Date on Master's Thesis/Doctoral Dissertation


Document Type

Doctoral Dissertation

Degree Name

Ph. D.


Civil and Environmental Engineering

Degree Program

Civil Engineering, PhD

Committee Chair

Rockaway, Thomas

Committee Co-Chair (if applicable)

Mohsen, J. P.

Committee Member

Sun, Zhihui

Committee Member

McGinley, W. Mark

Committee Member

Biles, William


Pavements, Porous--Testing; Neural networks (Computer science)


This dissertation is a numerical modeling study based on the findings of the two installed Permeable Interlocking Concrete Pavements (PICPs) in Louisville, KY and twenty one laboratory models. A new model derived to more accurately predict the captured surface runoff volume by the PICPs using Artificial Neural Networks (ANNs). The proposed model relates rainfall parameters and site characteristics to the runoff volume captured by the permeable pavements. The database used for developing the prediction models is obtained from the collected data of the monitored permeable pavements. The performance of the ANN-based models are analyzed and the results demonstrate that the model results compare satisfactorily with measured values. A parametric study is completed to determine the sensitivity of a variety of parameters on the captured runoff volume. The results indicate that the developed model is capable of estimating the captured runoff by the permeable pavements for different rain events and site characteristics. The ANN model considers all significant contributing factors and provides a more precise volume prediction than the linear model. Clogging, which is mainly caused by sediment deposition, is the other important factor that result in performance failure of PICPs. Measuring Volumetric Water Content (VWC) by Time Domain Reflectometers (TDRs) is an automated method to track the speed of clogging. Monitoring peak VWC during rain events has been used as an indication of clogging progression over the PICP. Five ANN models are developed from the recorded VWC in order to compute the peak VWC from the rainfall parameters and maintenance treatment. A comprehensive set of data including various rain events characteristics obtained from the rain gauge and the conducted maintenance on the PICP are used for training and testing the neural network models. The performances of the ANN models are assessed and the results demonstrate satisfactory model accuracy when compared to the measured values. A parametric study was completed and the results indicate that the models are capable of estimating the peak VWC of the permeable pavements for different locations. The models consider all the contribution factors and provide more precise prediction values than the linear model. Peak 5 minute intensity, the previous rainfall depth, and the cumulative rainfall depth from the installation are the most critical parameters with respect to the hydrologic performance of the PICP. Finally, twenty one model configurations with different combinations of slope, gap size, and joint filling material were built to study clogging progression and permeable pavement performance. In this study, a neural network model was used to predict the clogging progression rate with critical PICP characteristics. The results indicate that the model is accurately predicting the extent of clogging along the length of permeable pavement. Sensitivity analyses are completed and the results suggest surface slope and location as the most influential parameters on the clogging length. Moreover, the prediction model for infiltration edge progression is presented to estimate the rainfall depth with 99% accuracy on testing datasets. By predicting the precise cumulative rainfall depth based on the infiltration edge distance and the PICP specifications, the hydrologic operation for each configuration and at any rainfall depth is accessible. The results demonstrate that surface slope and gap size present the highest influence on the infiltration edge progression. By better understanding the effects of pavements’ specification and site characteristics and selecting the most efficient pavement configuration, improved future design and more effective maintenance operations can be achieved.