Page 450 - Marine Structural Design

426 Part IY Structural Reliability

Assuming that the component is subject to proof load, q* and that it survives the load. This
implies that the strength r2q'.
The updating of Pfcan be formed as follows
Pf,"p = PIR - s I OIR 2 4'1
= p[R-S I O]H 2 01 (23.34)
- P[R -S 5 On H 2 01 - P[R - S I On-H 01
-
-
P[H 2 01 P[- H I 01
where, H=R-q*
In general, different methods are available to update the structural reliability based on the new
information. Song and Moan (1998), Moan and Song (1998) presented the methods of
reliability updating for ships and jackets, which will be detailed in Part IV Chapter 27.

23.9 Target Probability

23.9.1 General
Guidelines are provided for structural designers on acceptable failure probability associated
with each failure mode, i.e. minimum acceptable reliability index PO, frequently referred to as
target probability. When carrying out structural reliability analysis, an appropriate safety level
should be selected based on factors like consequence of failure, relevant design codes,
accessibility to inspection and repair, etc. Target probability levels have to be met in design in
order to ensure that certain safety levels are achieved.
23.9.2 Target Probability
A design is safe if
P'PO (23.35)
where PO = target safety index
P = safety index as estimated from analyses
The regulatory bodies or classification societies andor professions agree upon a reasonable
value. This may be used for novel structures where there is no prior history.
Code calibration is to calibrate reliability levels that are implied in currently used codes. The
level of risk is estimated for each provision of a successful code. Safety margins are adjusted
to eliminate inconsistencies in the requirements. This method has been commonly used for
rules development.
Target probabilities are chosen to minimize total expected costs over the service life of the
structure. A cost-benefit analysis approach may be used effectively to define target probability
for design in which failures result in only economic losses and consequences. Although this
method is logical on an economic basis, a major shortcoming is its need to measure the value
of human live.
The target probabilities, for a reliability based design, are based calibrated values of implied
safety levels in the currently used design practice, as shown, e.g. by Bai et a1 (1997). The
argument behind this approach is that a code represents a documentation of an accepted