Malsperth

RESPONSE SPECTRA RECOMMENDED FOR AUSTRALIA
Malcolm Somerville, Kevin McCue and Cvetan Sinadinovski Australian Geological Survey Organisation, Canberra Response spectra suitable for intraplate regions such as Australia differ in shape from the familiar response spectral shapes derived from early accelerograph data from interplate regions. In applications such as building code provisions where the design-basis event is prescribed in probabilistic terms, rather than on maximum earthquake potential, the magnitude of the event will be smaller in intraplate regions, and the low-frequency response will be proportionately less. There is also the possibility of systematic divergence of earthquake source spectra between interplate and intraplate events. In this study we develop intraplate spectra empirically from statistical analysis of strong-motion data for magnitude ~6 intraplate earthquakes. 1 INTRODUCTION
The response spectra in the current loading code (AS 1170.4-1993) are based on the
earliest and most familiar accelerograph recordings of earthquakes, but these recordings
are atypical of the strong motion that is most likely to cause damage to structures in
intraplate regions such as Australia. The earliest codifications of response spectral shape,
such as those developed by the Applied Technology Council, did not consider earthquake
magnitude as a parameter determining the shape of the response spectrum. Also, early
codifications of response spectral shape were naturally directed toward critical facilities,
where the worst-case scenario (maximum credible earthquake) was under consideration.
This has led to an inherited tendency of codified response spectra to be representative of
high-magnitude earthquakes. Because response spectra are usually anchored to a
prescribed peak ground acceleration or acceleration coefficient, and because velocity and
displacement increase more rapidly with magnitude than does acceleration, the early
response spectrum codifications exaggerate the velocity and displacement response if the
earthquake under consideration is not of high magnitude.
Subsequent to the early codifications of response spectral shape, the importance of
earthquake magnitude (rather than just surficial site geology) has been recognised as a
determining parameter. Also, ground-motion codes have been developed for ordinary
facilities (as distinct from critical facilities), with probabilistic rather than worst-case
criteria, representing earthquakes with magnitudes less than maximum credible. The
response spectral shape should be appropriate for the magnitude of the most severe
ground-motion event that is likely to occur in the length of time under consideration. For
building codes, the usual span is 475 years (10% chance of exceedance in 50 years).
Because the magnitude of the probable event will be greater for tectonically active
regions than for stable regions, the response spectral shapes will differ. In this sense,
codified response spectral shapes will be different for interplate and intraplate regions.
Aside from stronger overall response, the interplate spectrum will exhibit a red shift
relative to the intraplate spectrum, all else being equal, because of the higher magnitude.
However, all else may not be equal, as it has been suggested[1] that interplate and
intraplate earthquake sources are systematically different, the latter having higher stress
drop and radiating more high-frequency energy (acceleration band) for a given
earthquake magnitude (measured in the velocity band) or seismic moment (measured in
the displacement band). An intraplate blue shift, affecting the acceleration band of the
spectrum, would have obvious engineering ramifications. Empirical evaluations of this
question, and the related issue of earthquake mechanism-dependence of strong motion,
have tended to be inconclusive[2]. Conflicting results in the literature as to
interplate/intraplate differences stem in part from variant definitions of the two
categories. Here we take an earthquake to be typically intraplate if its mechanism is
reverse or thrust, and if its repeat time is relatively long[1].
2 DATA
Our approach in this investigation was to delineate intraplate spectra empirically, by
statistical analysis of intraplate strong-motion data, rather than in relation to interplate
spectra. The main issues then had to do with the appropriate magnitude and whether
there was sufficient intraplate data to obtain statistically meaningful results. We
restricted our data selection to accelerograms recorded at rock sites, so as to bypass the
site-response issue. The data set comprises 13 accelerograms (26 horizontal components)
of strong motion recorded on rock at close range to earthquakes with magnitudes from
5.4 to 6.6: a narrower magnitude range would have yielded too few data. We accepted
the magnitude range 6 ± 0.6 as reasonably representative of the magnitudes of historically
damaging earthquakes in Australia and other intraplate regions. Data sources are the U.S.
National Geophysical Data Center's worldwide archive, and the California Strong Motion
Instrumentation Program. We accepted data for two earthquakes, Whittier and
Northridge, on thrust faults close to the San Andreas plate margin in southern California.
3 ANALYSIS
3.1 Normalisation scheme. Peak ground velocity (PGV) is the parameter that was
chosen for normalising all of the horizontal components of motion. Normalisation by
PGV was chosen as the best way of handling the problem of having a finite range of
magnitudes (it is not strictly appropriate to compare spectra for earthquakes of differing
magnitude, because of the systematic red shift of spectra with increasing magnitude). All
records were scaled to a common peak ground velocity of 5 cm/sec.
3.2 Statistics of response spectra. Normalised response spectra for all 26 horizontal
components of motion are illustrated in Figure 1. These spectra are for 5% of critical
damping, the conventional reference damping value. A few of the spectra have long-
period band limits less than 10 sec. Pseudo-relative velocity, plotted logarithmically in
Figure 1, is more nearly lognormally-distributed than normally-distributed for a given
oscillator period.
The statistics of the normalised response spectra are illustrated in Figure 2, which shows
the median and the 16th and 84th percentiles of the data. A noticeable aspect of the
statistics is that the standard deviation of the data is nearly uniform over the whole band.
The response, at any oscillator period, for this sample of velocity-normalised strong-
motion records can be predicted within about 60% (geometric s.d. of approximately 1.6).
3.3 Peak ground acceleration and displacement. The median peak ground acceleration
of the velocity-normalised records is 0.099 gravity, with a (geometric) standard deviation
of 1.54. For peak ground displacement, the result is 0.55 cm, with a standard deviation of
1.52. These results must be taken as being specific to an earthquake magnitude of 6.
4 RECOMMENDED SPECTRA FOR ROCK SITES
The shape of the median response spectrum lends itself to representation by the
trapezoidal-tent scheme[3]. This is illustrated in Figure 3, showing a straight-line
representation of the median spectrum in the manner of Newmark and Hall. The straight
lines have slopes of +1, 0 and -1 respectively in the acceleration, velocity, and
displacement bands of the response spectrum. These lines are parallel to the lines
representing peak ground acceleration, velocity, and displacement (dashed lines).
5 SITE RESPONSE AND 'S' FACTORS

The seismic response of surficial low-rigidity geologic strata is intrinsically frequency-
dependent, and so codification of the effect in terms of site amplification factors, S
factors, needs to be frequency-specific. With the Newmark-Hall trapezoidal-tent spectral
shape, it is convenient to divide the spectrum into three frequency bands (acceleration,
velocity, and displacement) and to determine S factors for the three bands. The S factors
in the current code (AS 1170.4-1993) are appropriate for the velocity band of the
spectrum, for frequencies between about 1 and 10 Hz (periods between 0.1 and 1 sec).
We propose S factors that are smaller at higher frequencies and larger at lower
frequencies.
A significant new empirical investigation of site response factors has been reported[4] in
the context of revision of the U.S. Uniform Building Code. Their site categories A, B, C,
and D have approximate correspondence with site factors of 1, 1.25, 1.5, and 2 in AS
1170.4-1993. These categories, in briefest terms, are rock, firm, soft, and very soft.
Crouse and McGuire obtained empirical site factors at oscillator periods of 0.3 and 1 sec,
and the proposed spectra are based on these results.
For the central (velocity) band, the S factors are 1, 1.3, 1.5, and 2 for site categories A, B,
C, and D, essentially the same as in the current code. For periods beyond 0.7 sec, the
amplification factors are 50% greater (S factors of 1, 1.95, 2.25, and 3 respectively),
while at high frequencies, beyond 10 Hz, the amplification is only half as much (S factors
of 1, 1.15, 1.25, and 1.5). The current code spectra are overly conservative at high and
low frequencies (acceleration and displacement bands of the spectrum). Proposed
dynamic amplification factors giving the ratio of spectral acceleration to peak
acceleration on rock are listed below for the four site categories.

6 DISCUSSION
The main assumption determining the shape of the proposed response spectra has to do
with earthquake magnitude. Specifically, the spectra are based on data for the magnitude
range 6 ± 0.6. Allowance for stronger long-period response should be made if higher
magnitudes are under consideration. The results herein are for elastic response spectra
and for 5% critical damping. Inelastic spectra, and results for other damping values, can
be readily derived[3]. The proposed response spectra can be anchored to any given peak
ground motion parameter, whether it be peak ground displacement, velocity, or
acceleration.
7 REFERENCES
1. Kanamori, H. and Allen, C.R., Earthquake repeat time and average stress drop,
Geophysical Monograph, AGU, 37, 1988, pp 227-235. 2. Boore, D.M., Joyner, W.B. and Fumal, T.E., A summary of recent results concerning the prediction of strong motions in western North America, Proceedings of the International Conference on Strong Motion Data, Menlo Park, California, 1993. 3. Newmark, N.M. and Hall, W.J., Earthquake spectra and design, Earthquake 4. Crouse, C.B. and McGuire J.W., Site response studies for purpose of revising NEHRP provisions, Earthquake Spectra, 12, pp 407-439.

Source: http://www.aees.org.au/wp-content/uploads/2013/11/19-Somerville.pdf