AI-powered design framework for open-graded friction course (OGFC) asphalt mixtures

Document Type

Article

Publication Date

1-1-2026

Abstract

An open-graded friction course (OGFC) is an asphalt mixture characterized by high air void (AV) content, providing functional benefits such as improved drainage and skid resistance compared to traditional road surfaces. Despite these advantages, OGFC has historically faced durability challenges, leading many highway agencies to evaluate mixture modifications on a case-by-case basis. This study addresses this issue by developing an AI-based tool that enables designers and contractors to estimate OGFC performance without relying solely on extensive laboratory testing. A comprehensive database was compiled by integrating volumetric properties, design parameters, and performance results of OGFC mixtures from multiple sources to support model development and validation. OGFC performance was described using five measures: AV content, Tensile Strength Ratio (TSR), rut depth at 5,000 passes, rut depth at 20,000 passes, and Cantabro loss. An Artificial Neural Network (ANN) was implemented using a Quantile Multilayer Perceptron (Q-MLP) architecture to generate median predictions and associated 95% prediction intervals. Model performance was evaluated using five-fold cross-validation and four statistical measures, achieving an average error of 10.83%, an average R² of 0.9034, 95.1% coverage of the prediction interval, and a pinball loss of 0.032. A Windows-based graphical user interface was developed to support practical use.

Publication Source (Journal or Book title)

International Journal of Pavement Engineering

This document is currently not available here.

Share

COinS