Corrosion of steel reinforcements is the leading causes of malfunction or even failures of reinforced concrete (RC) structures nationwide and worldwide for many decades. This arises up to substantial economic burden on repairs and rehabilitations to maintain and extend their service life of those RC public projects. The inherent natures of glass fiber-reinforced polymers (GFRP) bars, from their superior corrosion resistance to high strength-to-weight ratio, have promoted their acceptance as a viable alternative for steel reinforcement in civil infrastructures. Comprehensive understanding of the bond between GFRP bars and concrete, in particular under in-service conditions or extremely severe events, enables scientists and engineers to provide their proper design, assessment and long-term predictions, and ultimately to implement them toward the corrosion-free concrete products. This research aims to develop a holistic framework through an experimental, analytical and numerical study to gain deep understanding of the bond mechanism, behavior, and its long-term durability under harsh environments. The bond behavior and failure modes of GFRP bar to concrete are investigated through the accelerated aging tests with various environmental conditions, including alkaline and/or saline solutions, freezing-thawing cycles. The damage evolution of the bond is formulated from Damage Mechanics, while detailed procedures using the Arrhenius law and time shift factor approach are developed to predict the long-term bond degradation over time. Besides, the machine learning techniques of the artificial neural network integrated with the genetic algorithm are used for bond strength prediction and anchorage reliability assessment. Clearly, test data allow further calibration and verification of the analytical models and the finite element simulation. Bond damage evolution using the secant modulus of the bond-slip curves could effectively evaluate the interface degradation against slip and further identify critical factors that affect the bond design and assessment under the limit states. Long-term prediction reveals that the moisture content and elevated temperature could impact the material degradation of GFRP bars, thereby affecting their service life. In addition, the new attempt of the Data-to-Information concept using the machine learning techniques could yield valuable insight into the bond strength prediction and anchorage reliability analysis for their applications in RC structures. / ND NASA EPCoR (FAR0023941) / ND NSF EPSCoR (FAR0022364) / US DOT (FAR0025913)
Identifer | oai:union.ndltd.org:ndsu.edu/oai:library.ndsu.edu:10365/25864 |
Date | January 2016 |
Creators | Yan, Fei |
Publisher | North Dakota State University |
Source Sets | North Dakota State University |
Detected Language | English |
Type | text/dissertation, movingimage/video |
Format | video/mp4, application/pdf |
Rights | NDSU Policy 190.6.2, https://www.ndsu.edu/fileadmin/policy/190.pdf |
Page generated in 0.0018 seconds