Geoscience Reference
In-Depth Information
be seen that at the 500 year return period (or annual exceedence probability of 0.002)
thelossis$1.75Billion.Thelosslevelincreasesto$2.25Billionforthe1000yearreturn
periodand$2.75Billionatthe2000yearreturnperiodlevel.Itisapparentfromthiscurve
and the values reported in Figures 19.2 and 19.4 that consideration of structural damage
alone willgreatly underestimate the systemrisk.
3.6. INFLUENCE OF GROUND MOTIONANDDAMAGE CORRELATION
ON LOSS COMPUTATIONS
Ground motion and damage correlation will have a direct influence on the estimation
of loss uncertainty. In order to determine the importance of these correlations two small
networks within the study area were selected and the loss coefficients of variation for
directstructuraldamagewereestimated.Oneofthenetworkshas16bridgesandthesec-
ond consists of 9 bridges. Three levels of correlation were considered for ground motion
that include perfect correlation, partial and no correlation (i.e.
= 1, 0.5 and 0). Dam-
age correlation is conditional on ground motion, thus the effect of damage correlation
is computed in relation to the ground motion correlation. Figure 19.6 shows the results
of these analyses. From that figure it can be seen that the uncertainty in loss estimates
increasessignificantlywithincreaseinbothgroundmotionanddamagecorrelation.Fur-
thermore, the contribution of ground motion correlation and damage correlation to the
overalluncertaintyappeartohaveequalweights.Thesizeofthenetworkdoesnotappear
to have an influence on the conclusions with similar results obtained for the 16 and nine
component networks.
ρ
1.6
1.6
1.4
1.4
1.2
1.2
1
1
0.8
0.8
0.6
0.6
0.4
0.4
Independent
Partial
Perfect
Independent
Partial
Perfect
0.2
0.2
0
0
10
−
1
100
10
−
1
100
Damage Correlation
Damage Correlation
Fig. 19.6. Loss coefficient of variation for twonetworks,
N
= 16(a) and
N
=9(b).
Lowest marker points (
) result fromindependent ground motions. Highest marker
points (
) result fromperfect ground motion correlation
