Page 249 - Numerical Analysis and Modelling in Geomechanics

230 SEISMIC MICROZONING USING NUMERICAL MODELLING
Since most of the site amplification analyses required accelerograms as
reference input, artificial time-histories were simulated following the method
proposed by Sabetta and Pugliese (1996).
The method has features that make it particularly interesting:

• It reproduces the non-stationarity, in amplitude and frequency, of the real
ground motions;
• It allows the simulation of a family of time-histories, requiring only the
magnitude of the reference earthquake, the distance from source to site, and
the local site geology as input parameters;
• The simulated time-histories fit the recorded Italian accelerograms in terms of
several ground motion amplitude measures such as peak acceleration, peak
velocity, Fourier spectra, and response spectra.

To match the reference spectra and the peak ground accelerations of the three
reference villages, the non-stationary time-histories were further modified by
scaling the amplitude of each Fourier spectrum without modifying the phase; the
resulting accelerograms are shown in Figure 8.6.

Analysis methods
The present chapter, given the space limitations, cannot pretend to provide an
exhaustive survey of the whole literature but does provide a brief outline of the
computer programs used to estimate the “ground shaking” site effects in the
Umbria-Marche region, and to emphasise the main issues as yet unsatisfactorily
answered by the methods used and the limits of their application.
Three kinds of computer programs have been used. The first, SHAKE or
PSHAKE, is based on a simple model that was developed in the 1970s, and is
now almost routinely used in engineering practice (1D linear or linear equivalent
approach). The second, QUAD4 or QUAD4M, based on the Finite Element
Method, FEM, was used in our work to solve the site effect problems essentially
in the case of very soft valleys in 2D geometry. The third, BESOIL (Sanò,
1996), is based on the Indirect Boundary Element Method, IBEM, and allows
accounting for very complex 2D geometry and wave propagation.

1D model
The computer program SHAKE was written in 1970–71 by Schnabel and
Lysmer and was published in December 1972 by Schnabel et al. This has been
by far the most widely used program for computing the seismic response of
horizontally layered soil deposits. A new improved version, SHAKE91, has
recently been published (Idriss and Sun, 1992).