A Data-Driven Process to Determine CVN Value Considering Comparable Pipe Attributes

Abstract This paper presents a data-driven process to determine the statistical (i.e., probabilistic) characteristics and characteristic value of CVN. The derived probabilistic characteristics and characteristic value of CVN can be used as the input for probabilistic and deterministic crack assessment, respectively. The objective of this paper is two-fold. Firstly, it provides an interpretation of “comparable pipe” specified by 49 CFR § 192.712 e(2)(i)(A) and demonstrates the key pipe attributes for defining comparable pipe. Secondly, it describes the process of determining the characteristic value of CVN as well as the statistical characteristics of CVN using CVN data from “comparable pipe”. To that end, a four-tier protocol to collect CVN data samples from “comparable pipe” is established. For the crack assessment of a particular pipeline segment, the CVN values selected from “comparable pipe” following this protocol embeds necessary conservatism that ensures equivalent or higher safety than the approaches relying on limited CVN data points obtained from that specific segment. The proposed method explicitly considers data variability, reduces statistical bias in determining the characteristic value and is flexible to incorporate learnings from new research as a supplement to this approach. The proposed methodology will facilitate the integrity management program of pipeline with cracks.