Page 58 - Pipeline Risk Management Manual Ideas, Techniques, and Resources

Lessons learned in establishing a risk assessment program 2/37
The general lessons learned are as follows: Avoid complexity

Work from general to specific. Every single component of the risk model should yield more
Think “organic.” benefits than the cost it adds in terms of complexity and data-
Avoid complexity. gathering efforts. Challenge every component of the risk model
Use computers wisely. for its ability to genuinely improve the risk knowledge at a
Build the program as you would build a new pipeline. reasonable cost. For example:
Study your results.
Don’t include an exotic variable unless that variable is a
We now take a look at the specifics ofthese lessons learned. useful risk factor.
Don’t use more significant digits than is justified.
Work from general to specific Don’t use exponential notation numbers if a relative scale
can be appropriately used.
Get the big picture first. This means “Get an overview assess- Don’t duplicate existing databases; instead, access informa-
ment done for the whole system rather than getting tion from existing databases whenever possible. Duplicate
every detail for only a portion of the system.” This has two data repositories will eventually lead to data inconsistencies.
advantages: Don’t use special factors that are only designed to change
numerical scales. These tend to add more confusion than
I. No matter how strongly the project begins, things may their benefit in creating easy-to-use numbers.
change before project completion. If an interruption does Avoid multiple levels of calculations whenever possible.
occur, at least a general assessment has been done and some Don’t overestimate the accuracy of your results, especially in
useful information has been generated. presentations and formal documentation. Remember the
2. There are strong psychological benefits to having results high degree ofuncertainty associated with this type of effort.
(even if very preliminary--caution is needed here) early in
the process. This provides incentives to refine and improve Use computers wisely
preliminary results. So, having the entire system evaluated
to a preliminary level gives timely feedback and should Too much reliance on computers is probably more dangerous
encourage further work. than too little. In the former, knowledge and insight can be
obscured and even convoluted. In the latter, the chief danger is
It is easy to quickly assess an entire pipeline system by limit- that inefficiencies will result-an undesirable, hut not critical,
ing the number of risk variables in the assessment. Use only a event. Regardless of potential misuse, however. computers can
critical few, such as population density, type of product, operat- greatly increase the strength of the risk assessment process, and
ing pressure, perhaps incident experience, and a few others. no modem program is complete without extensive use of them.
The model can then later be “beefed up” by adding the variables The modem software environment is such that information is
that were not used in the first pass. Use readily available infor- easily moved between applications. In the early stages of a proj-
mation whenever possible. ect, the computer should serve chiefly as a data repository.
Then, in subsequent stages, it should house the algorithn-
Think “organic” how the raw information such as wall thickness, population
density, soil type, etc., is turned into risk information. In later
Imagine that the risk assessment process and even the stages of the project, data analysis and display routines should
model itself are living, breathing entities. They will grow he available. Finally, computer routines to ensure ease and con-
and change over time. There is the fruit-the valuable sistency of data entry, model tweaking, and generation of
answers that are used to directly improve decision making. required output should he available.
The ideal process will continuously produce ready-to-eat fruit Software use in risk modeling should always follow program
that is easy to “pick” and use without any more processing. development-not lead it.
There are also the roots-the hehind-the-scenes techniques
and knowledge that create the fruit. To ensure the fruit is 0 Early stage. Use pencil and paper or simple graphics soft-
good, the roots must he properly cared for. Feed and ware to sketch preliminary designs of the risk assessment
strengthen the roots by using HAZOPS, statistical analysis, system. Also use project management tools if desired to plan
FEMA, event trees, fault trees, and other specific risk tools the risk management project.
occasionally. Such tools provide the underpinnings for the Intermediate stages. Use software environments that can
risk model. store, sort, and filter moderate amounts of data and generate
Allow for growth because new inspection data, new new values from arithmetic and logical (if. . . then. . . else. . .)
inspection techniques, new statistical data sets to help deter- combinations of input data. Choices include modem spread-
mine weightings, missed risk indicators, new operating dis- sheets and desktop databases.
ciplines, and so on will arise. Plan for the most flexible Later stages. Provide for larger quantity data entry, manipu-
environment possible, Make changes easy to incorporate. lation, query, display, etc., in a long-term, secure, and user-
Anticipate that regardless of where the program begins and friendly environment. If spatial linking of information is
what the initial focus was, eventually, all company personnel desired, consider migrating to geographical information sys-
might he visiting and “picking the fruit” provided by this tems (GIS) platforms. If multiuser access is desired, consider
process. robust database environments.