LO stn’ eld Pal Dhl cde bale} Maintainability and Availability What You Will Learn In this chapter, you wil learn 1 Importance of suitable maintenance program for repairable systems to restore them to operating after failure 1 The two types of maintenance programs—preventive and corrective maintenances—that a Used nga and their objectives i The aspects of periodic inspections and replacement of worn-out parts in preventive maintenance, wicg conducted even before a failure 1& The aspects of bringing the failed system to an operating condition as soon as possible in the comsctie maintenance, which is undertaken after the failure of a compdent or system = The equations for the reliability and mean time to failure of a maintained system and their application sy. tems whose failure time follows exponential, normal, and Weibull distributions 1= The influence of imperfect maintenance 1m The optimal replacement strategy for minimum cost and the aspect of spare parts requirement t= The availabilty analysis using a Markovian approach and its application to systems with single componens series systems, and parallel systems Andrei Andreevich Markov (or Markoff) was a Russian mathematician who was born on June 14, 1855, and died on May 20, 1922. He graduated from St. Petersburg University in 1878 with a gold medal fers thesis on the solution of differential equations. He was a student of Chebyshev, and taught at St Petenbur, University, becoming a professor there in 1893. He developed the chain theory that has become poplarin ‘weather prediction, social sciences, biological sciences, molecular and statistical physics, quantum tear, ‘and genetics. A chain is associated with a process, which changes from stage to stage. The chain theory re vides the probability of being in the next stage once the probability of being in the current stage i kn ‘Markov chains are applied to random walks, which are a succession of walks along a particular path, Etre books are written on Markov chains and separate courses are offered in colleges and universities on te subject. Markov extended many of the works of Chebyshev and made many contributions to statisti He explained sampling theory within the framework of probability theory (15.17, 15.18]. 15.1 Introduction ‘The previous chapters were concemed with the problem of assessing the reli component or a system without considering the aspects of maintenance and Sd chapter deals with maintainable and repairable systems that can be15.2 15.2.1 cither at regular (rye scar St 152 Mamwrarnasiuity — 707 =e as ‘ Rea ps the maintainability ofa yg Mittin Both the aspects of msintinaity fails while the availabijie ot POV: Ames.) 2% tvalebilty are discussed. n general Ee r ; , use as any instant of time, "YM provides 4 hee epairability of a system when it © availability analysig sue of the readiness of a system for (1) the system conf Of a system, (2) the faire ap describing ne knowledge of the following aspects: wep fain! Poments, (3 he One re Functionally connected, and Tepe eh ANd (4) the repair thod of operation and the definition alr time distributions can be crt Mintenance poticy, A Tillman [15.1] gave a detailed oye oi the SUB adit Lie Pome a to both failure and repair time one cation of references on reli apd ile times) incl ne 2M feauently used disbuton for 00288 (Or rep ee etl, Weibull, and nomal dato Jognormal, gamma, Poisson, and uniforn of a system that fail frequently have ry components that fail infrequently, distribution. On the other hand, ure rate and the same repair distribution, In this chapter, we assume, mostly, that the failure times as well as repair times of indi- vidual components follow exponential distibution, This permits us to use Markov models for the analysis. the most prominent ones are exponential, distributions. For example, if the components atively short repair times compared to those the repair times can be assumed to follow exponential id, if every component of the system has the same fail- time, the repair times can be assumed to follow uniform Maintainability Overview i ioratit failed system to the normal Jntenance is to restore a deteriorating or failed eae pean te, Mas) ae ae Cerne i ystem in a speciied period of time. oa component or system in a mee Eee Tnaitenance. In preventive maintenance, sible: preventive eee S00 COO Se the system is periodical nara ade before he 5 enance sage are checked, and adjustment 1 gvoved during the comet sed afte the 5- is intended to eliminate costly PA, corrective maitene he reliability and ii when the system fails. tem fails. While the aim prolong the life ofthe s¥ objective ofCHAPTER 15 MAINTAINABILITY AND AVAILABILITY 15.2.2 Preventive Maintenance Reliability of the System, Let the reliability ofthe system with no maj by R(2) and let preventive maintenance be used at times i, 2,3, mente that each time a maintenance action is taken; the system is restored tg a, We tm, condition as indicated in Fig, 15.1 [15.4], The reliability of the maintaneg jo iy. during 0 = 1 = % can be stated as fee a R(t) = R(t); 0545 singe no maintenance action i taken until 1 = i. At any time during the ney time, 9 S 1S 2%, the reliability of the maintained system is given by the mae probability of the original system operating at time t = fy and the robatitity a tem, restored to new condition at fp, operating at t = t — fy ‘* (ts Rn(t) = R(o)R(t =); StS 2 89 Next, the reliability of the maintained system during 2% <1 < 3i can be obi y the product of the probability of the system operating at t = 2f) and the probability oe system, restored to new condition at 2io, operating at time t = t — 2ip Ru(t) = R() R(t )|mrgR(t = 2g) = P(g) R(t 2); %H S13By proceeding in a gin. in 315 (4 Ducane Mean Time to Failure, (MTTF,,) is given by MITE, =i ea Phdtes fm cy = pitt} ly As “a Rat) MITER Bro [ne ~ ity) dt bs (15.6) Defining r = t — ity, Eq. (15.6) can be rewritten as MITE, = Se) f° aya asn eae)! In view of the equivalence [159] pail! ~ Ro)CHAPTER 15 MAINTAINABILITY AND AVAILABILITY. Solution: Without Repair. The reliability of the system without repair (see Section 6:7) can as follows be ere 1 R(t) =e™ 2 RY) =1- o(*) 2 nnn (29) With Repair, When preventive maintenance is performed at time intervals the relish system is given by Eq. (15.4) ily oe R(t) = R()R(t~ io): it StS (i+ Ios §=0,1,2,.., 1. Here R(t) = e™ and Eq. (El) gives R(t) = (eB) ieAE M0) = Mog MelMo = eM for any ¢ | oRu(t) = R(t) = es 120 2. Inthis case, R(t) = 1 - @({t— 4 }/o), and hence Rn(t) = Ri(ta)R(t = io) wfie(ts8t ABB) ners ea 3. Since the reliability function of the system without maintenance (see Section 6.7.4) is givenby nol} the reliability of the maintained system becomes nst-weane-m) (ool (SF }owl-(9) } - GG Fe Aah 7 eySolution: The rtig OF reliabilig i Pale) lgya nm Ro hn Oe oN eee sinenan igen by 180 MaWTAWARETY — 711 +i BR ty ia 1a, ees eH 9M) oy Rig) } 1 =0,1,2,.. “ ) 2) Rin) ~ a=! This shows that preventive mai 1 ees es tribution, which has a constant failure rate, 4s not beneficial in the case of exponential flue time di Case 2: For Weibull failure time. Aistributon, the reliably function Gee Section 6.74 ven by wl (3 cs and Eq. (B2) leads to rioo [ATT as Ri), e{-( es seo () +6 = =i + > 0,oni#-! > 1. This inequality will be tate increases). This indicates that preventive pe cd a increases with time, This indicates that Ry(lo) > R(ie) satisfied only when > 1 (-e, when the fle maintenance is useful over the wear-out —_—— EXAMPLE 15.3CHAPTER 15 MAINTAINABILITY AND AVAILABILITY 15.2.3 ‘The expected value of can be found as re nee i oe a= Sa Sirwni't R()) = RCo) RU BARC) 1 Ray = RW) RG) (= RG) P= Rl) ‘The standard deviation of i can be evaluated as of = B[(i = BLi))?) = BP) = (et)? = Sen (ctayt = Ae) [1 R(m) Imperfect Maintenance Ic has been assumed in the previous section that maintenance is performed prety that the system can be considered to be “as-good-as-new” after the maintenance sci However, there is a possibility of performing imperfect maintenance due to human om. In such cases, the system will fail immediately after the preventive maintenance action ft the probability of performing imperfect maintenance is denoted as p the reliability oe system is to be multiplied by (1 — p) each time a preventive maintenance actions nen ‘Thus, the reliability of a maintained system can be expressed, by modifying Ea. (154), as Ra(t) = (1 - p)'R(g)R(t — ig); i StS (i + 1)g; 1=0,1,2,... SID In order to find whether the replacement of wearing parts during preventive mai worthwhile compared to the reduced reliability caused maintenance, the ratio Ry(t)/R(t) att = ig 7 ae : Ru(ito) _ (1 = p)'R'(w)R(0) _ (1 =p) R(itp) R(itp) ae since R(0) = 1, ce EXAMPLE 15.4Reality) 182” MaTeTamAnnery 113 epee ee iy" (1 = Sine the ratio piven by Nera em ™ (1 = py ts tive maintenance ha Ly 2. When the fae time opt i wont sn Soot (YI) *emette | = (1-pien((#)-(2). = 01,2... es For small values of p, we can use the approximation 3) (i-pyieer fi) and Eq. (BA) can be expressed as =e es wc aedefo (hh oe ‘beneficial only . ive maintenance will be be r ein Example 15:2 that ie ve ef wear pend Exton Ie was owen fue EES MO cay wet when B > preventive mantenanSr) Repair time (/) 0 MTTR 4 FIGURE 15.2 Probability density function of repair time. When Time to Repair Follows Exponential Distribution. The probability density function of repair time (7) is given by rt) = pe (15.12) where z = 1/MTTR is the repair rate and MTTR = mean time to repair. If, ,.... denote the observed repair times, MTTR can be computed as MTTR aa (15.13) When Time to Repair Follows Lognormal Dist be 1, 3, fy Then the sample mean (¥)ane EXAMPLE 15.5 The total number of. umber of maintenanee of repairing failure in (1) 1 Solution: The data re rt Joe n= 578/84 = 841g K = M/ST w 5 cay sae et in iat “ia i), 1 MC) = 1a . ai M(S) = 1 ~ eons) fi Me ey 3 M(20) = 1 ~ g-arasyayy ! Poa = Qi = 01352 00547 = Oty a EXAMPLE 15.6 The repair times 4; for a mainframe computer 40, 3.1, 3, and 39 hears eee ar eerine is mie S17 1822.26 following: aia determine L Maines od poe 2. Downtime required to achieve a maintainability of 0.99, Solution: Since the repair time 7 follows lognormal distribution, the mean of X = Intis given bya 1. Theminaiaiity othe yer comesonting 8 own tS oui pipe py M(S) = o( in 5 *) = of sees ons) = (20911) = ogg), lity of 0.99, the required downtime ¢ canbe determine fon the = o( Sau) i CHAPTER 15 MAINTAINABILITY AND AVAILABILITY 2. To achieve « mai relation M(t) = 0.99 = o( = 4) From standard normal tables, we obtain 0.3720 Int ~ 0.83152 oan 2927 that is, Int = 1.69176 or 1 = 5.4584 hours Dd EXAMPLE 15.7 ‘Solve Example 15.6 assuming that the repair times follow Weibull distribution. Solution: The 10 sample repair times (1) can be used to find the parameters of Weibull distributions (using the procedure outlined in Section 13.7.7) m = 2.75 and = 2.93 so thatthe maintanabiy M(t) is given by wel) 1. The maintainability corresponding to a downtime of 5 hours is given by5.2.6 time follows ex, is given by tial dist Tin With 18 MAMTANAMLTY 717 irate of a i 1 the maintainability function Cm Hence the ig ay alc ber of eit tbe tg us repaired fy Pome sy 5.18) ed failures in noe file in tine ine 1s sven by in Migs Will Nes anew, be equal to (15.19) A, EXAMPLE 15.8 The air-conditioning unit used j sed in a a system life of 12 months, how meg net istlation had is allowed for each repair, dere aires are key not et eae ve on ation wegen see! if a repair time of 2 oars . sper hour a 05, Solution: The expected number r failures. The number of failure whe is Hs, Ty is ve by Ay = 0412 ‘are not likely to get repaired is given by ftom me Na = Moe Wr = 1.260502) = 2 = 1.2(03679) = 0.4415 ‘Thus, 0.4415 failures every 12 months, onthe average, wil take longer than 2 hous to repair. . Optimal Replacement Strategy ‘As seen in the previous sections, preventive maintenance improves the reliability of a sys- tem during the wear-out period. In some systems, a component is automaticaly replaced afier an operational time off. This strategy is known as age replacement. Changing th and oil filter in an automobile after every "a months is an example of age replacement In some cases, a set or batch of components i replaced regulary times ty, 2p, 3b, --- and es a aire oocurs. This seg, known as bth repose, 5 wee aso whenever fle are involved such asa set of lightbabs in ever a large number of inexpensive comP® oe preventive maimte- a factory. In all the eases, : s nance should be compared with I Haas sa ‘ component be replaced at and cost epg a faled on ponent through preventive Inte 6? Cr ie through corrective H failed component > involves not only the ne with several addia HAPTER 15 MAINTAINABILITY AND AVAILABILITY and the total cost of replacement ¢ by C= CMa + CMe (152 ‘Since the total operating time 1 is very large compared to the mean time between ‘ments (MTBR), we have 1 ”* MTBR (15.29) where MTBR, similar to Eq. (6.28), is given by MTBR = i Rr) dr LeU) (1523) and R(x) is the reliability of the component which is automatically replaced afer atime Pa R(t) for r= R(7) ia fee oh (15.24) with R(r) denoting the reliability of the original (unmaintained) component. Equations (15.24) and (15.23) yield MTBR = [ames [Rove- [rove (isa Since the fraction of components that survive until the next preventive maintenance is given by R(%p), the number of surviving components, rg, that are replaced through pre- ventive maintenance, are given by Ny = R(ig)n (1526) ant similarly, the numberof failed component which are replaced theugh cots maintenance, are given by n= (1 ~ R())n = fn) ote53 1 Spare Parts Requirement When the failure times of the number of failures jn MPO of, probability of having pg dG System fo hres in tine an er ven by i 182 MaNrANABUY — 719 complex omental distibation, bon (152) Th, P( (ary a (a) OO where A = 1/MTBR and ‘ MTBF = (15.30) Tess than Or equal to rfaifur- Mea time failures. The probabitty of hay i P(nsr) = ar) (1531) ‘The value given by Eq. (15:31) is also the ppp ‘adequate replacement parts avaiable ifr spare parts are socked atthe begins ohn Equation can also be used to find the number of spare part, seedetee ra Ifa desired cumulative probability as 7 is given for a particular item, the number of spare pe C Summing the values of the individual probabil flue unl this sum is equal or gree thn he dese cumuateraomin a pler, but approximate, formula for determining the number of spare pars ina mission time Tyis given by ‘Number of spare parts = AT) + 21-_VATo (1532) where ATo and ‘VAT denote the mean and standard deviation of the number of failures in time To, according to Poisson distribution, and zy qs the value of standard normal variate corresponding to the confidence level a. EXAMPLE 15.9 is ii a die meg mission period of 30 days with 99 P= aloe = 720 hours athe ats sane ons eee ; probability i = 4240 eri(CHAPTER 15 MAINTAINABILITY AND AVAIL 15.3 Availability ‘The term availability is used to indicate the probability of a system or arom bet ‘operating condition at any time f, given that it was in operating condition at ¢ =o, 5". to be in operating condition at time f, the system must not have failed or, tis had failed -, the solutions of Eqs. (15.49) and (15.52) converge to -— ae 1 A(%) = ad Equations (15.53) and (15.54) indicate that the Df becomes independent ofits starting (or initial) st ‘The steady-state solution can also be fou since the derivative of the steady-state valu dP/dt = O-and dP, /dt = Oin Eqs. (15 Ab ibeiber rs nt at153 Avan Me Heady ice LY 795 Nm Atm) ae (15,58) ; As AO ) SOME SE, We Wil bey ny a specific period of time 7 in nai, (1559) and dividing by the total time: ‘ten ep ie the sytem over 1) over the time imerval 7 ‘A (23) cee Ate OT nesta Gter sé Similarly, if the system w: (15.60) yields as i aed ste at t= 0,4() is given by Ea (1551) and A(T) =e A a eae A cs (atu Au (tut twp (15.6) ‘The steady-state or long-term availability of the system can be determined by letting T— © in Eqs, (15.61) and (15.62). This gives ———— (1563) AO een Equation (15.63) cam also be expressed as an gover cae _ MTTF me eT + MTR coaaiesslhioCHAPTER 15 MAINTAINABILITY AND AVAILABILITY 15.3.5 GHEE FIGURE 15.5 A series system, Series Systems Consider a series system with two components A and B as shown in Fig. 15.5, Let the i. ure rate be A and the repair rate be ys for each component. The system willbe in any of ye following three possible states at any time t: 1. Both components are in operating condition (state 0) 2. One component isin operating condition and the other under repair (state 1) 3. Both components are under repair (state 2) Since both components must be in operating condition forthe system to be in operating condition, the availability of the system is given by the probability of the system being in state 0 at time ¢, that is, A(r) = P(r). The system will be down in states 1 and . One Repair Person, The availablity analysis of « multicomponent system depends on the number of repair persons available for servicing the failed components First we con- sider the case with one repair person. The resulting transition probabilities are summarized in Table 15.1 From this table, the transition matrix [P] can be identified as [15.3] 0 if € of (1 - 2a) 2A 0 DP 1 fee) (ee) (AS Alt bh (1-4) Thus the probability of the system being in state 0 or 1 or 2 at time ¢ + dt can be written as: Pot + de) = Py(0)(1 — 2adt) + P(t) at (136) Ault + dt) = A(t) ade + Py(t)(1—Adt— pat) +A(1\uar (156) Prt + dt) = P(t)Ade + P(r) (1 ~ dt) : These equations can be rewritten, using the definition given a, = PAR) + AP t) ‘ dP, BO any - a0 AP(1)ewer ee o~ (7) py ~ (p18 Pte 1 = (@r- Vien pa 0 ~ (PV) PNT ~ (PX — LPT WU 1 ~ LuPY 1) you vonodkno9 ont Jo seo posredox auuodwos ovo A110 pareydiuo> suauoduios sn0q 50 62% -aropdan0 ou sf Joyo au Jo s1edor uv strep wioioditi00 90, [ey sjusuoduios og. uonrpuos Sunerodo Ur sueuaz soqo xp pue sirey stusuodutoo atp Jo x9qRE, uontpuoa Sunesodo uy uewiar siusuodwio9 tog AywaeqosaCHAPTER 15 MAINTAINABILITY AND AVAILABILITY These equations can be solved for any specified iil condition Wo find P,P, ang p as functions of time, For example, if the system was in operating state at time ¢ = te initial conditions are given by z AO) P(0) = 0; 7(0) =0 (is) In most cases, the steady-state solution of gs. (15.69) t0 (15.71) willbe of interes, fg this, we set dPy/at = 0, dP/dt = 0, and dP/dt = 0 in Eqs. (15.69) t0 (15.71) and ye the condition Py + Fy + Py = 2A + MP, = 0 (isa) 2AR — (A+ WP + WP, = 0 (15.74) AA — wh = 0 (1575) At A+ h=1 (15:76) ‘The solution of Eqs. (15.73) to (15.76) gives 2 # A= >> % O° 2+ Det we Pa 2 4-5 a 5 (1578) 1 + 2dp + 28 2 eae om w+ Aw + 28 Thus the steady-state availability of the system is given by we 5.80) Ato) =F = = oe Be «ass (=) =R w+ dw + 2K ‘Two Repair Persons. In this case, each repair person can work on a different compo nent. If the system was in state 2 at time f, it can return to state 1 at time ¢ + dt if pai of one of the components is complete, while that of the other is incomplete. Since the rept of either of the two components can be completed by the time # + dt, the probability is 2 dt(1 ~ pdt) ~ 2y dr, On the other hand, ifthe system was in state 2at time fit coo {inues to remain in state 2 ifthe repair of both the components is incomplete. The P of this event is equal to (1 - wd)? ~ 1 — 2y dt. Thus, the transition matrix, [P} Of Eq, (15.65) will be modified as hi 0 1 of (1 - 2a) 2a Cle 1 aa (1-A~p) 2 ko mu aThis MALEIX CAM be Used 49 dg ri rete ANAIARILTy m9 We ationg hi Wy et Hh eo * WP + up, = 9 =6 A ~ tyr = 9 The solution of Eqs. (15.89) 45 ss A+R tna, 'S given by + wae rw +R (15.36, Pa z wt aR (1587) Aa e+e ie Thus, the steady-state availability ofthe system will be ACen (=) =P Pore (1589) Parallel Systems Consider a parallel system consisting of to components A and B as shown in Fig: 15.6 For this system, there ar thee posible sais: state (both components opersig), S88 1 (one component operating and the other under rep) ant sate 2 (both components under repair). One Repair Person. 1F there is only a snl rept person servic he two COMES nents, the transition matrix can be derived as: 0 ! i 0 1 - 2A) 1590) of ( oa i [P]=! A a-#) 2 A eu FIGURE 15:6 Apsall730° CHAPTER 15 MAINTAINABILITY AND AVALLABILITY This gives the steady-state equations as 2A + UP, = 0 AP (154 2AM) = (A+ p)P, + WP, = 0 a AP, = P= 0 ee A+R +R =1 ) (1559) ‘The solution of these equations is given by Eqs. (15.77) to (15.79), ce sates 9 constitute the operation of the system, the steady-state availability of the ‘yen cart found as w+ 2p 2 + Aw + 2 (595) Alo) = Fy + ‘Two Repair Persons. If there are two repair persons, one repair person canbe ase to each component and the transition matrix can be derived as 0 1 2 of (a -2a) 2a 0 LP | oie (l-A~a) a (1595) 2) ay ap (1- u) From this matrix, the steady-state equations can be obtained as 20K + MP, = 0 (53 2AR)— (A + w)P) + WP = 0 (1598 AP, ~ 2pP) = 0 (1599 A+h+R=1 (1s100) ‘The solution of these equations is given by Eqs. (15,86) to (15.88). The steady-state aval ability of the system can be determined as 2 A(e) = lt (15.0) a) matt w+ 2p + 2 15.4 Optimization Approaches Several optimization methods have been used for solving ity problems. The allocation of availability p to various components of a system for minimur using the Lagrange multiplier method in155 yy MERA aa son of MTBF, MTTR, and the py TIONS Us for minimum cost were ae of rei u ae Wundant unit state availability of a repainagy 8 & dynam nt” Mhieve ’ ble oy time is maximized under constea, With cold r od st MNTLAW AND EXCH, 731 achieve a pied avai * ae ha dana Sopronch, The steady- provision of spare pati Cn And nonzero replacement and availability of Papeetifcanty influence tt ela in Ret 1137) he fc a ee Operatir S or maximum av Optimal Ng oss, maintsinabiy ne ey reliability, md inaon Wei pice tan be con weight of Di and mini am be carted determination and some gt pee NE ben entre n toering co o agencies, and 60 on 15,8), mt re being outinly wed by cen ee 15.5 Numerical Solutions Using MATLAB and Excel a AMPLE 15.11 The failure time of a machine follows normal distribution witha mean value of « = 20 months and a standard deviation of o = 4 months. The machine undergoes preventive maintenance st regular intervals of 12 months Find the reliability ofthe machine att = 30 months without and with maintenance using MATLAB. Solution: ‘The reliability ofthe machine at time t with no maintenance is given by rjc lave o() e . undergoing preventive maintenance at Noting that the reliability ofthe machin a ine f (oven unre ic terse have for ners of ime ) is given by Ea (Blo Exam 2 ¥= 5° in sts (i+ I)mit=? a(t) = none melee = a ie m2) 1 - (PE - 9 a [ % (Nh een to find the ial ‘The MATLAB program shown below. ana! cheerHAPTER 15. MAINTAINABILITY AND AVAILABILITY Reino #1 = noxmodé (#10) Bie (eieo = mu) /0d FES (etme = redime0 «my fad Probl = normedt (11) Prob2 © normedt (32) etm = ((1 = Peobi)*2)#(1 = Prob} m0 = 2.8000 Relno = 0.0062 as 2 ae -3.5000 Probi = 0.0228 Proba = 2.32630-004 Role = 0.5548 a EXAMPLE 15.12 The failure time of a machine follows two-parameter Weibull distribution with parameters 6 = 20 and = 50 months. The machine undergoes preventive maintenance at regular intervals of 12 months. Find the reliability of the machine at t = 30 months without and with maintenance using Excel Solution: The reliability of the machine at time t with no maintenance is given by dec cla " Noting that the reliability of the machine at time + (When undergoing preventive maintenance tt intervals of time 1) is given by Eq, (E1) of Example 15.1, we have for = 12 and t = 30 mont Rn(t) = R() R(t ; fa 5 - “Gy ill-Ca)}) ‘The Excel program to find the reliabilities indicated in Eqs (El): Ss time uu : la 2 : 1 2S a Peck : = a : ‘ ae : : | REFERENCES AND BIBLIOGRAPHY 15.1 C,H. Lie, C. L. Hwang, and FA. Tilman, “Availabilty of Maintined Systems: A. , » of mms: A State-f- the-art Survey,” AIIE Transactions, Vol. 9, 1977, pp. 247-258. — 15.2. M.O. Locks, “Reliability, Maintainablity, and Availabilty Assessment" Hayden Book Co. Rochelle Park, NJ, 1973. 153 G.H., Sandler, System Reliability Engineering, Prentice-Hall, Englewood Cliffs, NJ, 1963. 1987 154 BE. Lewis, Introduction o Reliab Engineering, ohn Wiley, NewYork : i HL Moss, °A CostBused Aaaity Aloesvn Agseshns Se pats 20, 1971, pp. 178-182. Lambert, rms, “Optimal Redunan aA Mlssion 6 A. Walvela and PBI apy Vo R20, 197 pm 12-185 1" IEEE Transactions sem Avy and para SIE Units” 15.7. M. Sasaki, S. Kabura, 26, 197, po. 182-188 ’ IEEE Transactions 07 ReTAT ad isk Anas NY But Feuer, He 15.8 E. G. Frankel, Systems 1984. ‘Chemical Rubber Co, Cleveland Hague, The Netherlands: ables, iin eb THE 159. §.M, Selby, Standard of Sysem Availity: A Bayesian OF, cn 2 Mes peers 18.10 D. M. Brendef nme HEEBCHAPTER 15 MAINTAINABILITY AND AVAILABILITY 18.1 18.12 1813 184 DR. Gover J and M, Mazumdar, “Some Bayes stints of Long-Run Avi, ‘Two-State System,” EEE Transactions on Reliability, Vol. R-18, 1969, pp, 184 4gq"! a ‘A Bayes Analysis of Availabitit Subsystems,” IEEE Transactions W. B, Thompson and M. D Spr Consisting of Several Independ Vol, R-21, 1972, pp. 212-214 1W.E.Faragher ann HS, Watson, “Atay Analysis—A Realistic Methology 19, mpi on Reiland Quality Cont, Washingion, D.C, Samar 1968, p95 R.E Barlow and FProshan, Mathematical Theory of Reliability Sohn Wiley, New Ya, W faa - le Reliability Vol. R-15, 1966, pp. 58-68. M.A. Moss, Designing for Minimal Maintenance Expense, Marcel Dekker, New Yok, gg A.C. King and C. B. Read, Pathways to Probability, Holt, Rinehart & Winston, ig New York, 1963, C.C Gillespie (ed -in-chief), Dictionary of Scientific Biography, Vol. 9, Charles Sertne' Sons, New York, 1980, =O REVIEW QUESTIONS 1541 15.2 153 Give brief answers tothe following: Define the terms maintainability and availabilty. ‘What is the difference between preventive maintenance and corrective maintenance? What is MTTR? How is the reliability of a maintained system defined? Define instantaneous availability ofa system, ‘What is average uptime availability? How is steady-state availability defined? Explain the difference between inherent, achieved, and operational availailies. 9, What is transition matrix? 10, What is Markov analysis? 11, Whatis the effect of imperfect maintenance on the reliability of a system? 12. State the factors that influence the optimal replacement strategy. 13, How is the number of spare parts determined for a single-component system? Saar ee Indicate whether each of the following statements is true or false: 1. The purpose of maintenance is to restore a deteriorating or failed system to nora operating sate, i 2. If mean time to repair is given by MTTR, the repair rate (41) is given by # = jgr7R’ 3. Markovian approach is not useful forthe availability analysis. F 4, Markovian transition diagram and transition matrix provide differen information Fill in the blank withthe proper word(s): 1, Maintainability provides a measure of the____of the system. 2. Availability provides a measure of the __of the system for use at any i184 Select the replaced &. Before any fai After complete er complet fai After paral fit, 2. The ccmecive Afr fs not bh. Much before alu Just before fila 3. When the fai etme of eompon the numberof failures ponents of system fllow exponential dst aseifig pease ti val distibton « Noueaggeene b. Poisson distribution Binomial distribution 4. The long-term (steady ste) aaiilty, =), BAe) i a rps rue and failure rate (A) a: a oo H i Ate ea cent Nee ee # 15.5 Match the following avait ystantane lability a. Proportion cof time during which the syst 1. Instat ional by considering ready time, logistics. camp whore tem is time, and wating t preventive sina vam ine ding wich 982 a eam any cress enn vs uring hice se aly rl forse nsec! interval seis opr Pepsi ey ES herent availabilt) i nf time cing which the S188 ing only he coronsive 2, Average uptime availability 3. Steady-state av! ilailty ent 5, Achieved aval! mics sions xing iene sen ime we . Proportion ny tir Me 6. Operational availCHAPTER 15 MAINTAINABILITY AND AVAILABILITY. PROBLEMS ina thermal powerplant are givenby 42, 45 1 ‘The repaired fora oie na thermal power pl cia 5.6,5 3 1, 6.3, 6.4, 6.7, 6.8, 7.0, 7.3, and 7.5 hours. Find 2, 1. the maintainability ofthe boiler fran allowed downtine of 6 hours, 182 183 184 18s 15.6 18.7 158 159 15.10 D. the downtime required to achieve a maintainability of 0.95, 6 the rue! MTR. when the allowable downtime i 6 hous fr «maint (0,95, Assume thatthe repair times follow exponential distribution, Solve Problem 15.1 assuming that the repair times follow normal distribution Solve Problem 15.1 assuming thatthe repair times follow lognormal distribution, ‘Solve Problem 15.1 assuming thatthe repair times follow Weibull distribution, ‘The times required to restore power outages by an electric power company are 05,07, 9, 10,1.2,13,1.6,17,19,21,22,26 29,34, and 4.2 hous. Determine the following a, The maintainability ofthe power supply ifthe outage is permitted for 5 hous }, The duration ofthe outage corresponding to a maintainability of 0.99, Solve the problem assuming (i) normal distribution and (ii) lognormal distribution forte ‘outage restoring times. Find the reliability of a system for which preventive maintenance is performed afer every 500 hours of operation when the failure time Tin hours of the original system sven by the following distribution: a fr(t) = 0.001 e0# 1 (t= 1000)" ee) A-system has a mean time between failures of 200 hours, mean time to repair of 10 hours, ‘mean ready time of 5 hours, and a mean downtime of 15 hours. Find the inherent, achieved. and operational availabilities of the system. Find the time interval, , needed for replacing the components in preventive maintenance ifthe reliability ofthe original (unmaintained) components is given by wl) ne when cy and care given by 100 and 1000 dollars, respectively. Find the time interval, fo, needed for replacing the components in preventive maintenance ifthe reliability ofthe original (unmaintained) components is given by a9 = emn{-()"Ie sinus ‘where ¢), and ¢, are given by 100 and 5000 dollars, respectively, A system consists of three types of components with 10 units of type A, 20 units of typeB and 30 units of type C. The mean times before failure of type A, B, and Cc are, respectively, 2000, 1500, and 3000 hours. Find the number of spare parts Fe R(yHn time 99 of $9 09 me hour enn monies 1% ind the mean, Sentence fooned a mat (90 prent (695 felt) = dear 034 St 8 syste . *.6) Weitatrg* *BEH the te!” MHCh pre eae ah a inc ) ponent distin, Art) say ( “ “ye J ala) of Find the condition ym sysom for which te tae hte } lure time gt vent "toe diana At) = Bee ould be beneficial fora The number of failures ing restoring the system is 239 Power statin , is 220m in years 6. The al ne genta () 2 hours, nde) 10 heures he proba pace : ‘ ity of repairing a ure in (a1 Hou, A welding robot in an ass sembly shop is foundto fi and a repair takes, on the ws MoPiBfoundtofatonceinevery 10h caigemeree sof operation likely not to get rated sae a8® | our Determine the mune of furs so era pod of yearf area tng foreach repre Ain alts rtmiscyucd Suri yoda ene A system has & mean time between failures of 200 hours and es urs and a meantime to epi of 30 hours. Find the steady-state availability ofthe sytem, A series system consists of two identical components each with a flue rate of 0.005 per hour and a repair rate of 0.1 pe hour Find th steady-state evalabiity ofthe system if the number of repair persons is a) one and (b) two. ‘A parallel system consists of two identical components each with failure rate of 0.005 per hour and a repair rate of 0.1 per hour, Find the steady-state availabilty ofthe system ifthe number of repair persons is (a) one and (b) two ie i 9 failures bearing fal ‘A turbine is known to sustain two types of al 7 a ial di wth failure rate of 0. bearing failure times follow exponential dstnbuion wi ures and blade aires. The 0005 per hour and the blade failure times follow Weibull distribution with au (eee {-(: \; sas 0 alam) a raion ili pe ansnie sysem fe 1900 an a. Find the reliability of OP ics turbine. ie nid ae ine is tobe i Q ihe eal of he ib tim hours Ifthe design life of the esi if. onCHAPTER 15 MAINTAINABILITY AND AVAILABILITY 15.20. The MTTF and MTTR of a spot weldi stem assuming imperfect preventive maintenance after e & Reliability of om witha | percent probability of fllue ofthe system j hours of operation with after the maintenance, ery Loy Imiediaely robo are 60 hours and 3 hour, especie the wait ofthe robot. If the MTR is reduced to 1 hour, find the salle MITE which gives atleast the same value of availability. ying Value of : system failed and the times at which the 5 5.21 The times at which a multicomponent system fai ery daar ey iheugt cain overa mson perio of 1000 hous ae sha nthe felowigg age Time at Which Failure ‘Time at Which System ‘Occurred (Hours) Restored (Hours) 742 715 102.1 104.2 1934 194.1 2455 2478 381.0 381.9 4153 418.2 464.6 467.1 499.8 502.9 580.2 583.4 633.1 635.1 6789 6803 17 142 7885 791.6 891.0 893.3 982.4 985.8 Find the following: (a) average availability of the system, (b) average failure and rept rates of the system, and (c) availability of the system using the failure and repair fats found in part (b) 1522 Plot the availability functions given by Eqs, (15.48), (15.51), and (15.63) fora system hi ing a failure rate of 0.05 per hour and a repair rate of 0.5 per hou. MATLAB and Excel Problems 15.23 The failure time of a machine follows exponential distribution with a failure rate of 0.0001 failures per hour. The machine undergoes preventive maintenance at regular intervals % = 5000 hours. Find the reliability of the machine without and with maintenance 12,500 hours of operation using MATLAB, 15.24 Solve Example 15.6 using MATLAB, 15.25 15.26 Solve Problem 15.23 using Excel, Solve Example 15.6 using Excel.