Document Type



Doctor of Philosophy


Civil Engineering

First Adviser

Sause, Richard

Other advisers/committee members

Wilson, John L.; Ricles, James; Pakzad, Shamim; Wassef, Wagdy


The use of steel I-shaped girders with tubular flanges (i.e., tubular flange girders or TFGs) has been proposed for curved highway bridges. Compared with conventional steel plate I-shaped girders (IGs), TFGs have substantially greater torsional stiffness and strength, which improves significantly the behavior of an individual curved TFG compared with a corresponding IG. One type of TFG with a rectangular tube as the top flange and a flat plate as the bottom flange (i.e., the TFG1 section) is studied in this research. Compared to the previously studied TFG with rectangular tubes as the top and bottom flanges (i.e., the TFG2 section), a TFG1 has the advantage of avoiding the need for concrete infill or an internal steel diaphragm in the tubular bottom flange at the bearings, but still has sufficient torsional stiffness and strength to perform well as a curved bridge girder. This research focuses on developing and validating the use of TFG1s for curved highway bridges.Simplified FE models for curved TFG1 bridges are developed for use in the design of these bridges. Detailed FE models are developed for use in validating the simplified FE models and in verifying that the response of curved TFG1 bridges to various loading conditions is as intended. Design guidelines for curved TFG1 bridges are presented. Eight curved TFG1 bridges are designed. Detailed FE models of these bridges are used to evaluate the effectiveness of the design guidelines. The results show that the design guidelines can be used to safely design curved TFG1 bridge systems. In addition, comparisons between curved TFG2 bridges and curved TFG1 bridges are made. A comprehensive study of the response of curved TFG1 bridge systems compared with corresponding curved IG bridge systems is conducted. All comparisons are based on bridges that satisfy the current U.S. design recommendations for steel curved highway bridge girders. A parametric study is conducted, where the parameters of the bridges that were varied include the span length, the horizontal curvature, number of interior cross frames, and other factors. 28 different curved TFG1 bridges were designed. These bridges were used with detailed FE models in a study of the effectiveness of the design guidelines. Again, the results show that the design guidelines can be used to safely design curved TFG1 bridge systems. The results comparing TFG1 systems with IG systems show that curved TFG1 bridge systems have more efficient (lighter) girder cross sections and/or require fewer interior cross frames than corresponding IG systems.Detailing of cross frames for curved TFG1 bridges is studied. Conventional no load fit (NLF), steel dead load fit (SDLF), total dead load fit (TDLF) detailing methods were studied. In addition, two new detailing methods for curved TFG1 bridges, individual steel dead load fit (ISDLF) and remaining dead load fit (RDLF) were studied. The responses of example bridges with different detailing methods are compared, and appropriate detailing methods for curved TFG1 bridges are recommended. ISDLF detailing is recommended for cross frames of curved TFG1 bridges with a short span (90 ft or less). For longer span bridges, NLF detailing is recommended. A two-thirds scale two-girder test specimen and corresponding full size bridge was designed using the design recommendations. The test specimen was fabricated and erected. The procedure to erect and assemble the test specimen was studied. Detailed FE models were used to simulate the assembly of the test specimen and the results were compared with test data. The observations and analyses show that it is feasible to install the interior diaphragms as the individual curved TFG1s carry their own weight across the span without shoring or other temporary supports within the span.The test specimen was tested up to and beyond the maximum load capacity. An updated FE model, including accurate material properties, geometric imperfections, and test boundary conditions, was used to analyze the test specimen and make comparisons with the test data. The FE results and test data show that the design recommendations can be used to safely design curved TFG1 bridges for construction conditions. The close agreement of the FE results with the test data indicates that the FE models can be used to accurately predicate the response of curved TFG1 bridges.