Page 54 - Pipeline Risk Management Manual Ideas, Techniques, and Resources

Designing a risk assessment model 2/33
to reflect beliefs about frequency of certain failure types when important the risk will be until she sees the weighting of that
linking relative models to absolute calculations or when there is variable.
large variations in expected failure frequencies among the Confusion can also arise in some models when the same vari-
possible failure types. able is used in different parts of the model and has a location-
specific scoring scheme. For instance, in the offshore
environment, water depth is a risk reducer when it makes
Risk scoring
anchoring damage less likely. It is a risk increaser when it
Direction ofpoint scale increases the chance for buckling. So the same variable, water
depth, is a “good” thing in one part of the model and a “bad
In a scoring-type relative risk assessment, one of two point thing somewhere else.
schemes is possible: increasing scores versus decreasing to rep-
resent increased risk. Either can be effectively used and each Combining variables
has advantages. As a risk score, it makes sense that higher num-
bers mean more risk. However, as an analogy to a grading sys- An additional modeling design feature involves the choice of
tem and most sports and games (except golf), others prefer how variables will be combined. Because some variables will
higher numbers being better-more safety and less risk. indicate increasing risk and others decreasing, a sign conven-
Perhaps the most compelling argument for the “increasing tion (positive versus negative) must be established. Increasing
points = increasing safety” protocol is that it instills a mind-set levels ofpreventions should lead to decreased risks while many
of increasing safety. “Increasing safety” has a meaning subtly attributes will be adding risks (see earlier discussion of preven-
different from and certainly more positive than “lowering tions and attributes). For example, the prevention of perform-
risks.” The implication is that additional safety is layered ing additional inspections should improve risk scores, while
onto an already safe system, as points are acquired. This latter risk scores deteriorate as more soil corrosivity indications
protocol also has the advantage of corresponding to certain (moisture, pH, contaminants, etc.) are found.
common expressions such as “the risk situation has deterio- Another aspect of combining variables involves the choice
rated’ = “scores have decreased and “risk situation has of multiplication versus addition. Each has advantages. Multipli-
improved” = “scores have increased.” cation allows variables to independently have a great impact on
While this book uses an “increasing points = increasing a score. Adhtion better illustrates the layering of adverse condi-
safety” scale in all examples of failure probability, note that this tions or mitigations. In formal probability calculations, multi-
choice can cause a slight complication if the relative risk plication usually represents the and operation: If corrosion
assessments are linked to absolute risk values. The complica- prevention = “poor” AND soil comsivity = “high” then risk =
tion arises since the indexes actually represent relative proba- “high.” Addition usually represents the or operation: If depth of
bility of survival, and in order to calculate a relative probability cover = “good” OR activity levef = ‘‘low’’ then risk =“low.”
of failure and link that to failure frequencies, an additional step
is required. This is discussed in Chapter 14. Option 1
Risk variable = (sum of risk increasers) -(sum of nsk reducers)
Where to assign weightings
where the point scales for each are in the same direction. For
In previous editions ofthis model, it is suggested that point val- example,
ues be set equal to weightings. That is, when a variable has a
point value of 3, it represents 3% of the overall risk. The disad- Corrosion threat = (environment) - [(coating) + (cathodic protection)]
vantage of this system is that the user does not readily see Option 2
what possible values that variable could take. Is it a 5-point
variable, in which case a value of 3 means it is scoring Risk variable = (sum ofrisk increasers) + (sum ofnsk reducers)
midrange? Or is it a 15-point variable, for which a score of
3 means it is relatively low? Point scales for risk increasers are often opposite from the scale
An alternative point assignment scheme scores all variables of risk reducers. For example, in an “increasing points means
on a fixed scale such as CLlO points. This has the advantage of increasing risk” scheme,
letting the observer know immediately how “good” or “bad” Corrosion threat = (environment) + [(coating) + (cathodic protection)]
the variable is. For example, a 2 always means 20% from the
bottom and a 7 always means 70% of the maximum points that where actual point values might be
could be assigned. The disadvantage is that, in this system,
weightings must be used in a subsequent calculation. This adds (corrosion threat) = (24) + (-5 + -2) = 17
another step to the calculation and still does not make the point
scale readily apparent. The observer does not know what the Option 3
70% variable score really means until he sees the weightings In this approach, we begin with an assessment ofthe threat level
assigned. A score of 7 for a variable weighted at 20% is quite and then consider mitigation measures as adjustment factors.
different from a score of 7 for a variable weighted at 5%. So, we begin with a risk and then adjust the risk downward (if
In one case, the user must see the point scale to know that a increasing points = increasing risk) as mitigation is added:
score of, say, 4 points represents the maximum level of mitiga-
tion. In the alternate case, the user knows that 10 always repre- Risk variable = (threat) x (sum of% threat reduction through
sents the maximum level of mitigation, but does not know how mitigations)