A soft computing approach to predict and evaluate asphalt mixture aging characteristics using asphaltene as a performance indicator

Abstract

Prediction of long-term asphalt mixture aging using fundamental characteristics of asphalt binders and mixtures is a complex task. Asphaltene has been reported as one of the major chemical components of asphalt cement binder. Several research studies have established asphaltene content as the fundamental characteristic ingredient present in the asphalt binders required to understand aging-related performance. Dynamic complex modulus (|E*|) is recognized as the paramount performance response parameter for asphalt mixtures, routinely used in pavement design and evaluation exercises. Hence, there is a definitive need to develop mixture aging predictive models using asphaltene content as the fundamental parameter with its effect on the resulting |E*| performance of asphalt mixtures. The objective of this research study was to develop asphalt mixture aging predictive models with asphaltene content as a fundamental performance parameter, using soft computing techniques. Asphalt binders and corresponding asphalt mixtures were subjected to short- and long-term aging conditions. Asphaltene contents and rheological properties were measured for different asphalt binders. Volumetric properties and |E*| were conducted for corresponding asphalt mixtures. Artificial Neural Network (ANN) method was employed to develop a rational model for evaluating asphalt mixture aging behavior considering asphaltene content values from asphalt binders. A total of seven different dense-graded asphalt mixtures with virgin, polymer-, and rubber-modified binders with two different asphalt contents were produced for experimentation purposes. The results showed that the predictive model developed using the ANN approach provided a robust relationship with asphaltene aging indices, a fundamental asphalt property used to quantify asphalt mixture properties at various aging conditions.