Research on reliability of early-warning methods for collapse prediction of steel portal frames under fire

To consider the effects of uncertainties such as fire scenarios, load levels on the accuracy of early-warning methods for collapse in real fire, a method based on reliability theory was proposed herein to predict the remaining time of burning frame. A three-dimensional finite element model verified by experiments was used to simulate the collapse process of steel portal frames under fire. Based on the four collapse modes of steel portal frames under fire and the three-level early-warning methods, the influence of the heating condition of steel member, foot stiff, load ratio et al, on the prediction of remaining time before collapse was analyzed. Through Monte Carlo sampling, the statistical parameters with a certain probability guarantee rate were explored to ensure the reliability of the early warning index of steel portal frames under fire. The results show that the ratio of remaining time to collapse time, which can be determined before fire, is only related to the reliability level and collapse modes regardless of the values of uncertainties. The predetermined remaining time ratios with a reliability level of 60%-80% are advised to be adopted for fire rescue to predict the actual remaining time of the burning frame. © 2023 Science Press. All rights reserved.