65 Automated Methods to Estimate Transition Temperature, Upper Shelf Energy, and Uncertainty from Charpy V-Notch Data

Abstract

Pipeline feature toughness is a critical input for fitness for service assessments and MAOP reconfirmation based on the ECA approach. For these applications, toughness has historically been estimated from laboratory Charpy V-Notch (CVN) testing via the Upper Shelf Energy (USE) and 85% Shear Appearance (ductile to brittle) Transition Temperature (SATT). Two approaches are widely accepted for estimation of the USE and SATT from a set of CVN tests performed over a range of temperatures: (1) curve fitting a hyperbolic tangent to the experimental data, and (2) analytical solution of a system of empirical equations.

From a practical perspective, the unconstrained curve fitting method (1) is only accurate when the test data span a sufficient range of temperatures to sample both the lower and upper shelves and contain multiple points in the transition region. Additionally curve fitting is a manual, time-consuming procedure with poor repeatability.

Because of these challenges, pipeline operators often use the analytical solution method (2) that was originally proposed by Rosenfeld (Oil and Gas Journal, 1997) and later implemented in API-579 9F.2.2. However, this approach uses assumptions that can lead to over- or under-conservative estimates of toughness. Additionally, the uncertainty associated with the solution has not been quantified. As a result, both approaches often require the pipeline operator to accept unknown and potentially significant uncertainty in their toughness estimates.

This paper will provide a consistent and repeatable method to estimate the ductile to brittle transition temperature and upper shelf energy for CVN data that is applicable to any pipeline regardless of vintage. The method utilizes the hyperbolic tangent approach by applying several simplifying assumptions to effectively constrain the curve fitting to a physically meaningful range without compromising the accuracy of the solution. It will be shown that this approach increases the accuracy and reliability of the calculated values for incomplete datasets. We will identify the physical meaning of the assumptions and show how to estimate the uncertainty such that conservatism can be maintained for subsequent calculations compared to other methods.

Read/Download PaperGradeAvailable after the event