The finite element analysis of thin-walled box spine-beam bridges
This thesis considers the theoretical and experimental analysis of thin-walled box spine-beam bridges. Existing methods available for the analysis of spine-beam bridges have been reviewed, with special attention being paid to thin-walled box beam theories. A new approach combining the finite element technique and the thin-walled beamtheory, which is appropriate for design purposes, has been proposed. This approach is specially suitable for medium and long spans. It is intended to be a realistic and versatile method to be used during the preliminary analysis and design stage, when a full three-dimensional analysis is likely to be impractical. Special features related tothebending analysis of thin-walled members and the warping torsion theory of open and closed section members are summarized in the thesis. In addition, supplementary formulae for the calculation of the stress distributions and the thin-walled section properties are derived. The distortional effect on single-spined box beams subjected to torsion has been extended to a general form based on the principles of ordinary folded plate theory. A family of special one-dimensional sub-parametric elements has been developed. In addition to the usual truss and beam elements the family includes a general thin-walled box beam element which may be curved in space and may have a variable cross-section. Additional degrees of freedom have been included to account for the warping and distortion effects whichoccur in box beams. An inclined cable element with catenary action is included, and an approximate nonlinear process for the analysis of cablestayed bridges has been correlated with tests on an actual bridge structure. A finite element-grillage approach for the analysis of multibox structures with deformable sections has also been developed. The complete family of elements has been incorporated into a computer program called CUBAS. A supplementary program called PFRAN for calculating the distortional properties and the influence values of the equivalent Vierendeel frame has also been implemented. The accuracy of the results obtained is demonstrated by comparison with results obtained by other published methods. A series of model box beams were tested to •further substantiate the theoretical results. The model dimensions were chosen to highlight both warping and cross-sectional distortion effects. The degree of correlation obtained shows that the theoretical developments proposed in this thesis may be applied successfully to the analysis of box spine-beam bridges.