RELIABILITY ENGINEERING E BALAGURUSAMY Institute of Public Enterprise Hyderabad Bd : Tata McGraw-Hill Publishing. Company Limited eWorldDedicated to PROF. K B MISRA One of the pioneers in reliability engineering in IndiaPreface is now a well-recognized and rapidly developing branch of engineering. Since reliability study is considered essential for proper utiliza~ tion and maintenance of engineering systems and equipments, it has gained much importance among practising engineers and manufacturers. The subject of reliability engineering has started appearing in engineering curri- cula of many institutions and universities. This book is meant for those who wish to take up reliability as a subject of study. It should also be of parti- cular interest to all electrical, electronic, mechanical, chemical, industrial and systems engineers who have not had formal training in reliability engineering but are confronted with the urgent problem of quantitative treat- ment of reliability. The material is intended for an audience at the level of postgraduate or senior undergraduate students. It is assumed that the reader is acquainted with elementary differential equations and fundamental concepts of proba- bility law. The first chapter is of an introductory nature. In this chapter basic con- cepts of reliability and quality are explained and various modes and causes of failures are discussed. Chapter 2 presents a general procedure for reliability analysis and discu: a number of techniques that are available to improve Chapter 3 surveys the basic concepts of probability theory and Laplace transforms, a good working knowledge of which is necessary for any real appreciation of reliability. It provides probability distributions of interest in reliability engineering, In Chapter 4 the reliability characteristics of components are discussed. Failure curves and various forms of hazard models are analysed. Chapters 5 through 7 deal with reliability of systems and its improvement. Redundancy and repair, the most important methods of enhancing system reliabi considered in more detail. Chapter 8 is devoted to m-order systems involving non-identical units wl Chapter 9 presents stochastic models for systems with stress-dependent units. In Chapter 8, frequency and duration concepts are employed to determinethe parameters of resiable ystems. Many new techniques aresugasted in Chapter 9 to deat with diferent hinds of stessharand modes Liet of raintenance on sch system aso disused Both pale! and fractional eal Faces are considered. ‘An irspoctant class of sptons Keown a earch systems are dealt, swith in Chaper 10, Relbity and cost models are dscssed ad asteate. ‘The lst chapters focus on the main ius involved in she management and economiss of relabiiy engineering. Rely and wily ost models which would be wefl for bo the uses and manufacturers ate diseased. Relevant statieal tables and reliblity chars which couX! be wef for syers and dsienengeers ae lnctuded inthe Appendices. E macaconesuortAcknowledgements I was introduced to the subject area of the book by Prof. D.R. Kohli and Prof. K.B. Misra while I was a research fellow at the University of Roorkee. Iam greatly indebted to them. My special appreciation and thanks are due to Prof. Misra for his guidance, encouragement and support during my two years of research at Roorkee. I wish to thank the following journals for publishing some of my ideas: —IEEE Transactions on Reliability International Journal of Systems Science Microelectronics and Reliability —International Journal of Mathematical Education in Science and Technology —Electro-Technology —IEEE Power Engineers Society Proceedings 1 acknowledge gratefully the University of Roorkee and the Council of Scientific and Industrial Research for their support in carrying out research in this area. I owe a debt to the participants of many winter and summer schools and industrial short courses I conducted during the last five years. The greatest thanks are due to my wife Sushila who has been a source of great inspiration and encouragement throughout the project. E BALAGuRUSAMYNotation Am, Atk, tn) a BU, y) ae c Dk, m) d, E[X} S,f(k, m) fa) F() Gm, G(k, m) At) hy k ke Lt] m M Ne n P(x, y) Px) Pilly) , p(t) peice i Qrriy QmXt) * Availability (steady-state) of simple 1-order systems ‘Availability (steady-state) of k-out-of-m: ¢ systems Stress factor Binomial coefficient = x/(x~y)! y! ‘Syatem cost i oe Cost of each unit : Mean downtime of k-out-of-m: G systems Failure-rate derating factor of units Expected value of the random variable Y Frequency of failures” Probability-density function (pdf) Commulative-distribution function (caf) System MTTF gain “ Stress-dependent and time-varying bazard ‘function of units System failure rate at the state where x units have failed Minimum ‘fiamber of units to be good to keep the system operative Minimum number of units the failure of which would cause thé system to fail Laplace transform Total number of parallel redundant units in the system Number of minimal sets contained in the system Number of minimal cut sets contained in the system Total number of stages‘(levels) in a hierarchical system Probability of y successes in x trials Probability of the event x occurring Probability of the event x given the évent y has occurred Reliability of units (equipments, elements) ‘Effective reliability of units Reliabi of units at rated load Unreliability ‘of simple sh-order systems8 ean tk, m 1, aio) a a a a Bay Ral) Ri, Rm) Tow nretibilty of k-outom: G system Unreiabity of wits Fasice probability of wits dae to open cect Conditional probability of open file oft uit ive that a faite Bae occured Faire probabiliy of ueits due fo short circuit Conditional protabity of short filre of awit given hata faire bas osurred ny of siple monde systems [Number of repair Fsiisavaiate Mean ee time Expected time ores the system ltr an in-service fre Mean tine to fue of Lost of: @ ster Mean nec faite of sep order systems Enpeted fe to complet the schedled mstenance onthe ean optine of koutofn @ syees ‘Time-dependent flee rte of eit Efectie aurea of units, EEette tuore rate ‘Constant repaefte of waits Redundaey ellectvenesscoeficest Sires factor Ratio ofthe operaung ses tothe ated ses of us Redundaey fet Opec ratio 72, Service factor, 77s ‘Som ofthe erm indicated Prodect ofthe terms indated Buse elemeatof Union (eithe-o; logis on) Intersection (bothandlogieat Aso) Gana forcton ‘ean ane to fue of nonmatatined systems ‘Me time to at fale of maintained systems Mean time 1 inservice fare Te system ie good iF and only at least k of ts ris are 3008 ‘outofnn F The system is fled Wf and only it leas} of ts units are riledContents Preface vit Acknowledgements ix Notation xi 1 Basic Concepts of Reliability 1 1.1 Introduction 7 1.2 Reliability and Quality 2 1.3 Failures and Failure Modes 3 1.4 Causes of Failures and Unreliability 5 1.41 Causes 5 .4.2 Design, production and use 5 .4.3 System simplification 5 .4.4 Maintenance 6 .4.5 Communication and coordination 6 1.4.6 Human reliability 6 1.5 Maintainability and Availability 7 1.6 History of Reliability 8 1.7 Reliability Literature 9 1.7.1 Nonmaintained systems 9 1.7.2 Maintained systems 10 1.7.3 Trichotomous systems 10 1.7.4 Systems with nonidentical and dependent units 11 Exercises 11 Further Reading 12 2 Design for Reliabilit 21 Reliability Analysis 13 2.2 Mathematical Models and Numerical Evaluation /4 2.3 Designing for Higher Reliability 15 2.4 Redundancy Techniques 16 2,5 Equipment Hierarchy 17 Constraints and Considerations 1326 Retatilty and Cost 8 Exerches 19 Further Resting 19 3 Reliabitty Mathematics Introduction 20 32 Random Experients 21 321 Deloaionand dart 27 522 Sunolespae 20 333 news 2 BE Combntoa of ote 22 325 Caw of iodo nets 30 3. Peobabity 25 352 Gtinealoraprnd inten 26 333 Stati peatny 28 334 Aeomre petty 7 335 Beaentry rope potas 2 126 Conde repeat potby IP 339 Theromat eal pokeby 3 338 Inner cess 92 344 Random Variables 34 35 Ditbution Funetions 36 136 Dictte Diseibutions 37 3261 Desy end Geneon encom 7 BES Pomon boon” 4 137 Continvoes Distributions 42 22 Deny and dateton action 32 Unto antauoe 3053 Eapennal rian 43 Se Regia entce a 273 Won ann 307 Norm dba 38 Nomerieal Charac sof Random Variables 31 Maen epeaton 46 49 Laplace Testor 47 Becher Martie, Exercee 33 Further Reding $3Contents xv 4 Component Reliability and Hazard Models 56 4.1 Introduction 56 4.2 Component Reliability from Test Data 56 4,3 Mean Time to Failure 58 4.4 Time-Dependent Hazard Models 58 , 1 Field-data curves 58 2 Constant-Hazard model 59 3 Linear-hazard model 59 4 Nonlinear hazard model 60 ‘5 Gamma model 61 t 4, 4.6 Other models 67 4.5 Stress-Dependent Hazard Models 61 4.6 Derivation of Reliability Function using Markov Model 62 4.7 Treatment of Field Data 63 : Further Illustrations 64 7 e Exercises 66 . Further Reading 67 : sees 5 System Reliability Models 68 5.1 Introduction 68 5.2 Systems with Components in Series 68 5,3 Systems with Parallel Components 70 ° 5.4 k-out-of-m systems 75 : ! 5.5 Nonseries-parallel Systems 78 5.6 Systems with Mixed-Mode Failures 87 : 5.7 Fault-Tree Technique 83 ce Further Hlustrations 87 Exercises 89 Further Reading 92 : 6 Redundancy Techniques in System Design 93 6.1 Introduction 93 6.2 Component Versus Unit Redundancy 94 6.3 Weakest-Link Technique 96 6.4 Mixed Redundancy 98 6.5 Standby Redundancy 99 6.5.1 Simple standby system 99 6.5.2 k-out-of-m standby system J07 6.6 Redundancy Optimization 702 : 6.6.1 General 102 6.6.2 Problem formulation 102 6.6.3 Computational procedure 103 gg 6.6.4 Reliability versus cost 106 6.6.5 Optimum level of equipment reliability 107 :m1 coment 162 Double Failres and Redundeney 109 63.1 Efex of hor-iesk ares 19 Siesctetrennihtrmdaranrmetinehset WP (614 Masiniring he rcsbiy ofa singe wage system 111 {75 Ost redannnyconigasion 12 6 Seeger cype tt arngenet 172 Exercies 113 Further Resling 116 bitty Concepts wm inaiiy Fuetign 178 iy Fneion “119 Ta Frequency of ates 122 15 Toro Parallel System wih Repait 122 241 Stem atta 22 8 783 Sem monty as 26 kauvotm syiems 12 17 Preentve Natmensoce 129 Exncles 131 Fier Reve 132 8 m-dse Systens with Nonetial Us 1 S1 Intoduction 138 8.2 Nonmsitnined Stems 196 £24 Lasdagrm wpe 14 £22 Bremerton speach 29 . 2) Srmcmeante ofaate 12 8.3 Some Relay Approximations 102 84 Mainaled Systems 146 SA Arie and fare eeensy 16 B42 Socmaiantay 10 EA Feauey oft fae 15 15 Tvishotomous Sptems 151 £51 Stem teas 151 52 Renting ag oni 152 153 Renting an open 2 ESO Med ednaingy 13 Exercises 154 Further Reading 185 9 mOrser Spates with «Dependent Units 186 24 Ineoucion 136Contents xvii 9.2 Parameters of Dependency 157 (f 93 Analysis of Nonmaintained Systems 159 9.3.1 Events-space method 159 9.3.2 Renewal theory approach 162 9.3.3 Sandler's dependency model 166 9.3.4 Approximations 168 -5 Effective hazard-rate method 170 9.3.6 Conditional reliability approach 172 9.3.7 Derating chart 175 9.3.8 Mean hazard rate approach 178 9.3.9 Systems with nonidentical units 183 9.4 Systems with Repair 183 9.4.1 System availability 183 94.2 Failure frequency 192 9.4.3 Mean time to first system failure 194 9.5 An Optimal Maintenance Policy 197 9.5.1 Maintenance Policy 147 .5.2 Stochastic functions of the system 798 9.5.3 Optimal model 198 9.5.4 Mean time to in-service failure 207 Further Mustrations 203 Exercises 207 Further Reading 209 10 Hierarchical Systems 210 10.1 Introduction 2/0 10.2 Logic Diagram Approach 2/4 10:3 Conditional Probability Approach 2/5 10.4 System Cost 219 10.5 Ilustrations and Discussions 220 _10.6 Reliability Approximations 223 Further Mlustrations 224 * Exercises 226 Further Reading 228 -1 Economics of Reliability Engineering 229 11.1 Economic Issues 229 11.2 Manufacturer's Cost 229 11.3 Customer's Cost 230 11.4 Reliability Achievement Cost Models 237 11.5 Reliability Utility Cost Models 233 11.6 Depreciation Cost Models 233 11.7 Availability-Cost Model for Parallel Systems 236 Exercises. 237 Further Reading 240R Relahity Manogemert 12 Relabiliy Programe 248 122 Management Poies and Decisions 241 123 Relablity Management by Obeties 205 1244 Reliabitty Group 244 2.5 Relailty Data Acquistion an Arabs 246 1251 Panag da soliton ogame 266 1252 Doteaqnstoantboy 20 1253 Unetaaneia 20 4 Aes aang 24 256 Distant 3 : : 126 Managing People fr Retability ‘250 Rt Rewisecetoctange 20 164 Kerra theenoner ney 257 Furter Reoding 281 Appeetices 2s ‘Append A Refiabity Tables and Chars 257, ‘At Normal dutevton 297 ieemonerabity FP : 9 Retaity ot howctomsandy Sew 27 Append D Glorary 259 14x C Sle Bibhography 293 2991 Basic Concepts of Reliability 1.1 INTRODUCTION The development of science and technology and the needs of modern society are racing against each other. Industries are trying to introduce more and more automation in their industrial processes in order to meet the ever-increasing demands of society. The complexity of industrial systems as well as their products are increasing day-by-day. The improvement in effec tiveness of such complex systems has therefore acquired special importance in recent years. One would now like to know the answers to questions such as: What is the effectiveness of a nuclear power station ‘that is not operative formore than 100 days in a year? What is the use of a scientific satellite that cannot transmit any data to the ground? Of what use is a refrigerator that fails almost every week? ‘The effectiveness of a system is understood to mean the suitability of the system for the fulfilment of the intended tasks and the efficiency of utilizing the means put into it. The suitability of performing definite tasks is prima- rily determined by the reliability and quality of the system. An electronic company in USA producing space components rejected 60% of the integrated circuits delivered to it by its vendors, Faults included sur- face defects, foreign particles; and testing deficiencies. A report on the working of a thermal power station in India says that the load factor of the power station in 1979-80 was as low as 57% and the station had as many as 260 trippings, ie. an average of five trippings every week. A manufac- turer of domestic appliances had to replace more than 988 of 3060 products sold to its customers during the first six months. A failure in the brake system of a superfast train was the cause for its accident with another train resulting in death of 345 persons and hospitalization of more than 500. These are just a few examples that demonstrate failures in everyday life. Failures are a way of life in the modern technological world and the2 Retabiey Enameeing rep opin te en a ig suey ei te ete erred ble, Better understanding of fuluces, improved manufacturing techniques, ea a rl cr ee ee roca ake amma ase roe cama ita te ace eaten Sar ene Sy sun hematplatan Someta ck rakes se Deer htt ae orihe slaps, 4.2 RELIABILITY AND QUALITY ‘The concept of reliubily sas old as man himself. He bas long been Concerned with the problem of warehabiy of the products he wed, Will i tortion satisfactorily? Wil “that” lat long? Thee are some of familiar questions he has been aking all long. Bat its wince i new ands, {Sif under formulation and promt Relay ins simples form, means the probably that a faire may not oor na given tne interval. A more iprous efintion of raablity ‘os fllons: "Reliability ofa unit or prodae) isthe probability that the ‘it performs ts intended function adequately for a given period of tine ‘onder the stated operating condilons or environment. By 2 unit we mean fan element a system ora part ofa ajtem, or te ike. The elabilty df tion test four elements namely 1. Probabilty, 2, Intended fnvon, time, and 4. opertng condone, ITs the time lhe fae of the ult (a random variable) occurs, them the probability that wl not ulna given eniconment before tine (ot i eek) isBasic Concepts of Reliability 3 RO=P(T > 1) (1.1) Thus, reliability is always a function of time, It also depends on environ- mental conditions which may or may not vary with time. Since it is a pro- bability, its numerical value is always between one and zero, i.e. R(0)=1, R(o)=0 (1.2) and R(t) is a nonincreasing function between these limits. Quality of a device is the degree of conformance to applicable specifica tiots -andWorlamanstip standards: Ti not comcemed each the elements of time and environment. An equipment which has undergone all_quality tests may not necessarily be more reliable, Quality is associated with the manu- faeture_whereas_reliability is primarily associated with the design. In a way reliability is the ability of the unit to maintain its quality under specified conditions for a specified time. The “intended function” of the device is related to the quality. If an elec- tric generator is expected to provide an output of 5 kW at 220 V under certain conditions then the 5 kW output at 220 V is the intended function of the generator. Any deviation in either of these two output parameters will be termed as the failure of the generator. This may happen today or any time in the future. Environmental conditions such as temperature, humidity, vibrations, etc. affect the quality and in turn can cause the failure of the unit under opera- tion. Any change in the operating parameters might also cause failures. Another important distinction between quality and reliability is that one can build a reliable complex system using less reliable elements but it is impossible to construct a “good” quality system from “poor” quality ele- ments. For example, the reliability of a component can be increased by adding one or more similar components in parallel. 4.3 FAILURES AND FAILURE MODES A failure is the partial or total loss or change in those properties of a device (or system) in such a way that its functioning is seriously impeded or comple- tely stopped. Some components have well defined failures; others do not. For example, switches and electric bulbs have well defined failures. Either they are good or bad. Such devices are known as two state devices. On the con- trary, devices like voltage-stabilizers, resistors, etc, have a range of operating conditions. For instance, the output voltage of a stabilizer might lie within the limits of V+» and V—», The device is supposed to have failed only when the output voltage crosses either of these limits. In any case, the con- cept of failures and their details help in the evaluation of the quantitative reliability of a device. Many years of experience of failure data of various devices has shown that the failures, in general, can be grouped into different-1 -modes dependingetary Enginatng pan the nature of failure When we put a lzge collection of unit into operation, its Hkely that there area Tage number ffs intl. The ceil files ae called ina fare or ff matali. These flores te prinarily due to manufacturing defeas, such a5 weak pars, poor isu fion, bad asiembly, poor Bis, et. Since the defective unit re eliminated during thei atte peviod hs period is known as the debaging or amin peo. ‘Ae inital ares, fr along peiod of tine of operation fever falls are reported bat tis ditieult to determine their ctse. They ocear dv 10 the sharp chaage in paraiters determing the performance ofthe uni, cither asa esl ofthe chang in tbe working seses or environment con: ons Ba, ether eat, iis dealt to peediet the amplitude of ses ‘avtions and their time of ocuerence thos the fllres daring this period ae often called rendom fueron caranrophe flies. Ths i the period fof potmal operation and is chaacerved by (approximately) constant umber of fatres per un time, ‘As tie pases on he waits get outworn and begin to detrorate. A ‘radsal change m the values ofthe pazameers determining the performance of the unt rests and when these parameters go beyond the iit of a+ ‘istibity the unt fall This region scaled th weor-or region. The changes fe ieveable pysco-hemical in nature andthe prognosis of wear-out Tatlres i very diffe I this period the fllate rt increases ‘Atypical carve depicting tote thuce modes of bebaviout 1s shown Fig. LI and iti often elled the Barua carve in the relabiiy-berature (a6 ts shap). !Basic Concepts of Reliability 5 Many of the manufacturers who produce high-reliability components subject their products to an initial burn-in period of 11 h to eliminate the products having initial manufacturing defects, The period t2~ 11 is the useful life time of the item. After #2 h of operation, the item is to be replaced by another pretested item. 1.4 CAUSES OF FAILURES AND UNRELIABILITY 1.4.1 Causes The specific causes of failures of components and equipments in a system can be many. Some are known and others are unknown due to the comple- xity of the system and its environment. A few of them are listed below: . Poor design (component or system), 2. wrong manufacturing-techniques, 3. Jack of total knowledge and experience, 4, complexity of equipment, 5. poor maintenance policies, 6. organizational rigidity and complexity, and 7. human errors. 1.4.2 Design, Production and Use Poor design and incorrect manufacturing techniques are obvious reasons of the low reliability. Some manufacturers hesitate to invest more money on an improved design and modern techniques of manufacturing and testing. Improper selection of materials is another cause for poor design. Components and equipments do not operate in the same manner in all conditions. A complete knowledge of their characteristics, applications, and limitations will avoid their misuse and minimize the occurrence of failures. It is equally important to have sufficient knowledge and information about the environment (and its behaviour) where the component is being used. Often, this information is obtained from past experience, But before we have the time to experience, analyse the cause-effect relations and apply the re- sults, we launch new products and systems; our mistakes and ignorances continue to dominate their life. Many engineers and production managers say “we have never had a failure—just a few malfunctions”. They ignore the fact that any defect which can prevent successful accomplishment of a Mission is a failure. All failures have a cause and the lack of understanding these causes is the primary cause of the unreliability of a given system. 1.4.3 System Simplification In many cases a complex and sophisticated system is used to accomplish a task which could have been done by other simple schemes. The implica- tions of complexity are costly. First, it employs more components therebydecreasing overall rebbility of the sytem. Second, a complex scheme sens problems In lem of wir understanding and maintenance. Even fn enerienced designer wil ak time to comprehend all the process an Interactions and is key to commit makes during the design and deve lopment sage. On the other hand, snl costs ss, causesTesprobens, and has more reliably. A tase rule of elbilty with respect t0 com ‘lexty i Keep the system as simple sis compatible with the performance requirements. 1.4.4 Maintonance “The important ped in the Ue ejele of @ product ora system ists ‘opera Ing’ pried. Since no product s perfect its key to fai. However is ie time Ean be incensed if ean be repaired and put lato opeaton again. Tn ‘many ees preventive mentres are posible and a jsut designed pe ‘entive-masatsance policy cam elpeimioate Failures toa lage extent. The ‘Mage “prevention i beter than cue” apis to prodecte and equipments swell. 1.4.5 Communiestion and Coordination Rely is a concen of mot all departments of an organization. Ii ‘esetily a byntvto-death problem involving such areas as raw material and ‘pas, conceptual and detailed engneeing design, production, et and quae ity contol product shipment and. tora, installation, operation and rmineranos. A welborpanized management ith sa ecient stem af om tucieaton is equzed to sate th information and experiences about com. ‘ponents Saflelet opportunity shoud be availaie for the people concerned fo discus the causes of alae, In some organization, rig of ules and Drocedates peahibs the eeaiethinking aed desin, 4.4.8 Human Rolisity Jn spite of increased aplicction of automation techniques in mdustis and cher erganuations, it impose to completly elinate the bum it- ‘volvement fn te operation snd maintenance of systems. The contribution (orhumanrtors (othe unzetiulty may be at vatous sags ofthe product ‘le Palaces di to the heen can be det! 1, Lack ef understanding ofthe exigent, 2 lack of understanding ofthe proces, 3 excess, 4 orgenes, 5 oor judgemental sis, absence of coretoperting procedures and nsretions, and physical nabs. Althoush i not posbeto eliminate al uman-ertors, iis possible to risimize tome of them by the proper selection aed (ining of personel, Standardization ofprosuures, splifcation of contol schemes and thetBasic Concepts of Reliability 7 incentive measures. The designer should ensure that the operation of the equipment is as simple as possible with practically minimum probability for error. The operator should be comfortable in his work and should be free from unnecessary stresses. The following checklist should prove useful to the design engineer. 1. Is the operator position comfortable for operating the controls? 2. Do any of the operations require excessive amount of physical effort? 3. Is the lighting of the workplace and surrounding area satisfactory? 4. Does the room temperature cause any discomfort to the operator? 5. Are noise and vibration within the tolerable limits? 6. Does the Jayout ensure the required minimum movement of the operator? 7. Can the operator's judgement be further minimized? With all this cate, human operators are still likely to make errors. A human-error may or may not cause a failure. Consequently, the quantitative measurement of the human reliability is required in order to present a correct picture of the total system reliability. Human reliability is defined as the probability that a job or task will be Successfully completed by the personnel at any stage in the system operation Within the required minimum time (if the time requirement exists). Note the similarity of this definition with the definition of an equipment! Mathematically the reliability of human perforitiaiice cant be expressed as Ria 1-Qi=1-FiPr (1.3) where P; is the probability of producing error i, Fis the probability that failure will result if error i occurs, and Qj is the probability of failure as a result of error i, If two errors must combine to produce a failure, then, Pr= pip aa) where pi and p2 are representative probabilities of the two errors. When the system is subjected to # possible human-errors, then the total probability of failure due to human-errors is given by Qr=1-Rr=1-HU-99 5) To apply this model, estimates of P; for various kinds of errors and F; for various kinds of hardware are needed. 1.5 MAINTAINABILITY AND AVAILABILITY No equipment can be perfectly reliable, in ‘spite of the designer's best efforts. The equipment is likely to fail during its operation which might be costly in terms of money and time or sometimes dangerous in terms of safety. Maintenance, therefore, becomes an important consideration in the4 Ratatity Enainoeing long-term performance ofthe equipment. The equipment requis prveane ‘malntenanee (0 avold any posse failures) and clinatlon of fares te ing its operation (when they oceu:) Maintainabiiy i a peefozmance indie ‘sociated withsuch equipments orsjtens on which maintenance operation is performed. Maintinabilty can be defined asthe probably that a failed equipment is restoced to operable condition ia a spelfed time (called ‘dovavtine) when the maintenance f performed Under lated conditions characterizes the adapabity ofthe equipment othe detexton and elimination of fae tes as wll as thee prevention jow often an espment fis (elbiliy) and how fong i is down ine tsioaity) are tal eonsigcation in determing it worth. These two con caps have clove Tine with the cst, eomplenty, weight, and operational ‘equteents, The more relsble an equipment and beter is maitainabliy, the reer it ails and the shore it downtime. “Ausabity is anther measure performance of rsiatsned equipment 1e iterates both relabilay and malntanabiey paromecers and depends ‘the numer of flares that orcar and om how quickly any Tals are fected. The Tongan c steady-state avalsblty is defined es the propor- tion ofthe ime doing which an equipment is valsble for use. It can be expressed as Upsine Uptown 09 ‘The denominator is equal to the total time for whch the equipments required to fonction sod the upiume ie the actual period for wbich the ‘equipments avaiable Por use. The downctime ean pefud, In addon to Sve epartine, administrative aad other delays related to reps, Asnilabity= 1.6 HISTORY OF RELIABILITY ‘The growth and development of relabty Have strong Finks with ably contol and its development inthe st cuartrof his entry. the 19205, f team of workers m Bell Telepooe Laboratories Geveloped stated saathods to solve suse of thee quality control problems The provided the bass forthe farther dovlopmess inthe atea of statstieal quality coat! (QO). Subsequently, the Arverican Society foe Testing and Mater the ‘Anvrican Standards Association, andthe American Sovely for Mechasicl Engineers joined Bel Laboretones ta populating the quality conto tech ‘5s However, the rte of adoption of these techaigues among the ete ries wae vey slow bl Works War If Broke out in 1339 "The adage "Neowsty is the mother of fevenuon™ is true In the ease of telabilty foo. The Importance of reliably and quality ental was born ‘ut ofthe demands of modem technology uted i the Wold War I, Com Dlsty and automation of equipments tse in the war rested in severeBasic Concepts of Reliability 9 problems of maintenance and repair. Failures of equipments and compo- nents, particularly, electronic tubes, were more than expected, During the war the army and navy in the USA set up a joint committee known as the Vacuum Tube Development Committee in 1943 to study the failures of vacuum tubes which was considered to be one of the focal points of trouble. Quantitative techniques for reliability measurement were evolved and intro- duced. During the decade following the war many research laboratories and uni- versities initiated studies on failures of equipments and components. Bell Laboratories and Aeronautical. Radio, Inc were the two leading organiza- tions among those who contributed heavily in this area. Practising engineers and mathematicians took interest in the study of life-testing and reliability problems. The first major committce on reliability was set up by the US Department of Defence in 1950. This was later called the Advisory Group on Reliability of Electronic Equipment (AGREE). The AGREE published its first report in 1957 which included some reliability specifications such as minimum acceptability limits, reliability test requirements, etc. Since then many new organizations were formed to promote the concepts of reliability and quality among both manufacturers and users. During the 1950s other countries such as Britain and Japan began to take keen interest in the application of reliability principles to their products, In Britain, the British Productivity Council and the Institute of Production Engineers were independently engaged in promoting quality control concepts. » In 1961, the National Council for Quality and Reliability was formed with the main objective of creating an awareness of importance of achieving quality and reliability in the design, manufacture and use of products, ‘The last two decades have seen remarkable progress in the application of reliability’ principles in industries and government departments in almost all developed and developing countries. Today reliability has become a catch- word in many firms. Products carry reliability specifications, such as mean- time between failures (MTBF), failure rates, etc. National and international seminars and conferences on reliability are a regular features in most countries. 1.7 RELIABILITY LITERATURE 1.7.1 Nonmaintained Systems The mathematical theory of reliability has started receiving a great deal of attention from engineers and mathematicians. 1950s saw a large number of papers on component reliability, life-testing and failure distributions. The first significant paper to appear on mathematical models of multicomponent systems is by Moore and Shannon [1956]. This paper shows that by proper incorporation of redundant techniques a network of high reliability can be constructed using relatively less reliable components. Subsequently, severalpapers have been published on the reliahilty of redundant and majority Voting systems, Moskowite [1959], Mine 1959, Shoomsan {968} and Misra [1970] have described the use of eset approach to the reliably evaletion ‘of multicomponent systems. The cutset approach is discussed by Harlow ‘and Prosehan 1955, Jeon and Belesore [196] and Hanser 1971). Harley [1963] uses the protabiy-map method for geting simplified rea bilty expressions, Minra and. Reo [1570] bave presented a procedure for reliability evaltion of redundant networks using Rowe graphe which icone sidered as significant step forward in ehabily evaluation. Bening T1970} presen the eect of an increasing hazard rate on reabily and the mean IM of redundant stems. Banerjee and Rajamsni [1972] have tid to e- present probability in two dimensions and bave dersloped a parame Approach to analyte the refabiy characteris of general syste. Denis {1974} has tasted relbity foreuse for combined voting and standby redundancies and ha also provided many referees on voting ystems [03 rev paper on computertelabilty McClusky and Ogus [1975] describe the role of static, dynamic, and hybrid redundancy techniques in reliably modeling, Faulteee dagrans bave ban developed and extensively ured fot the reibiity an safety analysis of stems by Fosse [973] and Benaet 1995. 1.72 Mannainaa Systeme ‘The sluless of goodnes of a system is determined by its steady-state sailablity and fequeney of fires The concept of aality iswidly Gdscused la Tecate and the main contributors are Barlow and. Hustet 11960), Gaver [195], Sandler [1963], Myers {196d}, Datlow and Proschan [ug6s}, Rau {1970} and Beran (1974, However, the concept of fllure frequency fs not so well understood among rystemscegners. Mall ta [1968] ws Fequency of fares Tor pomersytem resblityexeubstios. Singh and Rilington{1975, 1975 supgetrethods for determining the fre- ‘quency of fallues of complex systema, The avaably and fale feqoeney ttepaiable redundant stems bave been discussed by Halagvisany and ‘Misr (1976, Buzacot 1970) has sugested sore speci teciies sch a ‘sate lump’ and system decompostion’ techniques for the aaljis of smainsind sjsems, Preventive and repaiewalatenanee policies are dete with by Barlow and Hunter (1960), Sander {1963}, Barlow and Proschan {0963} and Ra (197), Ralagurasiny 1977 hs proposed an opinal maine tenance policy fr redundant systems with dependent wits, 41.7.3 Tichotomevs Systems ‘ichotomous ystems posessing components with two modes of flues, ramely opencreit flare and shortelreui fae, have also been di ‘ised in Meratre, Creveling(1956] was one ofthe fist few’ authors to - ‘oss the double flues and. develop reinbiy and flare equavos for diode quads. During the following yeas Lipp (1957), Bariow 11960), PrinceBasic Concepts of Reliability 11 [1960], Polovko [1968], Henin [1972]. Proctor [1975], Dillon [1976] and others have contributed to the analysis of three-state device networks and systems. 1.7.4 Systems with Nonidentical and Dependent Units In analysing redundant systems most of the contributors have assumed that the redundant units are identical and statistically independent. These assump- tions do not hold true for many practical problems. Studies have been conducted to establish laws governing the failure rates with respect to work- ing stresses, such as voltage, current, force, torque, etc. Grange [1971] has discussed some of the stress-models represeuting the variations of failure rate with the thermal and electrical stresses, Shooman [1966, 1968] considered the problem of dependency and derived reliability expressions for a two-unit, parallel system using the joint-density approach and Markov models. A reliability model for 1-out-of-m system with linear stress-dependent units has been proposed and discussed by Sandler in his book [1963] but in a Jater study Mista and Balagurasamy [1975] disprove this model and its conclusions. Ninomiya and Harada [1973] have treated the dependency problem by the method of correlation and obtained correlation-coefficient assuming that the failure rate of units of a parallel-system increases linearly with the num- ber of failures. Yang and Lin [1974] have applied the concept of equivalent time to analyse a two-unit parallel system. Balagurusamy has considered the stress-dependency in detail in his Ph D work [1976] where he has proposed different approaches for evaluating the reliability and other related functions of redundant systems. He has also considered systems with unequal failure rates. The probabilistic behaviour of hierarchical systems that are encountered in industries and other areas frequently, have been analysed in detail. Hevesh [1967] and Muegglor and Martin [1973] have used nested binomial-equations for the reliability analysis of systems employing equipment hierarchy. Balagurusamy and Misra [1976] have considered m-order hierarchical systems and proposed models for the cost effective analysis of the systems. EXERCISES 1.1 Explain the importance of reliability in everyday life. 1.2 Distinguish between reliability and quality of a product. 1.3 What do you understand by the term “intended-function” of an equipment? 1.4 What do you mean by the useful life of a component? 1.5 Draw the failure rate curve for an industrial product and explain its shape, 1.6 How does the operating stress affect the failure rate? 8 1.7 State and explain the possible causes of low reliability of modern engineering systems. foes 1.8 List some of the environmental stresses which may reduce the reliability oft (a) An electronic measuring instrument, (®) an electronic computer, and @acar.12 Rel bry Enginesring 19. How doe taman or af hee lai? to Dette oi tems eRe, @ecmarbanys and (natn 1.1 Compre ad cosas the ftom {Redan aod wut ond (@hretaty ond muna. 112 From he da sate tour be, ete the ly ofthe ona: (Youre )yeut io ec TV, (hjourtetwuerent {your refraor ane le goar sate pacer 1s Youn reqed ones a dig frit Gof an eens tcc wo fo mate soir ste fheushou they. at ae Beatty om tari Scents rtocucea aby i oe G)Beare penser {Ghereone apeteedr and Weomenier ‘Sefacon wich 00 conde ol ath () ups aed i donee, FURTHER READING 14 tants, ta Refit They and Pre, rete Hal Eapeoo$ Ci Engencog Cu 2 1 Sande Gt, Shes Alley Exe, Pee, Eylenoed CM La ARINCRewash Coporaton Really Enpovere, Pence al, Eelsond Ls Rotene, NAL, Manematal Neds te Ray Pebeena, MeCrme New vor he NnsBt EL ors and HLM. Goh, Relay Pantie for Eecrove ‘Seon Jah Wey, ew Yor, 6 17 Buln Rivand Fontan, Mehematcal Teer of Reta, Joba ey 15 Ireon, WG, Rllbity Mandbook, NeGtas-Hll, New York, 196, 13 Wooo, Atte Fondament of Retitiy Ther eaten Prey New Yotk 119 Shooman AL, Peedi Retain: As Eyres Aprach, MeGree 1 Koti BA RAIEA. Ubi Rtalty Rebeca sod Wao 112 Kae O., Ornate and Ptaity In Systems Eber, Yoo Nos Rel Company, New York tay Arras Ds Riutiy Maman Meco, Ne York, 1A, TM Ginn, Atsand AD Bowe, Atay Tebuty, Wicy Toone, Lande 11S Balsgurseamy, 'Relatiy arabs of Afondet antes with SArpeodeat and Fonidwatenke, 2 Per, Dery ot Race, Ta,2 Design for Reliability Constraints and Considerations 2.1 RELIABILITY ANALYSIS A system designer, while planning and designing new systems with reliability as one of the design parameters of the system, is often confronted with several conflicting problems. Owing to increasing automation and com- plexity of tasks entrusted to various subsystems, they are often composed of an ever-increasing number of elements (components). This, in general, leads to a decrease in the overall reliability of the system. At the same time, the increasing importance of the tasks carried out by such systems imposes stringent reliability requirements. Resolution of this conflict necessitates a careful investigation of many facets of the problem of increasing the system reliability. Moreover, an increase in reliability cannot be obtained without investing money and achieving this objective requires both definite material expenditures and systematic and scientific analyses, The essential tasks of a reliability analyst would be: 1. Evaluation of system reliability and safety characteristics, 2. comparison of specific characteristics of alternative-designs objectively, and 3. location of weak spots (or subsystems) in the design and estimation of their contribution to the system unreliability. The above considerations need a thorough reliability analysis of the system at the design stage itself. In order to conduct an effective and effici- ent reliability analysis the following aids and tools should be available: 1, A concept and procedure for estimating system reliability and com- paring alternative designs, 2. a set of generally applicable mathematical models, and ; 3, efficient computer programmes to evaluate reliability models numerically.Since the relisbility Ggureofsmerit which fe consdered at 8 means 19 cevsuate the elletvenes of system ean be a he mont only es tsetmothy 25 its bass, a systematic and. ateuate procedure Is most impotent for ‘earning outs parposfa celiabliy analysis. ‘The quantitative analysis usually starts from the phyicol model, i esription of tbe fonction reltonsip betwen various subsystems and ‘components andthe expected fale mechanism for which tathematia sodels are develope, Io order to establish a reasonable physical model, ‘ualitstive analysis should be carried out. This ena the decomposition ot the gartcuae system into subsystems and components and determination of the pouble component and system operating states comstent with he rebabilty goals and criteria 2.2 MATHEMATICAL MODELS AND NUMERICAL EVALUATION For quantitate ana, appropriate fare merit and the factory that re responsible for cotsibuion tbe carly dene, The fgurevofemeit ‘shh are determined primary by the mision ofthe system, n general, atermune the system philosophy with regard to the use of the redundant fequpment and maintenance poiy. For noamainsied systems, the para ‘eters of erst are the probaly of the survital Teectio, the mean Tensth of te the systems tae (each the ale state forthe ft time, snd the system lore rate "The chove of rlaitty-measures for ravesinad yur requites the consieration of whether the main penalty or cost ofthe system fares epends 00 1. The total duration of fares, oF 2. the fequency of slr. 1 the total duration of fturessimnpotent, then the sppropriate measure sul be related to the svallabity of the system, he Fequeney of flues islimpoctan then the appropiate mentre wl be elated to the sytem tmean uptime or mean downtime, The importance ofthe mean uptine {or mean dovnaine) isles since avalabiy i cootered. with the total time the stem spends in sceptle ates aed does nt indicate how ths Aime is dstnbuted. For exami, a petiod of 1000 days the system may fl twin and be down for 3 days ead tine, yielding an avalabiiy oF 0.9. (Oa the other hand, the system may ail ve ies in the sme period and te down for 2 dsszach tine, azo yielding an avalabiity of 09, Some- times, the meant f0 Grit sem flare ean bea eaniafulmrasre 36 it indies the amount of tie the sytem can spend wih fll eapacity ‘efore the Fest complete system fae. ‘The factors that mainly contribute tothe above measures are: 4. The component flare process, the stem sractre,Design for Relis 3, the maintainability and its policies, and 4, the states in which the system is defined as success, A general procedure for the reliability analysis is as follows: 1. Definition of the reliability goals and criteria, 2. determination of physical model, basic component events and repair Tate, 2 3. collection of component and subsystem characteristics with respect to various stresses such as thermal, electrical, etc., selection of methods and mathematical models in view of (2) and (3), numerical evaluation of mathematical models, comparison of results, and development of alternative design if the reliability goals are not reached. mas It is obvious that in order to follow the procedure suggested above, a set of “appropriate” mathematical models and computer programmes for efficient and meaningful numerical evaluation must be available. Also, to develop alternative designs, various means of increasing the system reliabi- lity and constraints associated with them must be known. The models should be as far as possible more general and applicable to systems with: identical and nonidentical components, independent and dependent compo- nents, constant and time-dependent component failure rates; system and component redundancy, and repair and maintenance policies. 2.3 DESIGNING FOR HIGHER RELIABILITY A number of techniques are available to enhance the system reliability. Some of the important methods are: 1. Parts improvement method, effective and creative design, system simplification, use of over-rated components, structural redundancy, and maintenance and repair. In the parts-improvement method, either the reliability of all the consti- tuent components are improved or at least the most critical components are identified and their reliabilities improved. This involves application of im- proved production techniques and automation and, therefore, is a costly and difficult means of achieving the reliability. Nevertheless, it is quite effective up to a certain point. Since the production of a perfect component is almost impossible and the cost of part improvement is very high, the approach becomes unwieldy when one deals with large and complex systems. The effective and creative design approach needs some thinking on the part of the design engincer to create a new or improved circuit or system with SuwaywyDeter eiabilty. For instance, a propee design of protective schemes for shyt ean grealy enhance tbe reabty and ean life of thy comecors. Howerer, sill it may not be posible to elute flores completely. Ti cases whet the systems are poorly designed and only comples, the proper ue of the componens and redicing the complexity ean prove tobe fn important technigae for improving system relsiiy, However, over Simplfctios can lad 0 poor qualty and efciency ofthe syste. Failure atesof almost allcomporens change with ther operating street and, therefore, con be reduced sinicanly By the application of overrated ‘components. The depce of umprovement depends upon the (pe of com ponents. Although itis ofen possible to trade money and other resource {buy overzaed equipments, appheation ae obviously limvted by the ‘abi of equipment wit the reqired rating, ‘Stoctral eeduedaney provides avery efecive means of improving seme relly. Tht Involves dupheaton of paths at the component oe ube _pstom evel and appess fo be the only slation when high elit aor overrated components are not available Maintensace and pars, wherever posi, undoubtedly boot the system rehabilty. A maintained sytem when combined With redndaney may have 8 relay of almost one. 2.8 REDUNDANCY TECHNIQUES ‘Applicaton of redundancy in the system design i found ia almost al pes fof tstems due to ls nunerous advztoges over the other methods of tprovng system relay, The mporant ones ae enunratd below: 1 Any desired tee of ekablity ean be achieved I the avaiable re- Ssoures pom. 2, Frees in teil yer unt rexoute spent Right when eptinal redondaacytechniqes are employed 3. design through redundancy needs comparatively les si on the part of designe 4, provides a quck slauon, and 5. thismethod can be employed In the event of faltre of all eter rehods “The development of mnitorizaton techniques has made possible the pplication of redoadaney eater convenient n space vehicles alo, ns Allowing am intense thee ie ‘A few examples of sich spsems of subsystems where redundancy i cestesely used are 2, Interconnected poser systems, 2. poterhe system for mudesr actors,Design for Reliability 17 . aircraft propulsion systems, . satellite communication systems, ignition systems for rocket engines, . temperature control systems for space vehicles, and , data-processing systems. Various forms of redundancy—active (hot) redundancy, standby (cold) redundancy, warm redundancy, component redundancy, system redundancy, hierarchical redundancy, etc.—can be employed in a system, depending upon the feasibility. One has to select a suitable form considering such factors as the type of components, type of system, reliability requirements, resources available, etc. A redundant system having mt units connected in parallel is referred to asan m-order system. The performance requirements of such a system may impose a condition that at least a minimum of k of its m units should be operational for the system success, These k units are known as basic units and the remaining (m1 —k) units are known as redundant units (added for the purpose of increasing the system reliability). Such systems are classified as k-out-of-m systems. Series (k=m) and parallel (k= 1) are the special cases of a k-out-of-m model. Examples of this type of systems are as follows: 1. In an eight-cylinder automobile it may be possible to drive the car if only four cylinders are firing, but if less than four fire, then the auto- mobile cannot be driven. 2. In acommunications systems with three transmitters the average mes- sage load may be such that at east two transmitters must be opera- tional at all times otherwise critical message will be lost. 3. A four-engine aircraft needs only two engines to perform critical func- tions. 4. A bridge supported by 1 cables may require only r cables to support the maximum load, In incorporating redundancy into the system, the main problem is to determine the value of m and k under certain constraints of reliability, mean life, cost, performance, etc. MAMA 2.5 EQUIPMENT HIERARCHY Equipment hierarchy is a special class of redundancy which is often found in systems containing many stages in series, such as ¢ommunication, data processing, electrical power distribution, and many industrial systems, In all these systems, the operation of several equipments at the Jowest level of the hierarchy is contingent on the operation of one equipment at a higher level which is in turn contingent on an equipment at a still higher level, and so on. Thus the equipments of a hierarchical system are not statistically independent.18 tabi Engine 2.6 RELIABILITY AND COST ‘The cot of various methods ofeeievng the elit wil vay eccording tothe follow 1. Types of components, 2, eottof maimenane, 3. asestibity ofthe product forthe rsinterance, 4. time and manpower avaiable forthe desig, and 5 consis sacha weit, volume, ee A costatetineness analysis may be requted before choosing one parle colar method. The eect of incteme inthe reinbliyon he cont for any roduc i shown in ig. 21, The iii oat icreates butte operating cost ecreates withthe reibily andbenee thee existe a vale of the reaity for which the coe is minim, Fg, 24 Relay anon Design and development cost wil incense withthe iaceaed teal’ ‘sez of the need fo be more crits nthe dvgn and be need fOr more txts testing ofthe egupment. Production costs also intease with the Frproved relly, with the wee of beter eomponents and closet contol ff proseres, inspection 2nd test procedures, On the other hand cos of ‘ralaenance andreparfilland aloughia many asthe ony pat ofthese {hat devolve en the mannectre are those thst occur drig the guaranteeDesign for Reliability 19 period, nevertheless good reliability represents an intangible asset in the way of continued or increased sales. The development of methods of calculating reliability and expenditure factors is one of the more pressing and important problems in industrial economics. EXERCISES 2.1 Explain the functions of a reliability analyst in an organization. 2.2. State the importance of quantitative techniques for reliability analysis of compo- nents and systems. 2.3 Describe the various approaches available to increase the reliability of: (a) A single-component system, and (b) a multi-component complex system, 2.4 List three engineering devices or systems where redundancy has been employed to improve the reliability. 2.5 What are the advantages of using redundancy over other methods? 2.6 What is meant by a k-out-of-m system? Describe the behaviour of the system when (@) k=m, and (b) K=1 and m is greater than one. 2.7 Compare equipment hierarchy and equipment redundancy. 2.8 Describe the various components of the reliability-cost curve of a product. 2.9 Give one practical example where reliability is of paramount importance. Give Teasons. 2.10 Discuss the following issues: (a) Reliability versus weight of the power system of a space ship, (b) reliability versus safety of a locomotive engine, and (©) reliability versus customer satisfaction of a TV set. FURTHER READINGS 2.1 Lloyd, D.K. and M. Lipow, Reliability, Management, and Mathematics, Prentice- Hall, Englewood Cliffs, 1952. 2,2. Myers, R.H., K.L. Wong and H.M. Gordy, Reliability Engineering for Electronic ‘Systems, John Wiley, New York, 1964. 2.3 Dummer, G.W.A. and N. Griffin, Electronics Reliability—Calculation and Design, Pergamon Press, Oxford, 1966, 2.4 Shooman, M.L. Probability Reliability: An Engineering Approach, McGraw-Hill, New York, 1968, 2.5 Smith, C.O, Introduction to Reliability Design, McGraw-Hill, New York, 1976,