GNGTS 2014 - Atti del 33° Convegno Nazionale

GNGTS 2014 S essione 2.2 205 Methodology. Founded on the outcomes of the geologic survey, eighty measurement sites (yellow pins in Fig. 1b) were almost homogeneously spaced along a grid having size of about 600 m, taking into account the outcropping lithology. Moreover, six transects with four measurement sites each (red pins in Fig. 1b), where achieved along the fault lines and morphologic scarp, mostly located in the eastern part of the island. Ambient noise was also recorded in five buildings, selected as � ������ ������ �� ���� ���������� �������� ��� ������� a sample survey of both reinforced concrete and masonry edifices� �� ����� �� �������� ����� ����������� ������� ���� ��������� ��� ������� ��� ����� , in order to evaluate their fundamental period. Data recording was carried out using Tromino, a 3-component velocimeter particularly suitable for field measurements. Time series of 20 minutes length were recorded using a sampling rate of 128 Hz and processed through the Horizontal to Vertical Noise spectral Ratio technique (HVNR). Time windows of 20 s were considered and the most stationary part of the signal was selected excluding transients associated to very close sources. In this way the Fourier spectra were calculated in the frequency range 0.1-30.0 Hz and smoothed using a proportional 20% triangular window. Following the criteria suggested by ��� �������� ������� ���� ������� ���������� ����� ������� ����������� the European project Site EffectS assessment using AMbient Excitations (SESAME, 2004)� ���� ��� �������� ����� ����� ������ ��������� ������� ���� ��� ������ �� ��� , only the spectral ratio peaks having amplitude greater than two units, in the frequency range 0.5-15 Hz, were considered significant. Experimental spectral ratios, obtained in the measurement sites located along the transects, were also calculated after rotating the horizontal components of motion (Spudich et al. , 1996) by steps of 10 degrees starting from 0° (north) to 180° (south) in order to investigate about the possible presence of directional effects. �������� �� �������� �� ������� ��� �������� However, in presence of lateral and vertical heterogeneities or velocity inversion, the HVNR can be “non-informative” due to the occurrence of amplification on the vertical component of motion (Di Giacomo et al. , 2005). Thus in this study we also applied the time-frequency (TF) polarization analysis proposed by Vidale (1986) and exploited by �������� Burjánek et al. (2012�� ���� ��������� ��� ������� ����� ������ �������� ). This technique can provide quite robust results, overcoming the bias that could be introduced by the denominator spectrum in the HVNR calculation. Following Burjánek et al. (2010, 2012), the continuous wavelet transform [CWT, see Kulesh et al. , (2007)] is applied to signals in order to select time windows whose length matches the dominant period: signals are thus decomposed in the time-frequency domain and the polarization analysis is applied. For each time-frequency pair, polarization is characterized by an ellipsoid and is defined by two angles: the strike (azimuth of the major axis projected to the horizontal plane from North) and the dip (angle of the major axis from the vertical axis). Another important parameter is the ellipticity that is defined, according to Vidale (1986), as the ratio between the length of the minor and major axes: this parameter approaches 0 when ground motion is linearly polarized. As stressed by Burjánek et al. (2010, 2012), the chosen wavelet in the CWT analysis affects all the polarization parameters as well as the analysis resolution. Polarization strike and dip obtained all over the time series analyzed are cumulated and represented using polar plots where the contour scale represents the relative frequency of occurrence of each value, and the distance to the center represents the signal frequency in Hz. In order to assess whether ground motion is linearly polarized, the ellipticity is also plotted versus frequency. � ������ �������� �� ��� ������������ ����� ��� ��������� �������� ������� ��� ����� A direct estimate of the polarization angle was therefore obtained through the time- frequency polarization (TFP) analysis (Burjánek et al. , 2012). The dynamic properties of a building are usually described through its natural frequency and the damping ratio. The seismic performance of a building obviously depends on the progression of the frequencies along the input time-history, nevertheless the knowledge of its fundamental frequency at low amplitude values is of primary importance to characterize the initial seismic behaviour of a structure (Mucciarelli and Gallipoli, 2007). The engineering practice usually derives the dynamic behaviour of buildings through numerical or experimental methods (Gallipoli et al. , 2009a, 2009b; Oliveira and Navarro, 2009). The results obtained for different typologies of buildings are often processed, through statistical regression analysis, to achieve empirical relationships that let the estimate of building resonant period (T) as a function of the buildings geometry, usually either the height (H) or the