The objective of this paper is to demonstrate how detailed 3D laser scans of cold-formed steel members may be used to characterize cross-section dimensions, including complete information on the correlation between dimensions, as well as other manufacturing imperfections such as bow, camber, twist, crown, and flare. Advancing the use of simulation in design requires that manufacturing imperfections be understood, such that simulations are based on realistic geometry. This is particularly important for thin-walled members due to the well-known imperfection sensitivity of such members in certain failure mechanisms. C sections are selected for demonstration in this paper. The members are all scanned in a custom-built 3D laser scanner that results in a dense point cloud defining the true geometry of the outside surface of the scanned members. Algorithms are employed to post-process the point cloud into useful information including dimensions and geometric imperfections. The member dimensions (web height, flange length, corner radius, etc.) may be compared with nominally prescribed dimensions, and in addition the correlation across the dimensions is studied and the impact of typical manufacturing control is readily observed in the data. The imperfections (deviation from perfect) may be characterized in geometric terms: bow, camber, twist, crown of a given flat plate, flare of a given element; or may be characterized in terms of their modal buckling content: fit to flexural modes, torsional mode, local mode, and distortional mode. In addition, the geometric imperfections may be transformed into the frequency domain and power spectrum of the imperfection magnitudes can be obtained. This 1D spectrum approach provides a potentially novel means for generating realistic, but random geometric imperfections for use in shell finite element simulations. Shell finite element collapse analyses that compare the sensitivity in response to true, and various simulated imperfections are provided. The simulations indicate how simple modal imperfections are powerful for predicting strength conservatively, but the ID spectral approach more closely approaches the results from the true (scanned) members. In the future larger Monte Carlo simulations should be performed to assess the reliability of cold-formed steel members using these results.
展开▼
