Page 334 - Pipeline Pigging Technology

Assessment of pipeline defects

14, it is possible to calculate safety factors that, when applied to Eqn(l), will
give safe (95% confidence level*) predictions. Ref.3 suggests that a safety
factor of 0.97 should be applied to Eqn(l) and recommends the use of SMYS
and maximum defect depth.

SAFETY FACTORS ON FAILURE PRESSURES

The end product of a fitness-for-purpose calculation is a failure pressure for
a defect. Factors should then be applied to the failure pressure to accommo-
date uncertainties in the fitness-for-purpose analysis and also in the operation
of the pipeline (e.g. surges). A safety-factor philosophy directly related to
code requirements can be proposed[3]. Summarizing:

maximum operating pressure, Po = SM x SF x Pf (3)

where Pf = predicted failure pressure of corrosion (Eqn(l));
SM = safety margin related to pipeline codes; and
SF = safety factor to accommodate errors in failure criteria.

A value of SF = 0.97 is recommended to give a 95% confidence level on
failure predictions.
SM is obtained by considering the design and hydrotest pressures specified
in pipeline codes. Most codes, e.g. IP6[15], have a maximum design pressure
of 72% SMYS and a hydrotest pressure in excess of 90% SMYS. If we assume
that a defect-free pipeline will fail when the hoop stress reaches flow stress
C 1.15 x SMYS)[2], we obtain the following safety margins (Fig. 11):

hydrotest** safety margin = 0.72/0.90 = 0.8
defect-free pipeline safety margin = 0.72/1.15 = 0.63

Thus a new IP6 pipeline will have a safety margin between 0.8 (guaranteed
by the hydrotest) and 0.63. This latter defect-free safety margin is optimistic
because an operational pipeline, with its fittings, bends, etc., cannot be
expected to withstand a stress of 115% SMYS.
* The use of a 95% confidence level (mean minus 2 standard deviations) in failure calculations bos
been accepted as good practice for many years, with its adoption in BSI PD6493[10], the major defect
assessment code. The design curve (in effect the 'fracture' curve) in BSIPD6493 is a 95% lower confidence
level on a large full-scale test data base.
** Care should be taken in calculating these margins, as hydrotest and operating stress levels can be
based on minimum or nominal wall thickness.
315