UC San Diego

Three issues on steel truss bridge evaluation and design are investigated in this research. Part 1 deals with gusset plate buckling and the associated compression strength. Gusset plates are routinely used to connect members at panel points in steel truss bridges and braced-frame buildings. Current design practice is to treat a portion of the gusset plate at the end of the connected member (e.g., diagonal brace) as a compression member. This process requires a determination of the member length, an effective length factor, and the use of a Whitmore section with a proper dispersion angle to define the cross section of the compression member. The method, first proposed by Thornton (1984), was mainly based on engineering judgment. After the collapse of the I-35W Bridge in Minneapolis, MN in 2007 with 13 casualties, significant research efforts at both federal and state levels were made, which resulted in the new design requirements for checking gusset buckling in the AASHTO LRFD Bridge Design Specifications. Yet, the new requirements are still based on the Thornton’s column analogy concept with some minor modifications. In this dissertation, a novel approach that treats gusset buckling as a phenomenon of plate buckling under shear is developed. This proposed design method properly reflects the sway buckling mode, a failure mode that is always observed in laboratory testing but is not reflected in the column-analogy method. Based on finite element simulation, the model developed considers the effects of the free edge length of the gusset plate, the brace connected length, and the angle between the brace and the gusset plate. A correlation with the available test data shows that the scatter of data is drastically reduced when compared with the conventional design method.Part 2 of the research addresses issues faced by bridge designers on the load rating of existing steel truss bridges. Built-up members in many truss bridges have centroids that do not coincide with the working lines, and such eccentricity at panel points produces moments in the members. One common practice is to analyze the bridge as a pin-connected truss, and then the end moments are computed as the axial load multiplied by either 100% or 50% of the end eccentricity, depending on the type of the connection (either pin-connected or gusset-connected) at the panel point. Together with field load testing of a truss bridge in Southern California and the associated finite element simulation, this study concludes that the Caltrans practice of handling eccentricity has no technical basis. Instead, eccentricity should be explicitly included in the finite element models. The last part of the research deals with the behavior and design of steel pins. Using steel pins to connect members at the panel points is common in many existing steel truss bridges. The AASHTO Specifications treat the pin as a beam, and a moment-shear interaction check is used to check the pin strength. Since test data, especially pins with larger diameters, is very scarce, experimental testing of twelve 2-in. diameter pins with two steel grades was conducted. Together with finite element simulation, this research shows that the beam-analogy approach adopted in the AASHTO Specifications can be very conservative. Alternative equations to predict the strength of steel pins are proposed, which would potentially eliminate many unnecessary retrofits of steel pins in existing truss bridges.