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

Nasraoui, Olfa


Urban runoff--Management; Urban runoff--Measurement; Storm water retention basins; University of Louisville--Sanitation


In order to evaluate the performance of the stormwater best management practices (BMPs) installed on the Belknap campus at the University of Louisville, a comprehensive assessment on the stormwater BMPs’ the flow volume reduction, peak flow attenuation, and overflow area abatement was made. We used a two-pronged analysis based on 1) predictive modeling using data mining approach; 2) model-based hydraulic simulation. The novelty of study is that it not only assessed the stormwater BMPs’ performances on flow volume reduction, but also assessed their performance on peak flow attenuation which is neglected in previous studies and assessment practices. Flow volume reduction and peak flow attenuation were assessed through mining the rainfall and combined sewer flow data before and after the BMPs’ installation. The stormwater BMPs’ performances on overflow area abatement were assessed through contrasting the overflow areas before and after the BMPs’ installation. The radar rainfall data was verified using the local rain gauge data, and the rainfall event is sorted out using a 6 hour dry period. The data mining in this study includes rainfall data validation, data preparation, and modelling. The predictive Multiple Linear Regression Models (MLRMs) and Back Propagation Neural Network Models (BPNNM) were built. For the study area, flow volume in wet weather is mostly controlled by rainfall depth, and followed rainfall duration. Peak flow is decided by rainfall depth, peak rainfall intensity, duration and duration of above average rainfall intensity. Peak flow is negatively correlated with rainfall duration, while positively correlated with other three features. According to both model, the estimated volume of the flow diverted by the stormwater BMPs are approximately 30 million gallons per year, and the magnitude of the peak flow could be trimmed down by approximately 50%. Multiple linear regression and backpropagation neural network are evaluation methods which are not only applicable in the studied case, but also can be widely adopted. However, it shows that the BPNNM is only viable to predict for flow volume lower than 6 million gallons. The overflow area abatement was assessed through contrasting the overflow areas before and after the installation of the stormwater BMPs. Overflow areas were visualized by performing coupled 1D/2D hydraulic simulation. It shows that the overflow areas, which could be saved by the stormwater BMPs, depend on the magnitude of the rainfall event. The abatement in overflow areas is more evident at 4 inch rainfall event and the 1 inch rainfall event. In the 4 inch rainfall event, the overflow areas at the JB Speed School parking lot, Student Rec Center, and College of Business were significantly abated by the stormwater BMPs.