This thesis highlights the significance of using the Resilient Modulus (Mr) and Soil-Water Characteristic Curve (SWCC) coefficients within the scope of pavement design. It aims to demonstrate the importance of pavement quality and safety by developing a prediction model for bleeding and rutting depth in asphalt pavements using data from the Long-Term Pavement Performance (LTPP) database and the National Catalog of Natural Subgrade Properties Database. Additionally, the thesis builds predictive models using Artificial Neural Networks (ANNs), which include a single hidden layer and multiple neurons. The developed Mr and SWCC models in this thesis utilized the basic soil index properties from the state of Texas, and simple equations were extracted from the models. In fact, the developed models show potential implications for enhancing pavement design. Initially, an R2 value of 0.75 was obtained for bleeding prediction for a specific dry-no-freeze climate zone, considering traffic volumes, pavement thickness, asphalt content, and climatic factors. Furthermore, in predicting rut depth, factors like Temperature, Equivalent Single Axle Load, Resilient modulus, and Thickness of hot mixed asphalt layer were considered for the state of Texas, which resulted in a mathematical relationship between rut depth and the input variables.
Date of publication
Mena Souliman, Gokhan Saygili, Mayzan Isied
Masters in Civil Engineering
Khalifah, Rami, "PREDICTIVE MODELING OF ASPHALT PAVEMENT PERFORMANCE USING ARTIFICIAL NEURAL NETWORKS" (2023). Civil Engineering Theses. Paper 24.
Available for download on Saturday, December 06, 2025