A study on the reconstruction of 3D asphalt surface replicas for statistically reliable analysis

Abstract

Accurate replication of pavement textures is crucial for tire-pavement interaction studies in laboratory conditions, particularly at the macro-texture level. Existing replication methods often rely on costly equipment and lack standardized procedures for validating surface replicas. This study presents a novel methodology to replicate pavement surfaces and validation. The methodology integrates Mean Texture Depth (MTD) as a 3D texture indicator, silicone rubber casting technology, Finite Element (FE)-based 3D geometric calculations, and smartphone-video-based digital modelling using computer vision. Different asphalt mixtures with distinct surface textures were used for the development of the physical and digital replicas. Physical replicas were produced through silicone casting and epoxy molding, while digital replicas were generated using the Structure-from-Motion technique. Validation was conducted using the standard sand patch test for physical samples and FE-based MTD calculations for digital models. The results showed strong agreement between the replicas and the original pavement samples. Statistical analyses confirmed normal distribution, variance homogeneity, and no significant differences between digital and physical samples. The findings of this study are expected to benefit road authorities in the design and laboratory testing of pavement surfaces, with the aim of evaluating pavement performance.