Geoscience Reference
In-Depth Information
6. Conclusions
Parametric studies were conducted to evaluate the significance of the induced variability
arisingfromdifferenttypesandsetsofinputaccelerationrecords,differencesinsitechar-
acterisation, and due to the differences in the interpretation schemes adopted for micro-
zonation with respect to different parameters.
The results obtained support the use of previously recorded real acceleration time histo-
ries as input motion for site response analysis. Among the scaling options studied, PGA
scaling, in addition being on the conservative side, yielded more realistic ground motion
characteristics.Thefaultdistanceoftheaccelerationrecordsusedinsiteresponseanaly-
sis is one of the parameters that affect the outcome both with respect to peak ground or
spectral accelerations, thus acceleration records need to be selected compatible with the
regional hazard intermsof fault type, magnitude, and fault distance.
The Zeytinburnu case study was used to compare microzonation maps with respect
to PGA and spectral accelerations using real and simulated sets of input acceleration
records. For each set, microzonation maps were different even though the differences
may not be very significant. These results indicate the importance of the input motion
selectionscheme.Onepossibilityistouselarger setsofinput motionandadopt aproba-
bilisticprocedure assuming normally distributed PGA and spectral accelerations.
An interpretation scheme based on probabilistic evaluation of site response results with
respect to set of input acceleration records and site conditions would reflect the uncer-
tainty based on the selection of exceedance levels in accordance with the adopted ini-
tial probabilistic earthquake hazard level. Thus it may be more suitable for probabilistic
microzonation studies.
Site response analyses and two empirical formulation suggested in the literature were
used to calculate stress reduction factor for estimating liquefaction susceptibility with
depth. There were significant differences among the results obtained by the three
approaches. It was observed that in general the variation of CSR with depth, calcu-
latedbysiteresponseanalysisishighercomparedtoCSRcalculatedusingthesimplified
formulation for thestressreduction factor.
For the assessment of the vulnerability of the building stock, the average acceleration
response spectra obtained for each cell from site response analyses were evaluated for
determining the spectral accelerations for the short period corresponding to 0.2s (Ss)
and for the long period corresponding to 1s (S1). An approach was adopted to deter-
mine the best fitting envelope to the calculated average acceleration response spectra.
The results obtained are compared in terms of short period (0.2s) and long period (1s)
spectral accelerations obtained by NEHRP. The spectral accelerations determined by fit-
ting NEHRP spectrum to response spectrum obtained by site response analyses vary in
muchbroaderrange.InmostcasesspectralaccelerationsobtainedbytheNEHRPsimpli-
fiedprocedurewouldyieldvaluesontheunsafesidecomparedtothosedeterminedbythe
site response analyses. Thus it would be more reliable to perform site response analyses
