Development and Validation of a New Dielectric Profiling System Dielectric–Density Calibration Model Using Laboratory and Field Data

Abstract

Achieving reliable dielectric–density calibration is critical for accurate asphalt pavement compaction assessment using the dielectric profiling system (DPS). This study introduces and evaluates the average Hashin–Shtrikman (Ave-HS) model, or dielectric mixing model, designed to generate robust calibrations with minimal data. DPS measurements, field cores, and superpave gyratory compacted (SGC) specimens (pucks) from 22 projects in Michigan (29 datasets) were analyzed to assess model accuracy, variability, and bias. Puck-based calibrations were highly consistent: 28 of 29 datasets showed standard deviations below 0.5% voids, and as few as two pucks compacted in the laboratory provided reliable calibration results. Core-based calibrations exhibited greater variability, with over half of the datasets exceeding 1.0% voids standard deviation and several surpassing the ±1.25% measurement-to-measurement variability threshold. Day-to-day variability was minimal in both pucks and cores (≈0.4% voids), showing that calibrations remain stable across different asphalt mix production days under consistent mix conditions, allowing a single puck calibration to be applied over multiple paving days without repeated daily calibrations. Core–puck comparisons showed agreement within 2% voids for 23 of 29 datasets. Outlier datasets generally exhibited higher-than-expected core dielectrics, suggesting residual surface moisture at the time of DPS testing. Forensic re-testing of discrepant cores in the laboratory aligned with puck-based calibrations, further supporting the conclusion that field moisture influenced the initial measurements. The Ave-HS model’s physics-based approach, combined with minimal sampling requirements, provides a practical calibration method for DPS deployment. Puck-derived calibrations are recommended as the primary reference, with core measurements used for verification and forensic evaluation when discrepancies arise.