Page 181 - Dynamic Loading and Design of Structures

Page 156

●distance to source (or epicentral distance);
●closeness to the site under consideration.

It is beyond the scope of this book to discuss in detail these parameters and their relation to
the characteristics of the ground motion. In a practical context, the need becomes clear of
using a minimum number of criteria (ideally only one) for selecting ground motion records
for the purpose of time history analysis. An interesting proposal in this respect is the use of
the A/V ratio (Zhu et al., 1988), which is a simple parameter, easy to calculate from the
commonly available values of A(≡PGA) and V(≡PGV), and correlates well with the M–R
relationship, as well as with site conditions.
Another possible criterion is to select ground motion records whose spectra (Spa or
preferably S ,) are peaking in the vicinity of the fundamental period of the structure under
pv
consideration, irrespective of their other characteristics, which is generally a conservative
approach.

Scaling of recorded accelerograms
Whenever a careful selection of natural accelerograms has been made, for instance on the
basis of (M, R) pairs within a narrow range, one might argue that these could be directly used
for design purposes. In fact, if these records are used for analysis, significant variability in the
calculated response is found; Shome et al. (1998) reported dispersions of about 50 per cent to
60 per cent in the inelastic peak interstorey drifts of medium rise steel frames subjected to sets
of motions, each corresponding to a narrow magnitude range (e.g. 6.5–7.0) and distance range
(e.g. 50–70 km). This significant variability is attributed to the very different characteristics of
ground motions at a given location resulting from an earthquake of a given M, and is a clear
indication of the effect of neglecting the other important parameters characterizing the ground
motion. This points to the need for scaling (or normalizing) the selected earthquake
accelerograms before using them for time history analysis. In addition to the foregoing
considerations, scaling is also necessary whenever different limit states (serviceability,
ultimate, etc.) have to be considered, since it is generally impractical to select different sets of
records for each limit state.
The most commonly applied scaling procedure is based on the PGA (i.e. all records used
for design are scaled to the same PGA). Unfortunately this convenient procedure is one of the
most unsatisfactory ones, with the exception of structures with very low periods (not
exceeding about 0.2 sec). As discussed in Section 4.3.2, the spectral ordinates are proportional
to the PGA over the short period range only, whereas for longer periods (covering most of the
usual civil engineering structures) they are proportional to the PGV, and for very long periods
(more than about 3.0 sec) they are proportional to the PGD. The peak ground parameters have
indeed been used as scaling factors, and so have the integrals of their squared values and their
root-mean-square values (Nau and Hall, 1984). All these values