COMPARATIVE ANALYSES BETWEEN MONTE CARLO AND KER- NEL SMOOTHER METHODS TO EVALUATE STRUCTURAL RELIABILITY

Resumo

Among the methodologies widely used for structural reliability, some of them can be highlighted. The semi-
probabilistic approach, for example, considers the structure in a failure scenario, taking into account purely deter-
ministic variables and safety coefficients established by standards, without calculating the probability of failure.

Numerical methods, such as First-order reliability method (FORM), Second-order reliability method (SORM), and

First-order second-moment method (FOSM), calculate the probability of failure from the limit state equation consi-
dering stochastic factors. On the other hand, observational methods, such as Monte Carlo, simulate pseudo-random

scenarios and estimate the probability of failure by counting collapse occurrences. To guarantee the convergence of
this method, a sufficiently large number of simulations must be carried out, increasing the required computational
effort. Alternatively, the Kernel Smoother obsevational method may be used to calculate the probability of failure
(PoF) of structural systems demanding a computational effort considerably smaller than Monte Carlo. In that way,
the purpose of the present work is to establish a comparison between Monte Carlo and Kernel Smoother methods
by evaluating PoFs and reliability indexes calculated for numerical applications, by using both methodologies.