Sei sulla pagina 1di 3

In the Name of God

Structural Reliability and Probabilistic Modeling

Mojtaba Mahsuli

Assignment 3
Deadline: Monday, Esfand 19, 1392

Problem 1
Probability density functions
Comment on the validity of the following probability density functions:

Problem 2
Cumulative distribution functions
A random variable X has the mean =90 with 30% coefficient of variation.
a) Compute the probabilities for the events {X>120} and {X>160} assuming the following
distributions: Normal, lognormal, and uniform. It is recommended to employ Rt for this purpose.
b) Considering the results from Part (a), which of the probabilities P(X>120) or P(X>160) is most
sensitive to the choice of distribution type? Explain why this is the case.

Page 1 of 3

Structural Reliability and Probabilistic Modeling

Mojtaba Mahsuli

Problem 3
Bayesian updating
Consider the case of cracks in a weld of a structural member. The cracks have random lengths
denoted by a with the PDF

f (a ) e a
where =20mm-1. An X-ray device is used to detect the cracks. The probability that a crack will be
detected depends on its length and is given by

16a 2
P crack will be detected | A a
1

for 0 a 0.2mm
for a 0.2mm

a) Determine the PDF of the length of a crack that has been detected. You may use the following
relationship to compute the integral:

eax 2 2
x e dx a3 a x 2ax 2
2

ax

b) Determine the PDF of the length of a crack that has escaped detection.
c) Plot and compare the above two PDFs together with the PDF of the crack length before
detection.

Problem 4
Geometrical inference

A civil engineering facility is located in the vicinity of three seismic sources, as shown in the
figure below. An earthquake in the region may occur on any of the three sources. Historical
records indicate the following mean seismic activity rates (mean number of earthquakes of
magnitude greater than 4 per year): 0.4 per year for the area source, 0.3 per year for fault 1, and
0.1 per year for fault 2. Assume that the earthquake is modeled as a point. Furthermore, assume
that if the earthquake occurs in a source, then the point is equally likely to occur anywhere
within that source. Let R denote the distance of the site from an earthquake that may occur in
the region. Derive an expression for the CDF and PDF of R. Make a sketch of the PDF and CDF.

Fault 1

160 km

160 km
55 km
80 km

40 km

Facility
Fault 2

Page 2 of 3

120 km

Structural Reliability and Probabilistic Modeling

Mojtaba Mahsuli

Problem 5
Bayesian inference
Consider the occurrence of strong earthquakes in a particular region. There is scarce information
about past earthquakes, but it is tentatively suggested that they follow a Poisson process. That is,
assuming that the events occur randomly and independently of each other, the Poisson distribution
is used to model the number of earthquake events in time. The PMF for x number of events in the
time interval T reads

t
p ( x)
x!

e t

It is assumed that the return period is 100 years, i.e., the occurrence rate is =0.01. However, the
confidence in this value is low and a 100% coefficient of variation is assigned to this estimate.
a) To prepare for Bayesian updating of the occurrence rate, select a probability distribution for the
rate that yields a conjugate prior. Plot this distribution for the information that is given above.
b) Suppose it becomes known that there has been one earthquake in the past 150 years. Update
the probability distribution of and display it in the same plot.
c) Next, suppose by carrying out a more comprehensive research, it becomes known that, rather
than the information in Part (b), there have been a total of two earthquakes in the past 300
years. Update the probability distribution of and display it in the same plot.
d) Next, suppose more research reveals that, rather than the information in Parts (b) and (c), there
have been a total of four earthquakes in the past 600 years. Update the probability distribution
of and display it in the same plot.
e) Write the expression for the predictive PMF that incorporates the uncertainty in the
occurrence rate, i.e., the probability distribution of . Carry out the integration and write the
final expression for the predictive PMF.

Page 3 of 3

Potrebbero piacerti anche