A Predictive Model for Estimation of the Degree of Reclaimed Asphalt Pavement Bitumen Activity

Abstract

The proportion of reclaimed asphalt pavement (RAP) bitumen that gets activated and coats the virgin aggregate during the initial stage of plant production of RAP mixes, in which the RAP material is initially mixed with superheated virgin aggregate, designated as the degree of RAP bitumen activity (DoA), is an important parameter influencing the quality and performance of the RAP mix. DoA is influenced by a variety of mix and process parameters. This paper presents an empirical model, developed using the results from a previous experimental investigation and those obtained from the additional experiments conducted in the present investigation, for prediction of the DoA of RAP binder under varying combinations of mix and process variables. The experimental investigation involved mixing separately identifiable size fractions of RAP material (1.18 to 4.75 mm) and virgin aggregate (9.5 to 26.5 mm) and measuring the quantity of RAP bitumen transferred to virgin aggregate to estimate the DoA. The 41 sets of experimental data considered for the development of the empirical model covered different RAP material proportions (15% to 75%), RAP binder quality (softening point of 69°C to 78°C), RAP binder content (4.0% to 6.0%), mixing duration (0.5 to 3 min), mixing temperature (70°C to 180°C), and superheating temperature of virgin aggregate (155°C to 197.5°C). The gradation and shape of the RAP material used in different coating experiments were also different. Additional experiments were carried out in the present investigation to study the effect of superheating temperature of virgin aggregates. DoA decreased exponentially with RAP bitumen stiffness and RAP content, whereas it increased with mixing temperature, mixing time, RAP bitumen content, and the proportion of coarser RAP material. The superheating temperature of virgin aggregate did not affect DoA significantly. A regression model was developed for estimation of DoA with reasonable accuracy without having to carry out the coating experiments. Such a predictive model is useful for obtaining inputs for the estimation of the quality of final binder blend in the RAP mix.