A clustering approach for assessing external corrosion in a buried pipeline based on hidden Markov random field model

This paper describes the use of a clustering approach based on hidden Markov random field to extract potential homogeneous segments from a large length right-of-way of a pipeline structure with heterogeneous soil properties. This approach extends the conventional finite mixture model so that the spatial correlation of external corrosion sites can be taken into consideration. An algorithm is established for classifying corrosion defects using soil properties from an in-situ survey and location information from in-line inspection reports. The categorized corrosion defects reveal the hidden patterns of corrosion degradation in different segments along a pipeline structure. Stochastic simulation is employed to test this clustering approach. An example involving a 110-km pipeline interval is employed to illustrate the implementation of the clustering approach. The results indicate that the process of external corrosion propagation in a buried pipeline is position-dependent and is highly related to the soil environment. In addition, the results show that this phenomenon can be interpreted by segmentation using the proposed clustering method. A clustering-based inspection strategy. is discussed as a way to apply the present approach. (C) 2015 Elsevier Ltd. All rights reserved.