Semester of Graduation

Summer 2020


Master of Science in Civil Engineering (MSCE)


Civil and Environmental Engineering

Document Type



Dams are infrastructure assets of extreme importance; the failure of which can have catastrophic consequences on adjacent communities. Therefore, it is prudent for authorities to evaluate the safety of existing dams in seismically active regions. This thesis focuses on carrying out a reliability analysis of concrete gravity dams under earthquake loads using an enhanced finite element model. For many years, reliability studies on concrete gravity dams were performed using a finite element model that considered a massless foundation and assumed that water is incompressible. Moreover, the design ground motion was applied without any modification either at the bottom fixed boundary of the foundation domain or at the dam-foundation-rock interface. As these assumptions do not accurately represent real-life conditions, investigations incorporating realistic constraints are necessary.

In this study, a finite element model that considers three subdomains is employed - namely the dam, its foundation domain that includes mass, stiffness, and material damping, and a fluid domain that includes water compressibility. The interactions between the different subdomains is also included in the model. The truncated boundaries of the foundation-rock and fluid domains are modeled using standard viscous-damper boundaries. In addition, effective earthquake forces obtained by deconvolving ground motion are specified at these boundaries. Non-linear time-history analyses are performed using the developed model by considering uncertainties associated with material data as well as the aleatoric nature of earthquakes. Fragility curves are then obtained for the limit states - base sliding, excessive deformation, and tensile cracking at the upstream face and at the neck - with the goal of using them to assess the risk of the dam under earthquake loads.

The results are compared with that of Sen and Okeil [1]. It is observed that the differences in modeling assumptions have a significant impact on the probabilities of failures for the limit states. The critical limit state in this study was found to be excessive deformation. In contrast, in the study by Sen and Okeil [1], tensile cracking was identified to be the critical limit state. Among both the studies, the tensile cracking limit state was the least different. It was also observed that a loss of reservoir control could occur in the event of a moderate to strong earthquake.

Committee Chair

Okeil, Ayman