A predictive model for estimating the viscosity of short-term-aged bitumen

Abstract

For accurate prediction of pavement performance, it is necessary that the effect of ageing on different mix parameters is accounted for. The short-term- and long-term-aged bitumen viscosities estimated using Global Ageing System (GAS) models are used in the Mechanistic Empirical Pavement Design Guide for the estimation of dynamic modulus. Comparison of the viscosities of the bitumens extracted from field- and laboratory-aged asphalt mixes with the viscosities predicted using the GAS model suggests over-prediction for high temperature and lower bitumen film thickness (FT) combinations and under-prediction for low temperatures and high film thicknesses. That the ageing codes used in the GAS model are not sensitive to FT and temperature also has been established in this study. The new statistical model proposed in this study based on fixed effect regression modelling approach for estimating the short-term-aged bitumen viscosity as a function of bitumen type, FT and temperature has been found to predict the short-term-aged viscosity more accurately compared to the GAS model. The predictions with the new model are particularly better for certain combinations of temperature and FT for which large errors are observed in the predictions made with GAS models. The paper also proposes short-term-ageing conditions (sample weight and exposure time) to be adopted for bitumen in a Rolling Thin Film Oven to simulate the field ageing occurring in asphalt mixes having different film thicknesses.