Proof Testing a Bridge Deck Design with Glass Fiber Reinforced Polymer Bars as Top Mat of Reinforcement

Abstract

The primary objective of this project was to test a full-scale prototype of a bridge deck design containing glass fiber reinforced polymer (GFRP) bars as the top mat of reinforcement. The test deck mimics the design of the deck of one span of the new bridge over Gills Creek on Rt. 668 in Franklin County, Virginia. The purpose of the tests was to verify the deck design and provide assurance that the deck will behave as expected. Aspects of the behavior of the bridge deck, such as failure load, failure mode, cracking load, crack widths, deflections, and internal stresses, were examined. Four tests were performed on the deck, all of which tested the deck in negative moment regions. The tests comprised two overhang tests, one test of the deck over an interior girder, and one test of a cantilever section of the composite deck and girder. The cantilever test modeled the deck in a continuous bridge over an interior support. From the tests, it was concluded that the design of the deck was quite conservative. The secondary objectives of this project were to comment on the construction of a bridge deck reinforced with GFRP bars, note the advantages and disadvantages, and critique the current state of the art of designing bridge decks with GFRP reinforcement. It was found that the advantages of construction with GFRP bars easily outweighed the disadvantages and that the placing of the top mat of GFRP bars was much easier than the placing of the bottom mat of steel bars. The state of the art for the design of bridge decks reinforced with GFRP bars was found to be generally conservative. Three primary criteria dictate the deck design: strength, allowable stresses in the GFRP bars, and crack widths. For this deck, the size and spacing of the transverse GFRP bars were governed by crack control criteria. In testing the deck, however, it was found that the measured crack widths were far smaller than the calculated widths. The measured bar stresses, after cracking, were below those calculated, and below the allowable for all but the cantilever test. The ultimate failure loads were between 3.7 and 7.6 times the design wheel load plus impact. All failures were due to punching shear and were between 91% and 149% of the predicted failure load. Current methods for calculating one-way shear grossly under-predicted capacity. The current design is safe and should prove to be low maintenance. Improvements in design approach, particularly for crack widths and one-way shear, could result in more economical designs in the future. Although current methods for calculating strength and serviceability requirement do not result in accurate predictions of behavior, they do result in conservative designs.