Oral Presentation Ninth International Symposium on Life-Cycle Civil Engineering 2025

Time-dependent failure probability analysis considering discrete load based on surrogate model (112310)

Haopeng qiao 1 , Zhaohui Lu 1 , Chunqing Li 2
  1. Beijing University of Technology, Beijing, BEIJING, China
  2. Royal Melbourne Institute of Technology, Melbourne, Australia

Abstract: In time-dependent reliability analysis it is always the case that the structure subjected to a variety of loading conditions and environmental influence. Seismic action, characterized by its instantaneous duration and high intensity, will induce a violent structural response and consequently, makes the limit state function (LSF) non-differentiable with respect to the whole service period. This serve as one of the main obstacles for time-dependent structural seismic reliability evaluation based on outcrossing rate method which require the LSF continuously differentiable. Further, traditional methods may struggle with high computational costs when dealing with time-history analysis, which is necessary to obtain the structural response under the influence of seismic action.Therefore, a new formulation should be proposed to handle these two difficulties. On the one hand, based on the minimum value of LSF at the time of outcrossing, this study intends to propose a new algorithm to calculate the outcrossing rate within a period when structural performance alter drastically. With this, time-dependent reliability analysis could be conducted with non-differentiable LSF. On the other hand, the Kriging surrogate model, featured as a powerful tool for enhancing the estimation of uncertainty quantification based on Gaussian process interpolation, is adopted here to represent the relationship between seismic intensity and response as part of the LSF. By using the Kriging model, the numbers of time-history analysis can be significantly reduced. The objective of this study is that, first, precisely construct the LSF of a certain failure scenario with Kriging model represents the seismic response. secondly, using the introduced algorithm to calculate the outcrossing rate in the context of non-differentiable LSF and no need to call time-history analysis. Then, accurately and effectively estimate the time-dependent seismic failure probability of structures through the integration of outcrossing rate.

Key words: time-dependent seismic reliability; seismic action; outcrossing rate; Kriging model; time-history analysis.