Scribd
Upload
117 views66 pages

Reliability Given Notes - Compressed

reliability statistics notes

Uploaded by

Sheeba
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
117 views66 pages

Reliability Given Notes - Compressed

reliability statistics notes

Uploaded by

Sheeba
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
You are on page 1/ 66
wr 432 "Probability, Statistics and Queueing Theory istics of a system strongly depend on the design and topological layout of the subsystems in addition to the reliability characteristics of indi. vidual components of the subsystems. Jn the broader sense, reliability is associated with dependability with successful performance and with the absence of breakdown or failures. If a component is put into operation at some specified time and observed until it fails, then the time to failure or life length say 7(20) is a continuous random variable with some probability density function (pdf) f(t). The value of T cannot be predicted. If we use identical components under identical stress, still they may fail at differ- ent and unpredictable times. The failure is not uniform or constant. The word reliable or reliability is used in the qualitative sense. Wow- ever, from the engineering analysis point of view, we are in need to define quantitative approach to the qualitative description of reli- ability. Definition. Reliability of a component (or a system) is defined as the probability of the component (or the system) performing its pur- pose adequately for the period intended under given operating condi- tions. * In other words, reliability is defined as the probability that a given ” product or a system will successfully perform a required function with- out break or failure under specified environmental conditions, for a specified period of time. That is, the reliability R(‘) of a component (or a system) at time /, is defined as RO =P(9, where 7 is the life length of the component (or the system). RY) is called the reliability function of time t. ~ The above definition means that the reliability of a component equals the probability that the component does not fail during the interval (0,/) (which means that the component is still functioning at time #). For example, R(4)) = 0.99 means that approximately 99 per- cent of the items of the system, under given conditions, will be func- tioning, without failure, at time t). In terms of probability density function (pdf) of 7, namely. f(t), we get Re= [£@ ax @ i we recall the cumulative distribution function, F(x), as F@)= P(X t(ime) Period Period period In the figure, jure, We see there are three time-zones. In the time- » zone (0,E), the failure rate is decreasing but it is high. This Period of ” Ww ah ° ail - - - bewury Boe Wher failure rate is decreasing is called ‘infantile mortality pe- if £arly life period’, ‘debugging period’, or ‘break-in-period’, a) estes due to Lae manufacturing defect, poor design, poor rs in assembly or even poor knowledge of. operation etc, n€ Second time-zone (E, A) is called useful life time-zone. Here, the failure tate is approximately constant throughout this period of time. uring this period, the failure is random and unpredictable. From A. onwards, in the third time-zone, the failure rate increases and it is due to aging of the system. Our aim must be to reduce the first zone, early life period, by proper design, manufacture etc., and increase the middle Zone by reducing the third zone, the wear-out or aging period. Properties. The reliability R(¢) or probability of survival has the properties, @ O(to +t) /T >t} P {T>(t,+t)} P(T>t) Ro +t) R(&) tort - J z@a e* - -fzma es {Psa }eoa] =el? ° tort - J z@de ime» Note 1: Some authors use failure rate A(t) and hazard rate H(), without any difference. Note 2: General equation for failure-time distribution is given by t) dt ' FO=f (0) =2@€ 7 va ~an Reliability Engineering sah Number of failures during a particular unit interval Average population during that interval a Mean time to failure (MTTF): The MTTF is the average qr mean of the life distribution. That is, it is the average of life length of all the units in the population. The MTTF is usually used for products that have only one life, that is, not repairable. For products which are tepairable, the term mean time between failures (MTBF) is used to denote the average time between two successive component failures. Mean time to failure is the mean time to first failure Failure rate = = E(T)= j tf(0) de 0 =- [« RO) -fRae ] by integration by parts’ a E(T)= fr (de since R(0)=1, R (@)=0 and E (T) is finite o MTTF =&(T)= JR wae 3 Relation between MTTF and MTBF. Suppose there are n com- ponents in a system and m,is the MTTF of the i th component. Ifeach of the component is replaced (not repaired) on failure, then 1 oifB ag 1 Wrer 72 Dorota tee t+ MTBF om om om) m Mlustration of MTTF and MTBF . ; ‘MTBF is also a trieaSure of the average time of breakdown until the device is beyond repair. " — . A hy \ \ r ws a . Probabili isti ; ability, Statistics and Queueing Theory } Operating Haase 5, a eT Time : ; In the above figure, the time intervals ‘a,’, represent the times to failure of the system once it is put into operation. Then, by definition, 2 MitF=1 ya n kel On the other hand, observe that the time intervals (b,+a,) repre- sent the:times between two consecutive failures. Then, by defintion, - MTBF=1 n . Where == 4 represents the average repair time. fa en : Hence, Mean Time between Failures = Mean Time to Failure + Mean Repair Time. Note 1: If the sytem is first operated with all new components, the MTTF and MTBF are identical. . Note 2: For the useful life (chance failures), MITF and MTBF are equal. Note 3: If the repair time is negligible then MTBF ~ MTTF. Note 4: For. constant failure.rate 4,, MITE =2 MTBF 63 Failure Distributions soaps pass OE However inteliability theory, it has been found that the following prob- Henaony caymcenny ability distributions are more appropriate for failure process. Exponential distribution —_ § x Weibull distribution I & ) ee Normal distribution |\° // ‘ St Lognormal distribution ~ Gamma distribution Se shall now derive the reliability characteristics for the above failure distributions. a 1. Exponential distribution This law applies to the second phase of the bathtub curve. If the time to failure T follows an exponential distribution, then its p.d.f. is given by = & r= x B for 120, B>0 - 0 otherwise (i) Reliability RO= [7 (ae i tee “gle B dt ~ Gi) Hazardrate 2(y= LO RO parameter which shows # that when the failure distribution is‘an exponential di ibution'with:pa- 1 o . 7 . rameter P the failure rate-at any timé is constant, ons to 3 A UEN IN wen, ag teat pete dears Conversel ; ly, when hazard rate 2(¢) is a constant =. pihen by using cm ON EE EES DE -f zd FQ=zHee We set 4 sO=fe 5 which shows that if the hazard rate is constant, the distribution is expo- nential. Due to this property, in reliability contexts, the exponential distribution is often referred to as constant failure rate distribution. (ii) MTTF (Mean time to failure) is the average or mean of the life distribution. _ ie., MITF = E(Z) = B (already derived in chapter 3) _ 1 Mean life = MTTF (or MTBF) = Hazard rate B (vy) Var (N= (vy) Conditional reliability = R (¢/t) = R(oo*t) R (to) 5 (+40) 4 = [-x0-¢ ) t e F W which shows that the time to failure of a component is not depen- * dent on how long the component has been functioning) In other words, the reliability of the component for the next 1500 hours (say), is the same regardless of whether the component is brand new or has been - operating, for several ious. Tis property is called the memoryless property of the constant failure rate distribution. ~~"Note: If the parameter B and R(#) are given, we can find ¢ the number of hours of reliable operation. Reliability Engineering 2. Weibull distribution : Weibull distribution is very closely related to the exponential distr SQ bution and its p.d.f. is given by fis {eB Bt ea® 30, a>0,-B>0 ) , otherwisé 1 The above f(#) can be written in alternative form by putting % = rm p-1 (1) as ro=8 (5) e (s) ; 0>0,B> 0, 20 where B is called the shape parameter and 0 is called the character- istic life or scale parameter of Weibull’s distribution IO ‘The Weibull density function Zt) ae) ad >t Failure rate for the weibull model Pr. o @ Reliability =R(D= Jy (oat “Bey oy = YY = fe *dy by putting (5) =y r @ B nev ee (3) pt LM _B(t Gi) Hazard rate =z (¢)= G6 (é) = (ii) MTTF=E(7) =a ¥ (+4) (from chapter 3) ry be 1 fodse— Qo =o 7 {i+e [ dz ( “4 \ e” % 3 2 . ry" \ ® Gv) Vr (1) =0 lF (Hs (+5}} (from chapter 3) 0 |r (i (a}] R t (v) Conditional reliability-=* (/t). = Reto) Rh) “ 1 lz rc) | mo G4) (8) Note: @ 2(t) is constant if B = 1 di) 2(d) is increasing if B>1 Gi) 2(t) is decreasing if 00,a>0,8>0 1 f= 4BE(e) ; 0 otherwise As in the above distributions we can find, “Ws RO= 2G ef 50 rectly from MTTF = o B and hazard rate 2(f) can be found out di LO RO _- _Bxample 1 One thousand products were put on test at time zero. Four hundred of them were working at 2000 hrs, 50 of them failed in the next 100 hrs and another 50 failed in the next 100 hrs. Find the failure rate at 2000 hrs and 2100 hrs. eae The hazard rate for the products at 2000 hrs. Number failing per hour following 2000 hrs = Z (2000) = ¢ ) Number surviving at 2000 hrs G0/i00) =0,00125 units/hr Z (2100) = 0) —o.oo14 unite/hr Note: Between 2000 hrs and 2100 hrs, 50 numbers failed in 100 hrs. Therefore number failed during, one hour is 50/100. (after 2000 - hrs). ‘ : Example 2. The following table gives the result of life test of 665 electric motors giving the number of failures in time intervals. Find the hazard rate of motors which are (i) 10 months old (ii) 20 months old. Time interval in months Number of failures 0<1<10 347 — 10<1<520 61 20<1<30 69 30<1<40 87 40<1<50 101 Probability, Statistics and Queueing Theory erof failures in 50 months = 347+61+69+ 87+ 101=665 _” Ramber of failures in 10 months after 10 months = 61 Number of failures per month (after 10 months) =. Number of surviving motors after 10 months = 665 — 347=318 Z (10)= Number failing in one month after 10 months ee Biter, NY MODE ‘Number surviving at the end of 10 months 6.1 = >=— =0.0192 units/hr 318 w . From the above hazard rate, we understand that about, 1.92% of the 10 months old motors will fail within the next one month. Similarly Z (20) = no, = 0.0268 This shows that 2.68% of the motors which are 20 months old will fail in the next month. Example 3 Determine the MTTF for a mission time of 1000 hours life if the test data on 10 such components gave items to fail as shown in table. [Component number [1 T2 )3 1/4 5 [ 6 |7 [8s [9 |to | | Time to failure in hours] 807} 820] 810] 875 900 | 837|850 | 790| 866| 815 Also find the reliability. | MTTF = ae . 5 ey i, since the sample is small -4 [807 + 820+ 810+875+900+837+850+ 790 + 866+ 815] =837 . Reliability of the component 2 (¢)= e* bs r 1 L : = where A is the Hatlute rate = MMTTF 3377 0.0012 failures/hr ~ R(1000) = e 7 ©9012) (1000) = 9.3012 Example 4 The life testing data of 1000 specimens ofa particular device are given below. Determine the MTTF of the device for an operating period of 10 hours. Reliability Engineering 4aaT Time interval in hours Number of failures during the interval T< 1000 0 1000< 7s 1050 = 175 1050 Mo Mell 1 i.e., 11 t> [0t which is true, as 1 > 0. Hence one year burn-in period will improve the reliability. 121 (+11)? Note: Observe R (1/1) = 0.8403 > R (1) = 0.824. @ \Example 7\The density function of the time to failure of.an appli- eis f(t) = Newreliability =R (1/1) = = 0.8403 ues » £>0 isin years. (i) Find the reliability function R(®) Gi) Find the failure rate a, (iii) Find the MTTF i - (Anna April 2003) Al . @ Here fO= G anne t>0 __ (Anna AP. 2004 ) i - = 32 i RQ) = fro dt= Sarr dt x 16 | __ 16 -|-Ae , +4 2 Gi) Failure rate 4. =22@ RG) @+4) Probability, Statistics and Queueing Theory MTTF= Ro at a Example e reliability of a motor is given by 2 neo-(1-£) L0st0. Compute the instantaneous failure rate at t and show that it is decreasing. 3 What is the reliability if the engine to operate ? Gii) Given that the engine has operated for 50 hours, what is the probability that it will survive a second 50 hours‘of opera- tion? (iv) Determine the design life if a reliability of 0.95 is desired. (%) Givena 10 hour wear in period, compite the reliability for a 50 hour mission. . 452 Probability, Statistics and Queueing Theory Instantaneous failure rate = Hazard rate = z(t) ee VO00TT [0.001 1 zGant@.., 2t 1 0.001 RW) e- VO00Tt ? which is a decreasing function of t Gi) R (50) =e PFS _ QV 9519 P (72100) _ R (100) P (T2100/T250) ==) Gi? ¢ ) P(T250) R(50) - VO00T 100 _e V.00T 100 ea Voor so = 0.9048 0.9512 =0.9512 (iv) Design life = the time to failure (t, uy) Ry) =0.95 oe = 0.95 —J0.001 t, = log (0.95) =- 0.0513 0.001t,= (0.05 13 t-= 2.632 hours R(t+ >) (v) R(t) a R(60) NOOO 0.2449 RE0/10)=—0y ee = 0.8651 Example 10 The early failure rate of a component is given by 2(t) = ae*, Determine the probability of survival of the component Reliability Engineering Example 10 The early failure rate of a component is rive 2(t) = ae. Determine the probability of survival of the componers from age T for a mission time ¢ hours, given that the component has survived upto age T. Reliability of the component during the early failure period is = -fz@e R()=e The probability of component survival upto age Tis 2 (edt. Rwaor) Strlasly the probability of component survival upto age (7+/) is Rene oo } The probability of survival of the component in the interval 7° to T+ 1, given that it has survival upto Tis _R a+) Example‘11 The failure rate zis given by Z)=0 for0 Per year. ; “0 40 @ P(x )=0)=2)=1-[P @+P (+P 2] > =1- 0.2865 : v125+ 03] = 0.1315 d (iii) We have E [X()] = 24 : wee ~. E [number of failures per year] = 2.= 1.25 Eample 18 The life time in hours of a component is a random variable X which follows a Weibull distribution with a=0.1 » B=0.5. Obtain (i) mean life time of these components (ii) the probability that such a component will last more than 300 hours. @ . For Weibull distribution, Mean = MTTF =E(D) a t) omg Tr liek nat eink ae i gen ss 3 | P A Reliability Engineering 459 =(01)* F @) = 100 x 2 = 200 hours . , @ Wehave Rw =) where a “+ SS R@seu? Hence R (300) = ¢ © 6" =0.177 Example 19 A system follows Weibull distribution with.a shape parameter (B) of 1.4 and a scale parameter (8):of 550 days and also the system requires to have a design life reliability of 0.90. Compute (i) mean time to failure (i) the standard deviation ii) reliability for 200 days (iv) the design life, if there is no wear in period. (v) the design life, if there is a wear-in period of one month in the beginning. 1 @ Mean time to failure = MTTF =0T (1-2) = (550) r (+4) = (550) I (1.714) . = (550) (0.91057) using Gamma tables @ wore lates) ~ [oe (a8) | “= 6850)" [ (2.43) -(r amy] 460 Probability, Statistics and Queueing Theory = ($50)? [ (1.43) 11.43 = (ra.7y)’] = (550)? [1.43 x 0.88604 x (0.91057)?] Standard Deviation = Var (T) = 563.73 days (ii) Reliability for 200 days = R (200) -f (29) = 0.7847 We have, p¢y R(200) = (iv) Let t-be the required design life for R = 0.90 #5] j {é RUG )=e = 0.90 64 ys ie, ~ (550) = log (0.90) =-0.1054 iG t, =(0.1054)'" «550 = 110,25 days (v) Let t, be the wear-in period min =o [-($34F (3) Lett, be the required design life wi i 30 uatt be thereq esign life with wear-in period of one month... Then t i fo 30, 0= $50, B= 14 430)" 7 39) 2. OxXp | -]|] + — =— me [ ( 550 ) “(35 |-0s0 z _{t.+30)" 30) ie, (2) - t, +30)!" -(7S5-] =-0.1054-0.0170 0.1054 ltt oath inctrminaiy m= 123. 3 hours Example 21 CO mpute the operations peri ‘ng normal failure law pith ET) = 90 hours an} ied standard d 2 cliability is G) 0.90 Gi) 0.95 5 Gi) 0.99. rnrsifthe required edoperations Herem=90 and o = wehave RO=1- 1 R= oC) 4 (34) =RO [since igh reliabili so to get high reliability, the operating (g) and also oo= 1-o@ ime must be. less than Z Probability, Staustics ana Wueuemny iieury @ ie, #(* 90- a t, = 90 — (1.29) 5 = 83.55 hours Gio ("= } =0.95 “ty = 78.35 hours Example 22. Wear out of a componént follows a lognormal fail- ure ‘distribution with shape parameter 0.20 and the median time to failure is 5000 hours. Find the following @ the MTTF and Variance Gi) the reliability of the.compenent-for3000-hrs. - Here s='0:20, ¢,~'5000‘houts ' ‘ @ mrtr= , {3} = 5000 x ¢ 92 = 5101 hours ” RO= t o (i) We have R(t) [rou where f(i) is the pdf of the lognormal distribution. K(3000)= "f (oar ie | j 3 3 1 j j } 1 7 } “Reliability Engineering aes = f (z) dz Z 1 log ( t } . Zs Pa 1 spf 3000 7 a3"! 500) ‘ " “ = j (2) dz ws =0.9946 — (fromnormal tables) : 6.4 System Reliability % Introduction a In the earlier discussions, we ha tant failure models (distributions) for erally, all systems are made up of parts and components assembled to perform a certain function. The f: failure of a part or component may lead to the failure of the whole system in some cases and may not in others depending upon the functional relationship of components. A]- though variety of arrangements and analysis are available for reliability studies of systems, we concentrate here mainly on three configura- tions only which are considered as basic in such a study namely (i) series configuration (ii) parallel configuration Gii) mixed configu- ration. The system may vary from simple (consisting of one or two components connecting in series or in Parallel) to complex (involving thousands of components connecting in Series or in parallel or in com- bination of both). One approach for analysing such systems is to de- compose them into subsystems of convenient size, each rey specific function. Reliahilities of subsystems are then éstimated and combined to determine the reliability of the entire’system using certain Probability laws. This approach requires a complete knowledge of the Physical structure of the system and the nature of its functions suffi- ciently well to determine the behaviour of the system in the event of failure of a subsystem. The Teliability of the entire system is called as system reliability, 've considered a number of impor- r the individual components. Gen- he ccna presenting a paar tnt erate entree tanta tartetonte Series Configuration (Series Systems) A series (or non redundant) system is a system in which the vari- ous Components are interrelated in order to perform a required system operation in which the entire system will fail if one of its component fails or we can say that the components are connected in such a man- ‘er that the entire system will work successfully only ifall the compo- 460 Prohakite = Probability, Statistics and Queueing Theory on work without fail. ie., the information at the IN a reach the OUT end only ifall the components function success! ly. (We discuss the systems for n units in two ways. \(@ units dependent on each other other. series as i) units independent of each n’ unil Consider a system with * its which are connected in shown in the figure. System vale Let Xj, Xp) ++) Xn (events) be the successful operation of the individual units. P(X,), POG)»---» POX) be the probabilities of the suc- cessful operation of the individual units.)For the successful operation of the,system, all the ‘n’ units of the system should function success fully. {If P(S) is the probability of: ‘successful operation ofall units in the system then, ~ P(S) = P(X, and X, and X, and ....and X,,) Case (i) _ units dependent on each other. If the successful operation of each unit is dependent on the suc- cessful operation of the other units, then%,-X,-.-.» X, are events that . dependent on each other. Now (1) becomes P(S)= P(X) P (X,|%) P (%| X42)? (%,| 44a) If R(S) is the reliability of the entire system RS)=R(X)R(XQ|X,) R(X XX2)--R (X,PXX2--Xaa) -() . Case (ii) units independent of each other a es susessahi operation of each unit is independent of the suc- cessful operation of the remaining units, then X,, %; i pendent events. e , pela rea re es 2. (1) becomes P(S) = P(X,) PLX,) «.. PCX,)- Reliability Engineering je, RCS) = ROX) RUG) -- RO) IER(d denotes the reliability of the entire system denotes the reliability of the i component, then RAO = Ri RaQ --- Ru: R= tT RX) Hence we can say that in such series systems, the reliability of the system is equal to the product of the reliabilities of the independent components. This case is called product law of reliabilities (system connected in series) and it can be statéd as “f'n components, functioning inde- pendently, are connected in series, and if the i component has reliability R{t) then the reliability R,()) of the entire system is given by RO) = RO R(1)-~ R= BR @” Note: (fall units have the same reliability (j.e., units are identical) then R,(t) = R()- Gi) Since 0 < R(d) < 1, R,(0 decrease rapidly by increase in the number of components. Gil) RC) Smin {Ry RyRy} That is, the reliability of the entire system is less or equal to the reliability of any of its components. Theorem: Ifn independently functioning components having exponential failures with parameters Ay, Ag, «++ A, (in earlier discus- sions we denoted the parameter as dy are connected in series, the failure law of the resulting system is again exponential with 1 parameters (A, + 22+... + 2,) and MTTF = x Proof: Riyee™ el out Probability, Statistics and Queueing Theory RO =RO) R(t) R(t) Set edt gar melita an) IM =-R, SA Has te a.) eat tae) which shows that the failure law of the system is again exponential with parameters Aytagtit Ans Mean time to failure=MTTF = [R,(t)dt é _ Jette nay 3 - Cues iq)! | 2 i _1 1 Note 1: From the above MTTE = $a = ¥ where T, is the tat mt * mean life of the component i. Note 2: If components follow the Weibull failure law with the Parameters B, and 0, then we have, RO=RO RO ...R(t) wl’ el ey ale” oe SRP d ? Reliability Engineering 467 which shows that the system docs not follow Weibull failure law, even though every component follows a Weibull failure distribution. Parallel Configuration (parallel systems) or redundant con- figuration A parallel system is a system in which the various units are con- nected logically in parallel (the word parallel logically implies that the physical arrangement need not be parallel) in which the entire system will fail only when all the units in the system fails or we can state that the system will function successfully even when any one of the parallel units operates successfully i.e., the components are connected across each other and there are n parallel paths between IN end and the OUT end and the existence of any one of them is sufficient to transmit the information from the IN end to the OUT end. Consider a system with ‘ n’ units which are logically connected in parallel as shown in figure. Let ¥, X>, X;, ...X, be the (events) successful operation of the individual units. Then ¥,, ¥, individual units, In fail. i.e., for the un .. X, be unsuccessful operation of the parallel system, the system fails only when all units successful operation of the system, all the ‘n’ units of the system should fail. If P (3) is the probability of failure of the system, then P S)--@ and X; and... and X, Q) Probability, Statistics and Queueing Theory 468 Case 1: units dependent on each other. If the units are dependent on each other, that is, the system will fail only ifall the units fail simultaneously but the performance of an unit will be affected if any other unit fails. Hence (2) becomes, PCS) =P (%) P(% [X) P(X | HG)? (X, | Xn) Here P(X; | X,) denotes the probability of failure of unit 2 and the condition that unit 1 has failed. P(X, | X, Xz) represents the prob- ability of failure of unit 3 under the condition that unit 1 and unit 2 have failed. Case (2) : units independent of each other Here, in this case, the system will fail only when all the units fail simultaneously but the performance of an unit will not be affected by the failure of any other unit. Hence (2) becomes, P(S)-P (1) POG)? (®) But P(S)=1-P(S) =1-[? (%)-? (%) (%.)] =1-[(-P (4) (-P 2) OP OD] ie, R(S)=1-(-R (%,)) (1-8 (42). (1-8 (x,)) she, RO=1-(F-R, ()) (1-2: ©) O-2, CO) 1-N1-2,@) Hence, we can say that in a parallel system, the unreliability of the system is equal to the product of the unreliabilities of the compo- nents. This case is called product law of unreliabilities (for paral- lel system) and it can be stated as “If'n components functioning independently are operating in parallel, and if the. i” component has unreliability F(t), (reliability RW) then the unreliability FQ) (reliability R,(t)) of the entire system is given by FO=FORO-FO=HAO.” | | { I i ar Reliability Engineering Writing this interms of reliabilities, we have R(Q=1-(1- RO) (1-2) (1-8, CO) 1-R,(0) If all components have the same reliability, say R(t), Note 1: (i.c., identical components) then R(Q=1- (1- RY Note 2: Ifthere are only two components in parallel then R(t) =1-(1-R,() (1-R,(¢)) =RiO+ RO - RO RO Note 3: If the system consists of two components in parallel, each of whose failure time is exponentially distributed then RO = RO + RO - RO RO b -. The pdf of the failure time of the parallel system, say T is given ny FM=-R'O Haye aay Oe (ay tag) em OHH)! which shows that T is not exponentially distributed for Parallel system. MTTF = £(T)= fro at é = fone dt+ fe?" dt+ fe Or) an é 4 ° atig tt Ady Ata, a IfA,=Ay=A then MITF=—— If there are n components in parallel with equal failure rate ) then Probability, Statistics and Queueing Theory as mh 1 1 “2a aa , Note 4: The property of parallel systems is often called re- dundancy. i.¢., additional components existing within the system to help the system operate successfully in case of failure of one or more components. Mixed Configuration (Combination of systems connected in se- ries and parallel) Systems typically contain’components in both. serial and parallel relationships. This system may be analysed by considering each sub- system and then the whole system. Consider the system whose block diagram is MTTF >[ap>{B} te input [> output ~p} Ez Scries-parallel (Figure !) iP fv input—+t-+—_| Q — tT — output >—w ~—fu Parallel-series (Figure 2) “Ih figure 1, components A, B and C are in series forming a sub- system X and D, E are in series forming a second subsystem Y. Let R, denote reliability of the component A. Now subsystem, Y are in parallel Reliability of the entire system Rs=1-(1-Ry) 1-Ry) Ry=1-(1-R, Ry Ro) (1 Rp. Ry) Ih figure 2, P,Q, Ware in parallel forming a subsystem_X and V, T U are in parallel forming a second subsystem Y-and now X, Y are ‘in Reliability Engineering 471 scrics. Therefore, Ry = 1~(1—R,) (1 Rg) (1 Ry) —(1-R,) (1—R7) (1 — Ry) RA) = Ry. Ry = [1 -(1= Rp) (1= Rg) (= Ryd] x U-G-R) A-R) 1 - Ry] a_ High-level versus Low-level Redundancy System redundancy may be obtained in two ways. A system, in which m subsystems are connected in series where each subsystem has n components connected in parallel is called parallel series con- Siguration (or) low-level redundancy (see figure 2). IfR is the reliability of the individual component, (identical compo- nent) the reliability of each of the subsystems is equal to 1 — (1—R)". Since m subsystems are connected in series, the system reliability for the low-level redundancy is R,,,, = [1 — (1 -R)"]". } Asystem, in which m subsystems are connected in parallel where ‘each subsystem has n components connected in series is called a se- ries parallel configuration (or) High-level redundancy (see figure D. If R is the reliability of each component, the reliability of each of the subsystems is equal to R". Since m subsystems are connected in parallel, the system reliability for the high-level redundancy is given by Ryigh = 1-1 RY" Note : When m = n = 2, Rigy = Ryign SINCE Rigy — Ryign 2R? (1-R)? 20 k-out-of-n Redundancy A generalization of n parallel components occurs when a require- ment exists for k out of n independent components to function for the — system to function. Obviously k tand fails otherwise. Without knowing the nature of the random variable T it is not possible to give a general formula for the overall reliability of this system (as in case of series and parallel). Sup- Pose that the main unit and auxiliary unit have constant failure rates say A, and A, respectively, then it follows that T, and T, are exponén: tially distributed withp.d.f’s f,(=4,e™ 3 4, >0, ¢ >0 i F,.0=% e™ 5 2,>0,1 >0 1 (in earlier chapters we have taken the parameter as 7 instead of A). “rial unit [Auxihary ui ¥ (or) Stand by unit Since T, and T, are independent, we have T= T, + Ty = AO=f,O* 4,0 . = h 1,62 dy ‘ S prs oat ay a Reliability Engineering a7s Case (i) fA, =A, =A then (1) becomes SQ =VE™ fay 4 =Are™ and FO = [L004 3 ‘ = fye™ dy 6 _yplye™ eo =n] E> x 0 =-Ate™-e*¥ 41 =l-(e* sare”) But R()=1-F,0 R@)=e*(1-at) > (2) . Since 2,() =R,() = p=e™ then (2) becomes RQ) = p (1-log p) Case (ii) : If A, # A, then (1) becomes Sut) = Ay hg OM fo Dy = Aa -e*] and F@= [noe 3 476 Probability, Statistics and Queueing Theory tal (2) a-e) a] nn en aaalae™ me] But R()=1-F,() When (4) — 9 | = Pyand R,(t)=e™*' = p, @ then Ar =-1og p, and Agt = —log p, Now (3) gives, R(t) = 2eloe Ps ~ p2 log p, log p, —log p, Note: 1. When 4, =4,=A4, RU) = (+A) e* 13 2. When, A, =2, = 4, MTTF (for active redundant system) = 2 3. When 4, =A, = 4, MTTF (for standby redundant system) = 3 Example 23. Four units are connec performance. The probabilities are sho ted in series with independent wn below : Since the units are independent, let R(#) be the teliability of the #Pounit, ~ RO = RO Rl). RO RAO (0.97) (0.93) (0.90) (0.95) 7713 ca Reliability Engineering mple 24 A system has 100 units in series, enw J O98. “what is the reliability of the system ? . eliability, since the system has constant R= ROA P = (0.98)! = 0.1326 Example 25 Ata generating station of electrical energy furnace reliability Ry = 0.6, boiler reliability Ry =0-8, reliability of chinney Ro= 0.9, turbine reliability R-= 0.85 superheater reliability R, 0.9 and the reliability of the generator Rg =0.89. Find the total reliability of the power station. Jf the failure rate of each unit is given below, find {he total failure rate of the system and mean time between failures. Apa 8X10, Ae= 15 X 107, Ag= 12 * 1073, Ag =4 * 103, AG=2* 10°3, 7=3* 10%. Find the reliability after 100 hours when the each units follow exponential distribution. Reliability of the power station = Rp. Rg- Rc» Rs- Rr. Ra = (0.6) (0.8) (0.9) (0-9) (0.85) (0.89) 0.2941 Total failurerate A = Apthethyt asthe thr 103(8+154+12+4+2+3) 44x 103 a = 0.044 Mean time between failures =z—= Gag mate i= ‘ fe = 22.73 hr Reliability of the system after 100 hours = e =e- 0.044 = 100 =e 44= 0.0123 Example 26 An air-borne electronic s: ) A born system has a radar, a = puter and an auxiliary unit with MTBF’s of 83,167 and 500 ioue anid cayaient failure rates in failures per 1000 hours of 12,6 and 2 respec- Bel: Find the system reliability and MTBF for a five hour operatin: ime and also reliabilities of the sub-system. p ay Wie failure raves are, wp wens Ue used, = 4-2 6 2 y000” “ “To00 > * = F690 1 _ 1000 MTBF), =1 _ 1000 _ or ¢ ); 7 ag 7 83hrs,(MTBF), = Reliability of subsystems are Rt) 3 Ry(S) = e7 2125 _ y g4ig RQ) =e™ 5 Ry (5) = e795 *5 9.9704 Ry(t) =e" 5 Ry(5) = e- 99 *5 9.99 If either the radar or the computer or the auxiliary unit fail, the entire system fails. Hence the system is made up of series configura- tion. -. System reliability is; Rs (t) = Rt) R(t). RO) ay! aye ee Nc tes QytAg4agye oe wee 23)! _ 8 50 aii -. System reliability R,(5)=¢ 3° ** = 0.9048 1 1 MTBF of the system = $e(or Ra+atd, Example 27 A system has 5 identical components connected in series. If it is desired to have a system reliability of 0.95, how good should be the reliability of each component ? } = 50 hours Here RSD = 0.95, n= 5 and all are identical components RO =p" ifp is the reliability of each component ie., 0.95 = p> Reliability Engineering 479 Slogp =log (0.95) 1 log p = 5 log (0.95) = -0.0103 Pp =0.98975 Example 28 A system is comprised of four scrially related com- ponents cach has a Weibull failure distribution with parameters shown below Component Scale parameter | Shape parameter 1 100 1.20 2 150 0.87 3 510 1.80 4 720 1.00 Find the reliability of the whole system and also find the reliability of the system for 10 hours. If the components follow the Weibull failure law with the scale parameter 6, and shape parameters ; then RO) =exp [-% (<)] -o»[ oa) Gc)” N R,{10) = €-9172627 = 0.8415 \ ‘xample 29 A jet engine consists of 5 modules which are con- WP" nected in series each of which was found to have a Weibull failure jistribution with B = 1.5. Their characteristic lives are 3600, 7200, ~ 5850, 4780 and 9300 operating cycles. Determine the reliability func- tion of the engine and also MTTF. If R,(t) is the reliability of the engine then RD = RD. RO. RO. Ral. RO -() -6) 6-0) ta 480 Probability, Statistics and Queueing Theory we (015+ 07156 5-15, 91S, ots) As Here 8, = 3600, 8,=7200, 6,= 5850, 6,= 4780, 8,=9300 - 0. 12642 th5 Rg (t) = 7 900001 Ls ima) which implies that the engine failure time also follow Weibull’s distribution with B= 1.5 and @ = 1842.7. 1 “ MTTF of the engine = £(T)=0 Tr (+2) = 1842.7 *r(4) La = 1842.7 x 0.9033 = 1664.5 cycles Example 30 Suppose that the service life, in years, of a hearing- aid battery is a random variable having a Weibull distribution with &=%and B= 2. @) How long can such a battery be expected to lost? Gi) Whatis the probability that sucha battery will still be operat- ing after 2 years? @ A(r)= ehr(is) (ey =. (3) = [E=1254 years Gi) We require P(T > 2) = R(2) (ep For Weibull distribution, (1) =~ «" [i Roe @) _ a] Find the reliability of the total lamp in series, calculate the re! whole system is RO “Rp =e Hence, the =Ryg- Rep [1-0-8 (1-Re), peg-e) oN) fe) (-e")] fie 7) re connected in parallel. to produce f the lamps are 0.92, 0.95 and 0.96. system. If the lamps are connected liability of the three lamps system., reliability of the J 0-8) Gok =(2e"'-€”) (-" +e Example 32 Three lamps @ light in a hall. The reliabilities 0: Let R(f) be the reliability of the’ lamp. If the lamps are in parallel, Probability, Statistics and Queueing Theory RAD = 1-1 - RD) (1-20) 1 — RD) =1~(1~0.92) (1 - 0.95) (10.96) = 0.99984 If lamps are in series RO = RO ROR W (0.92) (0.95) (0.96) 0.83904 Sokcample 33 A system has 4 identical components connected in parallel and shows a system reliability of 0.90. How many more com- ponents should be added in parallel to get a system reliability of 0.99? For identical components R,(t) = 1—(1-—R()” Here R,{t) = 0.90 and n= 4. ae "2.0.90 =1-(1-R()* . (1-R())*= 0.1 ie., 1—R(D) = (0.10)! _- RO) = 0.4377 Now we find the number of components to be connected in.paral- lel to get Rg(t) = 0.99. x “Now we have R(t) = 0.4377, Ret) = 0.99 and x Rf) =1-(1- RO)" “ie, 0.99 =1-(1-0.4377)" \ _ 0.01 =(0.5623)" | - # * * nog (0.5623) = log (0.01) oy I 2 log (0.01) 7091 8 7” Tog (0.5623) _ Thus, to get the system reliability of 0.99, 8 components should be connected in parallel. ymple 34 Three electric lamps are connected in parallel. The failure rate for each‘lamp is 0.001 per hour. Find the mean time be- tween failures. If the same lamps are connected in series calculate the mean time between the failures. If-lamps are in parallel with constant failure rate 0 then Reliability Engineering 483 1 MTBF =~—+ at ha vl- ty 1 0.001 0.002 0.003 = 1000 + 500 + 333 = 1833 hours 1 If lamps are in series then MTBF “Ty 1 ~ 0.001 +0.001+.001 seals 0.003 =333 hours Example 35 A circuit consists of 4 transistors, 10 diodes, 20 resis- tors and 10 capacitors connected in series. Under certain stress con- ditions, each of these items has the constant failure rates 0.000002, 0.00001, 0.000001 and 0.000002 respectively. Find the (i) reliability of the entire system and (ii) the failure rate of the system. Find the (iii) expected time to failure for the system and (iv) the probability that the system will not fail in 10 hours. Since the failure rate is constant, the exponential distribution repre- sents the failure law for each of the components. As they are con- nected in series, the entire circuit is again exponentially distributed with @ Failure rate =a [iff()=ae™] =4 (0.000002) + 10 (0.00001) + 20 (0.000001) + 10 (0.000002) z(t) = 0.0001 (ii) Hence, reliability for the system e*'= e ~ 9.00010) Git) FO = Rw at =42—1 _~10,000 héurs a 0.0001 (iv) The probability that the entire system will not fail in 10 hours = ¢-0.0001(10) = 0.999 484 Probability, Statistics and Queueing Theory Example 36 Find the reliability of the component of a system consisting of 10 components in series cach of whose reliability is same, so that the system reliability is 0.999. Let the reliability ofa component be R. R= Ry Ry... Ry = R It is given R,= 0.999 R'°=0.999 R= (o.ssey0 = 0.9999 (OR) Let F=1 — R; hence R!° = 0,999 reduces to =(1—F)'°= 0,999; (F very small) 1—10F =0.999 approximately F =0,0001 R= 1—F = 0.9999 approximately Example 37 Two resistances series. The failure rates are aand system in both cases. If o=0.1, after 10‘hours. are connected in (i) parallel (ii) Bper hour. Find the reliability of the B= 0.2 per hour, find the reliability If they are connected in parallel, Rp (td) =1-(—e*) (1— eA =e eB oats If connected in series, Ry) = et, eB = eat pyt Ifa =0.1, B=0.2, then Rp(10) =e! +e-2 4 3 = 0.3679 + 0.1353 — 0.0498 = 0.4534 i.e., 45.34% Rs (10) = 3 =0,0498 ie., 4.98% Reliability in parallel system is roughly 10 times than that in series system. Keep 2 Ripe Bo =09)0 — 0.084] Powco =0: 0,916 = 0.8244 v, ctatkbeoo (oR Rec Ra Ro= 9-8 * 0.9=0.72 : — ka Roop Ryco=Rac* Rp-Rac Rp =0.72+0.7-0.72 * 0.7 = 1.420 - 0.504 = 0.916 Ruaco = Ru Reco = 0-9 * 0.916 = 0.8244 Example 39 Obtain the reliability of the following systems whose block diagrams are given below : ‘ Probability, Statistics and Queueing Theory my 08 -— output D ——+>— 0.9] B .2}+-fo3 oF 144 > yo 6 ; 0.4 {06 ‘ A B E F (ii) 0.2 {0.3} [0.3 {os Hy, I es | {0.6} ro Cc D G + i A B G > 0.1 10.2 > 0.2 - c H I—>| (0.3 +—l03 oO t —T HA. @ Ry = 0.6, Rp = 0.7, Ro = 0.8, Ry=0.9 4 Ryg = Ry + Ry— Ry Ry = 0.6 + 0.7 — 0.42 = 0.88 Rep = 0.8 + 0.90.8 x 0.9 = 1.7-0.72 = 0.98 _ Rysco’ = 0-88 * 0.98 = 0.8624 @) Ry = = 0.2 0.3 =0.06 = Ryyt Ro—Ryy- R= 0.06 + 0.4—0.06 x 0.4= 0.436 Rusc = 0.5 + 0.6-0.5 x 0.6 = 1.1 - 0.30 = 0.80 Roe Sat fos bef SES) S C@yY € Reliability Engineering We T asz Rs = Ryne X Rpg = 0.436 * 0.8 = 0.3488 Gil) Rasco = Raw + Rop~ Rag - Rep = (0.2 x 0.3) + (0.4 * 0.6) ~ (0.06) (0.24) = 0.06 + 0.24 -0.0144= 0.2856_ Rerg = (0.30.4)+0.5—0.5 x (0.12)=0.12-+0.5-0.06=0.56 Rs = 0.2856 0.56%0 j |= 0.09596 Gv) Rpge = 0.4 x 0.6 x 0.7 = 0.168; Ryg = 0.1 x 0.2 = 0.02 Ruscoer= | - (1 — 0.02) (1 — 0.3) (1 — 0.168) = 1-0.98. 0.7 x 0.832 = 0.42925 1-(1 -0.2) (1 - 0.3) (1 -0.4) — (0.8) (0.7) (0.6) = 0.664 R (for the system) = 0.42925 x 0.664 = 0.285022 Example 40 Find the unreliability of the system whose block dia- gram is given below: Ron -O Let Fp represent unreliability of the component D Fp = 0.2, F,=0.1, F,=0A Fpgp = 0.2 «0.1 = 0.1 = 0.002 Fyy = Fy. F,=0.2*0.2=0.04 Rascozr™ Ran Re» Rpgp= 0.96 * 0.9 x 0.998 = 0.862272 Unreliability F = 0.137728 since F+ R=1 ) 4 a7 Example 41 If the system unreliability is 0.8 for the following Circuit, find the reliability of the component C. Piuvaviny, tausuLS anu WucUsIy Heury 400 A >—{0.1 c I—) L>tx > B >—fo.s Systemreliability =1-0.8=0.2 «. Ry=0.2 Rs =[1-(1-0.1) (1-0.5)] x x (1-0.45)x =0.2; x = a = 0.3636 0.55 L Ro = 0.3636 . _-Example 42 Find the reliability of the system diagrammed below iC. F [07 +—.75} AB D 4 95-p.9' {0.7} {0.9} +0 E G »—fo7 +—.73 Let C, D, E be equivalent to a subsystem X, Let F, G be equivalent to a subsystem Y. Then A, B, X, Y, and Hare in series. Repe=1 —Fopg=1-0.3*%03*03=1— 0.027 = 0.973 7 Reg =1—Feg= 10.25 * 0.25 = 1 ~ 0.0625 =0.9375 + 7, Rs = 0.95 x 0.99 x 0.973.x 0.9375 x 0.9=0.7721211 / Example 43 Two shunt motors are running in parallel] and con- nected to a common supply of reliability 0.90. The reliability of fuse, switch, motor starter and motor itself are respectively 0:95; 0:99, 0.90 and 0.96. Find the reliability of the entire system. cae 2 RQOz IH S ingen a1- (eb! "ee a < il m, ine eye "a & p= ‘TR, ‘t) dt By definition, MITE = J 5 ( = pone fo Mae = zoe, (4) probability, Statistics and Queueing | neory we whenn=3, MTTF = —& When n=4, 0. * = 100 (4-3+$-3) =100%55 =208 =. the minimum value of 7 so that the MTTF is atleast 200 hour: Example 45 A refrigirator consists of three major components power supply, air cooling system, temperature control, having reliabilities of 0.8, 0.9 and 0.85 respectively. Compute system reliabilities for both high-level and low-level redundancy for systems with two parallel.com- ponents. ; Let R be the reliability of each component. aa We have Ryign = 1-1-2)" where m= 2 andn=3 = 1—[1— (0.8) (0.9) (0.85)? = 0.849 Row = ti-a-&)")" where m= 3 andn=2 = [1-d- 0.8)2] [1 — (1 - 0.9)7] [1-G - 0.85)7] = (0.96) (0.99) (0.9775) = 0.929 - Six identical components with constant failure rates: are connected in (i) high level redundancy with 2 components in each sub system (ii) low level redundancy with 3 components in each sub” system. Determine the component MTTF in each case, necessary to provide a system reliability of 0.90 after 100 hours of operation. (i) For high level redundancy, Rj,je, = 1 - (1 — Ry)" 1-(-RO)? , Example 46 Reliability Engineering 491 ou Since constant failure rate R() “ Ry () = 1- (1 - 28 &. Ry, (100) = 1-1 - e7200)3 ie. 0.90: = 1 — (1 — - 2004)3 (1 = €-2004)3 = 0.1 1 — 2004 = 0.4642 e 2004 = 0.5358 ~. 2004 = 0.6240 = 320.5 /hrs MTTF of each components = (i) For low level redundancy, ae =[1- 0-200 =[1-@-roy'] -[-¢-ey] Ree(100) =[1 -(-e™)'] oo0= [ee] ie, 1-(1-e1*) = 0.9487 G2") =00sis 1—e-100A =0,3716 eI 6284 100A = 0.4646 ; 1__100 <. MTTF of each component 7 = 9 4646 = 215.24 hours 488 Probability, Staticti~ Probability, Statistics and Queueing Theory 492 5 dancy with 3 compo- [Do the same problem for high level redunt ee th iw 2003) nents in each subsystem] Example 47 Ina power plant for a telephone exotee aed ont ofits four modules are required to operate the full load. I cae has a reliability of 0.97, determine the reliability of modules. We have, P(x) ="C, RX(1- RY" => P(x and R(t) 2 4¢,,(0.97)* (0.03)4-* Ms =0.9948 Example 48 Calculate the reliability of a 2 out of 4 having an identical constant failure rate of components hour for a mission time of 10 hour. For a constant failure rate, it follows exponential failure distribu- tion, the reliability for a k-out of-n configuration is given by 3 configuration A= 0.61 per RD = Etat (en _ 4 -0.01.x.(10) (,_ ,-0.01(10) )4-* = “gs (1-200) = 0.9999 Example 49 if n-components are hooked up in series arrange- ment, then & such series connections are hooked up in parallel to form the whole system. Assume that each of the components obey an ex- ponential failure law with parameter 0.05 and time operation is 10 hours andn=5. Find the value of k in order that the reliability of the whole system equals to 0.99. Here components (n) are in series, and sub systems (k) are in Parallel. itis a series parallel configuration * R= 1-(1- any _ Since components obey an exponential failure law, R= e~™ where Ais a parameter ua) RO gost sh x 2 oe) 253-17 254 " onst ge rate unit Of E ample 50 asysté pri ac ae ,00 0 nour a ani pian constant failut rat of M Fo 12,500 ne ve ‘what W° a be th S MIBE of a single constar failure 7! unit which without en dy eyould pave the $4 ie rob- abilitY mpleting * 10,000 jon a8 that of standby unit qatpey ore d=20,00° pours 1 aay = 39,000 9.00005 hours TB f , 1 m™ F)2 of standby unit = gee 500 hours ‘ L He os a, 7 12,500 0.00008 hours when t= 1 0000, eter =P, RO en ee R, pl , yom 0g Pa — P2 108 Ps log Pa = 1 2 1g Py RA - 03 (10,000) 20.8) + “(05 _— 0.8 + 0.5 = ) =0.8685334 Probability, Statistics and Queueing Theory > {Note: We can also use the formula . RW)= x 4 x (a ota, eM) J .IfAis the failure rate ofa single unit then "RQ =e™ ] —dt = log Rs(t) " qe Jee Rs ® 1 =- log —— 1 RO y =! _ jog (misx) “ 10,000 0.8685334 . = 0.141 « 10-4 failures / hour 1 ©. MTBF 70,948 hours Example 51 A system has a main unit and standby unit:, The main unit fails at a constant rate of 0.001 failure per hour and the standby unit fails when on line at constant rate of 0.005 failure per hour. (There are no failures while the unit is in standby mode). @ Compute the system reliability over a 72 hours period. _ Gi) Management desires to have a system MTTF of 2000 hours. Find the minimum MTTF of the main unit to achieve this goal, assum- ing that the standby unit MTTF does not change Here 4, = 0.001 per hour, 2, = 0.005 per hour 1 =i 4 @ R60 =Ro (Ae ~ ape") (0.001 eS‘ -0,005 ¢-.001 5 . RLTD= “goa (0.000697 - 0.00465) = 0,9882 ee TN AF . t Reliability Engineering 495 Gi) MTTF = J&s( ae 3 all ote) d el (2-4) “A-% AR ce a >» yA Management requirement is MTTF = 2000 1 1 i.e., 7 +7 «+=2000 Meo a I is. - (no change in standby unit qr = 2000 - 3 ¢ Bi y unit) = 2000- ios 1800 0.005 Hence MTTF of the main unit should be increased to 1800 hours. Example 52 If two identical elements cach has a reliability of 0.95 are in parallel redundancy in a system, determine the system reli- ability. If they are connected in standby redundancy, what would be the change in system reliability. Since two identical elements, reliability of the system RQ = 1-(1-R@)? = 1=(1-0.95)2 - = 1=(0.05)2 = 0.9975 Reliability of standby system R(t) =p (1 —lo; wi = =R,(0) = 0.95 BANE pag P ~ Ro + R(t) = 0.95 (1 — log 0.95) = 0.9987 Example 53 A device with an MTTF of 3000 h continuously on a 500 hour mission. () Compute the mission reliability Gi) Two such devices are put in standby parallel, and no failures ‘ours is to operate 496 Probability, Statistics and Queueing Theory in standby mode, what are the system MTTF and the mission reliability? 1 (i) By definition, MITF = 3 = 3000 1 “% = 3000 RQ=e* ‘00 R (500) =e (sen) = 0.84649 Gara Hae 00 RAQ= (1+ Ae me) « (ias) Rs (500: 1+—— ak »=( 3000 =0.98756 In MTTF => = 6000 hours a 6.5 Reliability by Markov Analysis We have determined the system reliability in the previous section from the knowledge of component reliabilities and their configuration. We can also apply Markov analysis to compute the system reliability. We derive the system reliability when.components are in series and in parallel. The fundamental assumption in a Markov process is that the probability that a system will undergo a transition from one state to another state depends only on the current state of the system and not on any previous states the system may have experienced. In applying i Markov analysis, we assume that each of 1 components will be in-one of two states (a) operating (or) (b) failed. For our convenience, we consider two component systems. Now the system will be in one of the four states at any time as given below: Reliability Engineering «ge State Component 1 Component 2 eT - T operating operating 2 failed operating 3 operating failed 4 failed failed . Our aimis to find the probability of the system being in each state as a function of time. We observe here that if the two components are in parallel (redundant), atleast one of the components function (i.e., state 1,2,3). If the two components are jn series, both components snust function (i.e., the system is in state 1). We denote the probability of being in state / at time ¢ as Pt). Now for series system R(t) = P,() and for parallel system RO =Pi(O+ PO + PO) System must be in any one of the four states at any given time. PQ +PAO+ PLO * PLO? a If we assume that the components have constant failure rates AY and A, we, can represent the system using the state transition diagram. Now we find the probability of the system being in state 1 at time (@+An. ie., P(t + Ad) = P [the system is in state 1. at time ¢ and neither component | nor component 2 fails during (t, f+ Ad). = P,(t). P (component 1 does not fail in At).P (com- ponent 2 does not fail in Ad) =P). (1-449 (1-4, 40) =P) {1- (A, +A.) 40, leaving (Ad)? ~ *Xercise 5c Lage 7 Probability, Statistics and Queueing Theory RG+AN-RO __ P(t ar (+4) FO Lt Att 4n-F@) aioe At sh +4) RO d at 240 --(442,) 20 _Q) Ina similar way we get other two equations for state 2 and state 3. B £80 ~ aR -2.2.0 d 280. AR) -ABO sane (2) weget BW Hak ero To find k, we use the fact that P,(0) = P (the system is in state 1 at 1=0) P (both the components function at = 0) B P\0)=1=> k=1 Buyse ta)" Now we solve equation (3) Using (5) in (3) 220 Ae efit) # -A,P,(0) 6) 42O , apweg Orn! at BRO= Ae , Integrating factor = ert Solution is P(t) e 4! =- e-Al 4k To find &, use P, (0)=0 k=1 Reliability Engineering 200 Hence P(e =- eM +1 Ait (Atay iLe., | (=e Similarly we get, P(t) Substituting P,(9, P,(0), P3(0) in (1) we get P,() <. Fora series system, r,(t)= P(e" Fora parallelsystem, (4) = P,(#) + P(t) + P39 eh ge Bol gg Oath which we have seen already. We can extend the result to 7 com- ponents. 6.6 Maintainability The objective of this section is to characterize and quantify the repair or restoration of a failed unit since no system can be perfectly reliable, only very few systems are designed to operate without main- tenance and for a large number of systems, maintenance is a must. Maintenance is one of the effective ways of increasing the reliability of a system. Therefore, we introduce the concept of maintainability and - availability of the system. Maintainability provides a measure of the repairability of a system where a certain amount of failures can be allowed, while the availabil- ity provides a prediction of the probability that a complex system will be ready for use at any moment in time (will not be down for repairs). Maintainability is an inherent design characteristics of a system. Maintainability is the ability ofan item to be maintained, whereas main- tenance constitutes a series of actions to be taken to restore or retain an item in an effective operational state. Maintainability is a design Parameter. Maintenance is a result of design.

You might also like