Page 31 - Pipeline Risk Management Manual Ideas, Techniques, and Resources

1/10 Risk: Theory and Application
from some commercially available existing models, cus- V. Data collection
tomized from existing models, or created “from scratch”
depending on your requirements. Multiple models can be Data and information are essential to good risk assessment.
run against the same set of data for comparisons and model Appendix G shows some typical information-gathering efforts
evaluations. that are routinely performed by pipeline operators. After sev-
eral years of operation, some large databases will have devel-
Step 2: Data collection and preparation oped. Will these pieces of data predict pipeline failures? Only
in extreme cases. Will they, in aggregate, tell us where risk hot
Data collection entails the gathering of everything that can be spots are? Certainly. We ohviously feel that all of this informa-
known about the pipeline, including all inspection data, origi- tion is important-we collect it, base standards on it, base regu-
nal construction information, environmental conditions, oper- lations on it, etc. It just needs to be placed into a risk context so
ating and maintenance history, past failures, and so on. Data that a picture of the risk emerges and better resource allocation
preparation is an exercise that results in data sets that are ready decisions can be made based on that picture. The risk model
to be read into and used directly by the risk assessment model. transforms the data into risk knowledge.
A collection of tools enables users to smooth or enhance data Given the importance of data to risk assessment, it is impor-
points into zones of influence, categories, or bands to convert tant to have a clear understanding of the data collection process.
certain data sets into risk information. Data collection is dis- There exists a discipline to measuring. Before the data gather-
cussed later in this chapter and data preparation issues are ing effort is started, four questions should be addressed
detailed in Chapter 8.
1. What will the data represent?
Step 3: Segmentation 2. How will the values be obtained?
3. What sources ofvariation exist?
Because risks are rarely constant along a pipeline, it is advan- 4. Why are the data being collected?
tageous to segment the line into sections with constant
risk characteristics (dynamic segmentation) or otherwise
divide the pipeline into manageable pieces. Segmentation What will the data represent?
strategies and techniques are discussed in Chapters 2 and 8,
respectively. The data are the sum of our knowledge about the pipeline sec-
tion: everything we know, think, and feel about it-when it was
Step 4:Assessing risks built, how it was built, how it is operated, how often it has failed
or come close, what condition it is in now, what threats exist,
Now the previously selected risk assessment model can be what its surroundings are, and so on-all in great detail. Using
applied to each segment to get a unique risk “score” for that the risk model, this compilation of information will be trans-
segment. These relative risk numbers can later be con- formed into a representation of risk associated with that sec-
verted into absolute risk numbers. Working with results of risk tion. Inherent in the risk numbers will be a complete evaluation
assessments is discussed in Chapters 8,14, and 15. of the section’s environment and operation.
Step 5: Managing risks How will the values be obtained?
Having performed a risk assessment for the segmented Some rules for data acquisition will often be necessary. Issues
pipeline, we now face the critical step of managing the risks. In requiring early standardization might include the following:
this area, the emphasis is on decision support-providing the
tools needed to best optimize resource allocation. Who will be performing the evaluations? The data can be
This process generally involves steps such as the following: obtained by a single evaluator or team of evaluators who will
visit the pipeline operations offices personally to gather the
Analyzing data (graphically and with tables and simple information required to make the assessment. Alternatively,
statistics) each portion of a pipeline system can be evaluated by those
Calculating cumulative risks and trends directly involved in its operations and maintenance. This
Creating an overall risk management strategy becomes a self-evaluation in some respects. Each approach
Identifying mitigation projects has advantages. In the former, it is easier to ensure consis-
Performing what-if’s tency; in the latter, acceptance by the workforce might be
greater.
These are fully discussed in subsequent chapters, especially What manuals or procedures will be used? Steps should
Chapter 15. be taken to ensure consistency in the evaluations.
The first two steps in the overall process, (1) risk model How often will evaluations be repeated? Reevaluations
and (2) data collection, are sometimes done in reverse order. should be scheduled periodically or the operators should be
An experienced risk modeler might begin with an exam- required to update the records periodically.
ination of the types and quantity of data available and from Will “hard proof” or documentation be a requirement in
that select a modeling approach. In light of this, the dis- all cases? Or can the evaluator accept “opinion” data in some
cussion of data collection issues precedes the model-selection circumstances? An evaluator will usually interview pipeline
discussion. operators to help assign risk scores. Possibly the most com-