed during use

11/11/02

SJSU Bus. 142 - David Bentley

19

THE MANAGEMENT AND CONTROL OF QUALITY, 5e, 2002 South-Western/Thomson LearningTM

Reliability Measurement

Failure rate (l) number of failures per unit time Alternative measures

Mean time to failure Mean time between failures

11/11/02

SJSU Bus. 142 - David Bentley

20

THE MANAGEMENT AND CONTROL OF QUALITY, 5e, 2002 South-Western/Thomson LearningTM

Cumulative Failure Rate Curve

11/11/02

SJSU Bus. 142 - David Bentley

21

THE MANAGEMENT AND CONTROL OF QUALITY, 5e, 2002 South-Western/Thomson LearningTM

Failure Rate Curve


Infant mortality period

11/11/02

SJSU Bus. 142 - David Bentley

22

THE MANAGEMENT AND CONTROL OF QUALITY, 5e, 2002 South-Western/Thomson LearningTM

Average Failure Rate

11/11/02

SJSU Bus. 142 - David Bentley

23

THE MANAGEMENT AND CONTROL OF QUALITY, 5e, 2002 South-Western/Thomson LearningTM

Reliability Function

Probability density function of failures f(t) = le-lt for t > 0 Probability of failure from (0, T) F(t) = 1 e-lT Reliability function R(T) = 1 F(T) = e-lT
11/11/02 SJSU Bus. 142 - David Bentley 24

THE MANAGEMENT AND CONTROL OF QUALITY, 5e, 2002 South-Western/Thomson LearningTM

Series Systems
1 2 n

RS = R1 R2 ... Rn

11/11/02

SJSU Bus. 142 - David Bentley

25

THE MANAGEMENT AND CONTROL OF QUALITY, 5e, 2002 South-Western/Thomson LearningTM

Parallel Systems
1
2

RS = 1 - (1 - R1) (1 - R2)... (1 - Rn)


11/11/02 SJSU Bus. 142 - David Bentley 26

THE MANAGEMENT AND CONTROL OF QUALITY, 5e, 2002 South-Western/Thomson LearningTM

Series-Parallel Systems
C
RA A RB B C RC

RC

RD D

Convert to equivalent series system


RA
A

RB
B C

RD
D
27

11/11/02

RC = 1 (1-RC)(1-RC)
SJSU Bus. 142 - David Bentley

Design optimization

Minimize variation in processes Seek robust design (Taguchi)

Insensitive to process variations or the use environment Taguchi loss function Optimizing reliability
02/25/06 SJSU Bus. 142 - David Bentley 28

Tools

THE MANAGEMENT AND CONTROL OF QUALITY, 5e, 2002 South-Western/Thomson LearningTM

Loss Functions
Traditional View loss no loss nominal tolerance loss

Taguchis View

loss

loss

11/21/02

SJSU Bus. 142 - David Bentley

29

THE MANAGEMENT AND CONTROL OF QUALITY, 5e, 2002 South-Western/Thomson LearningTM

Taguchi Loss Function Calculations


L(x) = k(x - T)2 Example: Specification = .500 .020 Failure outside of the tolerance range costs $50 to repair. Thus, 50 = k(.020)2. Solving for k yields k = 125,000. The loss function is: L(x) = 125,000(x - .500)2 Expected loss = k(2 + D2) where D is the deviation from the target.
11/21/02 SJSU Bus. 142 - David Bentley 30

THE MANAGEMENT AND CONTROL OF QUALITY, 5e, 2002 South-Western/Thomson LearningTM (Mod 11/11/02 DAB)

Optimizing Reliability

Standardization Redundancy Physics of failure

Rev. 02/26/06

SJSU Bus. 142 - David Bentley

31

Design Verification

Ensure that process capability meets the appropriate sigma level Meet specifications (AND customer requirements) Tools

Reliability testing Measurement systems evaluation Process capability determination


02/25/06 SJSU Bus. 142 - David Bentley 32

THE MANAGEMENT AND CONTROL OF QUALITY, 5e, 2002 South-Western/Thomson LearningTM (Mod 11/11/02 DAB)

Reliability Testing

Life testing Accelerated life testing Environmental testing Vibration and shock testing Burn-in

Rev. 02/26/06

SJSU Bus. 142 - David Bentley

33

Measurement System Evaluation

Variation can be due to:


Process variation Measurement system error


Random Systematic (bias)

A combination of the two

02/26/06

SJSU Bus. 142 - David Bentley

34

Metrology - 1

Definition: The Science of Measurement Accuracy

How close an observation is to a standard How close random individual measurements are to each other

Precision

02/26/06

SJSU Bus. 142 - David Bentley

35

Metrology - 2

Repeatability

Instrument variation Variation in measurements using same instrument and same individual Operator variation Variation in measurements using same instrument and different individual
02/26/06 SJSU Bus. 142 - David Bentley 36

Reproducibility

R&R Studies

Select m operators and n parts Calibrate the measuring instrument Randomly measure each part by each operator for r trials Compute key statistics to quantify repeatability and reproducibility

02/26/06

SJSU Bus. 142 - David Bentley

37

R&R Spreadsheet Template

02/25/06

SJSU Bus. 142 - David Bentley

38

R&R Evaluation

Acceptable: < 10% Unacceptable: > 30% Questionable: 10-30%

02/26/06

SJSU Bus. 142 - David Bentley

39

Calibration

Compare 2 instruments or systems

1 with known relationship to national standards 1 with unknown relationship to national standards

02/26/06

SJSU Bus. 142 - David Bentley

40

Process Capability

The range over which the natural variation of a process occurs as determined by the system of common causes Measured by the proportion of output that can be produced within design specifications

41

Types of Capability Studies


Peak performance study
conditions

How a process performs under ideal

Process characterization study How a process performs under actual Component variability study Relative contribution of different sources of
variation (e.g., process factors, measurement system) SJSU Bus. 142 - David Bentley
42

operating conditions

Process Capability Study



Choose a representative machine or process Define the process conditions Select a representative operator Provide the right materials Specify the gauging or measurement method Record the measurements Construct a histogram and compute descriptive statistics: mean and standard deviation Compare results with specified tolerances
SJSU Bus. 142 - David Bentley 43

Process Capability
(a) specification natural variation (c) specification natural variation (b) specification natural variation (d) specification natural variation

44

Process Capability Index


Cp = UTL - LTL 6 Cpu = UTL - m 3 Cpl = m - LTL 3 Cpk = min{ Cpl, Cpu }

45

Process Capability Ratios


Non-centered process (general case): choose cpk = the lower of: Upper spec process mean cpu = ---------------------------------3
Process mean lower spec cpl = ---------------------------------3
SJSU Bus. 142 David A. Bentley 09/30/02

or

46

Process Capability Ratios


Centered process (special case): specification width cp = ---------------------------process width Upper spec lower spec = -----------------------------6
SJSU Bus. 142 David A. Bentley 09/16/02

47

Process Capability Requirements


Process must be normally distributed Process must be in control Process capability result:

> < = >

1.34 = capable 1.33 = not capable 1.33 = barely capable 5 or 10 is overkill, excessive resource use
48

SJSU Bus. 142 David A. Bentley 0/24/06

Process Capability Spreadsheet Template

Rev. 02/26/06

SJSU Bus. 142 - David Bentley

49

Vous aimerez peut-être aussi