Page 367 - Pipeline Risk Management Manual Ideas, Techniques, and Resources

15/342 Risk Management
be assumed that some non-common causes are at work. That is, threshold requires more confidence in the risk assessment
some new “force” has been introduced into the system and accuracies than is probably prudent. Another problem is the
should be investigated. realistic premise that risk tolerance is not a fixed value in any
The following four example cases illustrate the response to company. In the complexities of the business world, it is influ-
some possible data distributions (often viewed by a histogram enced by economic conditions, public perception, and political
of risk scores or index sum scores). In these examples, lower conditions. A further complication is the need for a time factor
scores represent lower safety and, hence, higher risks-the in setting a risk tolerance. A certain level of risk is tolerable for
same scoring protocol used in Chapters 3 through 6 of this some period of time, until the situation can be efficiently
book. addressed. For instance, shallow cover may not requue imme-
diate attention and can be addressed in conjunction with other
Case 1: Extreme Outliers work planned in the area. At some level, however, the risk is
Description: Some low scores are more than 3 standard devia- seen to he so unacceptable that immediate action, even the shut
tions away from the overall average. The “3 standard devia- down of the pipeline, may be warranted.
tions” is a common decision point based on statistical Rather than setting a full range of definitive action points,
analysis that says measurements farther away from the aver- the more prudent approach to risk management is thought to lie
age in a normal distribution have a high probability of being in prioritizing. The choice of a relative risk assessment system
influenced by nonrandom effects. In other words, these reflects this. In a prioritization approach, the operator will
measurements are statistically different from other measure- always be ranking portions of their system, based on the level of
ments and probably warrant special attention. risk. This ranking in turn generates a list of possible projects to
Response: Implement immediate mitigation measures to bring reduce the risk level. More resources may then be allocated
higher risk scores to within 3 standard deviations within 6 toward changing the risk level of the worst sections first, and
months. then progressing down the list. In most cases, the amount of
Case 2: Frequent Outliers available resources will then set the defacto level of acceptable
Description: Lowest scores are more than 2.5 standard devia- risk, since money usually ms out before the list of “things to
tions away from average. do” is exhausted.
Response: Implement mitigation to bring these scores to within
2.5 standard deviations within 1 year.
Case 3: Infrequent Outliers IX. Risk mitigation
Description: Low scores do not meet 1 or 2 above, but are still
distinct outliers, based on visual examination of graphs. Where to start
Response: Implement mitigation to bring these scores to “the
edge” of the main population within 1 year. As discussed in previous sections, identifying the need for and
Case 4: Uniform appropriate aggressiveness of risk reduction efforts can be a
Description: Tight range of risk scores (no apparent outliers). very complex process. In the face of generous amounts of new
Response: Perform preferential mitigation with percentage of information, the new risk manager might well feel over-
total (mileage-normalized) mitigation proportional to the whelmed and need to ask “Where do I begin?” The experienced
distance from median. Pipeline segments farther from the risk manager will usually immediately see a host of things she
population average will receive proportionally higher levels can now do more quantitatively where previously she was
of mitigation. A formula can be developed to dictate pre- equipped with mostly opinion. She will see the advantages
cisely the level of mitigation for each section, even to the stemming from standardized valuations on risk conditions and
extent of reducing mitigation on the safest sections in order mitigations, avoiding competition for resources and much
to redirect resources to higher risk sections. uncertainty in decision making.
Under the continuous improvement philosophy noted ear-
Precedent-based criteria lier, a fundamental premise is that the risk management process
will not reach a conclusion. That is, there is no threshold level
Even without a formal risk management system, certain levels of risk that, once attained, will result in the end of the program.
of risk have always triggered immediate action. A trigger or Rather, it is assumed that some amount ofresources will always
action point can be seen as the risk level that is not tolerable, not be applied to risk reduction and a universally acceptable risk
even for a short time. One trigger point for many operators is an level will most likely never be attained. Issues of risk tolerance
active leak of such magnitude that damages could occur. This for individuals and the public in general can be examined in this
trigger point is ohviously a reaction to a failure that has already regard. The notion of a continuous effort of risk reduction is
happened. Nevertheless, it can also be viewed as a risk that is no also a realistic premise since the risk level tends to naturally
longer acceptable. increase with time. The aging of the infrastructure, time-
dependent failure mechanisms, and encroaching population
Continuous improvement density are mechanisms that help to increase risks over time.
Increasing competition and regulation are two mechanisms that
Establishing an absolute level of acceptable risk is often not the help to reduce the amount of available resources.
best approach to risk management. The first problem with an A simple and entirely appropriate initial approach is to create
absolute level is the inherent inaccuracies associated with fail- lists of system components rank ordered by overall risk and/or
ure probability and consequence calculations. Addressing risks specific failure modes. Then, focus attention and discretionary
above a certain threshold and ignoring any risks below that resources on the segments showing relatively higher risks. This