Incremental dynamic analysis (IDA) applied to seismic risk assessment of bridges - Seismic Risk Analysis and Management of Civil Infrastructure Systems - page 560

• Test data-related (TD) uncertainty. This type of uncertainty is related

to the quality of the test data used to defi ne the structural system.

• Modelling (MDL) uncertainty. This uncertainty is associated with the

quality and robustness of the nonlinear structural models and the ability
to accurately simulate the nonlinear behaviour of the structural ele-
ments subjected to seismic excitations.

The total uncertainty can be predicted by combining RTR, DR, TD and
MDL uncertainties in terms of lognormal standard deviation of β RTR , β DR ,

β TD , and β MDL using Eq. [21.1].

β = β

2

+ β

2

+ β

2

+ β

2

[21.1]

TOT RTR DR TD MDL

β RTR , can be obtained from the IDA results.
Where no detailed information is available, the other sources of uncertainty
may be estimated based on quality ratings. For example, in the ATC-63
provisions (ATC-63, 2008) the quality ratings of 'Superior', 'Good', 'Fair'
and 'Poor' are translated into quantitative values of uncertainty of 0.20,
0.30, 0.45 and 0.65, respectively. More details concerning the criteria to
determine the quality ratings for different sources of uncertainty are avail-
able in ATC-63 (2008). Other studies have also used similar values for
uncertainty. For example in a study by Mander et al. (2007), the uncertain-
ties due to modelling and predicting the capacity of the bridge columns
were considered as 0.25 and 0.2, respectively.
The reliability-based methods can be used for the detailed predictions of
the effects of uncertainty on different aspects such as structural modelling.
Such methods often include Monte Carlo simulation, the fi rst-order-second-
moment method (FOSM), the fi rst-order reliability method (FORM) and
the second-order reliability method (SORM). A summary of the methods
to estimate the modelling uncertainty is available by Liel et al. (2009). More
details regarding the incorporation of the uncertainty in probabilistic
seismic performance assessment and seismic risk analysis are available by
Cornell et al. (2002), Jalayer and Cornell (2003), Dhakal and Mander (2006)
and Ellingwood and Kinali (2009).

Record-to-record variability,

21.5 Record selection for incremental dynamic
analysis (IDA)
21.5.1 Spectral shapes of the records and epsilon values

The spectral shape of the records is especially important for collapse assess-
ment which can change the calculated collapse capacity of the structures
by 70% (Baker and Cornell, 2006a; Haselton and Baker 2006; Zareian,
2006). Therefore, neglecting the infl uence of the spectral shapes in collapse