Abstract

In the event of an earthquake, steel-frames are subjected to cyclic loads. Within the earthquake duration, the accumulated strain in ductile members may cause micro-scale fracture of steel material due to low-cycle fatigue, which manifests itself as early yielding and stiffness degradation at macro-scale stress-strain relations. Steel moment resisting frame members are made of thin-walled cross-sections, for which local or lateral-torsional buckling failure mode is considered in design. In order to capture local buckling, material yielding and lateral-torsional buckling modes sophisticated modelling approaches need to be adopted. When shell elements are used, multi-axial material models are needed and for this purpose Continuum Damage Plasticity (CDP) framework is used to build an inelastic multi-axial stress-strain relationship. The CDP has the capability of representing both the permanent deformations due to the plastic component and the degradation of elastic moduli due to the damage component. Early yielding and stiffness degradation of the material can be captured by adopting hardening/softening and damage accumulation criteria specific to the low-cycle behaviour of metals. In this research, an in-house computer program was developed. The program has the capabilities to impose Multi-Pont Constraints (MPC) which makes it flexible to generate the meshes for columns, beams and panel zones separately and then to bring them together. After validation of the program by using results of studies from literature, the possibilities of modelling different failure modes were investigated. MPCs were also used to simulate the beam behaviour as a special case of the shell model in which lateral-torsional buckling modes of thin-walled cross-sections of steel frames were captured under cyclic loads. It has been shown that considerations of local- and lateral-buckling modes as well as the low-cycle fatigue effect in the material cause significant differences in predicting the behaviour of steel frames under cyclic loads.