Geoscience Reference
In-Depth Information
5. Spectral accelerations for vulnerability assessments
A parametric study was conducted to observe the effect of averaging and adopting a
probabilisticapproach (i.e.10%exceedance probability)byassumingthatthecalculated
elastic acceleration spectra can be modelled by normal distribution for all the period
rangeconsidered.AnexampleofthebestfitNEHRPenvelopeobtainedbythisapproach
with respect to elastic acceleration response spectra with 10% exceedance probability
is shown in Figure 4.11 in comparison with the average and all acceleration response
spectracalculatedfortheselected(20)PGAscaledinputaccelerationtimehistories.The
NEHRP design spectrum is preferred because of its flexibility in defining short period
spectral accelerations and for vulnerability assessment of the building stock (Erdik and
Fahjan, 2005).
However, the question of conservatism again becomes important as can be observed in
Figure 4.12. One justification for using the probabilistic interpretation of the calculated
elastic acceleration response spectra from all site response analyses using larger number
of real input acceleration records is the capability of defining a hazard level in accor-
dance with the purpose of the microzonation. The criteria adopted in this scheme was to
use the same hazard level (10% exceedance probability) that was used to calculate the
earthquake hazard on the rock outcrop used for site response analyses. The approach is
stillnotcompletelyprobabilisticsincethevariabilityofthesoilprofilewasnottakeninto
account inaprobabilistic manner.
Theaverageaccelerationresponsespectraobtainedforeachcellfromsiteresponseanaly-
seswereevaluatedfordeterminingthespectralaccelerationsfortheshortperiod(Ss)cor-
responding to 0.2s and for the long period (S1) corresponding to 1s. An approach was
adoptedtodeterminethebestfitenvelopetothecalculatedaverageaccelerationresponse
spectra (Ansal et al., 2005b). All the requirements of the NEHRP design spectra were
appliedinobtainingtheshort(Ss)andlong(S1)periodspectralordinates.Thetwoinde-
pendent variables inthe developed optimization algorithm wereSs and S1.
Fig. 4.11. Typical bestenvelope NEHRPspectrafittedto10% probabilityofexceedance
elasticaccelerationresponsespectra,incomparisonwiththeaverageandallacceleration
response spectra calculated by siteresponse analysis
