112-2 Reliability and Risk Assessment

Homework #6 100 1100

May 06, 2024 (due May 20, 10 am)

Problem #6.1 (30 points)

Considering the given fault tree as below, list all the minimal cut sets and draw the equivalent
minimal cut set fault tree. (Hint: start from the Truth table by Boolean algebra)

Problem #6.2 (70 points)

The telecommunications tower of Fig. 6.2 is subject to seismic actions. In
particular, the monthly hazard curve in terms of peak ground acceleration is
given by:
𝑘𝑘𝑆𝑆
𝜈𝜈
𝐻𝐻𝑆𝑆 (𝑠𝑠) = 𝑃𝑃𝑟𝑟 (𝑆𝑆 > 𝑠𝑠) = 1 − exp �− � 𝑆𝑆 � � for 𝑠𝑠 ≥ 0
𝑠𝑠

with 𝜈𝜈𝑆𝑆 = 0.36 [m/s2] and 𝑘𝑘𝑆𝑆 = 4.9. The displacement at the top of the tower D
is related to the peak ground acceleration S through the following load effect
model:

𝐷𝐷 = 𝑎𝑎𝑆𝑆 𝑏𝑏

where 𝑎𝑎 = 0.15 [s2] and 𝑏𝑏 = 1.6.

The displacement level L at which the tower loses functionality is to be taken as

a normal random variable with mean 0.2 [m] and standard deviation 0.01 [m]:

Considering S and L to be Statistically Independent random variables:

(a) Estimate the probability of failure and reliability index for 𝑡𝑡𝑙𝑙 = 10 years
using first order reliability methods (FORM).
Fig. 7.1
6.2
(b) Check your solution using Monte Carlo simulation.
(c) Calculate the Mean Recurrence Interval of failure.

HW6-1
112-2 Reliability and Risk Assessment

Hints for Problem #6.2:

HW5

HW6-2
 6 1
.

0
T =
EIE2
=
( A + E 3 ) ( C+ E4 )
=
( A + ( B + C )λ ( + E 4 )
E1 × E2 ) =
A C + A [4 + ( B + C)
· - . C + ( Btc ) E 4
.

=
AC +
AE 4 + BC + CC + BE4 + CE 4
=
( AC + CB + C + CE4) + AE 4 + BE 4

( E4 )
=
C + AE 4 + BE 4
)
N + E3 = C + AB + AB
=
C +
AB
B
T

(B+ C)

A B C T

0 0 0 A B
1 0 0 0

+ 0 0

1 0 1
:
… 1

… (

1 " 1
。
 6 .
2 Hs ( s ) = P ( Ss )
)s

Fs ( 3 ) =
P ({≤ )
s =
exp [. 造 ] Type 丘

(G ) FD( d ) O .s
* * *
FB ( d ) 0s ,
k;

FORM
* *
G : L -
B

L ~
N (o .
2 ,
0 . 01 )
*
ks
*
D mexp [ -
[) a ]
6
D =
0 . 15
S

( 5d)
“
s =
0. i

[ ({品藏 ]9
六 vo

品絮 vos expFl 哇 j }
) *

zexp [ -
*s
Fr { d ) exp [ 管] ]
l ]
exp =
-

=
- = ]

= el [ 30 奶 l了 wg
P) -

exp [
*
FD (d )

γs
*
= 0. 1385 |ks * = 3 0625
.
,

{ x} { L d}
=

,
→
{u }
{u .
,
u0 }

恐
L L - 0 .
2

U,
=
=
0 . 01

L = 0 . 01 U + 0. r

*
1ks
[]
" “
Uz = Ψ [ exp [ -

] = 中 [ F0* (

ks *
中 ( uz ) =
e×
p [ [量]
-

*
ks
(張)
o
l ( ≡( ) ) u
=

此使 ( 2) )爽 u = (海

*
爽
d = os [ ln ( 更 \ 以 ]

Gu ( u , u 2 ) =
O O( U+
.
o .2 -0
s

中* \ 更 ( u)
爽
)

u ,
= 0 .
01

*+ )
{ ln llu
1

=
*
0 )} }
× <① luz)
[
Ψ lu)
r
「
子
》
》
 exp o 啦 ]
[ ). 0
lc) F* d ) =
( {

=
expE 照}
* *
Vd = 0. 065 ,
Ks = 3 . 0625

Pf =
Q .
330

MRI =
占310 =
30 26
.

yrs
*

According to (a)
, change Vd 比 1
yr
口
 (a)
25
clear all
% problem
-----------------------------------------------------------------
% L = Normal
% D = Weibull
mu_L = 0.2;
sig_L = 0.01;

v_D = 0.1385;
k_D = 3.0625;
mu_D = v_D*gamma(1 - 1/k_D);

% Step 1 - write limit state in terms of u --------------------------------

% Gu = (0.01*u1+0.2) - (v_D*((-log(cdf('normal',u2,0,1)))^(-1/k_D)));

% Step 2 - find partial dervatives of Gu ----------------------------------

DGu1 = @(u1,u2) (0.01);
DGu2 = @(u1,u2) ((-v_D*((-log(cdf('normal',u2,0,1)))^(-1/k_D-1))*pdf('normal',u2,0,1))/
(k_D*cdf('normal',u2,0,1)));

% Step 3 - pick initial guess for x* --------------------------------------

L_d(1) = mu_L;
D_d(1) = mu_D;

u1_d(1) = (L_d(1) - mu_L)/sig_L;

u2_d(1) = icdf('normal',exp(-(v_D/D_d(1))^k_D),0,1);

for i = 1:5
% step 4 - evaluate dervatives in u* ----------------------------------
D1 = DGu1(u1_d(i),u2_d(i));
D2 = DGu2(u1_d(i),u2_d(i));

alfa_1(i) = D1/sqrt((D1^2 + D2^2));

alfa_2(i) = D2/sqrt((D1^2 + D2^2));

% step 5 - solve limit state Gu = 0 for BetaHL ------------------------

Gu_B = @(Beta_FORM) ((0.01*(-alfa_1(i)*Beta_FORM)+0.2))-(v_D*((-log(cdf('normal',(-
alfa_2(i)*Beta_FORM),0,1)))^(-1/k_D)));
Beta_FORM_k(i) = fsolve(Gu_B,0);

% step 6 - given alfa_i and BetaHL_i find u*_(i+1) --------------------

u1_d(i+1) = -alfa_1(i)*Beta_FORM_k(i);
u2_d(i+1) = -alfa_2(i)*Beta_FORM_k(i);
 end

% Iteration history -------------------------------------------------------

figure(1)
plot(1:length(Beta_FORM_k),Beta_FORM_k,'-bs')
xlabel('Iteration')
ylabel('\beta_{FORM}')

% Beta and FORM estmate of Pf ---------------------------------------------

Beta = Beta_FORM_k(end)
Pf_FORM = cdf('normal',-Beta_FORM_k(end),0,1)
%
------------------------------------------------------------------------
-
% END
---------------------------------------------------------------------

Beta =

0.5880

Pf_FORM =

0.2783

(b)
5 clc
clear
close all
% basic random variables
N = 10000;

% Uniform numbers
r = rand(2,N);

% Simulation of the basic random variables

L_i = icdf('normal',r(1,:),0.2,0.01);

v_D = 0.1385;
k_D = 3.0625;
D_i = v_D*(-log(r(2,:))).^(-1/k_D);
 % Limit State Function
G_i = L_i - D_i;

% Failure Probability
n = 0;
for i = 1:length(G_i)
if G_i(i)<=0
n = n + 1;
end
end
n
Pf = n/N
% histogram(G_i,100,'Normalization','pdf')

Pf =

0.2804

⑳ (c)
clear all
% problem
-----------------------------------------------------------------
% L = Normal
% D = Weibull
mu_L = 0.2;
sig_L = 0.01;

v_D = 0.065841;
k_D = 3.0625;
mu_D = v_D*gamma(1 - 1/k_D);

% Step 1 - write limit state in terms of u --------------------------------

% Gu = (0.01*u1+0.2) - (v_D*((-log(cdf('normal',u2,0,1)))^(-1/k_D)));

% Step 2 - find partial dervatives of Gu ----------------------------------

DGu1 = @(u1,u2) (0.01);
DGu2 = @(u1,u2) ((-v_D*((-log(cdf('normal',u2,0,1)))^(-1/k_D-1))*pdf('normal',u2,0,1))/
(k_D*cdf('normal',u2,0,1)));

% Step 3 - pick initial guess for x* --------------------------------------

L_d(1) = mu_L;
D_d(1) = mu_D;

u1_d(1) = (L_d(1) - mu_L)/sig_L;

u2_d(1) = icdf('normal',exp(-(v_D/D_d(1))^k_D),0,1);
for i = 1:5
% step 4 - evaluate dervatives in u* ----------------------------------
D1 = DGu1(u1_d(i),u2_d(i));
D2 = DGu2(u1_d(i),u2_d(i));

alfa_1(i) = D1/sqrt((D1^2 + D2^2));

alfa_2(i) = D2/sqrt((D1^2 + D2^2));

% step 5 - solve limit state Gu = 0 for BetaHL ------------------------

Gu_B = @(Beta_FORM) ((0.01*(-alfa_1(i)*Beta_FORM)+0.2))-(v_D*((-log(cdf('normal',(-
alfa_2(i)*Beta_FORM),0,1)))^(-1/k_D)));
Beta_FORM_k(i) = fsolve(Gu_B,0);

% step 6 - given alfa_i and BetaHL_i find u*_(i+1) --------------------

u1_d(i+1) = -alfa_1(i)*Beta_FORM_k(i);
u2_d(i+1) = -alfa_2(i)*Beta_FORM_k(i);
end

% Iteration history -------------------------------------------------------

figure(1)
plot(1:length(Beta_FORM_k),Beta_FORM_k,'-bs')
xlabel('Iteration')
ylabel('\beta_{FORM}')

% Beta and FORM estmate of Pf ---------------------------------------------

Beta = Beta_FORM_k(end)
Pf_FORM = cdf('normal',-Beta_FORM_k(end),0,1)
MRI = 1/Pf_FORM
%
------------------------------------------------------------------------
-
% END
---------------------------------------------------------------------

Beta =

1.8378

Pf_FORM =

0.0330
MRI =

30.2612