Performing prior, posterior and pre-posterior decision analysis.
Setting acceptable levels for the probability of failure.
The two first points have already been addressed in previous lectures and will not be
considered here. The next two points, however, are essential for the understanding of the
framework of structural reassessment and will be described in some detail. The issue of
setting acceptable failure probabilities is central both for reliability based design and
reliability based assessment of structures. This issue is considered in more detail in a later
lecture.
12.4 Reliability Updating in Assessment of Structures
When assessing existing structures various types of information may be available. Examples
of information, which is available or can be made available at a given cost, are:
The structure has survived.
Material characteristics from different sources.
Geometry.
Damages and deterioration.
Capacity by proof loading.
Static and dynamic response to controlled loading.
In the assessment of existing structures such new information can be taken into account and
combined with the prior probabilistic models by reliability updating techniques. The result is
so-called posterior probabilistic models, which may be used as an enhanced basis for the
reassessment engineering decision analysis.
The following presents some general principles and formulations, which are useful in the
assessment of existing structures. The technical implementation is considered in Lecture 6
together with some of the available software tools. The benefit of the application of the
principles and formulations in the different situations encountered in practice is very much a
matter of the experience and creativity of the engineer. However, a sample of different
applications will be illustrated on simple examples in later sections.
When discussing updating techniques for structural reliability two types of quantitative
information should be distinguished:
information of the equality type and
information of the inequality type.
When information of the equality type is present, it means that for some basic or response
random variables the value has been measured. Examples are: the stress equals 200 MPa, the
crack length is 3.2 mm. Of course, these equality measurements are seldom perfect and may
suffer from some kind of measurement error. In a probabilistic evaluation procedure,