Environmental Engineering Reference
In-Depth Information
of the experts being used to assess performance
and therefore lacks repeatability.
.
Scientific models
- These are generally based
on analysis and synthesis of physical attributes of
the asset backed up by experiment. They enable a
calculation of performance to be made based on
equations comparing loading with resistance. A
scientificmodel, by definition, must be repeatable
and thereforemay provide amore reliablemeasure
of performance than a model based on limited
historical data or expert opinion. Problems with
the use of scientific models of performance are
often in obtaining accurate measurements of the
model parameters for an asset.
Drawing upon current research into the perfor-
mance modelling of defence assets (e.g. HR Wall-
ingford 2003), a set of relevant failure modes and
associated performance models was identified for
use with the new methodology being proposed.
The proposed method is described below and is
referred to as the
condition indexingprocess
. The
description starts with the key elements that form
the basis for the method.
method that whilst eliminating the limitations of
the current method also removes the benefits
provided by it.
A Performance-Based Methodology
for Visual Inspection
Some basic understanding of failure modes and
performance modelling of defence assets is neces-
sary to describe the method of visual inspection
being proposed. Defence assets can fail in several
ways. There are a number ofwell-known processes
for failure, some of which have been modelled
using physical or statistical models. These perfor-
mance models can be used to determine the like-
lihood of a specific failure mode occurring given a
set of data relating to the flood defence system
being analysed.
There are three methods for devising perfor-
mance models for an asset:
.
Historical data
- A statistical approach can be
used to analyse historical records, where available,
of asset failure. By comparing hydraulic loading
conditions and any failures that occur, rough
measures of likely performance can be produced.
The advantage of this approach is that it does
not require any detailed understanding of
the physical processes involved in asset perfor-
mance. However, it is highly dependent upon
the quality and amount of historical data available
for an asset.
.
Expert opinion
- This could be defined as being
ill-recorded historical data. Experts familiar with
the asset and local conditions could provide sim-
ple heuristics defining likely performance versus
hydraulic loading. Expert opinion iswidely used in
early definitions of fragility curves for asset types
under the Risk Assessment of flood and coastal
defence for Strategic Planning (RASP) project (HR
Wallingford & University of Bristol, 2004) and
the condition indexing system employed by the
United States Army Corps of Engineers (USACE)
(McKay et al. 1999). This method enables perfor-
mance to be assessed where there are few or no
recorded data. It also takes account of local con-
ditions. It is highly dependent upon the knowledge
Performance features (PF)
Performance features are the building blocks on
which the condition indexing process is based.
They represent the elements of an asset that are
to be inspected in order to produce the condition
index. Performance features must possess a num-
ber of attributes. These are listed below in order of
importance:
1
They must be related to at least one failure
mode, and therefore to the performance of the
asset - For a performance-based visual inspection,
the primary objective is to inspect visual indica-
tors of likely asset performance. If the visual con-
dition of a feature plays no role in asset
performance, it should not be inspected.
2
Visible - The feature to be inspected must be
easily assessed on a visual inspection. For exam-
ple, many of the performance models have geo-
technical parameters that cannot be observed on a
visual inspection. Their assessment requires there
being a visible feature on the asset surface that
directly or indirectly relates to their current value.
