3 Engineering Reliability
3 Engineering Reliability
3 Introduction to
Engineering Reliability
3.1 NEED FOR RELIABILITY
The reliability of engineering systems has become an important issue during their
design because of the increasing dependence of our daily lives and schedules on the
satisfactory functioning of these systems. Some examples of these systems are
aircraft, trains, computers, automobiles, space satellites, and nuclear power–generating
reactors. Many of these systems have become highly complex and sophisticated.
For example, today a typical Boeing 747 jumbo airplane is made of approximately
4.5 million parts, including fasteners [1]. Most of these parts must function normally
for the aircraft to fly successfully.
Normally, the required reliability of engineering systems is specified in the design
specification, and during the design phase every effort is made to fulfill this requirement
effectively. Some of the factors that play a key role in increasing the importance of
reliability in designed systems are the increasing number of reliability- and quality-
related lawsuits, competition, public pressures, high acquisition cost, past well-
publicized system failures, loss of prestige, and complex and sophisticated systems.
This chapter presents various introductory aspects of engineering reliability.
λ (t ) = θ λ β t β−1 + (1 − θ) b t b − 1 α eα t b (3.1)
23
Hazard
rate (time
dependent
failure
rate)
0 time t
TABLE 3.1
Reasons for the Occurrence of Failures in the Three Regions of the
Bathtub Hazard Rate Curve
Region Reason
dR(t )
f (t ) = − (3.2)
dt
where t is time, f(t) is the failure (or probability) density function, and R(t) is the
item reliability at time t.
Example 3.1
Assume that the reliability of an item is defined by the following function:
R (t ) = e − λ t (3.3)
d e− λ t
f (t ) = −
dt (3.4)
= λ e− λ t
f (t )
λ (t ) = (3.5)
R (t )
1 dR(t )
λ (t ) = − ⋅ (3.6)
R (t ) dt
Example 3.2
Using Equation 3.3, obtain an expression for the item’s hazard rate and comment
on the resulting expression.
Substituting Equation 3.3 into Equation 3.6 yields
1 d e− λ t
λ (t ) = − −λt
e dt (3.7)
=λ
1
− λ (t ) dt = ⋅ dR(t ) (3.8)
R (t )
Integrating both sides of Equation 3.8 over the time interval [0, t], we get
t R (t )
∫ ∫ R (t) ⋅ dR(t)
1
− λ (t ) dt = (3.9)
0 1
because at t = 0, R(t) = 1.
Evaluating the right-hand side of Equation 3.9 yields
∫
ln R(t ) = − λ (t ) dt
0
(3.10)
Thus, from Equation 3.10, we get the following general expression for reliability
function:
t
∫
− λ ( t ) dt
R (t ) = e 0 (3.11)
Equation 3.11 can be used to obtain the reliability of an item when its times to
failure follow any time-continuous probability distribution.
Example 3.3
Assume that the time to failures of an automobile is exponentially distributed and
its failure rate is 0.003 failures per hour. Calculate the automobile’s reliability for a
10-hour mission.
Using the data values in Equation 3.11 yields
10
∫ ( 0.003) dt
−
R (10) = e 0
= e − ( 0.003) (10 )
= 0.9704
This means there is an approximately 97% chance that the automobile will not
fail during the 10-hour mission. More specifically, its reliability will be 0.9704.
∫
MTTF = R (t ) dt
0
(3.12)
or
∫
MTTF = t f (t ) dt
0
(3.13)
or
where s is the Laplace transform variable, MTTF is the mean time to failure, and
R(s) is the Laplace transform of the reliability function R(t).
Example 3.4
Prove by using Equation 3.3 that Equation 3.12 to Equation 3.14 yield the same
result for MTTF.
Thus, by inserting Equation 3.3 into Equation 3.12, we get
∫
MTTF = e − λ t dt
0 (3.15)
1
=
λ
de − λ t
f (t ) =−
dt (3.16)
= λ e − λt
∫
MTTF = t λ e − λ t dt
0
∞
∞ e − λt
= − t e − λ t − − (3.17)
0
λ 0
1
=
λ
R (s ) = ∫e
0
−st ⋅ e − λ t dt
(3.18)
1
=
s+ λ
1
MTTF = s→0
lim
(s + λ )
(3.19)
1
=
λ
Equation 3.15, Equation 3.17, and Equation 3.19 are identical, proving that
Equation 3.12 to Equation 3.14 give the same result.
1 2 3 k
Rs = P ( E1 E2 E3 ......Ek ) (3.20)
where Ej denotes the successful operation (i.e., success event) of unit j for j = 1, 2,
3, …., k; Rs is the series system reliability; and P ( E1 E2 E3 ...... Ek ) is the occurrence
probability of events E1, E2, E3, …, and Ek.
For independently failing units, Equation 3.20 becomes
Rs = P ( E1 ) P ( E2 ) P ( E3 )...... P ( Ek ) (3.21)
Rs = R1 R2 R3 ...... Rk (3.22)
R j (t ) = e − λ j t (3.23)
k
− ∑ λj t
Rs (t ) = e j=1
(3.24)
1
MTTFs =
5 (0.0025)
= 80 hours
Thus, the system reliability and mean time to failure are 0.5353 and 80 hours,
respectively.
(
Fps = P E1 E2 .... Ek ) (3.26)
where Fps is the parallel system failure probability, E j denotes the failure (i.e., failure
(
event) of unit j, for j = 1, 2, …, k, and P E1 E2 E3 .... Ek is the occurrence proba-)
bility of events E1 , E2 , ......,and Ek .
Rps = 1 − Fps
(3.29)
= 1 − F1 F2 .....Fk
(
Rps (t ) = 1 − 1 − e − λ 1 t ) (1− e ) ...... (1− e )
−λ 2 t −λ k t
(3.30)
∫ ( ) dt
k
MTTFps = 1 − 1 − e − λ t
0
(3.31)
k
∑
1 1
=
λ j =1
j
where MTTFps is the parallel system mean time to failure and λ is the unit constant
failure rate.
Example 3.6
A system is composed of three independent and identical subsystems. At least one
of the subsystems must operate normally for the system to work successfully.
Calculate the system’s reliability if each subsystem’s probability of failure is 0.1.
By substituting the given data into Equation 3.29 we get
∑ ( ) R (1− R)
n
n− j
Rm = n
j
j (3.32)
n
j=m
where
( ) = ( n −nj!)! j !
n
j (3.33)
where R is the unit reliability and Rm/n is the m-out-of-n network reliability.
For constant failure rates of the identical units, substituting Equation 3.3 into
Equation 3.32 yields
∑ ( ) e (1− e )
n
n− j
Rm (t ) = n
j
−jλt − λt (3.34)
n
j=m
where Rm/n(t) is the m-out-of-n network reliability at time t and λ is the unit failure rate.
Substituting Equation 3.34 in Equation 3.12 yields
∞ n
∫ ∑( ) ( ) n− j
MTTFm = n e − j λ t 1 − e − λt dt
j
n
0 j=m (3.35)
n
∑
1 1
=
λ j=m
j
)∑j
1 1
MTTFm =
n ( 0.0035 j=3
1 1 1
= +
( )
0.0035 3 4
=166.67 hours
FIGURE 3.5 Block diagram of a standby system with one operating and k standby units.
or other means detect the failure and then replace the failed unit with one of the
standby units. Figure 3.5 shows the block diagram of a standby system with one
operating and k standby units. Each block in the diagram denotes a unit.
Using the Figure 3.5 diagram for independent and identical units, perfect detec-
tion, switching mechanisms and standby units, and time-dependent unit failure rate,
we write down the following expression for system reliability [6]:
K t − ∫ λ (t ) dt
j t
Rsb (t ) = ∑ ∫
j = 0 0
λ (t ) dt e 0
/ j!
(3.36)
where Rsb (t) is the standby system reliability at time t and λ (t) is the unit time-
dependent failure rate.
For constant unit failure rate, (i.e., λ (t) = λ), Equation 3.36 becomes
Rsb (t ) = ∑ (λ t ) e
j=0
j −λ t / j! (3.37)
∞ K
MTTFsb =
∫∑
0 j=0
(λ t ) j e − λ t / j ! dt
(3.38)
K +1
=
λ
Calculate the system reliability for a 100-hour mission and mean time to failure if
the standby unit remains as good as new in its standby mode and failure detection
and unit replacement mechanisms are 100% reliable.
By substituting the given data into Equation 3.37 we get
∑ ( 0.0045)(100) e ( ) ( ) / j!
j
− 0.0045 100
Rsb (100) =
j=0
= 0.9246
1+ 1
MTTFsb = = 444.44 hours
( 0.0045 )
Thus, the standby system reliability and mean time to failure are 0.9246 and
444.44 hours, respectively.
Rb = 2 R1 R2 R3 R4 R5 + R2 R3 R4 + R1 R3 R5 + R1 R4 + R2 R5
(3.39)
− R2 R3 R4 R5 − R1 R2 R3 R4 − R1 R2 R3 R5 − R3 R4 R5 R1 − R1 R2 R4 R5
1 2
4 5
Rb = 2 R5 − 5 R 4 + 2 R3 + 2 R 2 (3.40)
For constant unit failure rate, substituting Equation 3.3 into Equation 3.40 yields
where Rb (t) is the bridge network reliability at time t and λ is the unit constant
failure rate.
By substituting Equation 3.41 into Equation 3.12, we get
49
MTTFb = (3.42)
60 λ
49
MTTFb =
60(00.0075)
= 108.89 hours
Thus, the bridge network’s reliability and mean time to failure are 0.4552 and
108.89 hours, respectively.
Over the year, many reliability allocation methods have been developed [8–12].
One of the commonly used methods in the industrial sector is described below [1].
Failure
criticality Complexity/
time
Factors
Environment State-of-the-
art
assignment of 10 means the item under consideration is most affected by the influ-
ence factor in question, and 1 means the item is least affected by the same factor.
Subsequently, the reliability is allocated on the basis of the weight of these assigned
numbers for all influence factors considered.
Finally, reliability of an item is allocated by giving certain weights to both
similar familiar systems and factors of influence methods. The hybrid method is
more effective than both these methods used alone because it uses data from both
of them.
3.6 PROBLEMS
REFERENCES
1. Dhillon, B.S., Design Reliability: Fundamentals and Applications, CRC Press, Boca
Raton, FL, 1999.
2. Kapur, K.C., Reliability and maintainability, in Handbook of Industrial Engineering,
Salvendy, G., Ed., John Wiley & Sons, New York, 1982, pp. 8.5.1–8.5.34.
3. Dhillon, B.S., A hazard rate model, IEEE Transactions on Reliability, 28, 150, 1979.
4. Shooman, M.L., Probabilistic Reliability: An Engineering Approach, McGraw-Hill,
New York, 1968.
5. Dhillon, B.S., Reliability, Quality, and Safety for Engineers, CRC Press, Boca Raton,
FL, 2005.
6. Sandler, G.H., System Reliability Engineering, Prentice Hall, Englewood Cliffs, NJ,
1963.
7. Grant Ireson, W., Coombs, C.F., and Moss, R.Y., Eds., Handbook of Reliability
Engineering and Management, McGraw-Hill, New York, 1996.
8. Frederick, H.E., A reliability prediction technique, Proceedings of the Fourth National
Symposium on Reliability and Quality Control, 314–317, 1958.