Pipe drift estimation based on the measurements of geometrical parameters from a single pipe

Pipe drift is among the most relevant quality factors of pipes for deep water applications. This is estimated through empirical models encoding geometrical parameters. In a previous work, we relied on Gaussian fields to map these parameters onto the pipe drift conformance probability. The field kernel was estimated from a set of measurements gathered from pipes of the same lot, produced at the same mill. However, in practice it is very difficult to find this homogeneous dataset. The objective of this paper is to extend the previously developed framework to estimate the drift based on the geometrical data relevant to a single pipe, only. For this, we consider known analytical kernels and infer their parameters from the single pipe measurements. Then, we estimate the actual error on the drift conformance probability estimations, considering the best fitting kernel. This error turns out to be negligible.