Behavior of Buried Composite Arch Bridges.
Publication Name: Report to the Maine DOT
This dissertation presents experimental results and numerical analysis for short-span buried composite arch bridges whose main structural members are concrete-filled fiber reinforced polymer (FRP) tube arches (CFFTs). This investigation focuses on soil- structure interaction of the bridge system to expand understanding beyond isolated arch performance. Soil plays several roles: weight on the structure, restraining arch movement, and dissipating surficial truck loads. Investigation focused on experimental and numerical analysis of subscale bridge systems. Four bridge systems were constructed with two different span-to-rise ratios (5:1 and 2.67:1) and two different materials (linear-elastic steel and CFFT). Tested systems were half-scaled from commercial bridge dimensions, and cross sectional properties were selected for scaled for the same lateral earth pressure coefficient under service loading. Bridges were placed in a self-reacting timber soil box and were backfilled with compacted granular fill in alternating lifts to an elevation 610 mm above the apex. Bridges were subject to a series of live loads across the middle 60% of the span of the arches, replicating truck loading. Bridges were loaded at the apex until failure to find the capacity of the buried arch system.The experimental testing process was numerically replicated with two finite element solvers: a soil-spring model implemented for design of commercial CFFT arch bridges and a soil-continuum model that mimics the governing physics. The soil-spring model separates vertical applied force (soil weight, surficial live loads) and horizontal soil resistance (deflection dependent soil springs). The soil-continuum model uses depth- dependent elastic moduli and Mohr-Coulomb plasticity to model soil restraint and dilation in high-shear regions. Soil-structure interaction was governed by hard normal contact and frictional tangential contact. The soil-spring model was conservative yet adequate for most of the experimental backfilling and live load steps. Negative foundation moment in a short rise arch due to 60% offset load was excessively conservative and warrants future consideration. The continuum model was more accurate for live load moments away from the point of load, but was unconservative in several tests for peak positive moment. The continuum model was impractical as a design tool due to cost, development time, and run time.