‘Condition metrics’ are single-valued metrics used to describe unique aspects of a pipeline’s condition. For corroded pipelines, we can define simple condition metrics such as anomaly density and probability of exceedance, and calculate their values using in-line inspection (ILI) data.

In previous work, the authors have shown how condition metrics can be used to represent pipeline condition in a multidimensional space – namely, the ‘condition space’. Individual pipelines populate this space, and their relative positions give rise to the notion of ‘good’ pipelines and ‘bad’ pipelines.

While a pipeline’s current position within the condition space is useful information (and can form the basis of integrity management decisions), an equally important consideration is its trajectory through the condition space over time. The present study explores how these trajectories can be predicted using corrosion nucleation rates and corrosion growth rates derived from repeat ILI datasets.

In particular, we demonstrate how the future trajectory of a target pipeline can be simulated by learning from similar pipelines with repeat ILI data. The concept is trialed using a pilot data warehouse containing historical ILI data for over 5,000 pipelines.