Geoscience Reference
In-Depth Information
3. Hazard maps and seismic risk assessment of underground pipeline systems
Seismic risk assessment for underground structures has generally been performed based
on
PGV
hazardmaps,sincethecorrespondingvulnerabilityfunctionsareusuallydefined
in terms of
PGV
as well (NIBS, 2004). However, the large scatter of such relation-
ships and the close link between earthquake ground deformations and seismic loads
in underground structures has suggested replacing
PGV
with
PGS
as the parameter to
quantify ground motion severity. In this perspective, O'Rourke and Deyoe (2004) have
recently re-examined the available data on damage to pipeline systems and deduced a
new relationship between repair rate on the pipeline network and
PGS
. The drawback
of this approach is that its application for seismic risk assessment requires production of
hazard maps in terms of
PGS
. In the absence of well established approaches to evaluate
transient ground strains, or of sound formulas relating
PGS
to other peak parameters of
ground motion,
PGS
-based hazard maps are typically obtained in two ways, the alterna-
tive approaches involving either a large set of 2D numerical simulations along various
azimuths or a fully3D one being presently toodemanding:
a) byconstructinga
PGV
hazardmap(eitherbyprobabilistichazardstudiesorinterms
ofascenarioearthquake)andtransposingitinto
PGS
(eitheraxialorshear,seecom-
mentsinthesequel)throughastraightforwarddivisionof
PGV
byasuitablemeasure
of theapparent wave propagation velocity;
b) byperformingalargesetof1Dwavepropagationanalyseswiththebestinformation
available on the local ground properties and a selection of the input motion compat-
ible with the target hazard; these analyses may lead either to a map with the spatial
distributionof
PGV
,whence
PGS
canbededucedasatpoint(a),ortheycanprovide
the shear strainat a selected depth (theshear strainat surface being vanishing).
While the main drawback of approach (a) has been extensively discussed in the first part
of this paper, approach (b) involving a direct calculation of
PGS
through 1D ground
motion simulations, suffers of twomajor limitations.
Firstly,groundstrainsobtainedby1DanalysesofS-wavepropagationarepurelyofshear
nature,witharelativelysharpvariationwithdepth,andtheycannotbetranslatedstraight-
forwardly in terms of longitudinal strains. While shear strains
(γ )
are mostly interesting
forseismicdesignoftunnelsinatransversalcross-section,seismicdesignofpipelinesis
mainly governed by longitudinal strains
(ε)
. As shown in Figure 18.10, referring to the
resultsof2DnumericalsimulationsofSVwavepropagationalongtwogeologicalcross-
sectionsinThessalonikiforaM6.5earthquake,therelationshipbetweenpeakgroundval-
uesof
(
PGSa
)isveryscattered:ataround3mdepth,wheremostpipelines
are embedded, the ratio between
PGSs
and
PGSa
ranges between about 1 and 4, with a
mean value around 1.75. At larger depths, as shown in Figure 18.10 for a representative
depthof15m,thisratioisevenhigher,asexpectedforincidenceofSVwaves.Although
furtherstudiesarerecommendedtoassessageneralrelationshipbetween
PGSs
and
PGSa
as a function of depth, earthquake magnitude and site characteristics, the results shown
in Figure 18.10 support the use of calculated shear strains for both axial and transversal
γ
(
PGSs
)and
ε