Common Distributions

Physical quantities such as time of occurrence, completion, length, area,

temperature, pressure, load, intensity etc., when they need to be described
probabilistically, are modeled by discrete or continuous random variables
depending on the nature of the variables.
 Binomial/ Bernoulli distribution
events consist of repeated trials: occurrence or non occurrence, success or failure,
good or bad e.g. throw of coin: p+q=1

For a r.v. in total N trial k number of occurrence

 n  k (n  k )  n  k!
p X (k )    p q ,  , k  0,1, 2,3,...., n
k  k  k! (n  k )!

Mean & variance of

binomial distribution
For n=10
P=0.2
For n=5
 Poisson Distribution
The events which may occur
any point in time or in space
The events which may occur
any point in time or in space An earthquake can hit anytime
a structure in its life time

An accident can occur any

location in the highway

μ is the average occurrences of the event in x years

The Poisson distribution has the mean μ and the variance σ2 = μ.

It can be proved that Poisson distribution can be obtained as a limiting

case of the binomial distribution
if one let p → 0 and n → ∞ so that the mean μ = np approaches a finite
value.
 UNIFORM DISTRIBUTION
A continuous random variable X has a uniform distribution over an interval a to
b(b > a) if it is equally likely to take on any value in this interval.
Normal or Gaussian

Many RVs of practical

interest are normal
/Approximately normal
or can be transformed
to normal

Also It is a useful
approximation of more
complicated distributions
Standard Normal: u
X  u ( ,  X )      N (0,1)
X

1 u2
fU (u)= exp[ ]
2 2
u
1 u2
FU (u)=  exp[ ]du    φ(u )
-
2 2
 u Symmetric
1 u2
FU (u)=  exp[ ]du     (u )
-
2 2 (u )  1  (u )  p
u   1( p) or u   1(1  p)
when a = μ − σ and b = μ + σ , the right-hand side equals Φ(1) - Φ(-1); a = μ − 2σ
and b = μ + 2σ there corresponds the value Φ(2) - Φ(-2)

Using tabulated values of Φ function one thus finds

(a) P(μ -σ < X ≤ μ +σ) ≅ 68%
(b) P(μ -2σ < X ≤ μ +2σ) ≅ 95.5% (2.12)
(c) P(μ -3σ < X ≤ μ +3σ) ≅ 99.7%

(a) About 2/3 of the values will lie between μ -σ and μ +σ

(b) About 95% of the values will lie between μ -2σ and μ +2σ
(c) About 99¾ % of the values will lie between μ -3σ and μ +3σ.
The fractile xp that is defined as the value of the random variable X with
p non-exceedance probability (P(X ≤ xp) = p) is computed as follows:

x p = μ + k pσ
Log-normal distribution Hahn & Shapiro, 1967
In engg application r.v. can not be negative in most case - thus modeling as
lognormal is appropriate eliminating the possibility of negative value

If the r.v. ln X is normally distributed with mean μln X, standard deviation σ ln X,

then the r.v. X is log-normally distributed.

The CDF of r. v. lnX is of normal type

The pdf will be

m X and V X is the coefficient of variation of random variable X.

For computation we can use
Normal table, however

ln X  ln X
u
 ln X

Thus, probability of a lognormal

random variable having a value
between a and b can be obtained
as

P ( a  X  b)
 ln b  ln X   ln a  ln X 
      
  ln X
    ln X 
 GAMMA AND RELATED DISTRIBUTIONS
useful one-sided distributions gamma function

Incomplete gamma function

As it is one-sided, physical quantities that can

take values only in, say, the positive range are
frequently modeled by it.
It serves as a useful model as a wide variety of
shapes to the gamma density function can be
obtained by varying the parameters.
Exponential distribution
 BETA AND RELATED DISTRIBUTIONS
The lognormal and gamma distributions provide a diversity of one sided pdf

The beta distribution is rich in providing varied pdf over a finite interval

the beta function

 different combinations of the parameters of
their values permit the density function to
β=2
take on a wide variety of shapes.
 Distribution of extreme values Gumbel: “Statistics of extremes” 1958

The extreme values distribution is of

interest especially when one deals with
natural hazards like snow, wind,
temperature, floods, etc.

Let X j, j 1, 2, . . . , n, denote the jth gust

velocity of n gusts occurring in a year and
Yn denote the annual maximum gust
velocity.
We are interested in the probability
distribution of Y n in terms of those of X j .
CDF of Yn

Assuming Independent Xj

pdf of Yn

CDF of Zn

Assuming Independent Xj

pdf of Zn
Problem:
random variables Xj, j 1, 2, . , n, are independent & identically distributed
with CDF FX (x) and pdf fx (x) or pmf px(x), determine the forms of FYn (y)
and F Zn(z) as n-- α

Although the distribution functions FYn(y) and FZn(z) become

increasingly insensitive to exact distributional features of Xj as no
unique results can be obtained that are completely independent of
the form of F X (x)

Some features of the distribution function FX (x) are important, in

what follows, the asymptotic forms of FYn (y) & FZn(z) are
classified into three types based on general features in the
distribution tails of Xj.
 TYPE-I ASYMPTOTIC DISTRIBUTIONS OF EXTREME VALUES
Gumbel distribution

Distribution of max. values, limiting distribution of

Yn (y) from an initial distribution Fx (x) of which the
right tail is unbounded & is of an exponential type
A number of important distributions
fall into above category, such as the
normal, lognormal, and gamma
distributions
u and α are the location and scale
parameters of the distribution

It is interesting to note that the skewness

coefficient is approximately 1.1396 which
is independent of distribution parameters

This result indicates that the Type-I

maximum-value distribution has a fixed
shape with a dominant tail to the right
The fractile xp that is defined as the value of the random variable X with p non-
exceedance probability (P(X ≤ xp) = p) is computed as follows, given X follows
Gumbel distribution for maxima:

The values of k Gp for different non-exceedance probabilities

The distribution of Z n as n-α
Gumbel distribution for maxima in 1 year

mean μ x and
different cov Vx
Gumbel distribution for maxima in N years
Assuming that the occurrences of maxima are independent events

where: F(x)N years – CDF of r.v. X in N years, F(x)1 year – CDF of r.v. X in 1 year.

if the annual maxima in 1 year follow a Gumbel distribution for maxima then the
annual maxima in N years will also follow a Gumbel distribution for maxima:
TYPE-II ASYMPTOTIC DISTRIBUTIONS OF EXTREME VALUES
Such distribution of maximum values arises as the limiting distribution of Yn as n-α
from an initial distribution of the Pareto type, that is, the CDF FX(x) of each Xj is limited
on the left at zero & its right tail is unbounded

CDF

Extreme value I and II can be related as

Type-II asymptotic distribution of

minimum values arises under
analogous conditions. With CDF
FX (x) limited on the right at zero
& approaching zero on the left.
TYPE-III ASYMPTOTIC DISTRIBUTIONS OF EXTREME VALUES

The Type-III maximum-value asymptotic distribution is of limited practical interest

Type-III minimum-value asymptotic distribution is the limiting distribution of Z n as nα

for an initial distribution Fx (x) in which the left tail increases from zero near x
The distribution discussed is frequently used as a generalized time-to-failure
model for cases in which the hazard function varies with time. One can show
that the hazard function

its associated probability density function for T, the time to failure, is given by

Weibull distribution (Weibull, 1939). Clearly, it is a special case with ε=0

 Mean recurrence interval

The loads due to natural hazards such as earthquakes, winds, waves, floods
were recognized as having a randomness in time as well as in space.

The randomness in time was The recurrence interval (return period) is defined
considered in terms of the return as the average (or expected) time between two
period or recurrence interval. successive statistically independent events and it
is an estimate of the likelihood of events

The mean recurrence interval, MRI of a value x of the random variable

X may be defined as
 Thus the mean recurrence interval of a value x is equal to the reciprocal of the
annual probability of exceedance of the value x.

The mean recurrence interval or return period has an inverse relationship with
the probability that the event will be exceeded in any one year.

For example,

a 10-year flood has a 0.1 or 10% chance of being exceeded in any one year and
a 50-year flood has a 0.02 (2%) chance of being exceeded in any one year.

a 10-year earthquake will occur, on average, once every 10 years and that a 100-
year earthquake is so large that we expect it only to occur every 100 years.
The mean recurrence interval is often related with the
exceedance probability in N years. The relation among MRI,
N and the exceedance probability in N years, Pexc, N is: