Papers »

Predicting ductile fracture in steel column base plate connections using micromechanics-based models

Current trends in fracture assessment of steel connections typically rely on approaches which may involve either critical ultimate stress or strain parameters or traditional fracture mechanics parameters such as K, J or CTOD. While these approaches provide accurate predictions of fracture in many situations, they may not be reliable under situations involving large-scale yielding that invalidate the assumptions that underlie these methods. Micromechanics-based models, such as the Cyclic Void Growth Model (CVGM), are an attractive alternative to these approaches. Most prior work on these approaches has focused on coupon-scale specimens with largely homogenous properties. In this paper, these approaches are applied to large-scale structural details where significant material heterogeneity or other complicating effects (such as uncertain geometrical imperfections and boundary conditions) may be present. There are two objectives to this work: (1) to examine the efficacy of (and validate) the CVGM by comparing the simulation-based predictions to experimental data and (2) to apply the validated method through numerical parametric studies to generalize experimental findings. These objectives are realized through four physical tests and dozens of three-dimensional finite element simulations of large-scale column base plate connections with a complete joint penetration weld detail and subjected to cyclic loading. All simulations are conducted with the commercial code ABAQUS/Standard and incorporate both material and geometric nonlinear effects. The parametric study interrogates key geometric parameters of the column base plate connection including column size/shape and moment gradient. In addition to generalizing the experimental data, the numerical study demonstrates the efficacy of the approach, presenting it as a viable alternative to physical experiments.

Author(s):

Andrew Myers    
Northeastern University
United States

Amit Kanvinde    
University of California at Davis
United States

Gregory Deierlein    
Stanford University
United States