The macro-micro properties of concrete coated with modified polyurethane under the alternating action of abrasion and freeze-thaw cycles

Abstract

In this paper, the macroscopic and microscopic properties of concrete coated with modified polyurethane were studied under the alternating action of abrasion and freeze-thaw cycles. The protective effect of the coating was analyzed via hydrophobicity tests and microhardness tests. The macroscopic and microscopic properties of concrete were systematically studied by combining abrasion resistance freeze-thaw alternation tests, scanning electron microscopy (SEM) tests, and nuclear magnetic resonance (NMR) tests. The results indicate that under single abrasion action, small pieces of concrete coating peel off, resulting in increased mass loss. The single freeze-thaw effect is relatively small, and the mass increases first and then decreases. Especially, the mass loss rate remained negative throughout the entire freeze-thaw process. Under alternating action, the coating peels off severely and the quality loss is significant, with a maximum value of 0.56 %. Obvious aggregate particles and pore expansion could be seen on the surface of the coating. The size and volume of the gel pores and transition pores decreased, but the capillary pores and macropores increased significantly. As the number of alternating cycles increases, the abrasion resistance, bond strength, and compressive strength of the concrete coating gradually decrease, and the failure mode changes from cohesive failure to interface bonding failure. The addition of modifiers significantly improved the microstructure of the coating, thereby enhancing its mechanical properties in abrasion, freeze-thaw cycles, and alternating environments. With the addition of 0.5 % anti-aging agent and 2 % dispersant, the concrete coating achieves the optimal performance. A theoretical prediction model for concrete with modified polyurethane coating was established based on Kachanov's damage mechanics theory. The model has been validated experimentally with a correlation coefficient value exceeding 0.989, indicating a good fitting effect.