Experimental and analytical investigation of localization and post-peak behavior of steel members in tension

Abstract

Accurate simulation of extreme limit states in structures subjected to abnormal loads, earthquake, or blast requires detailed understanding of their post-peak behavior. In steel structures, these extreme-limit states are associated with local or global instability such as local buckling, tension necking, and fracture. Several methods have been developed to simulate such extreme limit states such as fiber models, hinge models, nonlocal models, and finite element analysis (FEA). However, disadvantages of these methods arise when the structural elements pass the peak response and experience post-peak softening. Furthermore, there is a lack of experimental data to validate the numerical solutions, especially for the nonlocal models. This research first looks at experimentally determining the localization length and its evolution throughout the loading history to study the effect of the length scale on the post-peak response of steel members. Second, linear, bilinear, and nonlinear post-peak localization models are developed and compared to experimental data. Third, a system of three steel members was investigated to evaluate how post-peak behavior in part of the system affects the remaining members. ... The response of a system of steel tension members was investigated analytically and experimentally to evaluate the effect of post-peak response on load redistribution and overall movements of the system. The analytical results showed that the members with even a slight imperfection (i.e. 0.001%) can cause lateral displacement or rotation when one member fails before the others. The experimental results showed that the system in which only one member softening did so at a load 10% higher and at displacement 21% less than the system with two softening members. Both of these results indicate that even slight variations can cause significant effects in the failure response of a structural system. The numerical localization model was applied to the system analysis and showed the difference in displacement at the failure load between the numerical simulations and the experiments was 4%.