Seismic records on existing dams assume great relevance due to the possibility to enhance the understanding of the seismic response of dams empirically compared with the characteristics of earthquake input. In this frame, the seismic analysis of Kasho gravity dam during the October 6, 2000 earthquake (Mw =7.1) and the analysis of the signals recorded at Shintoyone arch dam during the two earthquakes on March 16, 1997 (Mw = 5.8) and April 3, 2001 (Mw = 5.2) have been carried out.

In the study of Kasho dam, the results of the numerical simulations have shown as the dynamic response of the structure, under the effect of the seismic loading, remains within the limits of the elastic behaviour, in agreement with the evidence that no significant damages were observed after the earthquake. Moreover, the analysis has highlighted as the massless foundation approach, although fairly predicts the natural frequencies, leads to overestimate the acceleration amplitude both at the dam foundation and at the dam crest. In order to reduce these discrepancies, equivalent damping ratios (ranging from 5% to 15%) were considered in the simulations and their effect on the seismic response of the dam was evaluated.

The analysis of Shintoyone records was the object of the second part of this work. Baseline correction and filtering techniques were extensively used to remove the noise introduced by the instrumentation from the recorded accelerograms. Modal identification was performed by using, among the available methods, the Cross Spectrum, the Transfer Function, the Frequency Domain Decomposition and the Mean Transfer Function methods. Capabilities as well as advantages and drawbacks of each method were identified. Using the Fourier spectrum and the half power method an estimation of the structural damping ratio was also deduced.

In the last part of the report the viscous-spring boundary model is described; this technique is aimed to address the problem of overestimation of the seismic response resulting from the massless foundation approach. Some preliminary numerical tests were carried out using the spring and dashpot elements available in the finite element code Abaqus and calculating the effective earthquake forces to be applied at the truncated boundary of the foundation, following a procedure detailed in literature, based on the theoretical solution of the elastic wave problem in a half-space.