Session: 07-03-03 Risk Management

Paper Number: 87066

87066 - Distribution Pipeline Risk Framework 

By nature, gas distribution is a network system; it does not fit well with traditional pipeline risk models that assume a linear geometry.  Distribution system growth is multi‐generational and often leads to mixed assets in the same area where transmission pipeline segments are often constructed within shorter time frames and more uniform materials. Slow, sporadic growth leads to varied record availability and quality that might not readily support commercially available risk models. As a result, the project described in this paper was initiated to development of predictive methods to prioritize mitigation and replacement activities.  With geographically based predictive model, areas such as neighbourhoods or townships with distribution assets can be compared. Priority is assigned by equipment characteristics and environmental attributes similar to areas that have experienced high historical leak rates.

In a previous IPC paper, the authors developed a historical‐based predictive model but applied it to a single city area.  This work has been extended to cover the entire province of Saskatchewan.  The model relied logistic regression and a machine learning algorithm called XGBoost (XGB) to associate the historical failure rate with the asset type, age, pipe material, diameter, pressure, and an array of geographical-dependent attributes. Based on the output of the model, areas were prioritized by severity allowing integrity engineers to consider predictive failure rates while developing integrity activities. This model demonstrated the advantage of using available distribution system records to develop a historical‐based predictive model.

In addition, consequence estimates for distribution networks have been developed. Since risk is calculated as the probability multiplied by the consequence of failure, adding consequence to the model allows differentiation between assets with similar probabilities of failure. Distribution leaks are often classified into three hazard levels that differentiate operational response: A, B, and C. These records were used, with SME input to develop an event tree to evaluate the consequence of  a distribution leak.  This paper summarizes the work done in the project to calculate distribution asset probability of failure and consequences.

Presenting Author: Jason Skow Integral Engineering

Presenting Author Biography: Jason has a proven track record in a variety of engineering and leadership positions. He has 22 years of experience in the oil & gas industry with a focus on pipeline integrity management, data analytics, and risk & reliability.