You are on page 1of 10

First order reliability method (FORM)

Firstorderreliabilitymethod(FORM)
Limit
Limitstateg(X).Failurewheng<0.
state g(X) Failure when g<0
Linearapproximationoflimitstatetogether
with assumption that random variables are
withassumptionthatrandomvariablesare
normal.
Thenlimitstateisalsonormalvariable.
Th li i i l l i bl
Reliabilityindexisthedistanceofthemeanof
gfromzeromeasuredinstandarddeviations.
Approximation about mean
Approximationaboutmean
Predecessor
PredecessorofFORMcalledfirst
of FORM called firstorder order
secondmomentmethod(FOSM)
g ( X ) g ( X ) g ( X )T X X
Then, easy to show
1/2
n g 2

g g ( X ) g xi 2

i 1 xi
The
h reliability
li bili index
i d andd probabilityb bili off failure
f il are
g

g
Pf ( ) is the normal CDF
Beam under central load example
Beamundercentralloadexample
Probability
Probabilityofexceedingplasticmoment
of exceeding plastic moment
capacity g ( P, L,W , T ) WT PL / 4
P L W(plasticsection
W ( l i i T ( i ld stress))
T(yield
modulus)
mean 10kN 8m 0.0001m^3 600,000kN/m^2
Standarddeviation 2kN 0.1m 0.00002m^3 100,000kN/m^2
Reliability index for example
Reliabilityindexforexample
Usingthelinearapproximationget
Us g t e ea app o at o get
g g ( X ) 0.0001 600, 000 10 8 / 4 40kNm
g g
W 0.0001 T 600, 000
T W
g g
L / 4 2 P / 4 2.5
P L
1/2
n g 2
g x2i
i 1 xi

(0.0001100, 000) 2 (600, 000 0.000002) 2 2 2 (2.5 0.1) 2 16.13kNm


2

g
2.48
2 48
g
Chapter4ofCGCshowsthatifwechangeto
gg=T0.25PL/W
T 0.25PL/Wweget3.48instead(exactis2.46)
we get 3.48 instead (exact is 2.46)
Most probable point (MPP)
Mostprobablepoint(MPP)
Theerrorduetothelinearapproximationis
pp
exacerbatedduetothefactthattheexpansion
maybeaboutapointthatisfarfromthefailure
region (due to the safety margin)
region(duetothesafetymargin).
Hasofer andLindsuggestedremedyingthis
problembyfindingthemostprobablepointand
linearizing aboutit.
Thejointdistributionofalltherandomvariables
assigns a probability density to every point in the
assignsaprobabilitydensitytoeverypointinthe
randomspace.Thepointwiththehighestdensity
onthelineg=0istheMPP.
illustration
rr=randn(1000,1)*1.25+10;
randn(1000,1) 1.25+10;
c=randn(1000,1)*1.5+13;
20
f=@(x) x;
f=@(x)x;
fplot(f,[5,20])
hold on
holdon 15

plot(r,c,'ro')
c

xlabel('r') 10

ylabel('c')
5
5 10 15 20
r
RecipeforfindingMPPwith
independentnormalvariables
d d l bl
Transform
Transformintostandardnormalvariables
into standard normal variables
(zeromeanandunitystandarddeviation)
xi xi
ui
x i

Findthepointong=0ofminimumdistanceto
origin.ThepointwillbetheMPPandthe
i i Th i ill b h MPP d h
distancetotheoriginwillbethereliability
i d b d
indexbasedonlinearapproximationthere.
li i i h
min U TU
1/2

U g (U ) 0
Visual
Visual
General case
Generalcase
If
Ifrandomvariablesarenormalbutcorrelated,
random variables are normal but correlated
alineartransformationwilltransformthemto
independent variables
independentvariables.
Ifrandomvariablesarenotnormal,canbe
transformed to normal with similar probability
transformedtonormalwithsimilarprobability
offailure.SeeSection4.1.5ofCGC.
Project application
Projectapplication
The
Thetwothicknessvariablesarenormaland
two thickness variables are normal and
independent,sotransformingtoUsiseasy.
Tofindthemostprobablepointcanuse
To find the most probable point can use
optimizationineitherMatlab orEXCEL

You might also like