Enhanced oil recovery techniques and the improved methods for discovery of small pools of hydrocarbons have helped to maintain the potential for continued hydrocarbon recovery in the North Sea basin.
The probability of continued export infrastructure availability can have a significant impact on the development, and redevelopment of marginal fields. If the life of existing aging export infrastructure can be safely extended, the stranding of proven hydrocarbon reserves can be avoided.
To help determine if a pipeline is safe for continued operation an integrity assessment is typically performed. Historically the choice has been to perform either a deterministic assessment or a probabilistic assessment. Both methods have advantages, a deterministic assessment can be relatively quick and low cost, whilst a probabilistic assessment can help to demonstrate the relative probability of failure at various points in the future, and hence for how long the pipeline is likely to be acceptable for continued use.
Many probabilistic approaches use Monte-Carlo simulations to determine probabilities of failure. Typically, such simulations determine a failure probability for a specific time-interval. For a single defect this approach may be straightforward to carry out. However, for pipelines containing many defects this may not be the case, as performing multiple simulations can incur significant time and cost.
By using a blend of custom designed assessment and machine learning software the authors have developed a cost and time efficient method of assessing the risks associated with the life extension of existing assets.
Using modern computing techniques such as parallel processing and machine learning the authors have developed a form of defect assessment which aims to deliver the detail of a probabilistic approach with the speed and cost of a deterministic assessment.