Reliability engineering  theroy and practice

Considering that performance, dependability, cost, and time to market are key factors for today's products and services, but also that failure of complex systems can have major safety co

Alessandro Birolini

Seventh Edition

Reliability

Engineering

Theory and Practice

Tai ngay!!! Ban co the xoa dong chu nay!!!

Reliability Engineering Theory and Practice

Seventh Edition

With 190 Figures, 60 Tables, 140 Examples, and 70 Problems for Homework

123

Ingénieur et penseur, Ph.D., Professor Emeritus of Reliability Eng.

at the Swiss Federal Institute of Technology (ETH), Zurich

ISBN 978-3-642-39534-5 ISBN 978-3-642-39535-2 (eBook)

DOI 10.1007/978-3-642-39535-2

Springer Heidelberg New York Dordrecht London

Library of Congress Control Number: 2013945800

Ó Springer-Verlag Berlin Heidelberg 1994, 1997, 1999, 2004, 2007, 2010, 2014

This work is subject to copyright All rights are reserved by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed Exempted from this legal reservation are brief excerpts in connection with reviews or scholarly analysis or material supplied specifically for the purpose of being entered and executed on a computer system, for exclusive use by the purchaser of the work Duplication of this publication or parts thereof is permitted only under the provisions of the Copyright Law of the Publisher’s location, in its current version, and permission for use must always be obtained from Springer Permissions for use may be obtained through RightsLink at the Copyright Clearance Center Violations are liable to prosecution under the respective Copyright Law The use of general descriptive names, registered names, trademarks, service marks, etc in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use.

While the advice and information in this book are believed to be true and accurate at the date of publication, neither the authors nor the editors nor the publisher can accept any legal responsibility for any errors or omissions that may be made The publisher makes no warranty, express or implied, with respect to the material contained herein.

Printed on acid-free paper

Springer is part of Springer Science+Business Media (www.springer.com)

Louis Pasteur

"Quand on aperçoit combien la somme de nos

" ignorances dépasse celle de nos connaissances,

" on se sent peu porté à conclure trop vite." 2)

Louis De Broglie

"One has to learn to consider causes rather than

" symptoms of undesirable events and avoid

hypo-" critical attitudes."

Alessandro Birolini

1) "Opportunity comes to the intellect which is ready to receive it."

2) "When one recognizes how much the sum of our ignorance

The large interest granted to the 6th edition (over 2000 on-line requests per year) incited me for a 7th and last edition of this book (11 editions with the 4 German editions 1985 - 97).

The book shows how to build in, evaluate, and demonstrate reliability, maintainability, and availability of components, equipment, and systems It presents the state-of-the-art of reliability engineering, both in theory and practice, and is based on the author's more than 30 years experience

in this field, half in industry (part of which in setting up the Swiss Test Lab for VLSI, 1979 - 83 in Neuchâtel) and half as Professor of Reliability Engineering at the Swiss Federal Institute of Technology (ETH), Zurich Considering that performance, dependability, cost, and time to market are key factors for today's products and services, but also that failure of complex systems can have major safety consequences, reliability engineering becomes a necessary support in developing and producing complex equipment and systems.

The structure of the book has been conserved through all editions, with main Chapters 1 to 8 and Appendices A1 to A11 (A10 & A11 since the 5 th Edition 2007) Chapters 2, 4, and 6 deal carefully with analytical investigations, Chapter 5 with design guidelines, Chapters 3 and 7 with tests, and Chapter 8 with activities during production Appendix A1 defines and comment on the terms commonly used in reliability engineering Appendices A2 - A5 have been added to support managers in

answering the question of how to specify and achieve high reliability (RAMS) targets for complex

equipment and systems Appendices A6 - A8 are a compendium of probability theory, stochastic processes, and mathematical statistics, as necessary for Chapters 2, 4, 6, and 7, consistent from a mathematical point of view but still with reliability engineering applications in mind (demonstration of established theorems is referred, and for all other propositions or equations, sufficient details for complete demonstration are given) Appendix A9 includes statistical tables, Laplace transforms, and probability charts Appendix A10 resumes basic technological component's properties, and Appendix A11 gives a set of 70 problems for homework.

This structure makes the book self contained as a text book for postgraduate students or courses in

industry (Fig 1.9 on p 24), allows a rapid access to practical results (as a desktop reference), and offers

to theoretically oriented readers all mathematical tools to continue research in this field.

The book covers many aspects of reliability engineering using a common language, and has been improved step by step Methods & tools are given in a way that they can be tailored to cover different reliability requirement levels, and be used for safety analysis too A large number of tables (60), figures (190), and examples (210 of which 70 as problems for homework), as well as comprehensive reference list and index, amply support the text This last edition reviews, refines, and extends all previous editions New in particular includes:

• A strategy to mitigate incomplete coverage(p 255), yielding new models (Table 6.12 c & d, p 256).

• A comprehensive introduction to human reliability with a set of design guidelines to avoid human

errors (pp 158-159) and new models combining human errors probability and time to accomplish a task, based on semi-Markov processes (pp 294-298).

• An improvement of the design guidelines for maintainability (pp 154-158).

• An improvement of reliability allocation using Lagrange multiplier to consider cost aspects(p 67).

• A comparison of four repair strategies(Table 4.4, p 141).

• A comparison of basic models for imperfect switching(Table 6.11, p 248).

• A refinement of approximate expressions, of concepts related to regenerative processes, and of the

use and limitations of stochastic processes in modeling reliability problems(e.g Table 6.1, p 171).

• New is also that relevant statements and rules have been written cursive and centered on the text Furthermore,

• Particular importance has been given to the selection of design guidelines and rules, the opment of approximate expressions for large series-parallel systems, the careful simplification of

devel-exact results to allow in-depth trade off studies, and the investigation of systems with complex structure (preventive maintenance, imperfect switching, incomplete coverage, elements with more

than one failure mode, fault tolerant reconfigurable systems, common cause failures).

VII

• The central role of software quality assurance for complex equipment and systems is highlighted.

• The use of interarrival times starting by x= 0 at each occurrence of the event considered, instead of the variable t,giving a sense to MTBF and allowing the introduction of a failure rate λ ( )x and a

mean time to failure MTTF also for repairable systems, is carefully discussed (pp 5-6, 41, 175,

316, 341, 378, 380) and consequently applied Similar is for the basic difference between failure

rate, (probability) density, and renewal density or intensity of a point process (pp 7, 378, 426, 466,

524) In this context, the assumption as-good-as-new after repair is critically discussed wherever

necessary, and the historical distinction between nonrepairable and repairable items is scaled down

(removed for reliability function, failure rate, MTTF, and MTBF); national and international standards should better consider this fact and avoid definitions intrinsically valid only for constant (time independent) failure rates.

• Also valid is the introduction since the 1 st edition of indicesS i for reliability figures at system level

(e g.MTTF S i ) where S stands for system and , iis the state entered att= 0(system referring to the highest integration level of the item considered, and t= 0 being the beginning of observations, x= 0 for interarrival times) This is mandatory for judicious investigations at the system level.

• In agreement with the practical applications, MTBF is reserved for MTBF= 1 / λ

• Important prerequisites for accelerated tests are carefully discussed (pp 329-334), in particular to

transfer an acceleration factor A from the MTTF (MTTF A.MTTF)

1= 2 to the (random) free time τ ( τ1=A.τ2)

failure-• Asymptotic & steady-state is used for stationary, by assuming irreducible embedded chains; repair for restoration, by neglecting administrative, logistical, technical delays; mean for expected value.

For reliability applications, pairwise independence assures, in general, totally (mutually,

statisti-cally, stochastically) independence, independent is thus used for totally independent.

The book has growth from about 400 to 600 pages, with main improvements in the 4 th to 7 th Editions.

• 4th Edition: Complete review and general refinements.

• 5th Edition: Introduction to phased-mission systems, common cause failures, Petri nets, dynamic

FTA, nonhomogeneous Poisson processes, and trend tests; problems for homework.

• 6th Edition: Proof of Eqs (6.88) & (6.94), introduction to network reliability, event trees & binary

decision diagrams, extensions of maintenance strategies and incomplete coverage, refinements for large complex systems and approximate expressions.

The launching of the 6th Edition of this book coincided with my 70 th anniversary, this was celebrated with a special Session at the 12t h Int Conf on Quality and Dependability CCF2010 held in Sinaia (RO), 22-24 September 2010 My response to the last question at the interview [1.0] given to Prof Dr Ioan C Bacivarov, Chairman of the International Scientific Committee of CCF2010, can help

to explain the acceptance of this book:

" Besides more than 15 years experience in the industry, and a predisposition to be a self-taught man, my attitude to life was surely an important key for the success of my book This is best expressed in the three sentences given on the first page of this book These sentences, insisting

on generosity, modesty and responsibility apply quite general to a wide class of situations and people, from engineers to politicians, and it is to hope that the third sentence, in particular, will

be considered by a growing number of humans, now, in front of the ecological problems we are faced and in front of the necessity to create a federal world wide confederation of democratic states in which freedom is primarily respect for the other "

The comments of many friends and the agreeable cooperation with Springer-Verlag are gratefully acknowledged Looking back to all editions (1st German 1985), thanks are due, in particular, to K.P LaSala for reviewing the 4 th & 6 th Editions [1.17], I.C Bacivarov for reviewing the 6 th Edition [1.0], book reviewers of the German editions, P Franken and I Kovalenko for commenting Appendices A6 - A8, A Bobbio F Bonzanigo, M Held for supporting numerical evaluations, J Thalhammer for supporting the edition of all figures, and L Lambert for reading final manuscripts.

Zurich and Florence, September 13, 2013 Alessandro Birolini

1 Basic Concepts, Quality & Reliability (RAMS) Assurance of Complex Equip & Systems 1

1.1 Introduction 1

1.2 Basic Concepts 2

1.2.1 Reliability 2

1.2.2 Failure 3

1.2.3 Failure Rate, MTTF, MTBF 4

1.2.4 Maintenance, Maintainability 8

1.2.5 Logistic Support 8

1.2.6 Availability 9

1.2.7 Safety, Risk, and Risk Acceptance 9

1.2.8 Quality 11

1.2.9 Cost and System Effectiveness 11

1.2.10 Product Liability 15

1.2.11 Historical Development 16

1.3 Basic Tasks & Rules for Quality & Rel (RAMS) Assurance of Complex Eq & Systems 17 1.3.1 Quality and Reliability (RAMS) Assurance Tasks 17

1.3.2 Basic Quality and Reliability (RAMS) Assurance Rules 19

1.3.3 Elements of a Quality Assurance System 21

1.3.4 Motivation and Training 24

2 Reliability Analysis During the Design Phase (Nonrepairable Elements up to System Failure) 25

2.1 Introduction 25

2.2 Predicted Reliability of Equipment and Systems with Simple Structure 28

2.2.1 Required Function 28

2.2.2 Reliability Block Diagram 28

2.2.3 Operating Conditions at Component Level, Stress Factors 33

2.2.4 Failure Rate of Electronic Components 35

2.2.5 Reliability of One-Item Structures 39

2.2.6 Reliability of Series-Parallel Structures 41

2.2.6.1 Systems without Redundancy 41

2.2.6.2 Concept of Redundancy 42

2.2.6.3 Parallel Models 43

2.2.6.4 Series-Parallel Structures 45

2.2.6.5 Majority Redundancy 49

2.2.7 Part Count Method 51

2.3 Reliability of Systems with Complex Structure 52

2.3.1 Key Item Method 52

2.3.1.1 Bridge Structure 53

2.3.1.2 Rel Block Diagram in which Elements Appear More than Once 54

2.3.2 Successful Path Method 55

2.3.3 State Space Method 56

2.3.4 Boolean Function Method 57

2.3.5 Parallel Models with Constant Failure Rates and Load Sharing 61

2.3.6 Elements with more than one Failure Mechanism or one Failure Mode 64

2.3.7 Basic Considerations on Fault Tolerant Structures 66

2.4 Reliability Allocation and Optimization 67

IX

2.5 Mechanical Reliability, Drift Failures 68

2.6 Failure Modes Analyses 72

2.7 Reliability Aspects in Design Reviews 77

3 Qualification Tests for Components and Assemblies 81

3.1 Basic Selection Criteria for Electronic Components 81

3.1.1 Environment 82

3.1.2 Performance Parameters 84

3.1.3 Technology 84

3.1.4 Manufacturing Quality 86

3.1.5 Long-Term Behavior of Performance Parameters 86

3.1.6 Reliability 86

3.2 Qualification Tests for Complex Electronic Components 87

3.2.1 Electrical Test of Complex ICs 88

3.2.2 Characterization of Complex ICs 90

3.2.3 Environmental and Special Tests of Complex ICs 92

3.2.4 Reliability Tests 101

3.3 Failure Modes, Mechanisms, and Analysis of Electronic Components 101

3.3.1 Failure Modes of Electronic Components 101

3.3.2 Failure Mechanisms of Electronic Components 102

3.3.3 Failure Analysis of Electronic Components 102

3.3.4 Present VLSI Production-Related Reliability Problems 106

3.4 Qualification Tests for Electronic Assemblies 107

4 Maintainability Analysis 112

4.1 Maintenance, Maintainability 112

4.2 Maintenance Concept 115

4.2.1 Fault Detection (Recognition) and Localization 116

4.2.2 Equipment and Systems Partitioning 118

4.2.3 User Documentation 118

4.2.4 Training of Operation and Maintenance Personnel 119

4.2.5 User Logistic Support 119

4.3 Maintainability Aspects in Design Reviews 121

4.4 Predicted Maintainability 121

4.4.1 Calculation of MTTR S 121

4.4.2 Calculation of MTTPM S 125

4.5 Basic Models for Spare Parts Provisioning 125

4.5.1 Centralized Logistic Support, Nonrepairable Spare Parts 125

4.5.2 Decentralized Logistic Support, Nonrepairable Spare Parts 129

4.5.3 Repairable Spare Parts 130

4.6 Maintenance Strategies 134

4.6.1 Complete renewal at each maintenance action 134

4.6.2 Block replacement with minimal repair at failure 138

4.6.3 Further considerations on maintenance strategies 139

4.7 Basic Cost Considerations 142

5 Design Guidelines for Reliability, Maintainability, and Software Quality 144

5.1 Design Guidelines for Reliability 144

5.1.1 Derating 144

5.1.2 Cooling 145

5.1.3 Moisture 147

5.1.4 Electromagnetic Compatibility, ESD Protection 148

5.1.5 Components and Assemblies 150

5.1.5.1 Component Selection 150

5.1.5.2 Component Use 150

5.1.5.3 PCB and Assembly Design 151

5.1.5.4 PCB and Assembly Manufacturing 152

5.1.5.5 Storage and Transportation 153

5.1.6 Particular Guidelines for IC Design and Manufacturing 153

5.2 Design Guidelines for Maintainability 154

5.2.1 General Guidelines 154

5.2.2 Testability 155

5.2.3 Connections, Accessibility, Exchangeability 157

5.2.4 Adjustment 158

5.2.5 Human, Ergonomic, and Safety Aspects 158

5.3 Design Guidelines for Software Quality 159

5.3.1 Guidelines for Software Defect Prevention 162

5.3.2 Configuration Management 165

5.3.3 Guidelines for Software Testing 166

5.3.4 Software Quality Growth Models 166

6 Reliability and Availability of Repairable Systems 169

6.1 Introduction, General Assumptions, Conclusions 169

6.2 One-Item Structure 175

6.2.1 One-Item Structure New at Time t= 0 176

6.2.1.1 Reliability Function 176

6.2.1.2 Point Availability 177

6.2.1.3 Average Availability 178

6.2.1.4 Interval Reliability 179

6.2.1.5 Special Kinds of Availability 180

6.2.2 One-Item Structure New at Time t= 0 and with Constant Failure Rate λ 183

6.2.3 One-Item Structure with Arbitrary Conditions at t= 0 184

6.2.4 Asymptotic Behavior 185

6.2.5 Steady-State Behavior 187

6.3 Systems without Redundancy 189

6.3.1 Series Structure with Constant Failure and Repair Rates 189

6.3.2 Series Structure with Constant Failure and Arbitrary Repair Rates 192

6.3.3 Series Structure with Arbitrary Failure and Repair Rates 193

6.4 1-out-of-2 Redundancy (Warm, one Repair Crew) 196

6.4.1 1-out-of-2 Redundancy with Constant Failure and Repair Rates 196

6.4.2 1-out-of-2 Redundancy with Constant Failure and Arbitrary Rep Rates 204

6.4.3 1-out-of-2 Red with Const Failure Rate in Reserve State & Arbitr Rep Rates 207 6.5 k-out-of-n Redundancy (Warm, Identical Elements, one Repair Crew) 213

6.5.1 k-out-of-n Redundancy with Constant Failure and Repair Rates 214

6.5.2 k-out-of-n Redundancy with Constant Failure and Arbitrary Repair Rates 218

6.6 Simple Series - Parallel Structures (one Repair Crew) 220

6.7 Approximate Expressions for Large Series - Parallel Structures 226

6.7.1 Introduction 226

6.7.2 Application to a Practical Example 230

6.8 Systems with Complex Structure (one Repair Crew) 238

6.8.1 General Considerations 238

6.8.2 Preventive Maintenance 240

6.8.3 Imperfect Switching 243

6.8.4 Incomplete Coverage 249

6.8.5 Elements with more than two States or one Failure Mode 257

6.8.6 Fault Tolerant Reconfigurable Systems 259

6.8.6.1 Ideal Case 259

6.8.6.2 Time Censored Reconfiguration (Phased-Mission Systems) 259

6.8.6.3 Failure Censored Reconfiguration 266

6.8.6.4 Reward and Frequency / Duration Aspects 270

6.8.7 Systems with Common Cause Failures 271

6.8.8 Basic Considerations on Network-Reliability 275

6.8.9 General Procedure for Modeling Complex Systems 277

6.9 Alternative Investigation Methods 280

6.9.1 Systems with Totally Independent Elements 280

6.9.2 Static and Dynamic Fault Trees 280

6.9.3 Binary Decision Diagrams 283

6.9.4 Event Trees 286

6.9.5 Petri Nets 287

6.9.6 Numerical Reliability and Availability Computation 289

6.9.6.1 Numerical Computation of System's Reliability and Availability 289

6.9.6.2 Monte Carlo Simulations 290

6.9.7 Approximate expressions for Large, Complex Systems: Basic Considerations 293 6.10 Human Reliability 294

7 Statistical Quality Control and Reliability Tests 299

7.1 Statistical Quality Control 299

7.1.1 Estimation of a Defective Probability p 300

7.1.2 Simple Two-sided Sampling Plans for Demonstration of a Def Probability p 302 7.1.2.1 Simple Two-sided Sampling Plan 303

7.1.2.2 Sequential Test 305

7.1.3 One-sided Sampling Plans for the Demonstration of a Def Probability p 306

7.2 Statistical Reliability Tests 309

7.2.1 Reliability and Availability Estimation & Demon for a given fixed Mission 309 7.2.2 Availability Estimation & Demonstration for Continuous Operation (steady-state) 311 7.2.2.1 Availability Estimation (Erlangian Failure-Free and/or Repair Times) 311

7.2.2.2 Availability Demonstration (Erlangian Failure-Free and/or Repair Times) 313 7.2.2.3 Further Availability Evaluation Methods for Continuous Operation 314 7.2.3 Estimation and Demonstration of a Const Failure Rate λ ( or of MTBF= 1 / ) λ 316 7.2.3.1 Estimation of a Constant Failure Rate λ 318

7.2.3.2 Simple Two-sided Test for the Demonstration of λ 320

7.2.3.3 Simple One-sided Test for the Demonstration of λ 324

7.3 Statistical Maintainability Tests 325

7.3.1 Estimation of an MTTR 325

7.3.2 Demonstration of an MTTR 327

7.4 Accelerated Testing 329

7.5 Goodness-of-fit Tests 334

7.5.1 Kolmogorov-Smirnov Test 334

7.5.2 Chi-square Test 338

7.6 Statistical Analysis of General Reliability Data 341

7.6.1 General considerations 341

7.6.2 Tests for Nonhomogeneous Poisson Processes 343

7.6.3 Trend Tests 345

7.6.3.1 Tests of a HPP versus a NHPP with increasing intensity 345

7.6.3.2 Tests of a HPP versus a NHPP with decreasing intensity 348

7.6.3.3 Heuristic Tests to distinguish between HPP and Monotonic Trend 349

7.7 Reliability Growth 351

8 Quality & Reliability (RAMS) Assurance During Production Phase (Basic Considerations) 357 8.1 Basic Activities 357

8.2 Testing and Screening of Electronic Components 358

8.2.1 Testing of Electronic Components 358

8.2.2 Screening of Electronic Components 359

8.3 Testing and Screening of Electronic Assemblies 362

8.4 Test and Screening Strategies, Economic Aspects 364

8.4.1 Basic Considerations 364

8.4.2 Quality Cost Optimization at Incoming Inspection Level 367

8.4.3 Procedure to handle first deliveries 372

Appendices (A1-A11) A1 Terms and Definitions 373

A2 Quality and Reliability (RAMS) Standards 387

A2.1 Introduction 387

A2.2 General Requirements in the Industrial Field 388

A2.3 Requirements in the Aerospace, Railway, Defense, and Nuclear Fields 390

A3 Definition and Realization of Quality and Reliability (RAMS) Requirements 391

A3.1 Definition of Quality and Reliability (RAMS) Requirements 391

A3.2 Realization of Quality & Reliability (RAMS) Requirements for Complex Eq & Syst 393 A3.3 Elements of a Quality and Reliability (RAMS) Assurance Program 398

A3.3.1 Project Organization, Planning, and Scheduling 398

A3.3.2 Quality and Reliability (RAMS) Requirements 399

A3.3.3 Reliability, Maintainability, and Safety Analysis 399

A3.3.4 Selection and Qualification of Components, Materials, Manuf Processes 400 A3.3.5 Softwaer Quality Assurance 400

A3.3.6 Configuration Management 401

A3.3.7 Quality Tests 402

A3.3.8 Quality Data Reporting System 404

A4 Checklists for Design Reviews 405

A4.1 System Design Review 405

A4.2 Preliminary Design Reviews 406

A4.3 Critical Design Review (System Level) 409

A5 Requirements for Quality Data Reporting Systems 410

A6 Basic Probability Theory 413

A6.1 Field of Events 413

A6.2 Concept of Probability 415

A6.3 Conditional Probability, Independence 418

A6.4 Fundamental Rules of Probability Theory 419

A6.4.1 Addition Theorem for Mutually Exclusive Events 419

A6.4.2 Multiplication Theorem for Two Independent Events 420

A6.4.3 Multiplication Theorem for Arbitrary Events 421

A6.4.4 Addition Theorem for Arbitrary Events 421

A6.4.5 Theorem of Total Probability 422

A6.5 Random Variables, Distribution Functions 423

A6.6 Numerical Parameters of Random Variables 429

A6.6.1 Expected Value (Mean) 429

A6.6.2 Variance 432

A6.6.3 Modal Value, Quantile, Median 434

A6.7 Multidimensional Random Variables, Conditional Distributions 434

A6.8 Numerical Parameters of Random Vectors 436

A6.8.1 Covariance Matrix, Correlation Coefficient 437

A6.8.2 Further Properties of Expected Value and Variance 438

A6.9 Distribution of the Sum of Indep Positive Random Variables and of τmin ,τmax 438 A6.10 Distribution Functions used in Reliability Analysis 441

A6.10.1 Exponential Distribution 441

A6.10.2 Weibull Distribution 442

A6.10.3 Gamma Distribution, Erlangian Distribution, and χ 2 -Distribution 444 A6.10.4 Normal Distribution 446

A6.10.5 Lognormal Distribution 447

A6.10.6 Uniform Distribution 449

A6.10.7 Binomial Distribution 449

A6.10.8 Poisson Distribution 451

A6.10.9 Geometric Distribution 453

A6.10.10 Hypergeometric Distribution 454

A6.11 Limit Theorems 454

A6.11.1 Laws of Large Numbers 455

A6.11.2 Central Limit Theorem 456

A7 Basic Stochastic-Processes Theory 460

A7.1 Introduction 460

A7.2 Renewal Processes 463

A7.2.1 Renewal Function, Renewal Density 465

A7.2.2 Recurrence Times 468

A7.2.3 Asymptotic Behavior 469

A7.2.4 Stationary Renewal Processes 471

A7.2.5 Homogeneous Poisson Processes (HPP) 472

A7.3 Alternating Renewal Processes 474

A7.4 Regenerative Processes with a Finite Number of States 478

A7.5 Markov Processes with a Finite Number of States 480

A7.5.1 Markov Chains with a Finite Number of States 480

A7.5.2 Markov Processes with a Finite Number of States 482

A7.5.3 State Probabilities and Stay Times in a Given Class of States 491

A7.5.3.1 Method of Differential Equations 491

A7.5.3.2 Method of Integral Equations 495

A7.5.3.3 Stationary State and Asymptotic Behavior 496

A7.5.4 Frequency / Duration and Reward Aspects 498

A7.5.4.1 Frequency / Duration 498

A7.5.4.2 Reward 500

A7.5.5 Birth and Death Process 501

A7.6 Semi-Markov Processes with a Finite Number of States 505

A7.7 Semi-regenerative Processes with a Finite Number of States 510

A7.8 Nonregenerative Stochastic Processes with a Countable Number of States 515

A7.8.1 General Considerations 515

A7.8.2 Nonhomogeneous Poisson Processes (NHPP) 516

A7.8.3 Superimposed Renewal Processes 520

A7.8.4 Cumulative Processes 521

A7.8.5 General Point Processes 523

A8 Basic Mathematical Statistics 525

A8.1 Empirical Methods 525

A8.1.1 Empirical Distribution Function 526

A8.1.2 Empirical Moments and Quantiles 528

A8.1.3 Further Applications of the Empirical Distribution Function 529

A8.2 Parameter Estimation 533

A8.2.1 Point Estimation 533

A8.2.2 Interval Estimation 538

A8.2.2.1 Estimation of an Unknown Probability p 538

A8.2.2.2 Estimation of Param λ for Exp Distrib.:Fixed T, instant repl 542 A8.2.2.3 Estimation of Param λ for Exp Distrib.:Fixed n, no repl 543

A8.2.2.4 Availability Estimation (Erlangian Failure-Free and/or Repair Times) 545 A8.3 Testing Statistical Hypotheses 547

A8.3.1 Testing an Unknown Probability p 548

A8.3.1.1 Simple Two-sided Sampling Plan 549

A8.3.1.2 Sequential Test 550

A8.3.1.3 Simple One-sided Sampling Plan 551

A8.3.1.4 Availability Demonstr (Erlangian Failure-Free and/or Rep Times) 553

A8.3.2 Goodness-of-fit Tests for Completely Specified F ( )0t 555

A8.3.3 Goodness-of-fit Tests for F ( )0t with Unknown Parameters 558

A9 Tables and Charts 561

A9.1 Standard Normal Distribution 561

A9.2 χ2 -Distribution (Chi-Square Distribution) 562

A9.3 t-Distribution (Student distribution) 563

A9.4 F-Distribution (Fisher distribution) 564

A9.5 Table for the Kolmogorov-Smirnov Test 565

A9.6 Gamma Function 566

A9.7 Laplace Transform 567

A9.8 Probability Charts (Probability Plot Papers) 569

A9.8.1 Lognormal Probability Chart 569

A9.8.2 Weibull Probability Chart 570

A9.8.3 Normal Probability Chart 571

A10 Basic Technological Component's Properties 572

A11 Problems for Homework 576

Acronyms 582

References 583

Index 605

1 Basic Concepts, Quality and Reliability

and Systems

Considering that complex equipment and systems are generally repairable, contain

redundancy and must be safe, the term reliability appears often for reliability, maintainability, availability & safety RAMS (in brackets) is used to point out this wherever necessary in the text The purpose of reliability (RAMS) engineering is to develop methods and tools to evaluate and demonstrate reliability, maintainability,

availability, and safety of components, equipment & systems,as well as to support development and production engineers in building in these characteristics In order

to be cost and time effective, reliability (RAMS) engineering must be integrated inthe project activities, support quality assurance and concurrent engineering efforts,and be performed without bureaucracy This chapter introduces basic concepts,shows their relationships, and discusses the tasks necessary to assure quality and re-liability (RAMS)of complex equipment & systems with high quality and reliability

(RAMS) requirements A comprehensive list of definitions is given in Appendix A1.Standardsforquality and reliability(RAMS) assurance are discussedinAppendixA2

Refinements of management aspects are given in Appendices A3 - A5

Until the nineteen-sixties, quality targets were deemed to have been reached when

the item considered was found to be free of defects or systematic failures at the time

it left the manufacturer The growing complexity of equipment and systems, as well

as the rapidly increasing cost incurred by loss of operation as a consequence of

failures, have brought to the forefront the aspects of reliability, maintainability, availability, and safety The expectation today is that complex equipment and systems are not only free from defects and systematic failures at time t=0

(when they are put into operation), but also perform the required function failure free for a stated time interval and have a fail-safe behavior in case of critical or catastrophic failures However, the question of whether a given item will operate without failures during a stated period of time cannot be simply answered by yes

or no, on the basis of a compliance test Experience shows that only a probability for this occurrence can be given This probability is a measure of the item’s

A Birolini, Reliability Engineering, DOI: 10.1007/978-3-642-39535-2_1,

Ó Springer-Verlag Berlin Heidelberg 2014

1

reliability and can be interpreted as follows:

If n statistically identical and independent items are put into operation at time t=0 to perform a given mission and ν ≤n of them accomplish it successfully, then the ratio ν/ n is a random variable which converges for increasing n to the true value of the reliability (Appendix A6.11).

Performanceparameters as well as reliability,maintainability, availability,andsafety have to be built in during design & development and retained during production and

operation of the item After the introduction of some important concepts in Section1.2, Section 1.3 gives basic tasks and rules for quality and reliability assurance of

complex equipment and systems with high quality and reliability requirements

(see Appendix A1 for a comprehensive list of definitions and Appendices A2 - A5for a refinement of management aspects)

fail and be repaired (without operational interruption at item (system) level) The

concept of reliability thus applies to nonrepairable as well as to repairable items

(Chapters 2 and 6, respectively) To make sense, a numerical statement of reliability(e.g R=0 9 ) must be accompanied by the definition of the required function, the operating conditions, and the mission duration In general, it is also important to

know whether or not the item can be considered new when the mission starts

An item is a functional or structural unit of arbitrary complexity (e.g component,

assembly, equipment, subsystem, system) that can be considered as an entity for

investigations.+) It may consist of hardware, software, or both and may also include

human resources Often, ideal human aspects and logistic support are assumed, even if (for simplicity) the term system is used instead of technical system.

+) System refers in this book, and often in practical applications, to the highest integration level of the

item considered.

The required function specifies the item's task For example, for given inputs,

the item outputs have to be constrained within specified tolerance bands mance parameters should always be given with tolerances) The definition of the re-

(perfor-quired function is the starting point for any reliability analysis, as it defines failures Operating conditions have an important influence on reliability, and must there-

fore be specified with care Experience shows for instance, that the failure rate ofsemiconductor devices will double for operating temperature increase of 10 to 20°C.The required function and/ or operating conditions can be time dependent.

In these cases, a mission profile has to be defined and all reliability figures will be

related to it A representative mission profile and the corresponding reliability

targets should be given in the item's specifications.

Often the mission duration is considered as a parameter t, the reliability function

is then defined by R( )t R( )t is the probability that no failure at item level will

occur in the interval ( , ]0 t The item's condition at t=0 (new or not) influences nal results To consider this, in this book reliabilityfigures at system level will haveindices S i (e.g.RS i( )t ), where Sstands for system andiis the state entered at t=0(Tab.6.2) State 0, with all elements new, is often assumed at t=0, yielding RS0( ).t

fi-A distinction between predicted and estimated or assessed reliability is

important The first one is calculated on the basis of the item’s reliability structureand the failure rate of its components (Sections 2.2 & 2.3), the second is obtainedfrom a statistical evaluation of reliability tests or from field data by knownenvironmental and operating conditions (Section 7.2)

The concept of reliability can be extended to processes and services as well,

although human aspects can lead to modeling difficulties(Sections1.2.7,5.2.5,6.10)

1.2.2 Failure

A failure occurs when the item stops performing its required function As simple as this definition is, it can become difficult to apply it to complex items The failure- free time (hereafter used as a synonym for failure-free operating time) is generally a random variable It is often reasonably long; but it can be very short, for instance

because of a failure caused by a transient event at turn-on A general assumption in

investigating failure-free times is that at t=0 the item is free of defects and

systematic failures Besides their frequency, failures should be classified (as far as

possible) according to the mode, cause, effect, and mechanism:

1 Mode: The mode of a failure is the symptom (local effect) by which a failure

is observed; e.g., opens, shorts, or drift for electronic components (Table 3.4);brittle rupture, creep, cracking, seizure, fatigue for mechanical components

2 Cause: The cause of a failure can be intrinsic, due to weaknesses in the item

and/or wear out, or extrinsic, due to errors, misuse or mishandling during the design, production, or use Extrinsic causes often lead to systematic failures, which are deterministic and should be considered like defects (dynamic

defects in software quality) Defects are present at t=0, even if often they

can not be discovered at t=0 Failures appear always in time, even if thetime to failure is short as it can be with systematic or early failures

3 Effect: The effect (consequence) of a failure can be different if considered on

the item itself or at higher level. A usual classification is: non relevant, partial, complete, and critical failure Since a failure can also cause further failures, distinction between primary and secondary failure is important.

4 Mechanism: Failure mechanism is the physical, chemical, or other process

resulting in a failure (see Table 3.5 (p 103) for some examples)

Failures can also be classified as sudden and gradual In this case, sudden and complete failures are termed cataleptic failures, gradual and partial failures are termed degradation failures As failure is not the only cause for the item being

down, the general term used to define the down state of an item (not caused by apreventive maintenance, other planned actions, or lack of external resources) is

The failure rate plays an important role in reliability analysis This Section

intro-duces it heuristically, see Appendix A6.5 for an analytical derivation

Let us assume that n statistically identical, new, and independent items are put into operation at time t=0, under the same conditions, and at the time t a subset

ν( )t of these items have not yet failed. ν( )t is a right continuous decreasing step function (Fig 1.1) t1, ,t n,measured from t=0,are the observed failure-free

times (operating times to failure) of the n items considered They are independent

realizations of a random variable τ (hereafter identified as failure-free time) and

must not be confused with arbitrary points on the time axis ( t1*,t2*, ) The quantity

n – 1

t t

Figure 1.1 Number ν ( )t of (nonrepairable) items still operating at time t

still operating(or surviving) at timet Applying Eq (1.2) to Eq (1.3) yields

ˆ( ) R( )ˆ R(ˆ )

ˆR( )

The failure rate λ( )t given by Eqs (1.3) -(1.5) applies in particular to

nonrepairable items (Figs 1.1 & 1.2) However,

considering Eq (A6.25) λ( )t can also be defined for repairable items which are as-good-as-new after repair (renewal), taking instead of t the variable x starting by x=0 at each renewal (as for interarrival times);

this is important when investigating repairable systems, and holds in particular for λ( )x =λ (see remarks on pp 6, 40 - 41, 378, 380).

If a repairable system cannot be restored to be as-good-as-new after repair (with spect to the state considered), i.e., if at least one element with time dependent failure

re-rate has not been renewed at every repair, failure intensityz( )t has to be used(see

pp.378,426,524 for comments) The use of hazard rate for λ( )t should be avoided.

In many practical applications, λ( )t =λ can be assumed Eq (1.6) then yieldsR( )t =e− λt, (for λ ( )t = λ ), (1.7)and the failure-free time τ>0 is exponentially distributed (F( ) Pr{t = τ≤ =t} 1−e−λt);

for this, and only in this case, the failure rate λ can be estimated by

λˆ =k T/ , where T is a given (fixed) cumulative operating time and k the total number of failures during T (Eqs (7.28) and (A8.46)).

The mean (expected value) of the failure-free time τ>0 is given by (Eq.(A6.38))

MTTF =E[ ]τ =∞∫ R( )t d t

where MTTF stands for mean time to failure For λ( )t =λ it follows E[ ]τ =1/λ

A constant (time independent) failure rate λ is often considered also for

repairable items Assuming that the item is as-good-as-new after each repair, successive failure-free times are then independent random variables, exponentially distributed with the same parameter λ, and with mean

MTBF=1 /λ, (for λ ( )x = λ, x starting at 0 after each repair). (1.9)

MTBF stands for mean operating time between failures Also because of the cal estimate MTBF T kˆ = / used in practical applications (p.318), MTBF should be con-

statisti-fined to the case of repairable items with constant failure rate However, at

compo-nent level MTBF=108h for =λ 10− 8h− 1has no practical significance. For systemswith>2 states, MUT S (system mean up time) is used (p 278, Table 6.2). Finally,

it must be pointed out that for a repairable item, the only possibility to have successive statistically identical and independent operating times after each repair (interarrival times), giving a sense to a mean operating time between failures ( MTBF ), is to re-establish at each repair an as- good-as-new situation, replacing all parts with non constant failure rates The failure rate of a large population of statistically identical and independent items exhibits often a typical bathtub curve (Fig 1.2) with the following 3 phases:

1 Early failures: λ( )t decreases (in general) rapidly with time; failures in this phase are attributable to randomly distributed weaknesses in materials,

components, or production processes

2 Failures with constant (or nearly so) failure rate: λ( )t is approximately

constant; failures in this period are Poisson distributed and often cataleptic

3 Wear out failures: λ( )t increaseswithtime; failuresinthisperiodareable to aging, wear out, fatigue, etc (e.g corrosion, electromigration)

attribut-Early failures are not deterministic and appear in general randomly distributed in

time and over the items During the early failure period, λ( )t must not necessarily

decrease as in Fig 1.2,in some cases it can oscillate To eliminate early failures,

Figure 1.2 Typical shape for the failure rate of a large population of statistically identical and

inde-pendent (nonrepairable) items (dashed is a possible shift for a higher stress, e g ambient temperature)

burn-in or environmental stress screeningis used (Chapter8) Early failures must be

distinguished from defects and systematic failures, which are present at t=0,

deter-ministic, caused by errorsormistakes, and whose elimination requires a change in

design, production process, operational procedure, documentation or other Length

of early failure period varies in practice from few h to some 1'000h The presence of

a period with constant (or nearly so) failure rate λ( )t ≈λ is realistic for manyequipment & systems, and useful for calculations The memoryless property, whichcharacterizes this period, leads to exponentially distributed failure-free times and to

a time homogeneous Markov process for the time behavior of a repairable system if also constant repair rates can be assumed (Chapter 6) An increasing failure rate

after a given operating time (>10 years for many electronic equipment) is typical

for most items and appears because of degradation phenomena due to wear out.

A possible explanation for the shape of λ( )t given in Fig 1.2 is that the

popu-lation contains n p f weak elements and n(1−p f)good ones The distribution ofthe failure-free time can then be expressed by a weighted sum of the form

F( )t = p f F ( )1 t + −(1 p f)F ( )2 t , where F ( )1 t can be a gamma ( β < 1 ) and F ( )2 t a

shifted Weibull ( β > 1 ) distribution (Eqs.(A6.34),(A6.96),(A6.97)), see also pp 337,

355 & 467 for alternative possibilities

The failure rate strongly depends upon the item's operating conditions, see e.g.Figs 2.4-2.6 and Table 2.3 Typical figures for λ are 10− 10 to 10− 7h− 1 for

electronic components at 40°C, doubling for a temperature increase of 10 to 20°C.From Eqs (1.3)-(1.5) one recognizes that for an item new at t=0 and δt→0,

λ( )t δt is the conditional probability for failure in ( ,t t+δt] given that the item has not failed in ( , ]0 t Thus, λ( )t is not a density as defined by Eq (A6.23) and must

be clearly distinguished from the density f( )t of the failure-free time ( f( ) t δt is the

unconditional probability for failure in ( ,t t+δt]), from the failure intensity z( )t of

an arbitrary point process, and form the intensity h ( ) t or m ( )t of a renewal or Poisson process (Eqs (A7.228), (A7.24), (A7.193)); this also in the case of a homo-

geneous Poisson process, see pp 378,426,466,524 for further considerations.The concept of failure rate applied to humans yields a shape as in Fig 1.2

1.2.4 Maintenance, Maintainability

Maintenance defines the set of actions performed on the item to retain it in or to restore it to a specified state Maintenance is thus subdivided into preventive main- tenance, carried out at predetermined intervals to reduce wear out failures, and corrective maintenance, carried out after failure detection and intended to put the

item into a state in which it can again perform the required function Aim of a

preventive maintenance is also to detect and repair hidden failures, i.e., failures in

redundant elements not detected at their occurrence Corrective maintenance is also known as repair, and can include any or all of the following steps: detection, localization (isolation), correction, checkout Repair is used in this book as a syno- nym for restoration, by neglecting logistic and administrative delays To simplify calculations, it is generally assumed that the element in the reliability block diagram for which a maintenance action has been performed is as-good-as-new after mainte- nance This assumption is valid for the whole equipment or system in the case of constant failure rate for all elements which have not been repaired or replaced Maintainability is a characteristic of the item, expressed by the probability that a preventive maintenance or a repair of the item will be performed within a stated time interval for given procedures and resources (skill level of personnel, spare parts, test facilities, etc.) From a qualitative point of view, maintainability can be defined as the ability of the item to be retained in or restored to a specified state The mean (expected value) of the repair time is denoted by MTTR (mean time torepair (restoration)), that of a preventive maintenance by MTTPM Maintainability

has to be built into complex equipment and systems during design and development

by realizing a maintenance concept Due to the increasing maintenance cost,

maintainability aspects have grown in importance However, maintainabilityachieved in the field largely depends on the resources available for maintenance(human and material), as well as on the correct installation of the equipment orsystem, i.e on the logistic support and accessibility.

1.2.5 Logistic Support

Logistic support designates all actions undertaken to provide effective andeconomical use of the item during its operating phase To be effective, logistic

support should be integrated into the maintenance concept of the item under

consideration and include after-sales service

An emerging aspect related to maintenance and logistic support is that of

obsolescence management, i.e , how to assure functionality over a long operatingperiod (e.g 20 years) when technology is rapidly evolving and components need

for maintenance are no longer manufactured Care has to be given here to design aspects, to assure interchangeability during the equipment’s useful life without

important redesign (standardization has been started [1.5, 1.11, A2.6 (IEC 62402)])

1.2.6 Availability

Availabilityis a broad term,expressing the ratio of delivered to expected service

It is often designated by A and used for the stationary & steady-state value of thepoint and average availability (PA=AA) Point availability (PA(t)) is a characteristic

of the item expressed by the probability that the item will perform its required tion under given conditions at a stated instant of time t From a qualitative point of view, point availability can be defined as the ability of the item to perform its required function under given conditions at a stated instant of time (dependability) Availability evaluations are often difficult, as logistic support and human factors should be considered in addition to reliability and maintainability Ideal human and logistic support conditions are thus often assumed, yielding to the intrinsic (inherent) availability In this book, availability is used as a synonym for intrinsic availability Further assumptions for calculations are continuous operation and complete renewal of the repaired element in the reliability block diagram (assumed

func-as-good-as-new after repair) For a given item, the point availability PA( )t rapidly converges to a stationary & steady-state value, given by(Eq (6.48))

PA is also the stationary & steady-state value of the average availability (AA)

giving the mean (expected value) of the percentage of the time during which the

item performs its required function PA S and AA S is used for considerations atsystem level Other availability measures can be defined, e.g mission availability, work-mission availability, overall availability (Sections 6.2.1.5, 6.8.2) Application

specific figures are also known, see e.g [6.12] In contrast to reliability analyses for

which no failure at item (system) level is allowed (only redundant parts can fail and

be repaired on line), availability analyses allow failures at item (system) level.

1.2.7 Safety, Risk, and Risk Acceptance

Safety is the ability oftheitem not to cause injury to persons, nor significant materialdamage or other unacceptable consequences during its use Safety evaluation mustconsider the following two aspects: Safety when the item functions and is operatedcorrectly and safety when the item, or a part of it, has failed The first aspect deals

with accident prevention, for which a large number of national and international regulations exist The second aspect is that of technical safety which is investigated

in five steps (identify potential hazards, identify their causes, determine their effect, classify their effect as per Fig 2.13, investigate possibilities to avoid the hazard or at least to mitigate its effect), using similar tools as for reliability However, a distinc-

tion between technical safety and reliability is necessary. While safety assurance

ex-amines measures which allow the item to be brought into a safe state in the case of failure (fail-safe behavior), reliability assurance deals with measures for minimizing

the total number of failures Moreover, for technical safety the effects of external influences like human errors, catastrophes, sabotage, etc are of great importance and

must be considered carefully The safety level of the item influences the number of

product liability claims However, increasing in safety can reduce reliability.

Closely related to the concept of safety are those of risk, risk management, and risk acceptance; including risk analysis & assessment [1.3,1.9,1.21,1.23,1.26, 1.28]

Risk problems are often interdisciplinary and have to be solved in close cooperation between engineers and sociologists to find common solutions to controversial questions An appropriate weighting between probability of occurrence and effect (consequence) of a given accident is important The multiplicative rule is one among different possibilities Also it is necessary to consider the different causes (machine, machine & human, human) and effects (location, time, involved people, effect duration) of an accident Statistical tools can support risk assessment.

However,although the behavior of a homogenous human population is often known,

experience shows that the reaction of a single person can become unpredictable (see Section 6.10 for basic considerations on human reliability) Similar difficulties also arise in the evaluation of rare events in complex systems Risk analyses are ba-

sically performed with tools used for failure modes and effect analysis (Section 2.6).However, for high-risk systems, refinements are often necessary, for instance, using

the risk priority number concept with logarithmic scale [2.82].

Quite generally, considerations on risk and risk acceptance should take intoaccount that the probability p1 for a given accident which can be caused by one of n statistically identical and independent items, each of them with occurrence

probability p, is for n p small (n→∞,p→0) nearly equal to n p as per

p1=n p(1− p)n−1≈n p e−n p≈n p(1−n p)≈n p. (1.11)

Equation (1.11) follows from the binomial distribution and the Poissonapproximation (Eqs (A6.120) & (A6.129)) It also applies with n p= λtot T to thecase in which one assumes that the accident occurs randomly in the interval ( , ]0 T ,

caused by one of n independent items (systems) with failure rates λ1,…,λn, where

λtot =λ1+ … +λn This is because the sum of n independent Poisson processes is again a Poisson process (Eq (7.27)) and the probability λtot λ T

T e− tot for one

failure in the interval ( , ]0 T is nearly equal to λtot T Thus, for n p<<1 or

λtot T<<1 it holds that

Also by assuming a reduction of the individual occurrence probability p

(or failure rate λi), one recognizes that in the future it will be necessary either to

accept greater risks p1 or to keep the spread of high-risk technologies under tighter control Similar considerations apply to environmental stresses caused by mankind Aspects of ecologically acceptable production, use, disposal, recycling, reuse of pro- ducts should become subject for international regulations (sustainable development).

In the context of a product development, risks related to feasibility and time to market within the given cost constraints must also be considered during all develop-

ment phases (feasibility checks in Fig 1.6 and Tables A3.3 & 5.3)

Mandatory for risk management are psychological aspects related to risk awareness and safety communication As long as a danger for risk is not perceived, people often do not react Knowing that a safety behavior presupposes a risk awareness, communication is an important tool to avoid that the risk related to a

given system will be underestimated, see e.g [1.23, 1.26]

1.2.8 Quality

Quality is understood as the degree to which a set of inherent characteristics fulfills requirements This definition, given now also in the ISO 9000: 2000 family [A1.6], follows closely the traditional definition of quality, expressed by fitness for use, and

applies to products and services as well

1.2.9 Cost and System Effectiveness

All previously introduced concepts are interrelated Their relationship is best shown

through the concept of cost effectiveness, as given in Fig 1.3 Cost effectiveness is

a measure of the ability of the item to meet a service demand of stated quantitativecharacteristics, with the best possible usefulness to life-cycle cost ratio It is often

referred also to as system effectiveness Figure 1.3 deals essentially with technical

and cost aspects Some management aspects are considered in Appendices A2-A5

From Fig 1.3, one recognizes the central role of quality assurance, bringing together all assurance activities (Section 1.3.3), and of dependability (collective term

for availability performance and its influencing factors)

As shown in Fig.1.3, life-cycle cost (LCC)is the sum of cost for acquisition, ation, maintenance,and disposal of the item For complex systems,higher reliabilityleads in general to higher acquisition cost and lower operating cost, so that theoptimum of life-cycle cost seldom lies at extremely low or high reliability figures.For such a system, per year operating & maintenance cost often exceeds 10% of ac-quisition cost, and experience shows that up to 80% of the life-cycle cost is fre-quently generated by decisions early in the design phase To be complete, life-cycle

oper-cost should also take into account current and deferred damage to the environment caused by production, use, and disposal of the item Life-cycle cost optimization falls within the framework of cost effectiveness or systems engineering It can be

positively influenced by concurrent engineering[1.16, 1.22] Figure1.4shows anexample of the influence of the attainment level of quality and reliability targets

on the sum of cost of quality and operational availability assurance for two tems with different mission profiles [2.2(1986)], see Example 1.1 for an introduction

sys-Example 1.1

An assembly contains n independent components each with a defective probability p Let c k be

the cost to replace k defective components Determine (i) the mean (expected value) C( )i of the

total replacement cost (no defective components are allowed in the assembly) and (ii)the mean

of the total cost (test and replacement) C( )ii if the components are submitted to an incoming inspection which reduces defective percentage fromp to p0(test cost c t per component).

(ii) To the cost caused by the defective components, calculated from Eq (1.14) with p0 instead

of p, one must add the incoming inspection cost n c t

Using Eq (A7.42) instead of (A6.120), similar considerations to those in Example

1.1 yield for the mean (expected value) of the total repair cost C cm during thecumulative operating time T of an item with failure rate λ and cost c cm per repair

MTBF cm

(In Eq (1.16), the term λT gives the mean value of the number of failures during T

(Eq (A7.42)), and MTBF is used as MTBF=1 /λ.)

From the above considerations, the following equation expressing the mean C of

the sum of the cost for quality assuranceand for the assurance of reliability,maintainability, and logistic support of a system can be obtained

MTBF S cm S off d d

Thereby, q is used for quality, r for reliability, cm for corrective maintenance, pm

forpreventivemaintenance, l for logistic support, off fordowntime & d for defects.

Cost Effectiveness (System Effectiveness)

Life-Cycle

Cost (LCC)

Operational Effectiveness

Safety Capability Operational Availability

(Dependability)

Intrinsic Availability

Reliability Maintainability Human Factors Logistic Support Useful Life Injury to Persons Damage to Property Damage to Environment Acquisition Operation, Maintenance Disposal

Quality Assurance

(Hardw.& Softw.)

Reliability Engineering

ability Engineering

• Quality control

during tion (hardware)

• Required function

• Rel block diagr.

• Rel prediction

• Design reviews

• Maintainability targets

• Maintenance concept

• Partitioning

in LRUs

• Faults detection and localization

• Design guidelines

• Maintainability analysis

• Design reviews

• Safety targets

• Design guidelines

• Safety analysis (FMEA/FMECA, FTA, etc.)

• Design reviews

• Maintenance concept

• Customer/User documentation

• Spare parts provisioning

• Tools and test equipment for maintenance

• After sales service

Safety and Human- Factors Engineering

Cost Effectiveness Assurance (System Effectiveness Assurance)

Capability and

Life-Cycle

Cost

Logistic Support

Figure 1.3 Cost Effectiveness (System Effectiveness) for complex equipment & systems with high quality and reliability (RAMS) requirements (see Appendices A1 - A5 for definitions & management aspects; dependability can be used instead of operational availability, for a qualitative meaning)

MTBF S and OA S are the system mean operating time between failures (assumedhere =1 /λS ) and the system steady-state overall availability (Eq (6.196) with T pm

instead of T PM) T is the total system operating time (useful life) and n d is the number of hidden defects discovered (and eliminated) in the field C q, C r, C cm,

C pm, and C l are the cost for quality assurance and for the assurance of reliability,repairability, serviceability, and logistic support, respectively ccm , c off , and c d arethe cost per repair, per hour down time, and per hidden defect, respectively(preventive maintenance cost are scheduled cost, considered here as a part of C pm)

The first five terms in Eq (1.17) represent a part of the acquisition cost, the last three terms are deferred cost occurring during field operation A model for

investigating the cost C according to Eq (1.17)was developed in [2.2 (1986)], byassuming C q, C r, C cm, C pm, C l, MTBF S, OA S, T, ccm , c off , c d, and nd asparameters and investigating the variation of the total cost expressed by Eq (1.17)

as a function of the level of attainment of the specified targets, i.e., by ducing the variables g q=QA QA/ g, g r=MTBF S/MTBF S g, g cm=MTTR S g/MTTR S,

intro-g pm=MTTPM S g /MTTPM S, and g l=MLD S g/MLD S , where the subscript g denotes the

specified target for the corresponding quantity A power relationship

was assumed between the actual cost C i, the cost C ig to reach the specified target

(goal) of the considered quantity, and the level of attainment of the specified target

(0<m l <1 and all other m i>1) The following relationship between the number ofhidden defects discovered in the field and the ratio C q/C qg was also included inthe model

The final equation for the cost C as function of the variables g q, g r, g cm, g pm, and

g l follows then as (using Eq (6.196) for OA S)

T c g

g

pm pm

off q

MTTR MTBF

MLD MTBF

The relative cost C C/ g given in Fig 1.4 is obtained by dividing C by the value

C g form Eq (1.20) with all g i =1 Extensive analyses with different values for m i,

C ig, MTBF Sg, MTTR Sg, MLD Sg, MTTPM Sg, T pm, T, c cm, c off, and c d have shownthat the value C C/ g is only moderately sensitive to the parameters m i

1 2 3 4 5

Figure 1.4 Basic shape of the relative cost C Cg/ per Eq (1.20) as function of g q=QA QA/ g and

systems with different mission profiles (the specified targets g q= 1 and g r= 1 are dashed)

Product liability is the onus on a manufacturer (producer) or others to compensate

for losses related to injury to persons, material damage, or other unacceptable

consequences caused by a product (item) The manufacturer has to specify a safe operational mode for the product (user documentation) In legal documents related to product liability, the term product often indicates hardware only and the term defective product is in general used instead of defective or failed product.

Responsible in a product liability claim are all those people involved in the design,production, sale, and maintenance of the product (item), inclusive suppliers

Often, strict liability is applied (the manufacturer has to demonstrate that

the product was free from defects) This holds in the USA and increasingly

in Europe [1.10] However, in Europe the causality between damage and defect hasstill to be demonstrated by the user (see p 382 for further considerations)

The rapid increase of product liability claims (alone in the USA, 50,000 in

1970 and over one million in 1990) cannot be ignored by manufacturers.Although such a situation has probably been influenced by the peculiarity of

US legal procedures, configuration management and safety analysis (in particular causes-to-effects analysis, i.e., FMEA/FMECA or FTA as introduced in Section 2.6)

as wellas considerationsonrisk managementshouldbeperformed t o increase safety and avoid product liability claims (see Sections 1.2.7, 2.6 & 6.10, andAppendix A.3.3)

Table 1.1 Historical development of quality assurance (management) and reliability engineering

before 1940 Quality attributes and characteristics are defined In-process and final tests are

carried out, usually in a department within the production area The concept of

quality of manufacture is introduced.

1940 - 50 Defects and failures are systematically collected and analyzed Corrective actions

are carried out Statistical quality control is developed It is recognized that quality must be built into an item The concept quality of design becomes important.

1950 - 60 Quality assurance is recognized as a means for developing and manufacturing an

item with a specified quality level Preventive measures (actions) are added to tests

and corrective actions It is recognized that correct short-term functioning does not

also signify reliability Design reviews and systematic analysis of failures (failure

data and failure mechanisms), performed often in the research & development area, lead to important reliability improvements.

1960 - 70 Difficulties with respect to reproducibility and change control, as well as interfacing

problems during the integration phase, require a refinement of the concept of

configuration management Reliability engineering is recognized as a means of

developing and manufacturing an item with specified reliability Reliability

estimation methods and demonstration tests are developed It is recognized that

reliability cannot easily be demonstrated by an acceptance test Instead of a

reliabili-ty figure ( λ or MTBF= 1 / ) λ, contractual requirements are for a reliability assurance

program Maintainability, availability, and logistic support become important.

1970 - 80 Due to the increasing complexity and cost for maintenance of equipment and

systems, the aspects of man-machine interface and life-cycle cost become important.

Customers require demonstration of reliability and maintainability during the

warranty period Quality and reliability assurance activities are made project specific and carried out in close cooperation with all engineers involved in a project Concepts like product assurance, cost effectiveness and systems engineering are introduced Human reliability and product liability become important.

1980 - 90 Testability is required Test and screening strategies are developed to reduce testing

cost and warranty services Because of the rapid progress in microelectronics,

greater possibilities are available for redundant and fault tolerant structures.

Software quality becomes important.

after 1990 The necessity to further shorten the development time leads to the concept of

con-current engineering Total Quality Management (TQM) appears as a refinement to quality assurance as used at the end of the seventies RAMS is used for reliability,

availability, maintainability & safety, reliability engineering for RAMS engineering.

Methods and procedures of quality assurance and reliability engineering have beendeveloped extensively over the last 60 years For indicative purpose, Table 1.1 sum-marizes major steps of this development and Fig.1.5 shows the approximate distri-bution of the effort between quality assurance and reliability engineering during thesame period of time Because of the rapid progress of microelectronics, considera-

tions on redundancy, fault-tolerance, test strategy, and software quality gains in importance A skillful, allegorical presentation of the story of reliability is in [1.25].

Quality data reporting system Quality testing, Quality control,

effects / mechanisms analysis Fault causes / modes / System engineering (part)

Configuration management Software quality Reliability (RAMS) analysis

Quality assurance

Year Relative effort [%]

Figure 1.5 Approximate distribution of the effort between quality assurance and reliability (RAMS)

engineering for complex equipment & systems with high quality and reliability (RAMS) requirements

This section deals with some important considerations on the organization of quality

and reliability assurance in the case of complex repairable equipment and systems with high quality and reliability requirements In this context, the term reliability appears for reliability, availability, maintainability, and safety (RAMS) This minor

part of the book aims to support managers in answering the question of how to specify and realize high reliability (RAMS) targets for equipment and systems

Refinements are in Appendix A3 for complex equipment and systems for which tailoring is not mandatory, with considerations on quality management and total quality management(TQM)as well As a general rule,quality assurance and reliabil-ity (RAMS) engineering must avoid bureaucracy, be integrated in project activities,

and support quality management and concurrent engineering efforts, as per TQM

Experience shows that besides the prevention of defects and systematic failures,

which remains the primary task of a quality assurance system,

the development and production of complex repairable equipment and systems with

high reliability (RAMS) targets requires specific activities during all life-cycle phases

of the item considered Figure 1.6 shows the life-cycle phases and Table 1.2 gives the main tasks for quality and reliability (RAMS) assurance Depicted in Table 1.2 is

also the period of time over which the tasks have to be performed Within a project,the tasks of Table 1.2 must be refined in a project-specific quality and reliability(RAMS) assurance program (Appendix A3)

Table 1.2 Main tasks for quality and reliability (RAMS) assurance of complex equipment & systems with high quality and reliability requirements (the bar height is a measure of the relative effort)

Specific during

Main tasks for quality and reliability (RAMS) assurance of

complex equipment and systems, conforming to TQM

(see Table A3.2 for greater details and a possible task assignment;

software quality appears in tasks 4, 8-11, 14-16, see also Section 5.3)

1 Customer and market requirements

2 Preliminary analyses

3 Quality and reliability aspects in specs, quotations, contracts, etc.

5 Reliability and maintainability analyses

6 Safety and human factor analyses

7 Selection and qualification of components and materials

8 Supplier selection and qualification

9 Project-dependent procedures and work instructions

10 Configuration management

11 Prototype qualification tests

12 Quality control during production

13 In-process tests

14 Final and acceptance tests

15 Quality data reporting system

16 Logistic support

17 Coordination and monitoring

18 Quality costs

19 Concepts, methods, and general procedures (quality and reliability)

20 Motivation and training

Project-independent Conception Definition Design & Devel Evaulation Production Use

Conception, Definition,

Design, Development, Evaluation

Production (Manufacturing) Use Preliminary

Series production

Installation, Operation

• Qualified and released prototypes

• Technical documentation

• Proposal for pilot production

• Feasibility check

• Production documentation

• Qualified tion processes

produc-• Qualified and released first series item

• Proposal for series production

• Series item

• Customer documentation

• Logistical support concept

• Spare part provisioning Disposal, Recycling

Figure 1.6 Basic life-cycle phases of complex equipment and systems (the output of a given

phase is the input to the next phase), see Tab 5.3 (p 161) for software

Performance, dependability, cost, and time to market are key factors for today's

products and services Taking care of the considerations in Section 1.3.1, the basic rules for a quality and reliability (RAMS) assurance optimized by considering cost

and time schedule aspects (conforming to TQM) can be summarized as follows:

1 Quality and reliability (RAMS) targets should be just as high as necessary tosatisfy real customer needs

→ Apply the rule "as-good-as-necessary".

2 Activities for quality & reliability (RAMS) assurance should be performed

con-tinuously throughout all project phases, from definition to operating phase

→ Do not change the project manager before ending the pilot production.

3 Activities must be performed in close cooperation between all engineersinvolved in the project (Table A3.2)

→ Use TQM and concurrent engineering approaches.

4 Quality and reliability (RAMS) assurance activities should be monitored by acentral quality & reliability assurance department (Q & RA), which cooperates

actively in all project phases (Fig 1.7 and Table A3.2)

→ Establish an efficient and independent quality & reliability assurance department (Q & RA) active in the projects.

QI

Figure 1.7 Basic organizational structure for quality and reliability (RAMS) assurance i n a company

producing complex equipment and systems with high quality and reliability (RAMS) requirements

(connecting lines indicate close cooperation; A denotes assurance, I inspection, Q quality,

Rreliability (RAMS) )

Figure 1.7 shows a basic organization which could embody the above rules and

satisfy requirements of quality management standards (Appendix A2) As shown

in Table A3.2, the assignment of quality and reliability (RAMS) assurance tasks

should be such, that every engineer in a project bears his / her own responsibilities

(as per TQM) A design engineer should for instance be responsible for all aspects

of his/her own product (e.g an assembly) including reliability, maintainability andsafety, and the production department should be able to manufacture and test such

an item within its own competence The quality & reliability (RAMS) assurance department (Q & RA in Fig 1.7) can be for instance responsible for (see alsoTab A3.2)

• setting targets for quality and reliability (RAMS) levels,

• preparation of guidelines and working documents (quality and reliability(RAMS) aspects),

• coordination of the activities belonging to quality and reliability (RAMS)assurance,

• reliability (RAMS) analyses at system level,

• qualification, testing, and screening of components and material (quality andreliability aspects),

• release of manufacturing processes (quality and reliability (RAMS) aspects),

• development and operation of the quality data reporting system,

• acceptance testing (with customers)

This central quality and reliability (RAMS) department should not be too small(credibility) nor too large (sluggishness)

1.3.3 Elements of a Quality Assurance System

As stated in Sections 1.3.1, many of the tasks associated with quality assurance (in the sense of quality management as per TQM ) are interdisciplinary Inorder tohave a minimum impact on cost and time schedules, their solution requires the

concurrent efforts (close cooperation) of all engineers involved in a project.

To improve coordination, it is useful to group the quality assurance activities(see also Fig 1.3 and Appendix A3.3):

1 Configuration Management: Procedure used to specify, describe, audit &

release the configuration of the item, as well as to control it duringmodifications or changes Configuration management is an important tool for

quality assurance It can be subdivided into configuration identification, auditing (design reviews), control, and accounting (Appendix A3.3.6).

2 Quality Tests: Tests to verify whether the item conforms to specified

require-ments Quality tests include incoming inspections, as well as qualificationtests, production tests, and acceptance tests They also cover reliability,maintainability, safety, and software aspects To be cost effective, quality tests

must be coordinated and integrated into a test strategy.

3 Quality Control During Production: Control (monitoring) of the production

processes and procedures to reach a stated quality of manufacturing

4 Quality Data Reporting System (FRACAS): A system to collect, analyze &

correct all defects and failures (faults) occurring during the production and test

of the item, as well as to evaluate and feedback the corresponding quality andreliability (RAMS) data Such a system is generally computer-aided Analysis

of failures anddefectsmustb etracedtothe cause,toavoid repetitionof thesameproblem, and be pursued at least during the warranty period (Fig 1.8)

5 Software quality: Procedures and tools to specify, develop, and test software

(appears in tasks 4, 8-11,14-16 of Tables 1.2&A3.2,see also Section 5.3).Configuration management spans from the definition up to the operating phase(Appendices A3 & A4) Quality tests encompasses technical and statistical aspects(Chapters 3, 7, and 8) The concept of a quality data reporting system is depicted

in Fig 1.8 (see Appendix A5 for basic requirements) Table 1.3 shows an example

of data reporting sheets for PCBs evaluation

The quality and reliability (RAMS) assurance system must be described in an

ap-propriate quality handbook supported by the company management. Apossible

con-tent of such a handbook for a company producing complex equipment and systems with high quality & reliability (RAMS) requirements is:•General,•Project Organi-zation, •Quality Assurance (Management) system, •Quality & Reliability (RAMS)

AssuranceProgram, •ReliabilityEngineering, •MaintainabilityEng., •Safety & man Eng.,•SoftwareQualityAssurance,•LogisticSupport,•Motivation&Training

Short term Actions / Measures Long term

Long feedback loop (preventive measures)

Medium feedback loop (corrective actions and preventive measures)

Short feedback loop (corrective actions)

Table 1.3 Example of information status for PCBs (populated printed circuit board’s) from a

quality data reporting system

a) Defects and failures (faults) at PCB level

com-ponent

PCB short term long term pro- duction

Q A other areas

b) Defects and failures (faults) at component level

Period: PCB: No of PCBs:

type Same application

inspection

in-process test final test warranty

term long term

d) Correlation between components and PCBs

Period:

Com-ponent

PCB

1.3.4 Motivation and Training

Cost effective quality and reliability (RAMS) assurance/management can beachieved if every engineer involved in a project is made responsible for his/herassigned activities (e.g as per Table A3.2) Figure 1.9 shows a comprehensive,

practice oriented, motivation and training program in a company producing complex equipment and systems with high quality and reliability (RAMS) requirements.

Basic training

Special training

organization of the company's quality

managers, selected engineers

Documentation: ca 30 pp.

Engineering

Participants: Project managers, engineers

from marketing & production,

selected engineers from

development

(applications oriented and company specific) Participants: Design engineers, Q&RA

specialists, selected engineers from marketing and production

techniques Participants: Q&RA specialists, selected

engineers from development and production

Test and Screening Strategies, Software

Quality, Testability, Reliability and

Avail-ability of Complex Repairable Systems,

Fault Tolerant Systems with Hardware and

Software, Mechanical Reliability, Failure

Mechanisms and Failure Analysis, etc.

Figure 1.9 Example for a practical oriented training and motivation program in a company

producing complex equipment and systems with high quality and reliability (RAMS) requirements

(Nonrepairable Elements up to System Failure)

Reliability analysis during the design and development of complex components,

equipment, and systems is important to detect and eliminate reliability weaknesses

as early as possible and to perform comparative studies Such an investigation includes failure rate and failure mode analysis, verification of the adherence to design guidelines, and cooperation in design reviews This chapter presents meth-

ods and tools for failure rate and failure mode analysis of complex equipment and

systems considered as nonrepairable up to system failure (except for Eq (2.48)).

After a short introduction, Section 2.2 deals with series- parallel structures.Complex structures, elements with more than one failure mode, and parallel modelswithload sharing are investigated in Section 2.3 Reliability allocation with cost

considerations are discussed in Section 2.4, stress /strength and drift analysis inSection 2.5 Section 2.6 deals with failure mode and causes-to-effects analyses.Section2 7 gives a checklist for reliability aspects in design reviews.Maintainability is considered in Chapter 4 and repairable systems are investigated

in Chapter 6 (including complex systems for which a reliability block diagramdoes not exist, imperfect switching, incomplete coverage, reconfigurable systems,common cause failures, as well as an introduction tonetworkreliability,BDD,ET,dynamicFT, Petri nets,andcomputer-aidedanalysis) Design guidelines are inChapter 5, qualification tests in Chapter 3, reliability tests in Chapters 7 & 8.Theoretical foundations for this chapter are in Appendix A6

An important part of the reliability analysis during the design and development ofcomplex equipment and systems deals with failure rate and failure modeinvestigation as well as with the verification of the adherence to appropriate design

guidelines for reliability Failure modes and causes-to-effects analysis is considered

in Section 2.6, design guidelines are given in Chapter 5 Sections 2.2-2.5 are

devoted to failure rate analysis.

A Birolini, Reliability Engineering, DOI: 10.1007/978-3-642-39535-2_2,

Ó Springer-Verlag Berlin Heidelberg 2014

25

Investigating the failure rate of a complex equipment or system leads to the

calculation of the predicted reliability, i.e., that reliability which can be calculatedfrom the structure of the item and the reliability of its elements Such a prediction is

necessary for an early detection of reliability weaknesses, for comparative studies, for availability investigation taking care of maintainability and logistic support, and for the definition of quantitative reliability targets for designers and subcontractors.

However, because of different kind of uncertainties, the predicted reliability canoften be only given with a limited accuracy To these uncertainties belong

• simplifications in the mathematical modeling (independent elements, completeand sudden failures, no flaws during design and manufacturing, no damages),

• insufficient consideration of faults caused by internal or external interference(switching, transients, EMC, etc.),

• inaccuracies in the data used for the calculation of the component failure rates

On the other hand, the true reliability of an item can only be determined by reliability tests, performed often at the prototype's qualification tests, i.e., late inthe design and development phase Practical applications also shown that with anexperienced reliability engineer, the predicted failure rate at equipment or

system level often agree reasonably well (within a factor of 2) with field data.

Moreover, relative values obtained by comparative studies generally have a muchgreater accuracy than absolute values All these reasons support the efforts for a

reliability prediction during the design of equipment and systems with specified

reliability targets

Besides theoretical considerations, discussed in the following sections,

practical aspects have to be considered when designing reliable equipment and

systems, for instance with respect to operating conditions and to the mutualinfluence between elements (input/output, load sharing, effects of failures,transients, etc.) Concrete possibilities for reliability improvement are

• reduction of thermal, electrical and mechanical stresses,

• correct interfacing of components and materials,

• simplification of design and construction,

• use of qualitatively better components and materials,

• protection against ESD and EMC,

• screening of critical components and assemblies,

• use of redundancy,

in that order Design guidelines (Chapter 5) and design reviews (Tables A3.3, 2.8,

4.3, and 5.5, Appendix A4) are mandatory to support such improvements

This chapter deals with nonrepairable (up to system failure) equipment and

systems Maintainability is discussed in Chapter 4 Reliability and availability ofrepairable equipment and systems is considered carefully in Chapter 6