Superpave Design Aggregate Structure Considering Uncertainty: II. Evaluation of Trial Blends

Abstract

Abstract The current evaluation of Superpave design aggregate structure is deterministic. The design involves evaluation of selected trial blends based on volumetric, compaction, and dust proportion requirements. This article incorporates the uncertainties of all 17 variables involved in the process, measured by the coefficient of variation (CV), and develops a revised procedure for comparing mixture properties with the performance-based criteria. The uncertainties of eight measured properties are propagated through the calculation and make the uncertainty of some variables very large. Issues related to the reliability of some variables (CV > 25 %) are discussed and criteria to resolve them are established. The developed mathematical formulas of uncertainty were verified using Monte Carlo simulation. The results show that the potentially unreliable variables are as follows: volume of absorbed asphalt, percentage of absorbed asphalt, and percent voids in total mix at the design compaction level. A tool is provided to help the designer trace the uncertainty of the unreliable variables back to the measured properties so that their precisions may be revised. The current National Cooperative Highway Research Program–recommended precisions for specific gravities were found to be generally satisfactory. However, further research should continue to improve the precision of specific gravity measurements as some intermediate variables may still become unreliable.