Response Surface–Based Approach to Quantify Variability in Recycled-Asphalt Concrete Mixtures

Abstract

Due to the commercial and engineering value associated with old aggregates and aged asphalt binder, recycled asphalt pavement (RAP) usually is used in recycled asphalt concrete (RAC). However, the heterogeneity of RAP affects the quality of RAC. This in turn results in significant variation in the measured and predicted material properties. This paper presents a sensitivity and uncertainty–based approach to quantify this variation. Elaborate laboratory investigation was carried out on 17 different RAC mixtures. During this evaluation process, mixing viscosity, percentage of RAP, percentage of total binder, and percentage of virgin binder were considered as independent variables. These independent variables were used to predict the mixture volumetrics and Marshall properties through response surface models (RSMs). The results indicated that a second-order RSM can capture underlying patterns in the mean and coefficient of variation. Finally, uncertainty analysis was carried out to identify the combinations that resulted in the most and least scatter in the mixture properties. The uncertainty indicators were highly sensitive to mixing viscosity and RAP content. Sensitivity analysis was carried out using these RSMs to rank these independent variables. The results indicated that the percentage of RAP and mixing viscosity had the most and least influence on material properties. Overall results indicate that the variability in the volumetric and Marshall properties is highly influenced by the RAP content. Hence, the addition of RAP content should be monitored carefully.