Reliability Theory

What is reliability?
Why it is important to study?
Use of reliability in day to day life.
Reliability may be defined in several ways:
The idea that an item is fit for a purpose with
respect to time.
In the most discrete and practical sense: "Items
that do not fail in use are reliable" and "Items that
do fail in use are not reliable.
Reliability of a device is denoted as and
defined as
( ) R t
| |
( ) , 0. R t P T t t = > >
Where T is a continuous random variable
representing the life length of the device.
Some property of the reliability function :
( ) R t
( ) ( ) ( )
t
i R t f u du

=
}
Where f is probability density function (p.d.f) of T.
We assume
( ) 0 0 f t for t = <
( ) ( ) ( ) ii R t f t
'
=
( ) ( ) is a nonincreasing function of iii R t t
( ) (0) 1 , ( ) 0 iv R R = =
( ) ( ) 1 ( ) where ( ) is the cdf of v R t F t F t T =
( ) ( ) 1 ( ) where ( ) is the cdf of v R t F t F t T =
( )
( ) R t P T t = >
( )
, ( ) 1 or R t P T t = s
, ( ) 1 ( ) or R t F t =
Hence the results
( ) (0) 1 , ( ) 0
( ) ( )
iv R R
v iv
= =

( ) ( ) is a nonincreasing function of
( ) ( )
iii R t t
iv iii
( ) ( ) ( )
( ) ( )
ii R t f t
v ii
'
=

( ) ( ) 1 ( )
( ) ( ) ( )
v R t F t
R t F t f t
=
' '
= =
Hence the results
Conditional failure (hazard) rate function:
It is denoted by
( ) ( ) Z t or h t
| |
0
( ) lim
t
P t T t t T t
Z t
t
A
< s + A >
=
A
| |
For small
( )
t
Z t t P t T t t T t
A
A ~ < s + A >
| |
For small
( )
t
Z t t P t T t t T t
A
A ~ < s + A >
( )
For small , ( ) approximates the conditional probability
that the device will fail during the interval , given
that it has survived at the time
t Z t t
t t t
t
A A
+ A
In particular, if
( ) Z t =
Hazard rate is constant
In that case we can say that the device is as good as
new at any time of its life.
( )
( ) Exerci .
( )
se:
f t
Showthat Z t
R t
=
( )
0
( ) lim
t
P t T t t T t
Z t
t
A
< < + A >
=
A
( ) ( )
( )
0
( ) lim
.
t
P t T t t T t
Z t
t P T t
A
( < < + A >

=
A >

( )
( )
0
( ) lim
.
t
P t T t t
Z t
t P T t
A
( < < + A

=
A >
0
( )
( ) lim
. ( )
t t
t
t
f u du
Z t
t R t
+A
A
=
A
}
0
( ) ( )
( ) lim
. ( )
t
F t t F t
Z t
t R t
A
+ A
=
A
( ) ( )
( ) .
( ) ( )
F t f t
Z t
R t R t
'
= =
Exercise :
Find the failure rate function for Weibull Distribution:
( )
( ) 1
t
F t e


=
( ) ( )
( ) ( )
( ) 1 ( )
f t F t
Z t Z t
R t F t
'
= =

( )
( )
1
( )
t
F t e t

'
=
( )
( )
( )
( )
1
1
( )
t
t
e t
Z t t
e

= =
Failure Patterns
Theorem:
If T is the time to failure of a device with pdf f(t)
andcdf F, andZ(t) isthefailureratefunctionof the
device, then:
0 0
( ) ( )
( ) , ( ) ( ) , for 0.
t t
Z s ds Z s ds
R t e f t Z t e t

} }
= = >
( ) ( ) ( )
( )
( ) ( ) ( )
f s F s R s
Z s
R s R s R s
' '
= = =
0
Integrating, Note that (0) 1
( )
( )
( )
t t
o
R
R s
Z s ds ds
R s
=
'
==
} }
( )
0
( )
ln ( ) ( ) ( )
t
t
Z s ds
o
R t Z s ds R t e

}
= =
}
( )
( )
( )
f t
Z t
R t
=
( ) ( ) ( ) f t Z t R t =
0
( )
( ) ( )
t
Z s ds
f t Z t e

}
=
Theorem 6.3:
Expected system life time: Theorem 6.3 shows how
mean/average lifetime of a system can be
determined from a knowledge of the reliability
function R(t)
( )
0
( )
Where is the time to failure
E T R s ds
T

=
}
To prove that, it is sufficient to show that
( ) ( )
0
E T P T t dt

= >
}
0
. . ( )
t
R H S f u du dt

(
=
(

} }
0 0
( )
u
f u dt du

(
=
(

} }
0
( ) . . uf u du L H S

= =
}
Example:
( ) /10 0 10
1 10
( )
F t t t
t
Find E T
= s s
= >
:
( ) 1 ( )
1 /10 0 10
0 0
Clealy
R t F t
t t
t
=
= s s
= >
( )
0
( ) E T R s ds

=
}
( )
10
0
1
10
s
E T ds
| |
=
|
\ .
}
( )
10
2
0
10 5 5
20
s
E T s
| |
= = =
|
\ .
Reliability of a system
Series System:- A system whose
components(say n components) are arranged in
such a way that the system fails whenever any
of its components fail is called a series system.
s
1
s
i
R
R rel i abi l i ty of seri es system
R R e the i th component
n
i
i
R
w h e r e
l i a b i l i t y o f
=
=

[
Parallel System
A system whose components are arranged in
such a way that the system fails only if all its
components fail is called a parallel system
s
1
s
i
R ( ) 1 (1 ( ))
where R reliability of parallel system
R Reliability of ith component of the system
n
i
i
t R t
=
=

[
Example 1: Consider aseries systemof threeindependent
components and thelifelength of each component follows
uniformlydistributionover (0,10).
(i) Findthereliabilityof thesystem.
(ii) Findtheexpectedsystemlife.
Solution:
( )
3 3
1 1
( ) ( ) 1 ( )
s i i
i i
R t R t F t
= =
= =
[ [
( ) /10 0 10
1 10
i
F t t t
t
= s s
= >
( )
( )
3
3
( ) 1 ( )
1 /10 0 10
0 10
s i
R t F t
t t
t
=
= s s
= >
( )
0
( ) E T R s ds

=
}
( )
3
10
0
1
10
s
E T ds
| |
=
|
\ .
}
( )
5
2
E T =
Example 2:
Consider a system consisting of eight components .
The system consists of five assemblies in series, where
assembly 1 consists of component 1 with reliability
0.99, assembly II consists component 2 and 3 in
parallel with reliabilities 0.95 and 0.95, assembly III
consists component 4,5,6 in parallel with reliabilities
0.96, 0.92, 0.85, assembly IV and V consist of
component 7 and 8 with reliabilities 0.95, 0.82
respectively. calculate system reliability.
Exercise VI.7 :
Let Y be the life of an independent series system with n
component. Find thepdf and cdf of Y. Also provethat thehazard
ratefunctionof thesystemisthesumof thehazardfunctionsof its
components.
Solution:
1
( ) ( )
n
s i
i
R t R t
=
=
[
1
( ) 1 ( ) 1 ( )
n
Y s i
i
F t R t R t
=
= =
[
1
( ) ( ) ( )
n
Y s i
i
d
f t R t R t
dt
=
| |
'
= =
|
\ .
[
1 1
, ( ) ( ) ( )
n n
Y i j
i j
j i
or f t R t R t
= =
=
| |
|
'
=
|
|
\ .
[
1 1
1
( ) ( )
( )
( )
( )
( )
n n
i j
i j
j i
Y
s
n
s
i
i
R t R t
f t
Z t
R t
R t
= =
=
=
| |
|
'

|
|
\ .
= =
[
[
1 2
1 2
( ) ( ) ( )
( )
( ) ( ) ( )
n
s
n
R t R t R t
Z t
R t R t R t
' ' '
=
1 2
1 2
( ) ( ) ( )
( )
( ) ( ) ( )
n
s
n
f t f t f t
Z t
R t R t R t
= + + +
1 2
( ) ( ) ( ) ( )
s n
Z t Z t Z t Z t = + + +
1
( ) ( )
n
s i
i
Z t Z t
=
=

Exercise VI.8.
Consider an n-component independent series system.
Giventhat thesecomponentshaveconstantfailurerates
respectively. Findthe
(i) reliabilityof thesystem.
(ii) hazardrateof thesystem.
(iii) pdf of thelife-lengthof thesystem.
Alsoexpressthemeanlifeof thesystemintermsof
meanlifeof itscomponents.
Solution:
( )
1 1
( ) ( ) 1 ( )
n n
s i i
i i
R t R t F t
= =
= =
[ [
1 2
, , ,
n

1
1
( )
n
i
i
i
t n
t
s
i
R t e e

=
| |
|
|

\ .
=

= =
[
1 2
( ) ( ) ( ) ( )
n
Z t Z t Z t Z t = + + +
1 2 n
= + + +
1
1
( ) ( ( ))
n
i
i
t n
s s i
i
d
f t R t e
dt

=
| |
|
|
\ .
=

| |
= =
|
\ .

Themeanlifelengthof thesystemis
where isthemeanlife-lengthof component.
1 2
1
1 1 1
n

=
+ + +
1 2
1
( )
s
n
E T

= =
+ + +
i

th
i
Exercise VI.9.
Let T be the life length of an independent parallel
systemwithn-components.
(i) Findthepdf andcdf of T
(ii) Showthat
Solution:
Hint: Usepropertiesof reliability, f(t) andtheir
definitions.
0 0
( ) ( )
1
1 1
t t
i
Z s ds Z s ds
n
i
e e

=
| |
} }
|
=
|
|
\ .
[
Example:
Probability that a device can survive after
0,1,2,3,4 or more shocks are 1,0.8,0.4,0.2, 0
respectively. If the arrival of shocks follow
Poisson distribution with =0.15, find R(10).
Solution:
Put =0.15 and t=10 in the above.
2 3
( ) ( )
( ) 1 0.8( ) 0.4 0.2
2 6
t
t t
R t e t

(
= + + +
(

CERTAIN LIFE MODELS
Exponential: In reliability, this distribution plays a
vital role. ThisisthedistributionduringphaseII.
Useful life phases of the device, when failure
occurs due to an external causes, which may be
calledShock/shocks
Suppose that a device fails due to shocks, which occur in a Poisson process
with a rate . Assume that the device fails as soon as the first shock arrives.
Then:

( ) 1 , 0
t
F t e t

= >
,
( ) , 0
0
t
or
f t e t
else where

= >
=
Thus the failure-time distribution is exponential distribution.
Let us assume, p is probability that even after the
shock, no failure occurs.
| |
( ) No failure occurs in the time interval (0, ] R t P t =
| |
0
( ) shocks arrive in time (0,t] device survives all shocks
n
R t P n n

=
=

( )
Using:
( ) ( )
B
P A B P A P
A
=
| |
0
The device will survives all shocks
( )
t
t
n
n
R t P X n P
X n

=
(
= =
= (

| |
0
( )
n
t
n
R t P X n p

=
= =

0
( )
( ) , 0
!
t n
n
n
e t
R t p t
n

=
= >

Where is the number of shocks which arrives in the time-interval

(0, ]. Under the assumption that the shocks arrive in (0, ] is a Poisson
process, then has a Poisson distribution with parameter
t
t
X
t t
X t
( )
0
( )
!
n
t
n
p t
R t e
n

=
=

(1 )
( ) ...........(*)
p t
R t e

=
Differentiate (*) with respect to t we get
(1 )
( ) ...........(*)
p t
R t e

=
(1 )
( ) ( ) (1 )
p t
R t f t p e

'
= =
(1 )
( ) (1 ) , 0
p t
f t p e t

= >
Thus has exponential distribution with parameter (1 ) T p
Put p=0, we get
( )
t
f t e

=
Hazard rate of exponential distribution
( )
t
f t e

=
,
( ) 1 , 0
t
or
F t e t

= >
( ) ( )
( )
( ) 1 ( )
f t f t
Z t
R t F t
= = =

This is the phase-II in failure pattern graph

Gamma Distribution
Suppose that the shocks occur in a Poisson Process. And
the device fails as soon as the r-th shock arrives
( ) ( ) R t P T t = >

( )
There will be no failure during the interval (0,t] P
( )
There will be less than -shocks in the interval (0,t] P r
:
be the random varible "Number of Shocks that arrive in (0,t]"
t
Let
X
Then Clearly distribution of follows poisson with
parameter
X
t
( ) ( 1)
t
R t P X r = s
( )
1
0
!
k
r
t
k
t
e
k

=
=

Differentiating with respect to t we get

( )
1
( ) , 0
( 1)!
r
t
t
R t e t
r

'
= >

( )
1
( ) , 0
( )
r
t
t
f t e t
r

= >
I
Which is a Gamma distribution
Failure rate of Gamma distribution
1
1
( )
( )
( )
r t
r x
t
f t t e
Z t
R t
x e dx

= =
}
( )
1
( ) , 0
( )
r
t
t
f t e t
r

= >
I
2
1
L et .
r x
t
D x e dx

| |
=
|
\ .
}
2 1
( 1 )
( )
r t r x r t
t
t e r t x e dx t e
Z t
D

(
+
(

'
=
}
1
Now let (t)=( 1 )
r x r t
t
r t x e dx t e




+
}
2
(t)= ( 1) , (using by parts)
r x
t
r x e dx

'

}
Case I : If 0 1. r < <
Then ( ) 0 for all 0.
( ) is an increasing function and
increases from a negative value to zero.
( ) 0 for all 0.
( ) for all 0.
( ) is a decresing function of .
Thus Gamma distributio
t t
t
t t
Z t t
Z t t

'
> >

< >
'
< >

n for (0 1)
can be used as a life- model for any device
when the device in phase-I.
r < <
Case II : If 1. r >
1
,
r
let t

<
1
( ) 0 ( ) 0
r
t Z t for t

'
> > <
1
, then (t)<0 as r>1.
( ) is decreasing function and
decreases from a positive value to zero.
( ) 0 for all 0.
( ) 0 for all 0.
( ) is an incresing function.
Thus Gamma distribution
r
let t
t
t t
Z t t
Z t

'
>

> >
'
> >

for ( 1)
can be used as a life- model for any device
when the device in phase-I I I .
r >
Case III : If 1. r =
Then the Gamma distribution is changed
to exponetial distribution with parameter .
( ) (constant function).
Thus Gamma distribution for ( 1)
can be used as a life- model for any device
when the dev
Z t
r

=
=
ice in phase-II.
Failure rate of Normal Distribution
Let T be the life length of any device which follows
normal distribution with mean and variance
Assume that, the mean is so large.

is follows standard normal distribution.

T

( 0) is neglegible, so that ( 0) 1. P T P T < > ~

2
.
} {
2
2
1
2
1
2
2
2
1
1 ( )
2
( )
(1 ( ))
2
t
t
t
e F t
e
Z t
F t

| |

|
\ .

| |

|
\ .
(

| |
(

|
( \ .

'
=

}
{
2
1
2
1
Now let (t)= 1 ( )
2
t
t
e F t

| |

|
\ .

| |

|
\ .
Case I : If 0 . t < s
Then 0, and ( ) 0 .
( ) 0 for all 0 .
( ) is an incresing function.
t
t
Z t t
Z t

s >
'
> < <

Case II : If . t >
1
Then ( ) (1 ( )).
( ) 0 for all .
( ) is a decreasing function and
decreases from a positive value to zero.
( ) 0 for all .
( ) 0 for all .
( ) is a increasing function of 0.
t F t
t t
t
t t
Z t t
Z t t

'
=
'
< >

> >
'
> >
>
Thus Normal distribution for ( large)
can be used as a life- model for any device
when the device in phase-I II .

Statistical Methods of Determining Failure Time

Distribution
using correctionfor continuity.
0.5

( ) ,
i
i
F t
n

( )
( ) ,
i i i
d i i i
i
n t n t t
f t t t t t
n t

( )
( ) ,
( )
i i i
d i i i
i i
n t n t t
Z t t t t t
n t t

Data Analysis to check suitability of

exponential distribution
Ex. V11:
100 components were put on life-test. The test
was terminated as soon as the 11
th
components
failed. The life-length of those components which
failed were as follows:
8, 16.1, 24.3, 32.55, 40.9, 57.8, 66.4, 75.1, 83.9,
92.85. Guess which distribution fits the data.
Estimate the parameters.
11 10 9 8 7 Failure No.
49.3 40.9 32.55 24.3 16.1 8 Time to
failure
92.85 83.9 75.1 66.4 57.8 Time to
Failure
6 5 4 3 2 1 Failure No.
Case-I:
0.00111 83.9-92.85 0.00119 32.55-40.9
0.00113 75.1-83.9 0.00121 24.3-32.55
0.00116 57.8-66.4 0.00123 8-16.1
0.00117 49.3-57.8 0.00125 0-8
0.00119 40.9-49.3
0.00114 66.4-75.1 0.00121 16.1-24.3
Time
interval
Time
interval
( )
d
f t
( )
d
f t
Case-II:
0.1109 92.85 0.0460 40.9
0.0998 83.9 0.0356 32.55
0.0780 66.4 0.0151 16.1
0.0672 57.8 0.00501 8
0.0566 49.3
0.0888 75.1 0.0253 24.3

ln(1 ( ))
i
F t
i
t
i
t

ln(1 ( ))
i
F t
Case-III:
0.00124 83.9-92.85 0.00124 32.55-40.9
0.00125 75.1-83.9 0.00126 24.3-32.55
0.00125 57.8-66.4 0.00125 8-16.1
0.00125 49.3-57.8 0.00125 0-8
0.00125 40.9-49.3
0.00125 66.4-75.1 0.00124 16.1-24.3
Time
interval
Time
interval
( )
d
Z t
( )
d
Z t
Since is almost constant, we can assume
exponential distribution fits the data. Now using
least square method we can determine the
parameter
Calculation for .
( )
d
Z t
.

11
1
( )

0.00125.
11
i
i
Z t

Data Analysis to check suitability of

weibull distribution with =0.
Weibull distribution with parameter
A random variable T is said to have Weibull
distribution if its density function is given by
( , , ).
1
,
( )
0,
t
t
e t
f t
t

| |

|
\ .

| |

>
|
=

\ .

Special case of Weibull distribution when

It can be proved that the hazard rate and reliability
of the above Weibull distribution are
0. =
1
( ) , 0.
t
t
f t e t

| |

|
\ .
| |
= >
|
\ .
1
( ) , ( ) , 0.
t
t
Z t R t e t

| |

|
\ .
| |
= = >
|
\ .
Consider the reliability .
( ) R t
( ) , 0.
t
R t e t

| |

|
\ .
= >
ln( ( )) ln( ( ))
1
ln ln ln( ) ln
( )
t t
R t R t
t
R t



| | | |
= =
| |
\ . \ .
| |
| |
=
|
|
\ .
\ .
Now plot
1
ln ln ln( ) ln
1 ( )
1
ln ln ln( ) ln

1 ( )
i
i
t
F t
t
F t


| |
| |
=
|
|

\ .
\ .
| |
| |
=
|
|
|
|

\ .
\ .
1
l n( ) , l n l n

1 ( )
i
i
t
F t
| |
| |
| |
|
|
|
|
|
|

\ .
\ .
\ .
If the points
lies almost on a straight line, we can assume
Weibull distribution with =0 fits the data.
How to calculate the parameter and .
Case-I (using graphical method)
slope of the above line
1
l n( ) , l n l n

1 ( )
i
i
t
F t
| |
| |
| |
|
|
|
|
|
|

\ .
\ .
\ .

=
and can be obtained by putting any value for
in the equation given below:
1 1

ln ln( ) ln ln

1 ( )
i
i
t
F t

i
t
Case-ii (using least square method )
Least square method
Suppose given. We want to fit one
straight line to the above data. Least square
method estimates and by solving
(1)
( , ), 1,2, ,
i i
x y i n =
b x a +
a
b
( )
2
( , )
1
min .
n
i i
a b
i
y bx a
=

Using derivative concept, we can easily say that

equation (1) gives minimum when
( )
2
2

i i i i
i i
i i
n x y x y
b
n x x
y b x
a
n

=



In our problem,
ln( )
1
ln ln

1 ( )
ln( )
i i
i
i
x t
y
F t
b
a

=
| |
| |
=
|
|
|
|

\ .
\ .
=
=
Using least square method, we can estimate

a
b
e

=
=
Consider the reliability .
( ) Z t
1
( ) , 0.
t
Z t t

| |
= >
|
\ .
ln( ( )) ln ( 1)ln( ) ( 1)ln
ln( ( )) ( 1)ln( ) ln ln
Z t t
Z t t

| |
= +
|
\ .
= +
Now plot
( )
( )
ln( ), ln ( )
i d i
t Z t
ln( ( )) ( 1)ln( ) ln ln
d i i
Z t t = +
If the points
lies almost on a straight line, we can assume
Weibull distribution with =0 fits the data.
How to calculate the parameter and .
Case-I (using graphical method)
slope of the above line
1+slope of the above line
( )
( )
ln( ), ln ( )
i d i
t Z t

1 =

=
and can be obtained by putting any value for
in the equation given below:
( )

ln ( 1)ln( ) ln ( )

ln

i d i
t Z t

+
=

i
t
Case-ii (using least square method )
Least square method
Suppose given. We want to fit one
straight line to the above data. Least square
method estimates and by solving
(1)
( , ), 1,2, ,
i i
x y i n =
b x a +
a
b
( )
2
( , )
1
min .
n
i i
a b
i
y bx a
=

Using derivative concept, we can easily say that

equation (1) gives minimum when
( )
2
2

i i i i
i i
i i
n x y x y
b
n x x
y b x
a
n

=



Using least square method, we can estimate

l n

a
b
e

= +
=
In our problem,
( )
ln( )
ln ( )
1
ln ln
i i
i d i
x t
y Z t
b
a

=
=
=
=
Ex. V12:
100 components were put on life-test. The test
was terminated as soon as the 7
th
components
failed. The life-length of those components which
failed were 2, 3.3, 4.7, 5.9, 7, 8, 9.
(a) Using hazard rate, check whether Weibull distribution fits
the data or not ? Estimate the parameters using
(i) graphical method.
(ii) least square method.
(b) Using cdf/reliability, check whether Weibull distribution fits
the data or not ? Estimate the parameters using
(i) graphical method.
(ii) least square method.
8
6
9 7 5.9 4.7 3.3 2 Time to failure
7 5 4 3 2 1 Failure No.
0.0105 7 - 8
0.0106 8 - 9
0.0094 5.9 7
0.0085 4.7 5.9
0.0072 3.3 4.7
0.0077 2 - 3.3
0.005 0 2
Time interval
( )
d
Z t
Solution (a):
-4.55 2.07 8
-4.66 1.94 7
-4.76 1.77 5.9
-4.85 1.19 3.3
-5.29 0.69 2
-4.54 2.19 9
-4.87 1.54 4.7
ln( )
i
t
ln( ( ))
d i
Z t
i
t
( ) ( )
ln( ),ln ( )
i d i
t Z t
since the points lies almost on
a straight line, we can assume Weibull
distribution fits the data.
a(i). Calculation for parameters.

1 1 0.01 1.01 slope = + = + =

( )

ln ( 1) ln( ) ln ( )

ln1.01 ( 0.01) 0.69 5.29

1.01

188.79
i d i
t Z t
e
e

+
+
=
= =
a(ii). Calculation for parameters
( )
7 7
1 1
7 7
2
1 1
ln(1.44) 7.13
1.44
ln( ), ln ( ) , 1, 2, , 7.
11.39, 33.52
20.24, 53.74

0.44, 7.13

1.44, 182.32.
i i i d i
i i
i i
i i i
i i
x t y Z t i
x y
x x y
b a
e
= =
= =
+
= = =
= =
= =
= =
= = =