Page 358 - Pipeline Risk Management Manual Ideas, Techniques, and Resources

Cumulative risk 15/333
111. Measurement tool not to think we can find a 2% change in risk with a tool that is
only sensitive to *IO% changes. This is similar to the “accu-
The risk assessment model described in Chapters 3 through 7 racy” of the model, but involves additional considerations that
produces relative risk scores by combining the possible failure surround the high level of uncertainty associated with risk
mode scores and dividing this sum by the potential conse- management. However, it would not be reasonable to assume
quences score: that this tool cannot be continuously improved. Improvement
opportunities should be constantly sought.
Relative risk = [index sum]/[le& impact factor] See also Chapter 8 for a some simple statistical and graphi-
Index sum = [third party] + [corrosion] +[design] + [incorrect cal tools that can be used to further explore a risk model’s
operations] capabilities.
Having built a formal risk assessment system, it is useful to step
back and assess what information is now available. Recall that IV. Cumulative risk
the final risk numbers should be meaningful in a practical, real-
world sense. They should represent everything that is known Cumulative risk is a metric used to gauge the risk posed by any
about a specific piece of pipe-the collective intelligence of the length of pipeline. Because risk values are very location spe-
whole company including all knowledge gained over years of cific along the pipeline, a method of rolling up all of the risks
operating experience, all of the statistical data that can be gath- for a certain stretch ofpipeline is important.
ered, all intuitive beliefs, and all engineering calculations. Ifthe As noted in Chapter 2, the pipeline risk scores represent the
model has not captured all of this, then there is room for relative level of risk that each point along the pipeline presents
improvement. If any personnel are more knowledgeable in any to its surroundings. It is insensitive to length. If two pipeline
risk area than the model is, there is still work to be done. If the segments, 100 and 2600 ft, respectively, have the same risk
risk assessment results are not believable, then something score, then each point along the 100-ft segment presents the
is either wrong with the model or the perceptions of the dis- same risk as does each point along the 26004 length. Of
believers. In either case, the disconnect can be identified and course, the 2600-ft length presents more overall risk than does
resolved. the 100-ft length, because it has many more risk-producing
When, after careful evaluation and much experience, the points. A technique is needed to add the length aspect so that a
results are believable and trusted, the user will find many ways 100-ft length of pipeline with one risk score can be compared
to use the numbers that he perhaps did not foresee. against a 2600-ft length with a different risk score. This issue is
In creating a risk assessment system, a measurement tool has also discussed in terms of individual and societal risks in
been created. As with any measurement tool, it must have a Chapter 14.
suitable “signal-to-noise ratio” if it is to provide useful results. Many pipelines will have short lengths of relatively higher
This means that the “noise,” the amount of background vari- risk among long lengths of lower risk. In summarizing the risk
ability in the measurement (due to numerous causes), must be for the entire pipelhe, a simple average or median will hide the
low enough so that the “signal,” the risk value of interest, can shorter, higher risk sections. A cumulative risk-all of the
be read. Every system, especially complex systems such as a higher and lower portions with their respective lengths com-
pipeline existing in a natural environment, will show a great piled into a summary number-will produce the most mean-
deal of variation in many characteristics. Some of this natural ingful measure.
variation will be of interest-a signal-since it changes our The cumulative risk characteristic is measured in order to
risk perceptions. Some of the variation, however, will only be track risk changes over time, compare widely different types of
“background noise”-not of real risk interest and perhaps projects, and equate relative risks to absolute risks, if, for exam-
obscuring the real signals. ple, we want to compare the risk benefit of clearing 20 miles of
In the case of pipeline risk, some sources of variation that pipeline ROW and installing new signs to the value of lowering
must be filtered for signals include and recoating 100 feet of pipeline. On one hand, the failure
potential is being reduced significantly along a short stretch of
Varying static conditions along a pipeline and between com- ROW. On the other hand, amore widespread mitigation is being
pared pipelinesdifferent soils, vegetation, temperatures, broadcast over a long length. Even with relative risk scores at
pipe materials, pipe sizes, operating practices, etc. each location, the comparison is not intuitive unless a method
0 Varying dynamic conditions-activities of people, presence of equivalency is established.
of people, weather events, stress conditions, soil moisture This measure can be called cumulative risk (CR). With a rel-
content, etc. ative risk scale like the one presented in Chapters 3 through 7, a
0 The high level ofuncertainty associated with the modeling of simple formula can be used to calculate cumulative risk:
phenomena such as dispersion and explosion
0 Small amounts of statistical data from which to predict event CR =(]/risk score) x (length)
frequencies
Large numbers of variables that can contribute to risk The reciprocal of risk score is used because the “risk” score is
changes and which are often confounded with each other. really a “safety” score-higher points mean more safety-in
the model shown in Chapters 3 through 7.
A highly variable system limits the ability of the risk assess- Each pipeline and segment of pipeline has a CR value.
ment tool to distinguish real changes in risk level from changes Longer lines have higher CR values, reflecting higher risk.
that do not necessarily contribute to risk. We should be careful This is appropriate since a longer line logically has a higher