In order to understand the probability of failure (PoF) of any system, we must first quantify the uncertainties in the system we are modeling. Common industry practice is to assess potential pipeline defects deterministically and account for uncertainty by taking conservative estimates or adding margins of safety to each parameter. This creates very conservative values for use in integrity management. Instead, probabilistic modeling enables sources of uncertainty, such as measurement accuracy, and their influence on PoF to be quantified and independently accounted for.
Assessing threats probabilistically allows for ease of integration of likelihood with consequence for risk modeling and enables a quantitative comparison of crack, corrosion, deformation and other potential threats.
Enbridge is developing probabilistic models to obtain PoF results and an overall line condition for crack threats on its crude oil transmission pipelines. These results will enable Enbridge to quantify the effects of its integrity programs, and allow for a comparison with other potential mitigations such as hydro-testing and pressure reductions.
Enbridge’s probabilistic modeling methodology includes the selection and justification of model parameters, analysis methods, inputs, and uncertainties. This includes but is not limited to a selection of a failure model, failure criteria, pipe properties, feature sizing, and operating conditions. The selected input distributions are then sampled using a Monte Carlo method in order to calculate mean burst pressure and the resulting PoF.