Modified demand-capacity ratio methodology for seismic performance assessment of concrete gravity dams

Amongst several hazards a concrete gravity dam may experience during its lifetime, an earthquake can cause significant damage to the dam. Notable examples are Koyna dam, India, from the 11 December 1967, Koyna earthquake (Mw 6.6), and Shih-Kang dam, Taiwan, from the 21 September 1999, Chi-Chi earthquake (Mw 7.7). The Association of State Dams Safety Officials (ASDSO) in 2016 estimated that it would cost $60 billion to rehabilitate all the dams in the USA. Similarly, in dealing with ageing dams, the Government of India initiated the Dam Rehabilitation and Improvement Project (DRIP) of 257 dams at the cost of $533 million in 2012. Since the consequences of and after dam failures are significant, continuous research is essential for the safety assessment and design of dam structures. This study investigates the effects of strong ground-motion characteristics on the linear and nonlinear response of concrete gravity dams. In current practice, the Demand-Capacity Ratio (DCR) model developed by Ghanaat (2002) is widely used in the seismic safety assessment of concrete gravity dams. The allowable magnitude of stress excursions and the provision for the spatial extent of such stresses forms the basis of the model. According to the DCR model, the response is considered acceptable if the stress demand-capacity ratios vary between 1.0 and 2.0 linearly, and the cumulative duration of inelastic stress excursions falls below 0.3 seconds at DCR=1.0. Nonlinear analysis is recommended if the DCR values exceed 2.0. The methodology's main limitation is the lack of nonlinear damage correlation, which may lead the user to the inappropriate assessment of seismic performance. Furthermore, for concrete gravity dams, a lower cumulative inelastic duration of 0.3 seconds has been assumed based on a heuristic derivation from the arch dams of 0.4 seconds. Therefore, this thesis investigates the nonlinear response of a case study dam that has experienced tensile stress demands between DCR 1 and 2 to a suite of earthquake ground motions comprising near-fault earthquakes of fling step and forward rupture directivity characteristics. Obtained results are compared to a newly developed crack width criterion to identify the effects of ground motion characteristics. ABAQUS software has been used in the numerical modelling of the two-dimensional dam-reservoir-foundation system. The geometric nonlinearities have been applied at the dam-reservoir, reservoir-foundation, and dam-foundation contact interfaces in the modelling. In simulating the far-field effects, seismic radiation at the foundation's bottom and side boundaries were minimised by using infinite elements. Furthermore, the Concrete Damaged Plasticity model that includes the strain hardening and softening behaviour was employed for the material nonlinearity. Initially, the validation analysis was performed on the Pine Flat Dam to examine the boundary conditions and interactions' functional effectiveness. The Koyna dam, which experienced a peak horizontal ground acceleration of 0.47g during the 1967 Koyna earthquake (Mw 6.6) and suffered damage, is selected for the main study. In the first part of the investigation, linear time-history analyses on the Koyna dam is carried out using the selected sets of ground motions. For the analyses, twenty near-fault, twenty-five forward directivity, and twenty-five fling-step ground motions with varying amplitudes, durations, and frequencies are used. Ground motion accelerations are applied at the dam base in both +1 and -1 directions to characterise the most affecting scenario. In the second part, the demand-capacity ratio (DCR) model based on the allowable stress concept is used to evaluate the linear seismic performance of the Koyna dam. Analysis results of the dam subjected to the sets of ground motions are compared with each other. This includes the response accelerations and response velocities at the dam's neck and crest and the displacement response at the dam crest to assess the maximum response for the strong ground motion characteristics. The tensile stresses on the upstream and downstream are examined to identify the locations of stress concentrations. The linear seismic performance of the Koyna dam has been evaluated using the DCR model to identify the ground motions that have caused the dam's stresses to exceed the performance threshold curve. In the form of tensile cracking of the concrete, stiffness degradation has been observed at the change in the dam's cross-section at the neck region and the heel on the upstream face of the dam. In the third part, nonlinear analysis has been performed to assess the accumulated tensile cracking damage to the dam. Furthermore, the emphasis is given to study the dam's damage to the ground motions that have generated stress DCR's between 1 and 2. This is performed by varying the fracture energy of the concrete. A new response parameter based on the local damage indices calculated as the ratio between the tensile crack length to the potential crack pattern length has been established to correlate the damage. Results show that the near-fault ground motions with high-velocity pulses cause a higher dam response for the linear analysis and significant damage to the dam in the nonlinear analysis. DCR's and the total cumulative inelastic duration are higher when the dam experienced fling-step ground motions, mainly because of the long bracketed and significant durations of the ground motions. The magnitude of stresses at the dam's neck and the base is higher for the forward rupture directivity ground motions, representing their impulsive force characteristics, which can cause sliding and cracking in a short duration. The initiation and evolution of crack patterns at neck and heel areas are estimated appropriately to the laboratory tests and similar numerical studies. The local damage at the neck and heel region showed significant sensitivity to fracture energy. A new methodology by varying the fracture energy has been proposed to bridge the gap between the linear performance and the nonlinear damage assessment. This methodology will inform the existing demand-capacity ratio model to understand the dam's nonlinear behaviour for the DCR's between 1 and 2. According to the proposed Crack-Width Ratio (CWR) methodology, the performance assessment varies as a function of fracture energy and potential for cracking at a given location. Although this rationalisation is based on the Koyna dam, the structure's natural period must be considered in the calculation of stress excursions. Practitioners can select ground motions pertinent to a dam region and assess the potential for cracking normalised on the Koyna dam's response. Given the complexities of modelling dam structure response to earthquake ground motions, the framework provided in the thesis can be used for further research in this field.