Seismic risk analysis using Bayesian belief networks - Seismic Risk Analysis and Management of Civil Infrastructure Systems - page 192

Building damage

Site seismic hazard Building vulnerability

S a ( T 1 ) Liquefaction Lateral

spreading

Increase in
demand

Decrease in
resistance

Plan
irregularity

Vertical
irregularity

Year of
construction

Deterioration/
degradation

7.12 Hierarchical building damage assessment.

corresponding design response spectrum. Finally, using T 1 and correspond-
ing response spectra, spectral acceleration S a ( T 1 ) is obtained. As part of the
site seismic hazard, Fig. 7.12 also shows liquefaction and lateral spreading
and can be incorporated as shown in the previous section.
Building vulnerability to ground shaking and associated damage can be
grouped into two categories; factors contributing to an increase in seismic
demand (e.g. plan irregularity, vertical irregularities); and factors contribut-
ing to reduction in ductility and energy absorption capacity (e.g. construc-
tion quality, year of construction, structural degradation). In a preliminary
assessment of buildings, obtaining and incorporating exhaustive details of
those factors are not feasible. In this chapter, building vulnerability param-
eters considered are (i) building type, (ii) vertical irregularity (VI), (iii) plan
irregularity (PI), and (iv) year of construction (YC). In addition, reduction
in member capacity can occur to structures that were damaged from previ-
ous earthquakes and/or deteriorated (e.g; corrosion) (DD). PI and VI are
aggregated to compute increase in demand. Furthermore, YC and DD are
aggregated to compute decrease in capacity. Finally, the building type,
increase in demand and decrease in capacity are aggregated to quantify
building vulnerability. It should be noted that this framework is generic and
it can be applied to other regions; for example, Tesfamariam and Saatcioglu
(2008) applied it to US buildings, Tesfamariam and Saatcioglu (2010)
applied it to Turkish buildings, and Tesfamariam and Wang (2012) applied
it to schools and civil buildings in the state of Oregon, US.
The states considered for each input parameter, PI, VI, YC, DD and SSH,
are shown in Table 7.7. The intermediate nodes, are computed using CPTs
that are generated through expert knowledge and training from past earth-
quake damage. The discrete states considered for the intermediate nodes
are: