Failure rate

From Wikipedia, the free encyclopedia

Jump to: navigation, search Failure rate is the frequency with which an engineered system or component fails, expressed for example in failures per hour. It is often denoted y the !reek letter " #lam da$ and is important in relia ility engineering. %he failure rate of a system usually depends on time, with the rate varying over the life cycle of the system. For example, as an automo ile grows older, the failure rate in its fifth year of service may e many times greater than its failure rate during its first year of service. &ne does not expect to replace an exhaust pipe, overhaul the rakes, or have ma'or transmission pro lems in a new vehicle. In practice, the mean time etween failures #(%)F, *+"$ is often used instead of the failure rate. %he (%)F is an important system parameter in systems where failure rate needs to e managed, in particular for safety systems. %he (%)F appears frequently in the engineering design requirements, and governs frequency of required system maintenance and inspections. In special processes called renewal processes, where the time to recover from failure can e neglected and the likelihood of failure remains constant with respect to time, the failure rate is simply the multiplicative inverse of the (%)F #*+"$. , similar ratio used in the transport industries, especially in railways and trucking is -mean distance between failures', a variation which attempts to correlate actual loaded distances to similar relia ility needs and practices. Failure rates and their pro'ective manifestations are important factors in insurance, usiness, and regulation practices as well as fundamental to design of safe systems throughout a national or international economy.

Contents
.hide/

* Failure rate in the discrete sense 0 Failure rate in the continuous sense 1 Failure rate data o 1.* 2nits o 1.0 ,dditivity o 1.1 3xample 4 5ee also 6 7eferences o 6.* 8rint

6.0 &nline

9 3xternal links

[edit] Failure rate in the discrete sense

In words appearing in an experiment, the failure rate can e defined as %he total num er of failures within an item population, divided y the total time expended y that population, during a particular measurement interval under stated conditions. #(ac:iarmid, et al.$ While failure rate "#t$ is often thought of as the pro a ility that a failure occurs in a specified interval, given no failure efore time t, it is not actually a pro a ility ecause it can exceed *. It can e defined with the aid of the relia ility function or survival function R#t$, the pro a ility of no failure efore time t, as: where t* #or t$ and t0 are respectively the eginning and ending of a specified interval of time spanning ;t. <ote that this is a conditional pro a ility, hence the R#t$ in the denominator.

[edit] Failure rate in the continuous sense

3xponential failure density functions )y calculating the failure rate for smaller and smaller intervals of time , the interval ecomes infinitely small. %his results in the hazard function, which is the instantaneous failure rate at any point in time: =ontinuous failure rate depends on a failure distribution, , which is a cumulative distri ution function that descri es the pro a ility of failure prior to time t, where T is the failure time. %he failure distri ution function is the integral of the failure density function, f#t$, %he ha>ard function can e defined now as (any pro a ility distri utions can e used to model the failure distri ution #see List of important probability distributions$. , common model is the exponential failure distribution, which is ased on the exponential density function.

For an exponential failure distri ution the ha>ard rate is a constant with respect to time #that is, the distri ution is ?memoryless?$. For other distri utions, such as a Wei ull distri ution or a log@normal distri ution, the ha>ard function may not e constant with respect to time. For some such as the deterministic distri ution it is monotonic increasing #analogous to ?wearing out?$, for others such as the 8areto distri ution it is monotonic decreasing #analogous to ? urning in?$, while for many it is not monotonic.

[edit] Failure rate data

Failure rate data can e o tained in several ways. %he most common means are:

Aistorical data a out the device or system under consideration. (any organi>ations maintain internal data ases of failure information on the devices or systems that they produce, which can e used to calculate failure rates for those devices or systems. For new devices or systems, the historical data for similar devices or systems can serve as a useful estimate. !overnment and commercial failure rate data. Aand ooks of failure rate data for various components are availa le from government and commercial sources. (IB@A:)C@0*DF, Reliability Prediction of Electronic Equipment, is a military standard that provides failure rate data for many military electronic components. 5everal failure rate data sources are availa le commercially that focus on commercial components, including some non@electronic components. %esting. %he most accurate source of data is to test samples of the actual devices or systems in order to generate failure data. %his is often prohi itively expensive or impractical, so that the previous data sources are often used instead.

[edit] Units
Failure rates can e expressed using any measure of time, ut hours is the most common unit in practice. &ther units, such as miles, revolutions, etc., can also e used in place of ?time? units. Failure rates are often expressed in engineering notation as failures per million, or *EF9, especially for individual components, since their failure rates are often very low. %he Failures In Time (FIT) rate of a device is the num er of failures that can e expected in one illion #*EG$ device@hours of operation. #3.g. *EEE devices for * million hours, or * million devices for *EEE hours each, or some other com ination.$ %his term is used particularly y the semiconductor industry.

[edit] dditi!it"
2nder certain engineering assumptions #e.g. esides the a ove assumptions for a constant failure rate, the assumption that the considered system has no relevant redundancies$, the failure rate for a complex system is simply the sum of the individual failure rates of its components, as long as the units are consistent, e.g. failures per million hours. %his permits testing of individual components or su systems, whose failure rates are then added to o tain the total system failure rate.

[edit] #xample
5uppose it is desired to estimate the failure rate of a certain component. , test can e performed to estimate its failure rate. %en identical components are each tested until they either fail or reach *EEE hours, at which time the test is terminated for that component. #%he level of statistical confidence is not considered in this example.$ %he results are as follows: Component =omponent * =omponent 0 =omponent 1 =omponent 4 =omponent 6 =omponent 9 =omponent D =omponent H =omponent G =omponent *E Totals $ours Failure *EEE <o failure *EEE <o failure 49D Failed *EEE <o failure 91E Failed 6GE Failed *EEE <o failure 0H6 Failed 94H Failed HH0 Failed %&'( )

3stimated failure rate is or DGG.H failures for every million hours of operation.

[edit] *ee also

Statistics portal Failure Failure mode 7elia ility 7elia ility theory 7elia ility theory of aging 7elia ility engineering 5urvival analysis

and longevity

Wei ull distri ution (%)F ,nnuali>ed failure rate )urn in 2ser 7eengineering #products$

[edit] +eferences
%his article includes a list of references, related reading or external links, ut its sources remain unclear because it lac,s inline citations- 8lease improve this article y
introducing more precise citations where appropriate. (November 2009

[edit] .rint

)lanchard, )en'amin 5. #*GG0$, Lo!istics En!ineerin! and "ana!ement# $ourt% Ed., pp 09I10, 8rentice@Aall, Inc., 3nglewood =liffs, <ew Jersey. 3 eling, =harles 3., #*GGD$, &n 'ntroduction to Reliability and "aintainability En!ineerin!, pp 01I10, (c!raw@Aill =ompanies, Inc., )oston. Federal 5tandard *E1D= Capur, C.=., and Bam erson, B.7., #*GDD$, Reliability in En!ineerin! (esi!n, pp HI1E, John Wiley J 5ons, <ew Kork. Cnowles, :.I.,#*GG6$, )%ould *e "ove &+ay $rom ,&cceptable $ailure Rate,, =ommunications in 7elia ility (aintaina ility and 5upporta ility, Lol. 0, <o. *, 8. 01, International 7(5 =ommittee, 25, (ac:iarmid, 8restonM (orris, 5eymourM et al., #no date$, Reliability Tool-it. /ommercial Practices Edition, pp 16I1G, 7elia ility ,nalysis =enter and 7ome Ba oratory, 7ome, <ew Kork. 7ausand, (. and Aoyland, ,., #0EE4$, )ystem Reliability T%eory0 "odels# )tatistical met%ods# and &pplications, John Wuiley J 5ons, <ew Kork. we site %urner, %., Aockley, =., and )urdaky, 7., #*GGD$, T%e /ustomer Needs & "aintenance1$ree 2peratin! Period, *GGD ,vionics =onference and 3xhi ition, <o. GD@EH*G, 8. 0.0, 37, %echnology Btd., Beatherhead, 5urrey, 2C

[edit] /nline

(ondro, (itchell J, #June 0EE0$, ?,pproximation of (ean %ime )etween Failure When a 5ystem has 8eriodic (aintenance?, 'EEE Transactions on Reliability, v 6*, no 0. #availa le from (I%73 =orp.$ Reliability Prediction of Electronic Equipment, (IB@A:)C@0*DF#0$, #:&: download site.$ )athtu curve issues y ,5N=. Fault %olerant =omputing in Industrial ,utomation y Au ert Cirrmann, ,)) 7esearch =enter, 5wit>erland

[edit] #xternal lin,s

2senet F,N a out (%)F 7elia ility and ,vaila ility )asics 8roduct failure ehaviour and wear out .hide/

vOdOe

*tatistics
.show/ 0escripti!e statistics 1ocation Continuous data (ean #,rithmetic, !eometric, Aarmonic$ 2 (edian 2 (ode

7ange 2 5tandard deviation 2 =oefficient of variation 2 0ispersion 8ercentile 2 Interquartile range *hapeLariance 2 5kewness 2 Curtosis 2 (oments 2 B@moments

Cate3orical dataFrequency 2 =ontingency ta le )ar chart 2 )iplot 2 )ox plot 2 =ontrol chart 2 =orrelogram 2 Forest *tatistical 3raphicsplot 2 Aistogram 2 N@N plot 2 7un chart 2 5catter plot 2 5templot 2 7adar chart .show/ 0ata collection 0esi3nin3 studies 3ffect si>e 2 5tandard error 2 5tatistical power 2 5ample si>e determination

*ur!e" methodolo3"5ampling 2 5tratified sampling 2 &pinion poll 2 Nuestionnaire Controlled experiment :esign of experiments 2 7andomi>ed experiment 2 7andom assignment 2 7eplication 2 )locking 2 &ptimal design

Uncontrolled studies<atural experiment 2 Nuasi@experiment 2 & servational study .show/ *tatistical inference

4a"esian inference Classical inference

8rior 2 8osterior 2 =redi le interval 2 )ayes factor 2 )ayesian estimator 2 (aximum posterior estimator =onfidence interval 2 Aypothesis testing 2 5ampling distri ution 2 (eta@analysis

P@test #normal$ 2 5tudent-s t@test 2 F@test 2 =hi@square test 2 8earson-s *pecific testschi@square 2 Wald test 2 (annIWhitney 2 2 5hapiroIWilk 2 5igned@ rank (ean@un iased 2 (edian@un iased 2 (aximum likelihood 2 (ethod 5eneral of moments 2 (inimum distance 2 (aximum spacing 2 :ensity estimation estimation .show/ Correlation and re3ression anal"sis 8earson product@moment correlation 2 7ank correlation Correlation#5pearman-s rho, Cendall-s tau$ 2 8artial correlation 2 =onfounding varia le 1inear re3ression 6on7standard predictors 5eneralized linear model 5imple linear regression 2 &rdinary least squares 2 !eneral linear model 2 ,nalysis of variance 2 ,nalysis of covariance <onlinear regression 2 <onparametric 2 5emiparametric 2 Isotonic 2 7o ust 3xponential families 2 Bogistic #)ernoulli$ 2 )inomial 2 8oisson .show/ 0ata anal"ses and models for particular data t"pes (ultivariate regression 2 8rincipal components 2 Factor analysis 2 =luster analysis 2 =opulas :ecomposition 2 %rend estimation 2 )oxIJenkins 2 ,7(, models 2 5pectral density estimation 5urvival function 2 CaplanI(eier 2 Bogrank test 2 Failure rate 2 8roportional ha>ards models 2 ,ccelerated failure time model

8ulti!ariate statistics Time series anal"sis *ur!i!al anal"sis

#n!ironmental statistics!eostatistics 2 =limatology 8edical statistics3pidemiology 2 =linical trial 2 =linical study design *ocial statistics ,ctuarial science 2 8opulation 2 :emography 2 =ensus 2 8sychometrics 2 &fficial statistics 2 =rime statistics Cate3or" 2 .ortal 2 /utline 2 Index 7etrieved from ?http:++en.wikipedia.org+wiki+FailureQrate? =ategories: ,ctuarial science R 5tatistical terminology R Failure R 7elia ility engineering R 5urvival analysis R (aintenance R 5tatistical ratios Aidden categories: ,rticles lacking in@text citations from <ovem er 0EEG R ,ll articles lacking in@text citations R 5tatistics articles with navigational template
.ersonal tools

<ew features Bog in + create account

6amespaces

,rticle :iscussion

9ariants 9iews ctions *earch

7ead 3dit Liew history

5earch
6a!i3ation

(ain page =ontents Featured content

=urrent events 7andom article

Interaction Toolbox

, out Wikipedia =ommunity portal 7ecent changes =ontact Wikipedia :onate to Wikipedia Aelp

What links here 7elated changes 2pload file 5pecial pages 8ermanent link =ite this page

.rint:export

=reate a ook :ownload as 8:F 8rinta le version

1an3ua3es

:eutsch 3spaSol 3uskara FranTais Italiano UVWWXYZ 5imple 3nglish 5venska [X\]^_W`X] %his page was last modified on 9 July 0E*E at E*:14. %ext is availa le under the =reative =ommons ,ttri ution@5hare,like BicenseM additional terms may apply. 5ee %erms of 2se for details. Wikipediaa is a registered trademark of the Wikimedia Foundation, Inc., a non@ profit organi>ation. =ontact us

8rivacy policy , out Wikipedia :isclaimers