Determination and quantitative representation of three-level dispersion system in asphalt mixture interface area

• The digital twin model of asphalt mixture interface was established. • The improved U-NET algorithm can effectively segment mineral powder and asphalt in CT images of asphalt mixtures. • Within the asphalt mixture, a discernible layer of pure asphalt delineates the boundary between the coarse aggregate and the asphalt mastic. • For OGFC-13, the surface of the coarse aggregate is sequentially coated with an average 15 μm layer of pure asphalt and a 66 μm layer of asphalt mastic. To establish a theoretical foundation for the quantitative design of asphalt mixtures based on mastic theory, this study investigates the interfacial zone of OGFC-13 asphalt mixtures. A three-dimensional digital twin model of the interfacial zone in OGFC-13 was developed using X-ray computed tomography (CT) and an enhanced U-Net deep learning algorithm. The model delineated a three-level dispersion system (TDS) within the interfacial zone, outlining the distribution ranges of each system. Furthermore, the distribution of each zone within the interfacial zone’s three-level dispersion system was validated through micro-mechanical, morphological analyses, elemental analyses, and theoretical calculations. The reliability of the enhanced U-Net deep learning algorithm in processing mesoscopic images was confirmed by segmenting asphalt mixture images. The digital twin model reveals that within the three-level dispersion system of the interfacial zone in OGFC-13, the pure asphalt area spans 8–25 µm, the asphalt mastic area ranges from 30 to 108 µm, and the asphalt mortar area extends from 300 to 1360 µm. These distribution ranges agree with the results obtained from nanoindentation, energy-dispersive spectroscopy (EDS), and washing tests, thereby confirming the consistency of the digital twin model with experimental evidence.