Investigation of changing SARA and fatigue properties of asphalt bitumen under ageing and analysis of their relation based upon the BP neural network

Thermal oxidation has a significant effect on asphalt binders’ saturates, aromatics, resins, asphaltenes (SARA) fractions and fatigue properties, but its effect on the SARA of different asphalt binders and the relation between SARA and the fatigue performance is not clear. In this study, thin layer chromatography/flame ionization detection (TLC/FID) was applied to analyze the SARA of the 36 specimens of original, short-term aged, and long-term aged asphalt binders. And then the viscoelastic continuum damage (VECD) model was applied to analyze these asphalt binders’ fatigue. Finally, the relation between SARA and the binders’ fatigue was determined by back-propagation (BP) neural network. The results showed that asphalt ageing can be considered to be the conversion of the light constituents to heavy constituents but the conversion ratios differed significantly for different asphalt binders. Asphalt binders of the same grade but different origins may have completely different mechanical properties and exhibit an entirely different composition conversion during ageing. Saturates can be considered the most inert constituent in asphalt. The oil source is the critical factor that determines the SARA and its degree of change during ageing. Ageing increased all of the asphalt binders’ fatigue life under a low strain level, but decreased it under high strain. The fatigue performance of asphalt binders with the same oil origin and grade but different manufacturers, did not differ significantly, either in the original or the aged states. Further, training the BP neural network showed that asphaltenes had the most significant effect on the asphalt’s fatigue behavior, followed by the resins and saturates, while aromatics had a minimal effect. And the BP neural network can forecast asphalt binders’ fatigue performance accurately based upon the SARA.