Time-history deformation analyses of upstream-raised tailings dams use seismic records as input data. Such records must be representative of the in-situ seismicity in terms of a wide range of intensity measures (IMs) including peak ground acceleration (PGA), Arias intensity (AI), cumulative absolute velocity (CAV), source-to-site distance, duration, among others. No single IM is a sufficient descriptor of a given seismic demand (e.g. crest settlement) because different records, all of them compliant with any IM, can produce a very wide range of results from insignificant damage to global failure. The use of brute force, where hundreds of seismic records compliant with a set of IMs are employed, has proven to be a reasonable workaround of this limitation, at least able to produce a probabilistic density function of demand indicators. This procedure, however, requires a large number of runs, and is therefore expensive and time-consuming. Analyses can be optimized if an a priori simple tool is used to predict which seismic records would yield a given demand, thus obtaining estimations with much fewer runs. In order to perform a more precise selection, a semi-analytical screening procedure is presented in this paper. The procedure makes use of the spectral properties of the seismic record, considering only the intensity of the frequency content which is not filtered by the dam to obtain an a priori estimate of demand, expressed in this case in terms of displacements. The tool is validated using analytical and numerical models that prove insensitivity to the constitutive model used in the analysis, and is applied to a large tailings dam subjected to strong earthquakes.
Download