Degree

Doctor of Philosophy (PhD)

Department

Civil and Environmental Engineering

Document Type

Dissertation

Abstract

The study was carried out using finite element (FE) models to assess the advantages of using geosynthetic layers as reinforcement in flexible pavements. A comprehensive parametric study was performed to determine the impact of various factors like asphalt layer thickness, base thickness, geosynthetic type, geosynthetic stiffness, and subgrade stiffness. The FE simulations were run for 100 load cycles and permanent deformation (PD) was used to adjust the ME rutting equation parameters for each layer. The PD data was extrapolated to determine the pavement's service life. The comparison of the PD curves of the unreinforced and reinforced sections was used to calculate the Traffic Benefit Ratios (TBR) at different levels of rutting targets. I addition, by calculating the ratio between the calibrated rutting curves of the base and subgrade layers, the reduction coefficients of 𝛼𝑏 and 𝛼𝑠 are derived, which can directly be used in the design of geosynthetic reinforced pavements. Then, the TBR values are used as input parameters to AASHTOWare to derive other benefits like effective resilient modulus (MR-eff) and Equivalent Base Thickness (EBT). The results of the study showed that incorporating one layer of geogrid or geotextile at the base-subgrade interface greatly reduces pavement rutting. The study found that using geogrid is more effective than using geotextile in reducing rutting as it interlocks with the base aggregates. The TBR values increased with deeper rutting and higher geosynthetic stiffness. The optimal combination of TBR, MR-eff, and EBT for low-volume roads with asphalt thickness of 3.5 in. was found to be at a base thickness of 10 inches. For medium (6.5 in. asphalt thickness) and high (10 in. asphalt thickness) volume traffic roads, the benefits are reduced with an increase in base thickness. In addition, to predict the benefits of using geosynthetics in flexible pavements in terms of TBR, MR-eff, EBT, 𝛼𝑏 and 𝛼𝑠 based on the influential parameters, several prediction models are developed. These prediction models include regression and machine learning models, including Multiple Linear Regression (MLR), Multivariate Exponential Regression (MER), Decision Tree (DT), Support Vector Machine (SVM) Random Forest (RF), and Gradient Boosting (GB). The prediction accuracy of the models was evaluated x using different metrics of the coefficient of determination (R2), Mean Square Error (MSE), and Mean Absolute Error (MAE). The results showed that the RF and GB models for a given combination of unseen parameters of pavement structure are able to accurately predict the benefits of using geosynthetics in pavement structure in terms of TBR, MR-eff, EBT, 𝛼𝑏 and 𝛼𝑠. Furthermore, FE-ME results are used to propose a new rutting model for base and subgrade layers. The proposed rutting model can capture the reinforcing effect of geosynthetics and changes in different variables of pavement structure by changing the calibration coefficient of π›Όβˆ—.

Date

2-27-2024

Committee Chair

Abu Farsakh, Murad

Download

Available for download on Tuesday, February 25, 2031

Share

COinS