Abstract
Stress corrosion cracking (SCC) on pipelines has been extensively studied over the past three to four decades. Various models have been developed to predict where and how fast SCC occurs on pipelines. However, due to the complexity of SCC, no general models are currently available to accurately predict SCC on pipelines. Models developed based on operating experience for one geographic location has often performed poorly in another region. For example, the SCC soils model developed in the past predicts that low-pH SCC will occur in poorly drained, anaerobic soils; however, in the same general geographic region, low-pH SCC has also been found to occur preferentially in well-drained soils. It is therefore critical to collect all related data and understand the actual SCC mechanism to develop an effective SCC susceptibility model that will be more generally applicable. This paper introduces a data mining methodology, a decision tree approach, for identification of the correlation between the presence of SCC and environmental/ loading conditions and further refinement of SCC susceptibility with mechanistic understanding.
