Titel_SS06

variables R and are acting as resistance variables as their -A 5 values are negative. The
failure probability for the steel rod is determined as PF, ( 3.7448) 9.02 10  ^5.

6.6 Correlated and Dependent Random Variables

The situation where basic random variables X are stochastically dependent is often
encountered in practical problems. For Normal distributed random variables, the joint
probability distribution function may be described in terms of the first two moments, i.e. the
mean value vector and the covariance matrix. This is, however, only the case for Normal or
Log-normal distributed random variables.

Considering in the following the case of Normal distributed random variables these situations
may be treated completely along the same lines as described in the foregoing. However,
provided that, in addition to the transformation from a limit state function expressed in
variables to a limit state function expressed in Uvariables, a transformation in between is
introduced where the considered random variables are first standardized before they are made
uncorrelated. I.e. the row of transformations yields:

X

XYU

In the following it will be seen how this transformation may be implemented in the iterative
procedure outlined previously.

Let us assume that the basic random variables are correlated with covariance matrix given
as:

X

     

X 1

1

1

n



112 , ... ,

,

n

nn

XCovXXCovXX

XVarX

^8

CX 8 

 

 1

9

9

(^89) 



Var

Cov

1

(6.20)

and correlation coefficient matrix X:

1

1

n

1

1

X







(^89) 

8

8 

9

 9 (6.21)

If only the diagonal elements of these matrixes are non-zero clearly the basic random
variables are uncorrelated.

As before the first step is to transform the -vector of basic random variables into a vector
of standardised random variables with zero mean values and unit variances. This operation
may be performed by

n X

Y

i , 1,2,..

X

Yini

i

iX

X







  (6.22)