Page 394 - Pipeline Risk Management Manual Ideas, Techniques, and Resources

Appendix E
Sample Pipeline

Risk Assessment

Algorithms

The intent ofthis appendix is not to provide complete risk algo- Rehabititation,” a report to the American Gas Association.
rithms to the reader, but rather to convey a sense of how model- September 28,1990.
ing has been done in the past. Presentation of complete models
would necessitate the inclusion of full documentation of all the- This report presents the development of PIMAR, the A.G.A./PRC
ory and rationale implicit in the model. That would take a book sponsored ranking algorithm for pipeline mamtenance. One focus of
of this size for each comprehensive model evaluated. By pre- this paper is the development of the risk assessment algorithm,
PIMAR, for prioritizing pipeline maintenance and rehabilitation.
senting examples, it is hoped that the reader will gain confi- Each of the contributing factors associated with probability of failure
dence in setting up his or her own models. This confidence is and the consequences are defined. The parameters chosen for the
gained through the knowledge that there is no “magic algorithm in terms ofprobability of failure werecategonzed into eight
approach’ that guarantees better results than any other. A good different groups: type ofpipe, soil stability, coating integrity, cathodic
risk model will be firmly rooted in engineering concepts and protection, damage susceptibility, hydrostatic test history, leakhp-
consistent with experience and intuition. That is why there are ture history, and pipeline condition. Those related to consequences of
so many similarities in the efforts of many different modelers failure were class location, security of throughput, product type,
examining many different systems at many different times for propensity for ductile fracture propagation, and transition tempera-
differing objectives. Beyond compatibility with engineering ture. The rationales for the selection of these parameters was
provided.
and experience, a model can take many forms especially in dif-
fering levels of detail and complexity. The end result of each of these contributing factors was an
algorithm for probability of failure, given below.
GRI model reviews PF = Pt(JtSS + CaP*SCA + DS + HT + LR +PiCo)
The following discussions of two published pipeline risk mod- where
els are extracted from preliminary work performed by Kiefner
and Associates, Inc., on behalf of the Gas Research Institute Pt = risk as a function ofthe type of pipe
[GRI]). Many other references on these models are available in JtSS = risk associated with longitudinal stresses
the technical literature, These two were chosen as being fairly caused by soil-induced forces and older join-
representative of many systems developed by consultants and ing methods
by operating companies themselves. CaP*SCA = risk associated with corrosion
susceptibility
DS = damage susceptibility
Model 1
HT = hydrostatic test history
Kiefnrr: .I F, Vieth, H., Orbun, J E., and Feder; P I., LR = service IeaWrupture history
“Mt.thod5 for Prioritizing Pipeline Maintenance and PiCo = pipe condition.