| Accurate prediction of fracture is critical for ensuring the safety of civil infrastructure, as evidenced by damage to buildings and bridges in earthquakes such as the 1994 Northridge Earthquake and the 1995 Kobe Earthquake, and other recent events such as the 2007 collapse of the I-35W steel bridge in Minneapolis. The recent developments of so-called "local" approaches to assess fracture are major steps in the evolution of methodologies to predict fracture. In contrast to single-parameter fracture mechanics models, which necessitate limiting assumptions about the crack tip stress and strain field, the local or micromechanics-based approaches permit the assessment of fracture under a broader range of stress and strain states as determined through detailed finite element analysis. However, the use of "local" models to predict fracture in large civil structures is not prevalent, mainly owing to the lack of well-established calibration protocols and validation exercises.;Motivated by this need, this study addresses two "local" models for ductile fracture initiation, i.e. the Stress Modified Critical Strain (SMCS) criterion and the Cyclic Void Growth Model (CVGM). Several refinements to these models are presented which aim to increase the accessibility and accuracy of the models when applied to large-scale structural steel components. Specifically, this study aims to (1) develop a simplified calibration procedure for the SMCS which enables more widespread use of the model, (2) establish the size independence of the SMCS over a range of sizes and geometries, (3) propose a unified probabilistic framework which incorporates various sources of randomness to predict fracture under cyclic or monotonic loading with confidence limits, (4) refine the damage counting algorithm of the CVGM to offer improved accuracy, (5) examine the effect of material heterogeneity on ductile fracture in weldments, and (6) demonstrate the efficacy of the modeling approach through validation against a series of large-scale tests on column base plate connections.;The above objectives are accomplished through an extensive experimental program, including a series of 6 large-scale tests (2/3rd scale) on steel column base plate connections, eighty-six coupon-scale tests on circumferentially notched tension bars (CNTs) which includes thirty-nine CNTs extracted from weldments, data synthesis of eighty-five CNT tests from previous studies, and fractographic analysis. All experiments are complemented by comprehensive non-linear finite element simulations in ABAQUS/Standard through which the fracture models are implemented. The data from CNTs extracted from weldments indicate that the mean ductile fracture toughness in the HAZ is approximately 50% smaller than that of typical base metal and approximately 15% smaller than that of typical weld metal. Moreover, the variability of the HAZ toughness (expressed in terms of the coefficient of variation) is approximately twice that of base and weld metal.;In addition to providing validation data, the large-scale tests provide insights that are relevant to the design of column base connections. This test series examined a complete joint penetration (CJP) weld detail and a partial joint penetration (PJP) detail, and the data indicated that the ductility of the PJP connection exceeded that of the CJP connection. The PJP detail sustained cyclically imposed hinge rotation of up to 0.08 to 0.09 radians prior to ductile crack initiation, wheras the CJP detail only sustained rotations up to 0.05 to 0.06 radians. The results from these tests are complemented by the validated modeling approach to generalize the findings to untested details. |
