Page 25 - Pipeline Risk Management Manual Ideas, Techniques, and Resources

1/4 Risk: Theory and Application
threats can be generally grouped into two categories: time- case of service interruption discussed in Chapter 10, the gen-
dependent failure mechanisms and random failure mechanisms, eral definition of failure in this book will be excessive leakage.
as discussed later. The term leakage implies that the release of pipeline contents
The phrases threat assessment and hazard identification are is unintentional. This lets our definition distinguish a fail-
sometimes used interchangeably in this book when they refer to ure from a venting, de-pressuring, blow down, flaring, or other
identifying mechanisms that can lead to a pipeline failure with deliberate product release.
accompanying consequences. Under this working definition, a failure will be clearer in
some cases than others. For most hydrocarbon transmission
Risk pipelines, any leakage (beyond minor, molecular level emis-
sions) is excessive, so any leak means that the pipeline has
Risk is most commonly defined as the probability of an event failed. For municipal systems, determination of failure will not
that causes a loss and the potential magnitude of that loss. By be as precise for several reasons, such as the fact that some
this definition, risk is increased when either the probability of leakage is only excessive-that is, a pipe failure-after it has
the event increases or when the magnitude of the potential continued for a period of time.
loss (the consequences of the event) increases. Transportation Failure occurs when the structure is subjected to stresses
ofproducts by pipeline is a risk because there is some probabil- beyond its capabilities, resulting in its structural integrity being
ity of the pipeline failing, releasing its contents, and causing compromised. Internal pressure, soil overburden, extreme tem-
damage (in addition to the potential loss of the product itself). peratures, external forces, and fatigue are examples of stresses
The most commonly accepted definition of risk is often that must be resisted by pipelines. Failure or loss of strength
expressed as a mathematical relationship: leading to failure can also occur through loss of material by
corrosion or from mechanical damage such as scratches and
Risk = (event likelihood) x (event consequence) gouges.
The answers to what can go wrong must be comprehensive in
As such, a risk is often expressed in measurable quantities order for a risk assessment to be complete. Every possible fail-
such as the expected frequency of fatalities, injuries, or economic ure mode and initiating cause must be identified. Every threat
loss. Monetary costs are often used as part of an overall expres- to the pipeline, even the more remotely possible ones, must be
sion of risk however, the difficult task of assigning a dollar identified. Chapters 3 through 6 detail possible pipeline failure
value to human life or environmental damage is necessary in mechanisms grouped into the four categories of Third Par&
using this as a metric. Corrosion. Design, and Incorrect Opemtions. These roughly
Related risk terms include Acceptable risk, tolerable risk, correspond to the dominant failure modes that have been
risk tolerance, and negligibie risk, in which risk assessment historically observed in pipelines.
and decision making meet. These are discussed in Chapters 14
and 15.
A complete understanding of the risk requires that three Probability
questions be answered: By the commonly accepted definition of risk, it is apparent that
probability is a critical aspect of all risk assessments. Some
1. What can go wrong? estimate of the probability of failure will be required in order
2. How likely is it? to assess risks. This addresses the second question of the risk
3. What are the consequences? definition: “How likely is it?”
Some think of probability as inextricably intertwined with
By answering these questions, the risk is defined. statistics. That is, “real” probability estimates arise only from
statistical analyses-relying solely on measured data or
Failure observed occurrences. However, this is only one of five defini-
tions of probability offered in Ref. 88. It is a compelling def-
Answering the question of “what can go wrong?’ begins with inition since it is rooted in aspects of the scientific process and
defining a pipeline failure. The unintentional release of the familiar inductive reasoning. However, it is almost always
pipeline contents is one definition. Loss ofintegrity is another woefully incomplete as a stand-alone basis for probability esti-
way to characterize pipeline failure. However, a pipeline can mates of complex systems. In reality, there are no systems
fail in other ways that do not involve a loss of contents. A more beyond very simple, fixed-outcome-type systems that can be
general definition is failure to perform its intended function. fully understood solely on the basis of past observations-the
In assessing the risk of service interruption, for example, a core of statistics. Almost any system of a complexity beyond a
pipeline can fail by not meeting its delivery requirements simple roll of a die, spin of a roulette wheel, or draw from a
(its intended purpose). This can occur through blockage, deck of cards will not be static enough or allow enough trials for
contamination, equipment failure, and so on, as discussed in statistical analysis to completely characterize its behavior.
Chapter 10. Statistics requires data samples-past observations from which
Further complicating the quest for a universal definition of inferences can be drawn. More interesting systems tend to have
failure is the fact that municipal pipeline systems like water and fewer available observations that are strictly representative of
wastewater and even natural gas distribution systems tolerate their current states. Data interpretation becomes more and more
some amount of leakage (unlike most transmission pipelines). necessary to obtain meaningful estimates. As systems become
Therefore, they might be considered to have failed only when more complex, more variable in nature, and where trial obser-
the leakage becomes excessive by some measure. Except in the vations are less available, the historical frequency approach