Page 337 - Pipeline Risk Management Manual Ideas, Techniques, and Resources

14/314 Absolute Risk Estimates
0 More complete data are available (larger historical failure database severity of consequences (solely from a public safety perspective)
and data are better characterized). associated with a pipeline’s failure depends on the extent of the prod-
0 Strong influence by a major US. operator on design, operations, uct release, thermal effects from potential ignition of the released
and maintenance. product, and the nature of any damage receptors within the affected
0 Similar regulatory codes, pipeline environments, and failure expe- area. The area affected is primarily a function of the pipeline’s diame-
riences. ter, pressure, and weather conditions at the time of the event.
0 Apparently similar failure experience between the countries. Secondary considerations include characteristics of the area includ-
ing topography, terrain, vegetation, and structures.
Since the combined experience of all US. pipelines cannot realisti-
cally represent this pipeline’s future performance (it may “encom- Failure discussion
pass” this pipeline, hut not represent it), a suitable comparison
subset of the data is desired. Variables that tend to influence failure The potential consequences from a pipeline release will depend on the
rates and hence are candidates for criteria by which to divide the failure mode (e.g., leak versus rupture), discharge configuration (e.g.,
data, include time period, location, age, diameter, stress level, wall vertical versus inclined jet, obstructed versus unobstructed), and the
thickness, product type, depth ofcover, etc. Unfortunately, no data- time to ignite (e.g., immediate versus delayed). For natural gas
base can be found that is complete enough to allow such characteri- pipelines, the possibility of a significant flash fire or vapor cloud
zation of a subset. Therefore, it is reasonable to supplement the explosion resulting from delayed remote ignition is extremely low due
statistical data with adjustment factors to account for the more sig- to the buoyant nature of gas, which prevents the formation of apersist-
nificant differences between the subject pipeline and the population ent flammable vapor cloud near common ignition sources.
of pipelines from which the statistics arise. Rationale supporting the ACME applied a “Model of Sizing High Consequence Areas
adjustment factors is as follows: (HCAs) Associated with Natural Gas Pipelines” [83] to determine the
potential worst case ACME Pipeline failure impacts on surrounding
0 Larger diameter is 40% of failures in the complete database people and property. The Gas Research Institute (GRI) funded the
(90+% benefit from higher diameter is implied by the database but development of this model for U.S. gas transmission lines in 2000, in
only 25% reduction in failures is assumed) association with the US. Office of Pipeline Safety (OPS), to help
Lower stress decreases failure rate by 10% (assumption based on define and size HCAs as part of new integrity management regula-
the role of stress in many failure mechanisms) tions. This model uses a conservative and simple equation that calcu-
New coating decreases failure rate by 5% (assumption note the lates the size of the affected worst case failure release area based on
well-documented problem with PE tape coatings in Canada) the pipeline’s diameter and operating pressure.
New IMP (integrity management program) procedures decreases
failure rate 10% (assumption based on judgment of ability for IMP Failure scenarios
to interrupt incident event sequence)
Deeper cover (2 fi of additional depth is estimated to be worth 30% There are an infinite number of possible failure scenarios encompass-
reduction in third-party damages according to one European study ing all possible combinations of failure parameters. For evaluation
so a 10% reduction in overall failures is assumed) purposes, nine different scenarios are examined involving permuta-
More challenging offshore environment leads to 10% increase in tions of three failure (hole) sizes and three possible pressures at the
failures (somewhat arbitrary assumption, conservative since there time of failure. These are used to represent the complete range of pos-
are no known unresolved offshore design issues). sibilities so that all probabilities sum to 100%. Probabilities of each
bole size and pressure are assigned, as are probabilities for ignition in
Combining these factors leads to the use of a -50% reduction from the each case. For each of the nine cases, four possible damage ranges
average US. gas transmission failure rate. This is conservative- (resulting from thermal effects) are calculated. Parameters used in the
accepting a bias on the side of over-predicting the failure frequency. nine failure scenarios are shown in Table 14.38.
Additional conservatism comes from the omission of other factors that
logically would suggest laver failure frequencies. Such factors include
Table 14.38 Parameters for the nine failure scenarios under
Initial failure frequency is derived from pipelines that are pre- discussion
dominantly pre- 1970 construction-there are more stringent
practices in current pipe and coating manufacture and pipeline
construction Probability of
Better one-call (more often mandated, better publicized, in more occurrence (99) Comments
common use)
Better continuing public education Hole sue (in.)
Designed and mostly operated to Class 3 requirements where Class 50% to full-bore 20 Possible result of third-party
3 pipelines have lower failure rates compared to other classes from rupture(8-16) damage or land movement
which baseline failure rates have been derived 0.5-8 40
Leaks versus ruptures (leaks less damaging, but counted if report- <0.5 40 Corrosion or material defect
ing criteria are triggered) related
Company employee fatalities are included in frequency data, even Pressure (psig) 20 The contract delivery pressure
though general public fatalitieshjuries are being estimated 1800-2220 is 1800 psig; current
Knowledge that frequency data do not represent the event of (2220 psig is used) connection pressures
“one or more fatalities,” even though that is the event being normally are -800 psig;
estimated. > 1800 psig would not be
normal.
150&1800 70
Model-based failure consequence estimates (1 800 psig is used)
4500 IO
An analysis of consequence, beyond the use of the historical (1 500 psig is used)
fatalityhnjury rate described above, has also been undertaken. The