A hybrid method for bond-slip behavior of reinforcement rebar in steel-polypropylene hybrid fiber reinforced concrete structures in marine environments

Abstract

Fiber-reinforced concrete is an advanced class of construction materials that display higher mechanical and durability performance, but limited information is available on the bond behavior of reinforcement rebar in steel-polypropylene hybrid fiber reinforced concrete (SPHFRC) in marine environments, which produce crack states commonly seen in reinforced concrete structural components. In this study, a hybrid method for predicting the bond response behavior of SPHFRC under monotonic and cyclic loading was developed. The proposed method combining mechanical-driven model and generative adversarial networks (e.g., GAN) method. Experimental databases were constructed. Two empirical models were developed. The GAN method enhanced its prediction accuracy through adversarial training support by database and hyper-parametric analytics. Finally, to verify the accuracy and effectiveness of the developed model, a total of 65 SPHFRC specimens under cyclic loading were conducted. The comparison shows that this model can reproduce the corrosion-induced bond deterioration of SPHFRC. The model can also capture the cyclic bond degradation of SPHFRC and other types of ultra-high-performance concrete with an acceptable range.