Investigation of bond-slip behavior of steel reinforcement in coal gangue concrete

The bond-slip behavior between coal gangue sand concrete and steel reinforcement serves as the fundamental mechanism for their composite action, significantly influencing the structural safety. A systematic analysis was conducted using 15 sets of pull-out tests to evaluate the effects of the coal gangue sand replacement rate, concrete strength, steel reinforcement diameter, and anchorage length parameters on the bond-slip behavior. The results indicated that the failure modes of the pull-out specimens were pull-out failure and splitting pull-out failure. When the restraint of concrete on steel reinforcement is weak, pull-out failure occurs. The coal gangue sand replacement rate, anchorage length, and steel reinforcement diameter exhibited negative correlations with the ultimate bond strength, whereas concrete strength demonstrated a positive correlation with it. Compared with the specimens with a coal gangue sand replacement rate of 0%, when the coal gangue sand replacement rate reaches 100%, the ultimate bond strengths of the C30, C40, and C50 grade coal gangue concrete are decreased by 33.6%, 30.2%, and 26.2% respectively. The bond stress distribution exhibited a characteristic bimodal pattern in most specimens. Based on the tested results, a bond-slip constitutive model of the coal gangue sand reinforced concrete was developed for numerical modelling. The results indicated that compared to the influence of steel reinforcement diameter and anchorage length, the coal gangue sand replacement rate had a relatively limited impact on the ultimate load and bond strength. A design methodology for the critical anchorage length of ribbed steel reinforcement and coal gangue sand concrete had been proposed, and the error was mostly concentrated within the 20%. A prediction model for the bond-slip behavior of steel reinforcement and coal gangue sand concrete based on BP neural network had been developed. The determination coefficient of the test set prediction reached 0.8603, demonstrating that the model exhibits high predictive accuracy in predicting bond-slip curves.