The prediction of collapse of structures has gained growing attention in recent years to enable the structural engineering community to predict possible extreme loads that precipitate collapse. To predict collapse of steel structures, finite element deletions strategies have been used successfully in the past to account for fracture in steel members. Prior work has of-ten used a constant critical strain approach (CS), which deletes an element when it achieves a specific level of strain (e.g., 0.2, which is used in this work), typically without modeling of material softening. This approach requires frequent re-calibration depending on the configuration of the structural components and systems. This research proposes a more robust and general-purpose approach to collapse modeling of steel structures through the use of a Void Growth Model (VGM) to simulate the initiation of softening and the Hillerborg model for modeling the subsequent material softening, followed by an element deletion strategy that is developed in this work. In addition, a second approach is investigated that adds a Bao-Wierzbicki model to the VGM strategy (VGM-BW) in order to better account for lower and negative triaxiality regions in determining softening. The parameters of the VGM strategy were calibrated to a comprehensive set of experimental test results of circumferentially notched tensile (CNT) coupon specimens, while the Bao-Wierzbicki parameters in VGM-BW strategy were deter-mined analytically. These strategies were then validated without recalibration through comparison with a comprehensive range of experimental test results of material characterization specimens and full-scale structural steel connection tests (Figures 1 and 2), moment resisting frame experiments (Figure 3), and multi-story braced frame experiments. The VGM strategy provided most accurate prediction, while VGM-BW has better potential if it is calibrated to experimental results directly in the low and negative triaxiality range (Figures 1 and 2). In general, the constant strain strategy did not compare well to experiments (Figures 1 and 2). The VGM and VGM-BW approaches thus enable high-fidelity parametric simulation capabilities of interest to researchers, practitioners, and code developers who address collapse of structures.
• Rentschler, G. P., Chen, W.-F., and Driscoll, G. C. (1978). Tests of Beam-to-Column Web Moment Connections, Report #405.9, Fritz Engineering La-boratory, Bethlehem, PA.
• Sadek, F., Main, J. A., Lew, H. S., Rob-ert, S. D., Chiarito, V. P., and El-Tawil, S. (2010). An Experimental and Com-putational Study of Steel Moment Con-nections under a Column Removal Sce-nario, NIST Technical Note 1669, Na-tional Institute of Standards and Tech-nology, U.S. Department of Commerce, Gaithersburg, Maryland.
• Saykin, V. V. (2014). “A Validated Approach for Modeling Collapse of Steel Structures,” Ph.D. Dissertation, Civil and Environmental Engineering, Northeastern University, Boston, Mas-sachusetts.
• Schutz, F. W., Schilling, C. G. J., and Beedle, L. S. (1953). The Collapse Strength of a Welded Single Bay Frame, Report #205D.5, Fritz Engi-neering Laboratory, Lehigh University, Bethlehem, Pennsylvania.
Jerome F. Hajjar, Northeastern University
Junho Song, Seoul National University
Vitaliy Saykin, Wentworth Institute of Technology
Derya Deniz, University of Illinois at Urbana-Champaign
Tam Nguyen, KTP Consultants Pte Ltd.
Sponsored by National Science Foundation