Banca de DEFESA: RODRIGO SANTOS DE OLIVEIRA

Uma banca de DEFESA de MESTRADO foi cadastrada pelo programa.
STUDENT : RODRIGO SANTOS DE OLIVEIRA
DATE: 04/02/2022
TIME: 10:00
LOCAL: Auditório Programa de Pós-Graduação em Engenharia Civil (on line)
TITLE:

Structural Reliability Analysis Using Selective Monte Carlo Method

KEY WORDS:

Monte Carlo, Reliability Analysis, Selective Monte Carlo, Structural Analysis

PAGES: 77
BIG AREA: Engenharias
AREA: Engenharia Civil
SUMMARY:

The Monte Carlo (MC) method has as main advantages its robustness and simplicity. In practical problems of structural reliability, however, the sample size can make the MC unfeasible due to the high computational cost for calculating the failure probability. Several methods based on MC were developed in order to reduce this problem, such as Importance Sampling (IS), Subset Simulation (SuS) and Separable Monte Carlo (SepMC). In this work, the selective Monte Carlo method (SMC) is proposed, aiming at problems whose fault function is locally monotonic in the region of interest, which happens in most structural engineering problems. The SMC reduces the number of failure function evaluations () of the MC by calculating only a part of the points, performing a search for the failed points in the failure direction of each variable. Two structural problems and 13 benchmark examples are analyzed, where the for a given Coefficient of Variation is counted and compared with the MC, IS, SuS and SepMC. Furthermore, a parametric study of the impact of the number of variables of the failure function in the proposed method is carried out. The results showed that the reduction in by SMC can reach more than 99.9% in relation to MC and was greater than that of SuS and SepMC for all problems, being also greater than that of IS for most problems. It is worth noting that when the failure function is monotonic, for the same sample, the SMC and the MC result in the same failure probability value.

BANKING MEMBERS:
Externo à Instituição - ANDRÉ JACOMEL TORII - UNILA
Interno - 2216046 - RAMIRO BRITO WILLMERSDORF
Presidente - 2250444 - RENATO DE SIQUEIRA MOTTA