ENGI 9411

MODELING UNCERTAINTY
PART IV
Discrete Random Variables
Binomial Distribution
In many engineering applications, events consisting of repeated trials can be formulated in terms of
occurrence or nonoccurrence, success or failure, good or bad, etc. Only two outcomes are possible,
representing the behavior of a discrete random variable. In addition, if the events satisfy the additional
requirements of a Bernoulli sequence, that is, if they are statistically independent and the probability of
occurrence or nonoccurrence of events remains constant, they can be mathematically represented by the
binomial distribution. If the probability of occurrence of an event in each trial is p and the probability of
nonoccurrence is (1-p), then the probability of x occurrences out of a total of n trials can be described by
the PMF of a binomial distribution as:

= , ) = 1 = 0, 1, 2, , 5

Where the binomial coefficient:

n!
= 6
[x! n x !]

2
Binomial Distribution
CDF:

1 = 0,1,2, , (7)
=
=0
1 >
The mean of a binomial distribution:
E X = 8

The variance:
= 1 (9)

3
Example 1
Suppose the probability of failure of a structure due to earthquakes is estimated as 105 per year.
Assuming that the design life of the structure is 50 years and the probability of failure in each year
remains constant and independent during its life time, calculate the probability of
i) No failure
ii) At least a failure
throughout the design life.

Solution
i) 50 = = 0, 50 105 )
50
= 105 0 1 105 500
0
50!
= 1 105 50 = 0.99950
0! 50 0 !

ii) 50 = 1 50
= 1 0.99950 = 0.00050
4
Example 2
The drainage system of a city has been designed for a rainfall intensity that will be exceeded on an average
once in 50 years. What is the probability that the city will be flooded
i) 2 out of 10 years?
ii) at most 2 out of 10 years?
Solution
1
i) = = 0.02
50
2 10 = = 2, 10 0.02)
10
= 0.02 2 1 0.02 102
2
= 0.015
ii) 2 10

= = 0, 10 0.02 + = 1, 10 0.02 + = 2, 10 0.02)

10 0 10
10 1 9
10
= 0.02 0.98 + 0.02 0.98 + 0.02 2 0.98 8
0 1 2

5
= 0.817 + 0.617 + 0.015 = 0.999
Geometric Distribution
If an event occurs in a Bernoulli sequence and p is the probability of occurrence in each trial, then the
probability that the event will occur for the first time at the ith trial which implies that there was
no occurrence in the previous (t - 1) trials is given by the geometric distribution as:
PMF:
1
= = 1 = 1,2, 10
0
CDF:

= 1
= 0,1,2, , 11
=0

The mean of a geometric distribution:

1
= 12

The variance of a geometric distribution:
1
= 2 (13)

6
Poisson Distribution
The Poisson distribution is commonly used in engineering and the sciences that deal with the occurrence of
some random event in the continuous dimension of time or space. For example, the number of occurrences
of natural hazards, such as earthquakes, tornadoes, or hurricanes, in some time interval, such as 1 year, can
be considered as a random variable with a Poisson distribution. In these examples, the number of
occurrences, x, in the time interval is the random variable. Therefore, the random variable is discrete,
whereas its reference space (i.e., the time interval) is continuous.
A Poisson process has the following properties:
1) Memoryless: the occurrence of an event does not affect the occurrence probability of an event in the
next time instance that means the occurrence rate of events (i.e., events per year) is constant
2) The time between events, T, follows an exponential probability distribution
3) Bernoulli process as n, the number of time interval, approaches infinity

PMF:
t
= = 0,1,2, (14)
!
0
7
Poisson Distribution
CDF:

t
= = 0,1,2, 15
!
=0

The mean of a Poisson distribution :

= 16

The variance of a Poisson distribution:

= (17)

8
Example 3
From records of the past 50 years, it is observed that tornadoes occur in a particular area an average of two
times a year. Determine:
i) The probability of no tornado in the next year
ii) The probability of 2 tornadoes in the next year
iii) The probability of no tornadoes in the next 50 years
iv) The probability of at least one tornadoes in the next 50 years

9
Example 4
For a large construction project, the contractor estimates that the average rate of on the job accidents is
three times per year. From past experience, the contractor also estimates that the cost incurred for each
accident may be modeled as a lognormal random variable with a median of $6,000 and COV of 20%. The
cost of each accident can be assumed to be statistically independent
i) What is the probability that there will be no accident in the first month of construction?
ii) What is the probability that only 1 out of the first 3 months of construction is free of accidents?
iii) What is the probability that an accident will incur a loss exceeding $4,000?

10