Evaluation of Superstructure Response and Load Distribution in Skewed Steel I-Girder Bridges
Advisor: Professor Larry A. Fahnestock and Professor James M. LaFave
Abstract
Highly-skewed steel I-girder bridges are commonly used, especially in congested areas, despite complications in their analysis, design, and construction. Standard line girder analysis (LGA) considers skew effects through the application of a live load distribution factor (LLDF) (for skew exceeding 30°) and additional flange lateral bending stress (for skew exceeding 20°). For bridges with skew exceeding 60°, a higher level of analysis is often required. Although these design requirements and associated skew limits have some basis in prior research, more study is needed to establish a comprehensive understanding of load distribution in skewed steel I-girder bridges and to potentially refine the associated analysis and design procedures.
To understand practices used and challenges faced when designing and constructing skewed steel I-girder bridges, a survey was formulated and distributed to state transportation agencies across the U.S. Findings from the responses illuminate issues, concerns, and current practice related to design, construction, and service-life of such bridges. The agency survey also informed the selection of two bridges in the vicinity of each other in Champaign, Illinois for field monitoring, in order to provide new understanding of skew effects on bridge superstructure behavior. Mattis Avenue over I-74 (Mattis-74) and Mattis Avenue over I-57 (Mattis-57) are both two-span continuous bridges using cross-frames in a staggered layout throughout the bridges. Mattis-74 is a stub abutment bridge with a 41° skew, and Mattis-57 is an integral abutment bridge with 45° skew.
Three-dimensional finite element analysis (3D FEA) was conducted to guide field instrumentation, and sensing equipment capable of high sampling frequency (up to 20 Hz) was selected. Key girder cross-sections and cross-frames were instrumented with strain gauges, the end rotations of critical girders were measured with tiltmeters, global bridge movements were captured with displacement transducers, and temperature was simultaneously recorded with the structural response measurements. Data collection was initiated during construction, and the instrumented bridges were evaluated in the field during concrete deck placement and after each bridge was in service through isolated truck tests at various speeds – 3.2 kph (2 mph), 32 kph (20 mph), and 56 kph (35 mph).
3D FEA was carried out to provide enhanced understanding of bridge behavior; good agreement was observed between the numerical simulation results and field monitoring data for both deck placement and live load testing. Numerical parametric studies were conducted on the instrumented bridges to evaluate effects of bridge geometry, especially skew and abutment type, on skewed steel I-girder bridge superstructure response. Load distribution during deck placement and under live load was investigated, and standard design guidelines regarding girder strong-axis bending and flange lateral bending have been evaluated.
Under concrete dead load, a staggered cross-frame layout provided additional load paths that transferred lateral load from the deck overhang internally from an exterior girder, which was observed to be most significant near midspan of a skewed bridge. Development of flange lateral bending near supports of a skewed bridge was found to be primarily caused by superstructure stiffness along the skewed bearing lines; large flange lateral bending stress occurred when the lateral movement induced from girder rotation along the skew is restrained at the support. For a skewed bridge with a staggered cross-frame layout along its length, standard design recommendations from the AASHTO LRFD Bridge Design Specification regarding skew-related lateral bending stress were observed to be effective for interior girders, except near the ends of some bridges with skew over 45°, but inadequate for exterior girders.
Under isolated truck live loads, exterior girders are prone to larger strong-axis bending stress than interior girders when directly loaded. Estimating girder strong-axis bending stress using LGA with the LLDF calculated according to AASHTO requirements is overall conservative for isolated truck loads, but the common practice to use a controlling LLDF for all girders can be overly conservative for interior girders. On the other hand, interior girders are more critical regarding flange lateral bending stress than exterior girders. The distribution of lateral bending stress on a skewed bridge is dependent on the live load positioning, and limiting the distance between the first intermediate cross-frame to bridge ends is important to avoid lateral stress concentration near bridge obtuse corners. The dynamic load allowance of 33% used in design is mostly conservative, except for a few monitored locations near obtuse corners of the integral abutment bridge.
