1Integral Engineering, Edmonton, Canada, 2Flint Hills Resources, Wichita, USA
The current guidance in API 1183: 2020 directs analysts to estimate dent indentation formation strains using Appendix R of ASME B31.8, relying on the local radii of curvature extracted from the surface contours of measured dents, once signal noise has been removed using smoothing techniques. Since the publication of the current recommended practice, several alternative strain estimation methods have been proposed in IPC2022-87168, PRCI MD-5-2, and IPC2024-133905 to address the limitations in the B31.8 equations. These alternative methodologies improve upon the B31.8 approach using some combination of advanced techniques: incorporating the circumferential membrane strain for a more complete characterization of the effective strain; focusing on the deeper portions of the dent (ex. below 85% depth) to define reference undeformed lengths; using a higher-order function to characterise the local dent curvature or deformed lengths while reducing sensitivity to in-line inspection (ILI) tool resolution; and incorporating regression models fit to data generated from finite element analysis (FEA) that account for the change in profile geometry of unrestrained dents due to re-rounding from internal pressure. Although these methods have individually demonstrated improvement over ASME B31.8, they have not been benchmarked against each other. In addition, the vast majority of the FEA cases used by the original authors to validate these models are limited to idealized single-apex, symmetrical dents. In this paper, we evaluate the accuracy of each proposed method against a comprehensive set of FEA cases that feature a range of realistic geometries, including asymmetric single-apex and multi-apex dents. We propose guidance on the application of these models for both restrained and unrestrained dents. Ultimately, this guidance supports more reliable and conservative screening of dents to limit the number of dents that require costly and time-consuming assessment using FEA.
Available after the event



