EP 1130-2-500
27 Dec 96
of occurrence or analytical procedures. Expert Elicitation should only be used in consultation
with CECW-E.
(c) Calibration of Reliability Models. Performance function models used to evaluate the
project component or feature reliability should be calibrated by applying the model to at least
two similar components whose performance is known. Reliability should be calculated for a
similar component known to have suffered distress, and for another similar component known to
meet current design criteria. If the performance function model does not accurately predict the
known structural performance levels, the assumptions, conditions, simplifications and parameters
used in the model should be reexamined and adjusted to realistically provide an accurate
prediction.
(d) Time Dependent Reliability. The reliability of a component or feature varies with
time due to many factors including environmental conditions, component stress history,
corrosive resistance of the materials, as well as maintenance history. Therefore, a time-
dependent reliability analysis must be conducted in order to consider the impact of these factors
on project performance and service life. Projections of future changes in reliability should be
based upon the calculation of performance functions using the procedures outlined in Appendix
D. Rates of degradation in random variable properties should be based upon available existing
data, industry practice and experience at similar projects. If available project data is scarce or
non-existent, then estimating rates of degradation will require that considerable engineering
judgement be exercised in consultation with CECW-ED.
(e) Engineering Characterization of Structural Features. The complex nature, time and
cost of reliability analyses require that the number of elements analyzed for any project or feature
be reduced to the critical elements, or to representative groups or sections. In some cases this can
be done by grouping together elements or components which are similar and can be represented
by a single element, or a small portion of a large element, i.e., the reliability of an entire length of
lock wall might be represented by a typical one foot section of the wall. In other cases, it may be
possible to group related elements together and represent the group by a single critical element.
The reliability of the critical element (and its associated probability of unsatisfactory
performance) would then be assumed to govern the reliability of the entire group of elements and
they would then be considered as one element or component in the economic risk analysis. An
example of this is in a steel miter gate or a steel truss which is dependent on the satisfactory
performance of all members to resist loads. The unsatisfactory performance of one or more
critical members would lead to the unsatisfactory performance of the entire structure.
(2) Engineering Consequences. The engineering, or physical, consequences of the
expected level of performance should be described in detail for each performance function
evaluated. The sequence of events caused by the unsatisfactory performance of a component
should be reasonable, with consideration given to the importance of the component to the overall
performance of the structure or feature. While "worst case" unsatisfactory performance scenarios
need to be evaluated and described, lesser events should also be included since these higher
probability events may have greater impact upon the service life of the structure and the
economics of the project. If the unsatisfactory performance of components or features is
expected to result in emergency repairs, provide an assessment of the impact of the repairs upon
both the reliability of the repaired elements and those elements not included in the repairs.
(3) Engineering Evaluation of Alternatives. Alternative schemes for repair and
rehabilitation must be fully investigated. Alternatives investigated should include the use of new
B-7
Previous Page
