Xiang, Shengli
(2025)
Advanced techniques for characterizing and predicting asphalt macrotexture.
PhD thesis, University of Nottingham.
Abstract
In the realm of asphalt mixture research, traditional texture measurement methods, such as laser scanning and sand patch tests, have long been plagued by high costs, complex operations, and time-consuming procedures. These limitations have hindered the efficient assessment of asphalt surface characteristics crucial for road safety and maintenance. To address these challenges, this thesis embarks on a comprehensive investigation to establish an efficient texture prediction framework by leveraging advanced techniques like X-ray computed tomography (CT), virtual asphalt modeling, and deep learning.
The first study utilizes X-ray CT to delve into the intricate relationship between the void properties of asphalt mixtures and the Mean Profile Depth (MPDCT) of Marshall core samples. Departing from the conventional focus on air void content alone, this research uncovers a strong linear correlation between MPDCT and topological characteristics of voids, including average void diameter (with a correlation coefficient of 0.89). This breakthrough reveals that these properties can serve as more informative predictors of MPD, offering a more nuanced texture control in asphalt design.
Building upon the first study's findings, the second study explores the nonlinear relationship between MPDLS measured via laser scans of asphalt slabs and Mean Texture Depth (MTDSPT) obtained from sand patch tests. By employing the Intersected Stacked Air Voids (ISA) method, this research validates the feasibility of virtual asphalt modeling as an alternative to CT, reducing the reliance on expensive imaging equipment. The study successfully establishes a nonlinear model for MPDLS-MTDSPT, elucidating how maximum aggregate sizes and air void content influence this relationship.
The third study represents a significant leap forward, introducing a deep learning (DL) approach based on Visual Geometry Group (VGG)-based Convolutional Neural Network (CNN) architecture. By training the model on two-dimensional images captured by consumer-grade cameras, this innovative method enables cost-effective and non-contact measurement of asphalt surface texture. The model classifies MPDLS ranges with high accuracy and predicts exact MPDVGG values through regression analysis, showcasing the potential of digital techniques to replace traditional field measurement methods.
These three studies present a “CT - Virtual - DL” three-tier substitution solution, which has demonstrated the ability to reduce measurement costs by up to 90%. Integrating classification and regression techniques, this efficient framework provides precise and actionable insights, making it applicable across the entire asphalt mixture design, acceptance, and monitoring lifecycle. By advancing from laboratory analyses to real-world pavement assessments, this research paves the way for automated, cost-effective texture evaluations, ultimately enhancing road safety and maintenance strategies through deep learning optimization and non-contact measurement techniques.
| Item Type: |
Thesis (University of Nottingham only)
(PhD)
|
| Supervisors: |
Albertini, Gabriele
Thom, Nick |
| Keywords: |
Asphalt Mixtures, Mean Profile Depth (MPD), Mean Texture Depth (MTD), Volumetric Properties, Topological Properties, X-ray Computed Tomography (CT), Digital Image Processing, Virtual Asphalt, Macrotexture Assessment, Laser Scanning, Sand Patch Test, Nonlinear Relationship, Intersected Stacked Air Voids (ISA) Method, Deep Learning (DL), Visual Geometry Group (VGG), Convolutional Neural Networks (CNNs), Automated Measurement Techniques, Regression Analysis, Image-Based Prediction, Asphalt Surface Characterization, Non-Contact Measurement |
| Subjects: |
T Technology > TE Highway engineering. Roads and pavements |
| Faculties/Schools: |
UK Campuses > Faculty of Engineering > Department of Civil Engineering |
| Item ID: |
81477 |
| Depositing User: |
XIANG, Shengli
|
| Date Deposited: |
31 Dec 2025 04:40 |
| Last Modified: |
31 Dec 2025 04:40 |
| URI: |
https://eprints.nottingham.ac.uk/id/eprint/81477 |
Actions (Archive Staff Only)
 |
Edit View |
Tools
Tools
