Nonlinear analyses of earth dams at the ultimate limit state with different selection of acceleration input signals

Dams are strategic infrastructures that can be highly susceptible to earthquakes. The nonlinear time-history (NLTH) analysis is currently the most realistic approach to evaluate the seismic response of earth dams through performance-based criteria. An adequate characterization of the seismic motion is a prerequisite of the overall modelling procedure. Spectrum-compatibility is usually invoked by the most advanced technical codes and standards worldwide, that is the difference between the average spectrum of the selected accelerograms and the reference code spectrum must not exceed a determined tolerance in a specific range of periods. The selected time series must also satisfy the requirement of seismic compatibility with respect to the regional seismotectonic and seismogenic context. In addition, other specific requirements may be needed, such as a limitation on the scaling factor of natural accelerograms. To overcome all these limitations, a good solution could be the use of simulated time series, which can address the lack of sufficient natural accelerometric recordings, especially at higher return periods (thousands of years) as required at the ultimate limit state for strategic infrastructures. This paper focuses on a sensitivity analysis to ascertain how a different choice of the input motion might impact the seismic performance of a bituminous-faced rockfill dam. In the first approach, the natural records were retrieved from the Engineering Strong-Motion database (ESM). In the second, a novel code developed for simulating non-stationary stochastic ground motions was employed. By comparing the results obtained from these different sets of analysis, indications will be provided for dam safety assessment.