2011 IR&D Annual Report

Development of Computational Methodology to Assess Structural Damage in
Spillway Sections of Dams, 20-R8131

Principal Investigators
Biswajit Dasgupta
Kaushik Das
Debashis Basu
Ron Green

Inclusive Dates:  01/01/10 – 10/01/11

Background — Damage from scour and erosion of rock downstream of dam spillways is a safety concern. Spillways are overflow structures and high-velocity flows over unlined spillways can cause significant damage by erosion and headcutting of the bed rock risking instability of the dam. Examples of rock scour and erosion of hard rock from high discharges during flood inflicting severe damage are Bartlet Dam, Arizona; Tuttle Creek and Milford Dams, Kansas; Kariba Dam, Zimbabwe; Ricibayo Dam, Spain; and Srisailam Dam, India. High Flows caused significant erosion of the rock at the toe of the Midskog Dam in Sweden and on the unlined spillway at Canyon Lake Dam, Texas. Scouring of riverbed rock is induced when the erosive capacity of the water exceeds the resistive capacity of the rock mass. The erosive capacity of water is related to the flow conditions. Indicator parameters most often used are flow velocity, shear stress and stream power. The resistive capacity primarily relates to the ability of the rock mass to withstand the hydro-mechanical forces from the flowing water. During the erosion process, the rock blocks are ejected or fractured from the riverbed and swept away by the flow of water. The process stops when the energy is insufficient to cause further rock breakage or raise the broken material out of the scour hole. Currently, the dam industry relies primarily on physical model testing and empirical methodologies to predict scour potential and guide the hydraulic design of the spillways. These approaches, however, cannot accurately predict scour location, extent, and depth in the prototype structure or help understand erosion dynamics.

Approach — The dominant failure modes for erosion of unlined spillways are rock block removal, brittle fractures of rock and fatigue fracturing. All the failure modes are caused by transient water pressure in the rock joints or in preexisting closed fissures. In this research, a methodology is developed to explicitly evaluate the erosion mechanism by computational modeling. The methodology integrates fluid dynamics modeling techniques that are used to simulate the water flow with a geotechnical model that is used to assess rock damage caused by the impacting fluid. The erosive capacity of water is estimated by fluid dynamics simulation of the turbulent water flow and resistive capacity by geomechanical modeling of the rock. Computational fluid dynamics (CFD) is used to model the unsteady turbulent flow field caused by spillway discharge. Time-dependent pressure and shear stress on the spillway surface were obtained from the CFD calculations that were subsequently used in the geomechanical discontinuum model. The geomechanical model explicitly includes geologic structural features of jointed rock media, and the intact rock blocks between the joints were modeled as micro blocks by Voronoi tessellation. The hydrodynamic response of the rock caused by transient pressure and shear stress modeled the scour damage simulating block removal and brittle fracture modes. The modeling approach is a coupled process between CFD and geomechanical analyses. Therefore, after each cycle of CFD and geomechanical analysis, the geometry of the model is modified based on scour profile and the process is continued until the water jet fluctuations have insufficient energy to induce further erosion. Based on a review of the literature and interactions with peers at conferences, a coupled approach using fluid dynamics and discontinuum modeling to predict spillway erosion has not been demonstrated before.

Accomplishments — The plunge pool of the Kariba Dam in Zimbabwe and the emergency spillway of Canyon Lake Dam in Texas were used as examples to demonstrate that the methodology can model the critical failure modes of spillway erosion. Kariba Dam is a concrete arch dam on the Zambezi River. There are six rectangular-shaped gates near the crest of the dam. The water jet emerging from these gates when open impinges on the plunge pool from a height of 75 m. The plunge pool progressively scoured from 1962 to 1982, yielding a scour profile. CFD simulations have been carried out for a single gate opening with velocity inlet boundary condition replicating the measured velocity of the water jet. The three-dimensional flow-field pattern of the jet and its interaction with the plunge pool water was simulated and pressure fluctuation from jet impact on the plunge pool rock bed was obtained. Geomechanical modeling of the erosion process of the riverbed material consists of several stages that involve initialization of the model and application of jet impact loading from CFD modeling. The analysis showed the scour depth and profile are controlled by the joint orientation, while the damage of rock is controlled by the brittle fracture.

During the record flood in 2002, the emergency spillway at Canyon Lake Dam was overtopped for the first time, resulting in excavated soil and trees as well as significant erosion and transport of rock, thus creating the Canyon Lake Gorge. The time-dependent velocity profile similar to the flood hydrograph was used to specify the inlet condition for the CFD model. The CFD model provided the unsteady pressure and shear stress distribution at several points along the spillway floor. During overtopping of the spillway, the pressure fluctuations on the spillway water rock interface were transmitted in the rock joints. The geomechanical analysis demonstrates that the block separation and removal or plucking failure modes are similar to that observed at site.

The combined approach provides a significant advantage over existing physical and empirical models because multiple full-scale numerical model analyses can be performed to study changes in flow characteristics, geometry, and material properties at a significantly lower cost and turnaround time and with greater certainty and accuracy. Additionally, geomechanical modeling provides an understanding of the dominant mechanism of scour and the extent of damage allowing study of the effectiveness of mitigation system design (e.g., rock bolts, cut-off walls). The results were presented in three conference papers. One received the Outstanding Poster Presentation Award at the U.S. Society of Dams 2011 Annual Meeting and Conference, held April 2011 in San Diego, Calif.

Benefiting government, industry and the public through innovative science and technology
Southwest Research Institute® (SwRI®), headquartered in San Antonio, Texas, is a multidisciplinary, independent, nonprofit, applied engineering and physical sciences research and development organization with 9 technical divisions.