Abhijit Roy

ORCID Identifier(s)


Graduation Semester and Year




Document Type


Degree Name

Doctor of Philosophy in Civil Engineering


Civil Engineering

First Advisor

Mohsen Shahandashti


Past earthquakes have revealed that earthquakes disrupt the operations of underground water infrastructure systems. Assessment of the seismic vulnerability of underground water pipe networks plays a critical role in formulating preventive rehabilitation decision-making to ensure maximum serviceability after an earthquake event and avoid high repair costs. Although existing seismic vulnerability assessment methods and seismic rehabilitation decision-making models are sensitive to water pipe network uncertainties (e.g., uncertainties in nodal demand, reservoir head, and pipe roughness coefficient) and pipes’ degradation, the extent of the effects of the network uncertainties and pipes’ degradation on the postearthquake serviceability of the network and seismic rehabilitation decision-making has not been examined. The serviceability and damage of a water network after a seismic event depend on the hydraulic properties and physical properties of the network. The hydraulic properties and physical properties of the network are sensitive to network uncertainties and degradation. So, it is necessary to investigate the effects of network uncertainties and degradation of proactive seismic rehabilitation decision-making of water distribution networks. This research is divided into three sections to investigate the effects of water network uncertainties and pipes’ degradation on seismic vulnerability assessment models and seismic rehabilitation decision-making. In the first section, this research investigates the effects of water pipe network uncertainties on the seismic vulnerability assessment of networks. The approach was tested on two networks (New York Tunnel Network and Oberlin Network). The statistical analysis results indicated that the combined impact of the three selected water pipe network uncertainties on the seismic vulnerability assessment of networks is statistically significant. Nodal demand and pipe roughness coefficient uncertainties did not individually have a statistically significant effect. The individual effect of reservoir head uncertainty was statistically significant. Sensitivity analysis determined the minimum value of the coefficient of variation to have a statistically significant effect. The results from sensitivity analysis showed that a small uncertainty in the reservoir head results in a statistically significant effect on seismic vulnerability assessment. By contrast, the coefficient of variation for uncertainties in nodal demand and pipe roughness has to be quite large to significantly affect seismic vulnerability assessment. The next section aims to explore the impacts of water network uncertainties on proactive seismic rehabilitation decision-making. Pipe roughness coefficient, demand, and reservoir head were selected as uncertain network parameters for this study. Critical pipes were identified for a limited budget constraint considering these three network uncertainties individually and combinedly. Sensitivity analysis was performed to quantify selected network uncertainties. A stochastic combinatorial optimization problem was formulated considering network uncertainties and seismic ground motion intensities to identify the most critical pipes of a network for a limited rehabilitation budget. A simulated-annealing algorithm was used to solve the stochastic combinatorial optimization problem. Modena network was used to demonstrate the method. The optimization results showed that the selected network uncertainties significantly affect the identified critical pipes of the water pipelines. Also, the maximum achievable serviceability index for the selected rehabilitation budget reduces significantly if network uncertainties are considered. This index has been reduced by 3%−4% due to the consideration of all three network uncertainties. This third part of the research aims to investigate the effects of the degradation of pipes on the seismic rehabilitation decision-making of water distribution networks. Simulation experiments were designed to investigate the effects of degradation on the inside surface of pipes and on the outside surface of pipes individually and combinedly. The seismic repair rate was calculated considering the effects of degradation based on the probabilistic stress change of the pipe with age. The probabilistic nature of the pipes’ outside degradation rate was considered to determine the probabilistic value of stress change. A probabilistic pipe roughness growth rate model was used to modify the hydraulic modeling of pipe considering pipes’ inside degradation. A simulated annealing-based optimization approach was used to identify the critical pipes and associated maximum serviceability for each experiment and each budget constraint. The Analysis of Variance (ANOVA) test and Tukey statistical tests were conducted to identify the statistical significance of the effect of integrating degradation. The application of the proposed approach was illustrated on the Modena network. Five rehabilitation budget constraints were selected for this study. Critical pipes were identified for each rehabilitation budget constraint based on the optimization algorithm. The results for each simulation experiment showed that the identified critical pipes were different. The associated maximum serviceability was reduced for the same budget constraints if outside and inside degradation were considered individually and combinedly. The changes in identified critical pipes and associated maximum serviceability due to the consideration of outside and inside degradation imply the dependency of a proactive seismic rehabilitation decision-making model on outside and inside degradation. The statistical test results imply that the degradation of pipes in the water distribution network has an impact on seismic rehabilitation decision-making models of water distribution networks. Therefore, it is recommended to integrate the degradation effect with existing seismic rehabilitation decision-making models.


Proactive seismic rehabilitation, Network uncertainty, Degradation of pipes


Civil and Environmental Engineering | Civil Engineering | Engineering


Degree granted by The University of Texas at Arlington

Available for download on Saturday, February 01, 2025