Page 284 - Pipeline Risk Management Manual Ideas, Techniques, and Resources

Station risk assessment 13/261
ciated with the storage of large volumes of flammable liquids, cal data. It is acknowledged and accepted that in using most risk
one station section may consist of all components located in a assessment models, some realism is being sacrificed in the
bermed storage tank area, including tank (floor, walls, roof), interest of understandability and usability. This is consistent
transfer pump, piping, safety system, and secondary contain- with the intent ofmost models.
ment.This section wouldreceiveariskscorereflectingtherisks The ideal risk assessment methodology works well under
specific to that portion of the station. The risk evaluations for conditions of “very little data” as well as conditions of “very
each section can be combined for an overall station risk score or extensive data.” An overview assessment, where data are
kept independent for comparisons with similar sections in other scarce, might base an assessment on only a few variables such
stations. as
Often, a station’s geographical layout provides a good oppor-
tunity for sectioning. There are usually discrete areas for Nearby population density
pumps, manifold, truck loadinghnloading, additives, tanks, Presence of special environmental areas
compressors, etc., that provide appropriate sections for risk Quantity of storedproducts
assessment purposes. Further distinctions could be made to Type of products handled
account for differences in tanks, pumps, compressors, etc., Incident history at the facility
thereby creating smaller sections that have more similar char- Date of last API 653 out-of-service inspection (for tanks)
acteristics.
In certain cases, it might be advantageous to create contigu- In this case, the model would not provide much guidance on
ous or grouped station sections. In the above example, a specific equipment or procedural changes for a specific tank.
section could then include all piping, independent of the It could, however, point to areas where the greatest amounts
tank, pump, or process facility to which it is connected. of resources are best sent. A more detailed version of the
Another approach could be to include all liquid pipeline sta- methodology, designed to help in detailed decision making,
tion tanks in one section, independent of their type, location, might use a data set including all of the above as well as the
and service. following:
The sectioning strategy should take into account the types of
comparisons that will be done for risk management. If individ- Tank surface area
ual tanks must be compared (perhaps to set specific inspection Tank profile (heighdwidth ratio)
frequencies), then each tank should probably have its own eval- Tankjoint type (bolt, rivet, weld)
uation. If all “compressor areas,” from station to station, are to Tank year of construction
be compared, that should lead to an accomodating sectioning Tank foundation type
strategy. Tank level alarms
A sectioning strategy should also consider the need to pro- Tank level alarm actions (local, remote, automatic, etc.)
duce cumulative, length-sensitive scores for comparison to Tank corrosion rate
pipeline lengths. This is discussed on page 287. Staffing level
Traffic flow patterns
Data requirements Traffic barriers
Security fences
As noted in Chapter 1, a model is a simplified representation of Visitor control
the real world. The way to simplify real-world processes into an Programmable logic controller (PLC) usage
accurate facilities model is to first completely understand the Critical instrument program
real-world processes in their full complexity. Only then are we Management of change program
able to judge which variables are critical and how they can be Operator training specifics
logically combined into a valid model. The objective is not to Use of SCADA systems
simulate reality, but to model it accurately. The ideal station risk UT inspection program
model must be able to withstand a critical engineering evalua- MF inspection program
tion, in addition to its application in real-world risk manage- Pump type
ment decision making. Pump speed
As with line pipe, the quality and quantity of safety data are Pump seal type
limited for pipeline station facilities. Therefore, few statisti- Pump seal secondary containment
cally based correlations can be drawn from all of the factors Fatigue sources
believed to play a significant role in failure frequency and con- Material toughness
sequence. The contributing factors, however, can be identified Etc.
and considered in a more qualitative sense, pending the acquisi-
tion of more statistically significant data. This list can easily extend into hundreds of variables as shown
Concepts from statistical failure analysis are useful and at the end of this chapter. The risk assessment methodology
underlie portions of this station risk model. However, given the should work for operators who wish to work with limited data
unavailability of data, the uncertainty associated with the rare as well as those with extensive, pre-existing databases that need
event data, and the complexities of even the simplest facility, a to be incorporated.
departure from strict statistical analysis is warranted. This Figure 13.3 provides an example of an overall station risk
departure requires the inclusion of experience and judgment, model, showing some of the variables chosen for one of the
even when such judgment is only weakly supported by histori- facility modules.