Page 268 - Pipelines and Risers
P. 268
Remaining Strength of Corroded Pipes 24 1
Calibration of Partial Safety Factors
0 Perform safety verification, formulated as a design equation utilizing the characteristic
values and partial safety factors.
The Load Resistance Factor Design (LRFD) method provides engineers with rational tools for
achieving consistent levels of safety in the design of structural components. A partial safety
approach is:
(14.29)
where, yli are load factors by which the characteristic loads Q.i are multiplied to obtain the
design loads, cp is a resistance factor by which the characteristics strength R. are multiplied to
obtain the design resistance.
The load factor, yli, and resistance factor, cp, serve the same purpose to account for the
uncertainties in the determination of the strength and load effects. Their values are to be
calibrated so that the implied safety level of a structure has a failure probability which is close
to a target failure probability.
14.5.1 Target Failure Probability
The target failure probability is developed based on the historical failure data and the safety
level implied in the existing B31G criteria. The target safety level should be determined
considering the consequence of failure as well as the effects of inspection, maintenance, and
repair. The safety level to be applied in the new criteria should be the same level as the safety
level in the existing B31G criteria. Based on the historical data, reliability analysis of the
existing B31G criteria, and other factors, an annual target safety level of IO4 is used in the
development of the reliability-based criteria.
14.5.2 Design Equation and Limit State Function
For the sake of simplicity, only internal pressure is considered in the design equation. The
LRFD approach leads to:
PR ?TpL (14.30)
where, PR is the characteristic strength of the pipe based on a criterion, P, is the
characteristic load (internal pressure), y = YL is referred to as the partial safety factor.
(PU
A bias factor X is introduced to reflect the confidence in the criterion in prediction of burst
strength:
true burst strength
X= (14.31)
predicted burst strength
Normalized random variables in the design equation are:
