Environmental Engineering Reference
In-Depth Information
Table 15.2
A structured procedure for the assessment of asset fragility (adapted from Simm et al. 2008)
Step
Description
1. De
ne asset function
A
ooding; often functioning as a valuable environmental
habitat, navigation or amenity asset. Understanding the multi-functionality of the asset is an important
precursor to understanding how to manage it
ood defence asset rarely acts solely to protect from
2. Establish incident
loading
An asset may be subject to a range of loading conditions
joint wave and water levels, marginal high or low water
-
levels, groundwater levels or perhaps a combination
3. Identify failure modes
The failure mechanisms (processes that can lead to ultimate failure) and the failure modes (a process that
de
nes ultimate failure) also need to be described. To avoid unnecessary effort, conventional deterministic
approaches can be helpful to eliminate unrealistic failure mechanisms (i.e. relative low-probability individual
events in comparison with the likely overall reliability of the asset)
Research into failure mechanisms continues to be vital to better understand asset performance (e.g. Allsop
et al. 2007; Dyer et al. 2009; Sentenac et al. 2009)
4. Prepare a fault tree
Fault trees provide a useful visual, and formal, encapsulation of the failuremechanisms and their relationship to the
failure modes
5. Identify/establish
appropriate Limit
State Equations
An appropriate model needs to be selected to represent each failure mechanism\mode. In many cases empirical
relationships will exist and these can be easily translated into the form of a Limit State Equation (utilized in the
reliability analysis
-
see below). In some cases, the failure mechanisms are complex (e.g. slip failure) and demand
the use of more sophisticated models (e.g. traditional slope stability analysis or
nite element model). It is
possible to link such models within the reliability analysis (Lassing et al. 2003; Vrouwenvelder 2001a, 2001b)
but this is often dif
cult and can incur an unacceptable runtime overhead. Emulation of these more complex
models, through Arti
cial Neural Networks, for example, provides an ef
cient and effective means to enable
such complete mechanisms to be incorporated into the reliability analysis (Kingston and Gouldby 2007)
6. Document
uncertainty in model
variables and
parameters
The engineering parameters, and the empirical variables, within the Limit State Equations will not be perfectly
understood. Describing the uncertainty within these relationships and the supporting data on the asset of
interest is an important task. In describing the uncertainty it is important that this process is comprehensive
(ignoring uncertainty at this stage is to assume the data are perfectly known). Two groups of uncertainties can
typically be distinguished (USACE 1999; Environment Agency 2002):
.
Natural variability (aleatory uncertainty)
Uncertainties that stem from known (or observable) populations
and therefore represent randomness in samples
-
(Continued)
