System-based reliability analysis of stainless steel frames under gravity loads

Itsaso Arrayago,Kim J.R Rasmussen

doi:10.1016/j.engstruct.2020.111775

Itsaso Arrayago, Kim J.R Rasmussen

Open Access

https://doi.org/10.1016/j.engstruct.2020.111775

Copy DOI

Abstract

Current structural codes for steel and stainless steel structures such as AISC360-16, AISC370-21, AS/NZS4100 and Eurocode3 are based on the traditional two-step member-based design approach, in which internal actions are first obtained from a structural analysis, usually elastic, and the strength of each member and connection is subsequently checked using a structural design standard. However, the most recent versions of these standards already incorporate preliminary versions of the direct, or one-step, system-based design alternative, which is based on the design-by-analysis concept and allows evaluating the strength of structures directly from numerical simulations, although the standards in their current form do not provide reliability requirements for structural systems. Therefore, it is necessary to build a rigorous structural reliability framework to investigate acceptable target reliability indices for structural systems and to provide adequate system safety factors and system resistance factors. While this framework has been developed based on advanced Finite Element analysis for hot-rolled and cold-formed carbon steel structures in recent years in the form of the Direct Design Method (DDM), the framework does not exist for stainless steel structures. This paper presents an extension of the DDM to the analysis of stainless steel structures, in which system reliability calibrations are presented for six stainless steel portal frames under gravity loads covering the three most common stainless steel families and different failure modes using advanced numerical simulations. From the derived reliability calibrations, suitable system safety factors γM,s and system resistance factors ϕs are proposed for the direct design of stainless steel frames in the European, US and Australian design frameworks under gravity loads.

Highlights

The most commonly accepted approach for determining structural reliability, i.e. the probability of failure of a system of structural members, is to assess the reliability of a structural system by requiring the probability of loads Q exceeding the structural resistance R to be less than an adopted target value, in which models for resistance and loading are individually defined
This paper presents the derivation of appropriate system safety factors γM,s and resistance factors φs to adopt in the design-by-advanced analysis of stainless steel cold-formed portal frames subjected to gravity load combinations
The ultimate loads of the frames were determined from the load-vertical deflection curves obtained from the finite element simulations for the apex joint reference points: for those frames showing load-deflection curves with a clear peak point, this peak load was adopted as the ultimate load; on the contrary, where no clear peak point existed the load at which the stiffness of the curve fell below a value of 5% of the initial stiffness was defined as the ultimate load

Summary

Introduction

The most commonly accepted approach for determining structural reliability, i.e. the probability of failure of a system of structural members, is to assess the reliability of a structural system by requiring the probability of loads Q exceeding the structural resistance R (or reaching a certain limit state more generally) to be less than an adopted target value, in which models for resistance and loading are individually defined. Current design of structures is based on traditional limit state criteria following a two-step approach where internal actions are first determined from a structural analysis, usually elastic, and limit states are subsequently checked for members and connections such that the ultimate capacity of the structure is reached when the first member or connection reaches its ultimate limit state Provisions for these checks are given by structural codes for the particular materials, e.g. steel [1,2,3,4], in which design parameters and their random variations are modelled by nominal or characteristic values based on partial coefficient methods. The exponential growth of the computational power of standard desktop computers and rapid advances in design software during the past decades have made it possible to accurately predict the behaviour and failure of complex structural systems, providing engineers with the opportunity to progress from the current two-step member-based design method to a direct system-based design-by-analysis approach.

Methods

Results

Conclusion