Evaluation of redundancy in trapezoidal box-girder bridges using finite element analysis
The AASHTO Bridge Design specifications define a fracture-critical member as a component in tension whose failure is expected to result in the collapse of a bridge. The tension flanges of twin box-girder bridges are thus labeled as fracture-critical. In order to avoid the catastrophic collapse suggested by the AASHTO specifications, fracture-critical bridges, constituting 11% of all steel bridges in the country, are subjected to frequent and stringent evaluation and inspection. The Texas Department of Transportation, interested in reducing the cost of an otherwise attractive bridge design, is now questioning the validity of the original statement by AASHTO. In particular, it is not clear whether or not a single localized fracture can lead to the collapse of a bridge. Contrary to this belief, there have been multiple instances of fracture-critical bridges with two tension flanges that have experienced fracture without collapse. This project was designed to determine the level of redundancy that can be found in twin box-girder bridges. To achieve this goal, a full-scale test specimen of a box-girder bridge was built at the Ferguson Structural Engineering Laboratory in Austin, Texas. In unison, a finite element model of the bridge was built using ABAQUS/Standard. A fracture was initiated in one bottom flange of the test specimen. The data gathered during the test were compared to the calculated response from the model to verify the predictive capabilities of the model. If able to predict response accurately, a computer model could be used during design to indicate the presence of redundancy and the decreased need for frequent inspection of a bridge. The computer model was used to simulate a full-depth web fracture event in the exterior girder of a twin-girder bridge with a very large horizontal radius of curvature. The model was then modified to consider the influence of several parameters, including radius of curvature, structural redundancy through continuous spans, and external bracing. Results obtained from the finite element model indicate that adequate redundancy exists in the bridge design to maintain stability after the fracture of one girder. The most significant design change is to add continuity through spans, as adding structural redundancy greatly reduced the expected deflections and stresses that would be induced in the system. Further study using the modeling techniques presented in this thesis should begin by verifying or improving upon the assumptions that were made. Specifically, the concrete material model and the shear stud modeling method should be examined in more detail and should be used to predict the response of smaller-scale laboratory tests. With further refinement, this model could be utilized during the design phase to verify the presence of redundant load paths and thus reduce the necessity for frequent inspections.