Stochastic modelling of non-stationary environmental loads for reliability analysis under the changing climate

Abstract

In recent years, a rapid pace of climate change is becoming evident by a marked increase in the frequency and intensity of weather extremes, and this trend is expected to continue with an increase in global warming for decades to come. The paper presents the non-homogeneous Poisson process as a basis to model an environmental load process induced by non-stationary climate conditions. The distribution of maximum load generated by a non-stationary process is derived, which explicitly accounts for the loading frequency and intensity amplification over time. Using this distribution, structural reliability can be computed for any given time interval in future. The paper shows that traditional measures like the return period, extreme percentiles, and the annual probability of failure, would vary with time under the changing climate. Examples presented in the paper show that a modest amplification of the load magnitude has a drastic negative effect on reliability than that of the frequency of loading. Probabilistic models presented in this paper will be useful in revising currently used “stationary” design codes and ensuring a consistent level of safety in the non-stationary climate.