Online Archive of University of Virginia Scholarship
Development of Performance Models for Asphalt Overlay Design Using Traffic Speed Deflectometer and Pavement Management Data159 views
Author
Smith, Bryan, Civil Engineering - School of Engineering and Applied Science, University of Virginia0000-0002-5278-0892
Advisors
Smith, Brian, EN-CEE, University of Virginia
Abstract
Transportation agencies are increasingly prioritizing cost-effective preservation treatments to maintain pavements, with asphalt overlays being a primary strategy. The pavement engineering process for treatment selection has a disconnect between network-level management data and project-level decisions. This research aims to develop a data-driven framework to link pavement management, structural evaluation, and pavement design through asphalt overlay performance models.
The objectives are to model asphalt overlay performance, validate structural data informs performance models, and predict site-specific performance. This study uses network-level data from Virginia Department of Transportation (VDOT) including over 15 years of automated pavement condition survey data and Traffic Speed Deflectometer (TSD) structural testing to build performance models.
Findings demonstrate the TSD Surface Curvature Index (SCI) exhibits superior correlation with pavement performance models compared to Falling Weight Deflectometer. The TSD structural data provided the most important features to predict cracking and rutting distress following an asphalt overlay. The machine learning models achieved site-specific prediction accuracy, standard error of 1.5% for cracking and 0.03 inches for rutting, meeting the Mechanistic-Empirical Pavement Design (MEPDG) local calibration criteria. By addressing data imbalance, synthetic data generation and quantile random forest machine learning approaches improved prediction accuracy at higher, critical distress values.
The findings advance a conceptual framework for the pavement maintenance life cycle, demonstrating how moderate-detail agency data can inform thin asphalt overlay design decisions using network-level data. This new approach enables improved overlay treatment selection, enhances network-level planning, and supports longer-lasting maintenance investments.
Smith, Bryan. Development of Performance Models for Asphalt Overlay Design Using Traffic Speed Deflectometer and Pavement Management Data. University of Virginia, Civil Engineering - School of Engineering and Applied Science, PHD (Doctor of Philosophy), 2025-07-27, https://doi.org/10.18130/76xq-tj75.
