Multiscale power spectrum analysis of 3D surface texture for prediction of asphalt pavement friction

Abstract Pavement surface friction plays a crucial role in reducing traffic accidents. The friction performance largely depends on the texture characteristics of the pavement surface. However, the contribution of effective contact texture at different areal ratios and wavelengths to pavement friction has not been fully understood due to lack of high-quality texture data and effective evaluation framework. A multiscale power spectrum analysis of a three-dimensional (3D) surface is conducted in this study to predict asphalt pavement friction. This method novelty stratifies 3D texture data into multiple layers with different areal ratios. Twenty-seven testing sites within three different types of pavement rut boards were selected as the lab testing beds. Pavement surface friction and texture data were collected in parallel using a British Pendulum Tester and a portable ultra-high-resolution 3D laser scanner. The pavement surface power spectrum indicator (C(q)) was investigated to characterize texture attributes. The texture wave vector (q) was applied to extract texture features at eight wavelengths ranging from 0.1 to 55.1 mm. The effective contact area and optimal wavelength of the 3D texture were identified based on the Pearson correlation coefficient between C(q) and friction. Finally, considering the effective contact area, pavement friction was predicted as a function of C(q) under both optimal macrotexture (5.06 mm) and microtexture (0.33 mm) wavelengths. The results indicate that the developed power spectrum-based friction prediction model can predict pavement friction more accurately than previous methods.