Advanced analysis methods are presently being considered for adoption in U.S. design specifications for steel frames. Strength predictions by advanced analysis are expected to be more accurate than those made by LRFD provisions, which can ultimately lead to more efficient designs. This paper compares the structural reliability of 16 two-bay, two-story frames designed by conventional LRFD methods to those designed with advanced analysis. Gravity loads and member yield stresses are modeled as random variables. Non-linear structural simulations are used to generate strength distributions, from which reliabilities are estimated based on first-order methods and importance sampling. Bias factors and resistance factors necessary to achieve a specific target reliability are calculated for each frame. The results provide insight into the probabilistic differences between design by advanced analysis and elastic-LRFD methods, and begin to provide probabilistic justification for resistance factors used with advanced analysis.
展开▼
