Seismic vulnerability analysis of a complex interconnected civil infrastructure - Seismic Risk Analysis and Management of Civil Infrastructure Systems - page 494

centroids, PGV is used as a secondary IM as an input to pipes' fragility
model. Both simulation of PGA and the conditional simulation of PGV are
carried out using the GMPE by Akkar and Bommer (2010), recently devel-
oped for the European context. Spatial correlation with correlation param-
eter (range) is estimated based on European records by Esposito et al.
(2010). For within-site cross-IM correlation 0.754 is used. No site

amplifi cation, nor geotechnical hazard, is considered for this simplifi ed
example. The rectangular grid for the prediction of the primary IM consists
of 71

ρ PGA , PGV =

× 61 = 4331 points.

18.8.2 Results for a single scenario event

The impact of seismic events on the considered region is assessed through
a simulation carried out with the importance sampling method by Jayaram
and Baker (2010), without performing, however, the K-means clustering. A
total number of 1000 runs has been carried out. This section describes
results of one of these runs, i.e. the impact of a single 'scenario event', a
magnitude 7 earthquake on source 2 (for this magnitude this is the only
source that can generate the event, as shown in Fig. 18.14), quite close to
city B. The corresponding shake-fi eld has already been shown in Plate I
(between pages 452 and 453) and Fig. 18.7.
Physical damage caused by the event is illustrated in Fig. 18.12. Plot (a)
shows damage to the EPN. In general this system tends to be 'stiffer' than
the WSS, with damage states mostly resulting in either a functional or not
functional network. This is particularly true for this very small system
lacking much redundancy. The shown scenario, however, is a relatively rare
one in which damage occurred in a sub-station (in city B, the one closest to
the epicentre) but circuit-breaker prevented its propagation within the
station and to other stations, resulting only in a single line being off-line.
Under these conditions the system of equations in [18.2] could still be
solved for an equilibrium condition, though with lower voltages than
needed in many stations. Plot (b) shows damage to the WSS. In this sce-
nario there is no pipe rupture but leaks occur at more locations, in cities B
and A.
Plots (c) and (d) show damage to URM and RC buildings, respectively.
In the maps, a shade of increasingly darker grey, which is proportional to
the damage state, is assigned to cells. White cells are outside urban areas
and have no buildings. Damage is mostly limited to city B, with the larger
number of collapsed buildings in the URM typology, as expected, though
light damage (exceedance of the yield limit state) is widespread among RC
buildings.
As observed in Cavalieri et al. (2012), where more details are given on
the way large groups of buildings in urban areas are lumped into cells within