Using a Neural Network Model to Assess the Effect of Antistripping Agents on the Performance of Moisture-Conditioned Asphalt

Abstract

Moisture damage in asphalt is one of the prime concerns for flexible pavements degradation worldwide. Many of the pavement distresses are the direct and indirect outcomes of the moisture intrusion in asphalt pavement. This study focuses on developing a neural network (NN) to determine the effect of types and percentages of chemical antistripping agents (ASAs) on the adhesion forces of polymer-modified dry and wet asphalt binder samples. Atomic force microscopy (AFM) test is conducted to determine the adhesion and cohesion forces of asphalt samples with varying contents of polymer and ASAs using four different functionalized and industrial tips. A NN adhesion force prediction model is developed on the basis of AFM laboratory data with varying percentages of ASAs. Except for adhesion loss measured by the NH3 tip, all results show improvement in adhesion loss attributed to the addition of ASAs. Among all the chemical ASAs, Morlife shows the best performance in the presence of 3% styrene-butadiene and 5% styrene-butadiene-styrene in reducing moisture effect (average 18% reduction for all sample types) on adhesion and cohesion forces. In all cases, an increase in percentage of additives of greater than 1% does not aid in resistance to the damage caused by moisture in polymer-modified asphalt.