Benchmarking Recycled Binder Blends Using Statistical Analysis: A Case Study of Virginia and North Carolina

Abstract

Recycling agents (RAs) are added to asphalt binder blends containing recycled materials to mitigate the negative effects of adding oxidized material to asphalt mixtures. Several methods exist to select a particular RA dosage to restore a specific Superpave performance parameter. However, the rheological similarities between the RA-modified systems and typical virgin binders across the complete range of temperatures remain to be ascertained. In this study, two RA dosage selection procedures, one aiming to restore high-temperature performance grade and the other aiming to restore low-temperature performance grade were statistically compared. Quality assurance datasets from Virginia and North Carolina were used to benchmark the typical virgin binder properties, and 17 recycled binder blends using constituent materials from the two states were studied with respect to Superpave binder properties. Both univariate and bivariate statistical analyses were applied to compare and contrast the rheological properties of these blends with the benchmarked virgin binder properties for the two states. Then, a statistical similarity analysis was done using the Mahalanobis distance to evaluate the rheological properties of these blends with those of the reference binders. The study found that both dosage selection procedures were successful in achieving rheological similarity at target-specific temperatures and for specific binder properties but might not necessarily yield similar properties at other temperatures. Finally, an example application is presented using statistical analysis to select an RA dosage that yields a rheological fingerprint more similar to that of virgin binder systems.