On the correlation of analyses and tests of the inelastic flexural behavior of wide-flange steel beams

Aspects of numerical modeling which influence the correlation between tests and finite element analyses of the inelastic stability of wide-flange beams are discussed. Results from a series of fully non-linear shell finite element studies are compared to data from tests used to determine the inelastic rotation capacity of wide-flange steel beams. The modeling issues addressed include: (1) the distribution and magnitude of residual stresses, (2) imperfection sensitivity, in view of possible yield surface corner-type effects, (3) representation of the yield plateau and strain-hardening response of structural steel, and (4) the sensitivity to variations in the yield stress. Numerical problems encountered in the incremental-iterative analyses and the analysis strategies used to overcome these problems are also outlined. For the examples analyzed and the incremental-iterative solution approach employed, the experimental and the numerical results have been found, in general, to be in good agreement. It is shown through numerical studies that, for the particular examples considered, variations in the assumed residual stress pattern have practically no effect on the predicted rotation capacity. The representation of the yield plateau and variations in the magnitude of the initial imperfections only slightly affect the analysis results. Variations in the yield stress significantly change the model's flexural response, but they only slightly affect the predicted rotation capacity. It is shown that small errors in measurement of the yield stress can result in large changes of the rotation capacity calculated from experimental data.