RELIABILITY ENGINEERING AND APPLICATIONS TO POWER SYSTEMS

Reliability
Reliability is the probability of a device performing its purpose adequately for
the period of time intended under the operating conditions encountered.
This definition breaks down into four basic parts:
-Probability,
-Adequate performance,
-Time,
-Operating conditions.
The ·first part, probability, provides the numerical input for the assessment
of reliability and also the first index of system adequacy. In many instances it
is the most significant index, but there are many more parameters calculated
and used. These are discussed later.
The other three parts--adequate performance, time and operating conditions-
-are all engineering parameters, and probability theory is of no assistance in
this part of the assessment. Often, only the engineer responsible for a
particular system can satisfactorily supply information relating to these.
The 'time intended' may be continuous or very sporadic and the 'operating
conditions' may be perfectly uniform or extremely variable, as in the
propulsion phases associated with space rockets and in the take-off, cruising
and landing of commercial flights.
Probability
The word 'probability' is used frequently in a loose sense implying that a
certain event has a good chance of occurring
It is important to realize that it has a strict technical meaning and is a
scientific 'measure of chance', i.e., it defines quantitatively the likelihood of
an event or events. Mathematically it is a numerical index that can vary
between zero which defines an absolute impossibility to unity which defines
an absolute certainty.
For example, the probability that a man will live for ever is zero, and the
probability that one day he will die is unity.

MUQTHIAR ALI 1
 RELIABILITY ENGINEERING AND APPLICATIONS TO POWER SYSTEMS

Example 1. Consider a coin and the probability of getting a head or a tail

in a single toss.
In this example s = f = 1.
Therefore the probability of getting a head or a tail in a single throw is 1/2.
Example 2.Consider a die and the probability of getting a 4 from a single
throw.
If a 4 is called success, then s = 1 and f = 5 since there are five ways of not
getting a 4.
Therefore the probability of getting a 4 is 1/6 and the probability of not
getting a 4 is 5/6.
Example3 Consider 2 dice and the probability of getting a total of 9 spots
in a single throw of both dice.
In this case, the successful outcomes are: (3+6),(4+5),(5+4) and (6+3)=4
ways=s
The failed outcomes are:
(1 + 1), (1 +2), (1 +3), (1 +4), (1 +5), (1 +6)
(2+ 1), (2+2), (2+3), (2+4), (2+5), (2+6)
(3+ 1), (3+2), (3+3), (3+4), (3+5)
(4+ 1), (4+2), (4+3), (4+4), (4+6)
(5 + 1), (5 +2), (5 + 3), (5 + 5), (5 +6)

MUQTHIAR ALI 2
 RELIABILITY ENGINEERING AND APPLICATIONS TO POWER SYSTEMS

(6+1),(6+2), (6+4),(6+5),(6+6)=32 ways=f

Therefore, the probability of getting a total of 9 spots in a single throw of two
dice = 4/36 = 1/9
and similarly, the probability of not getting a total of 9 spots is 32/36 = 8/9.
Example 4.From 6 men and 5 women, how many committees of 6
members can be formed when each committee must contain at least 3
women? The condition that at least 3 women must be present on each
committee is satisfied in three ways, when (a) there are 3 women and 3
men, or (b) there are 4 women and 2 men, or (c) there are 5 women and
1 man.
This example is not concerned with the order, only the total composition of
the committees.
a) It can be selected in 5C3 x 6C3 ways
b) (b) can be selected in SC4 X 6C2 ways
c) (c) can be selected in sCs X 6C1 ways

Giving the total number of committees as (sC3 x 6C3) + (sC4 x 6C2) + (sCs x
6Cl) = 281
This number of possible committees can be compared with 462 committees
when no conditions are imposed on the composition of each committee.

Rules for combining probabilities

Rule 1-Independent event.

Two events are said to be independent if the occurrence of one event does not
affect the probability of occurrence of the other event.

Example 5. Throwing a die and tossing a coin are independent events since
which face of the die is uppermost does not affect the outcome of tossing a
coin.

Rule 2-Mutually exclusive event.

Two events are said to be mutually exclusive (or disjoint) if they cannot
happen at the same time

Example 6 : When throwing a single die, the events 1 spot, 2 spots, 3 spots,
4 spots, 5 spots, 6 spots are all mutually exclusive because two or more
cannot occur simultaneously.

Rule 3-Complementary events

Two outcomes of an event are said to be complementary if, when one outcome
does not occur, the other must

MUQTHIAR ALI 3
 RELIABILITY ENGINEERING AND APPLICATIONS TO POWER SYSTEMS

Example 7. When tossing a coin, the outcomes head and tail are
complementary
since: P(head) + P(tail) = 1 or P(head) = P(tail)
Rule 4-Conditional events
Conditional events are events which occur conditionally on the occurrence of
another event or events.
Example : Consider two events A and B and also consider the probability of
event A occurring under the condition that event B has occurred.

This is described mathematically as P(A I B) in which the vertical bar is

interpreted as GIVEN and the complete probability expression is interpreted
as the 'conditional probability of A occurring GIVEN that B has occurred'.

Rule 5-Simultaneous occurrence of events

The simultaneous occurrence of two events A and B is the occurrence of BOTH
A AND B.
Example
In this rule there are two cases to consider, the first is when the two events
are independent and the second is when they are dependent.

(a) Events are independent

If two events are independent, then the probability of occurrence of each
event is not influenced by the probability of occurrence of the other.
In this case P(A / B) = P(A) and P(B / A) = P(B) ,
the probability that they both occur is: P(A n B) = P(A) . P(B)

MUQTHIAR ALI 4
 RELIABILITY ENGINEERING AND APPLICATIONS TO POWER SYSTEMS

Example 8 An engineer selects two components A and B. The probability that

component A is good is 0.9 and the probability that component B is good is
0.95.
Therefore the probability of both components being good is:
P(A good n B good) = P(A good) . P(B good) = 0.9xO.95 =0.855
(b) Events are dependent
If two events are not independent, then the probability of occurrence of one
event is influenced by the probability of occurrence of the other.
Thus: p(AnB)=p(B / A) ·P(A) =P(A/B) ·P(B) (2.11)
Example 9. One card is drawn from a standard pack of 52 playing cards.
Let A be the event that it is a red card and B be the event that it is a court or
face card. What is the probability that both A and B occur
P(A) = 26/52 Given that A has occurred,
the sample space for B is 26 states of which 6 states are those of a court or
face card
Therefore P(B I A) = 6/26
P(A n B) = 6/26 x 26/52 = 6/52
Alternatively P(B) = 12/52
Given that B has occurred,
the sample space for A is 12 states of which 6 states are those of a red card,
Therefore P(A I B) = 6/12
and p(AnB) = 6/12 x 12/52 = 6/52
Rule 6-Occurrence of at least one of two events
The occurrence of at least one of two events A and B is the occurrence of A
OR B OR BOTH. and is expressed as: (AUB), (A OR B) or (A+B)
(a) Events are independent but not mutually exclusive
There are two ways in which the appropriate equation can be deduced; using
an analytical method and using the Venn diagram. Consider first the
analytical technique.
P(A U B) = P(A OR B OR BOTH A AND B)
= 1-P(NOT A AND NOT B)
=1-p(AnB) =1-P(A) 'P(B)
= 1-(1-P(A»· (1-P(B)

MUQTHIAR ALI 5
 RELIABILITY ENGINEERING AND APPLICATIONS TO POWER SYSTEMS

= P(A) + P(B)- P(A)· P(B)

= P(A) + P(B) - P(A n B)
Example 10: An engineer selects two components A and B. The probability
that component A is good is 0.9 and the probability that component B is good
is 0.95.
The probability that component A or component B or both is good:
P(A good U B good) = P(A good) + P(B good) - P(A good) . P(B good)
= 0.9+0.95-0.9x 0.95 =0.995
(b) Events are independent and mutually exclusive
In the case of events A and B being mutually exclusive, then the probability
of their simultaneous occurrence P(A) . P(B) must be zero by definition.
Therefore P(A U B) = P(A) + P(B)
Rule 7-Application of conditional probability
This principle can be extended to consider the occurrence of an event A which
is dependent upon a number of mutually exclusive events Bi
p(AnB)=p(A IB)· P(B)
Using Equation
, the following set of equations can be deduced for each Bi
P(A n B1) = P(A I B1) • P(B1)
P(A n B2) =P(A I B2) • P(B2)
P(A n B;) =P(A I B;) . P(BJ)
P(A n Bn) = P(A I Bn) . P(Bn)
On combing

Example 11
A certain item is manufactured at two plants. Plant 1 makes 70% of the
requirement and plant 2 makes 30%. From plant 1, 90% meet a
particular standard and from plant 2 only 80%.
Evaluate, (a) out of every 100 items purchased by a customer, how many
will be up to standard and, (b) given that an item is standard, what is the
probability that it was made in plant 2?

MUQTHIAR ALI 6
 RELIABILITY ENGINEERING AND APPLICATIONS TO POWER SYSTEMS

consider A as the event that the item is up to standard, Bl as the event that
the item is made in plant 1 and B2 as the event that the item is made in plant
2.
Therefore, p(AIB1)=0.9,
p(AIB2)=0.8,
P(B1) =0.7,
P(B2) =0.3
P(A) =0.9xO.7+0.8xO.3
=0.63 + 0.24 = 0.87
and out of every 100 items purchased by the customer 100xO.87 = 87
b) The probability that the item comes from plant 2 given that it was standard,
is p(B2 1 A).
in this case when i = 2).
From part (a), the probability that the component is standard AND comes
from plant 2,
P(A n B2) = 0.24 and the probability that it is standard, P(A) = 0.87. Therefore
P(B I A) = p(AnB2) 2 P(A) = 0.24/0.87 = 0.276

Example 11
The probability that a 30-year-old man will survive a fixed time period
is 0.995. An insurance company offers him a $2000 insurance policy for
this period for a premium of $20. What is the company's expected gain?
Gain = + $20.00 if man lives = -$1980.00
if man dies Probability that he lives = 0.995
Probability that he dies = 0.005 /1.000
expected gain = (+20) xO.995 +(-1980) xO.005 =$10
This expected gain must be positive and greater than some minimum value
for the company to make a profit.
Properties of the binomial distribution
From the previous section, the binomial distribution can be represented by
the general expression: (p+q)"
For this expression to be applicable, four specific conditions are required.
These are: (a) there must be a fixed number of trials, i.e., n is known,

MUQTHIAR ALI 7
 RELIABILITY ENGINEERING AND APPLICATIONS TO POWER SYSTEMS

(b) each trial must result in either a success or a failure,

i.e., only two outcomes are possible and p + q = 1,
(c) all trials must have identical probabilities of success and therefore of
failure, i.e., the values of p and q remain constant, and
(d) all trials must be independent (this property follows from (c) since the
probability of success in trial i must be constant and not affected by the
outcome of trials 1,2, ... , (i -1))

AND

Example 12
.A coin is tossed 5 times. Evaluate the probability of each possible
outcome and draw the probability mass (density) function and the
probability distribution function.
In this example
n = 5,
p = q = 1/2.

MUQTHIAR ALI 8
 RELIABILITY ENGINEERING AND APPLICATIONS TO POWER SYSTEMS

Example 13.
Consider the case in which the probability of success in a single trial is
1/4 and four trials are to be made. Evaluate the individual and
cumulative probabilities of success in this case and draw the two
respective probability functions.
In this example
n = 4,
p = 1/4,
q = 3/4,

Expected value and standard deviation

Two of the most important parameters of a distribution are the expected, or
mean value, and the standard deviation. The binomial distribution is a
discrete random variable and therefore the expected value and standard
deviation can be evaluated
First consider the expected value

MUQTHIAR ALI 9
 RELIABILITY ENGINEERING AND APPLICATIONS TO POWER SYSTEMS

As the contribution to this summation made by x = 0 is zero then

But

Therefore, for a binomial distribution, the expected number of successes is

equal to the number of trials multiplied by the probability of success or
conversely, the expected number of failures is equal to the number of trials
multiplied by the probability of failure.
Now consider evaluation of the standard deviation

MUQTHIAR ALI 10
 RELIABILITY ENGINEERING AND APPLICATIONS TO POWER SYSTEMS

Example 14 A product is claimed to be 90% free of defects. What is the

expected value and standard deviation of the number of defects in a
sample of 4?
In this example n = 4, p(defect) = 0.1, q(no defect) = 0.9.
E(defects) = 4 x 0.1 = 0.4
u(defects) = √4 x 0.1 x 0.9 = 0.6
The same result can be achieved in a more tedious way by the direct
application

MUQTHIAR ALI 11
 RELIABILITY ENGINEERING AND APPLICATIONS TO POWER SYSTEMS

Poisson distribution
The Poisson distribution represents the probability of an isolated event
occurring a specified number of times in a given interval of time or space
when the rate of occurrence (hazard rate) in a continuum of time or space is
fixed. The occurrence of events must be affected by chance alone. A particular
characteristic feature of the Poisson distribution is that only the occurrence
of an event is counted, its non-occurrence is not. This is one of the essential
differences between the Poisson and binomial distributions because in the
latter, both the occurrence and non-occurrence of an event must be counted.
Examples of this are
-Number of lightning strokes in a period
-Number of telephone calls in a period
-Number of faults on a system.
Derivation of the Poisson distribution
The Poisson distribution which has the hazard rate is constant, the hazard
rate is generally termed the failure rate, a term which is more widely
recognized by system engineers.
ʎ = average failure rate or average number of failures per unit time.
Let dt be a sufficiently small interval of time such that the probability of more
than one failure occurring during this interval is negligible and can be
neglected.
Therefore, ʎ dt = probability of failure in the interval dt, i.e., in the period (t,
t+dt)
(a) Zero failures
Let Px(t) be the probability of failure occurring x times in the interval (0, t),
then because:

MUQTHIAR ALI 12
 RELIABILITY ENGINEERING AND APPLICATIONS TO POWER SYSTEMS

Probability of zero failures in the interval (0, t + dt)

= probability of zero failures in the interval (0, t) x probability of zero failures
in the interval (t, t+dt).
Po(t+dt) = Po(t)(1- A dt)
Assuming event independence

This is the first term of the Poisson distribution and gives the probability of
zero failures occurring in a specified time period.
It shows that if

MUQTHIAR ALI 13
 RELIABILITY ENGINEERING AND APPLICATIONS TO POWER SYSTEMS

It is assumed however that the interval dt is sufficiently small that the

probability of more than one failure in this interval is negligible.

Expected value
The expected value of a discrete distribution is given by

since the summation of the probabilities for all x must be unity

Therefore the equation is also written as

if µ is defined as the expected value E(x).

EXAMPLE 15
In a large system the average number of cable faults per year per 100 km of
cable is 0.5. Consider a specified piece of cable 10 km long and evaluate the
probabilities of 0, 1, 2, etc., faults occurring in (a) a 20 year period, and (b) a
40 year period. Assuming the average failure rate data to be valid for the 10
km cable and for the two periods being considered, the expected failure rate
, is,

MUQTHIAR ALI 14
 RELIABILITY ENGINEERING AND APPLICATIONS TO POWER SYSTEMS

(a) for a 20 year period, E(x) = 0.05 x 20 = 1.0

Relationship of Poisson and the binomial distribution

The probability of an event succeeding r times in n trials was given in

Therefore
np = ʎt and r = x

MUQTHIAR ALI 15
 RELIABILITY ENGINEERING AND APPLICATIONS TO POWER SYSTEMS

The standard deviation of the binomial distribution was

If it is assumed that the value of p is very small, i.e., the condition which
makes the Poisson and binomial distributions equivalent, then

Example 16 The probability of success in a single trial is 0.1. Calculate

the probability that in 10 trials there will be exactly two successes using
(a) the binomial distribution, and (b) the Poisson distribution
(a) P(2) = 10C2 0.12 x 0.98 10! 2 8

= 0.1937
(b) np=10xO.l=1.0 1.02
therefore

Repeat Example 16. when the number of trials is 20 and the probability
of success in a single trial is 0.005.

The normal distribution

The normal probability distribution, sometimes referred to as the Gaussian
distribution, is probably the most important and widely used distribution in
the entire field of statistics and probability. Although having some important
applications in reliability evaluation, it is of less significance in this field than
many other distributions.
The probability density function of a normal distribution may be expressed in
general terms as:

MUQTHIAR ALI 16
 RELIABILITY ENGINEERING AND APPLICATIONS TO POWER SYSTEMS

The discussion concerning probability density functions indicates that for a

continuous distribution

The exponential distribution

The exponential, or strictly the negative exponential, distribution is probably
the most widely known and used distribution in reliability evaluation of
systems.
The most important factor for it to be applicable is that the hazard rate should
be constant, in which case it is defined as the failure rate ʎ. This is essentially
the same requirement as that of the Poisson distribution and it can be argued
that the negative exponential distribution is only a special case of the Poisson
distribution.

Exponential reliability functions. (a) Areas showing Q(t) and R(t).

MUQTHIAR ALI 17
 RELIABILITY ENGINEERING AND APPLICATIONS TO POWER SYSTEMS

(c) Failure density function. (c) Cumulative failure distribution. (d) Hazard
rate

The failure density function f(t), cumulative failure distribution Q(t) and
hazard rate ʎ(t)are shown in Figures, c and d, respectively.
Mean value and standard deviation
The expected value of a continuous random variable having a range (0, ∞) is
given by

MUQTHIAR ALI 18
 RELIABILITY ENGINEERING AND APPLICATIONS TO POWER SYSTEMS

The Weibull distribution

The Weibull distribution, in common with a small number of other
distributions such as the gamma and lognormal distributions, has one very
important property; the distribution has no specific characteristic shape

MUQTHIAR ALI 19
 RELIABILITY ENGINEERING AND APPLICATIONS TO POWER SYSTEMS

There are two particular cases that can be deduced from the Weibull
distribution;
the first is when β= 1 and the second when β = 2.
a) For β = 1.
In this case, the above equations reduce to

Equations are identical to those for the exponential distribution if a = 1/ ʎ,

that is, the value of a represents the mean time to failure (MITF).
(b) For β = 2.
In this case, Equations 6.39 and 6.42 reduce to

The expected value of the Weibull distribution is given by

MUQTHIAR ALI 20
RELIABILITY ENGINEERING AND APPLICATIONS TO POWER SYSTEMS

MUQTHIAR ALI 21