Abstract: Several operators have increased the number of pipe cutouts and associated laboratory toughness testing for vintage assets to fill gaps in material records the original testing certificates did not include fracture toughness, a properties that the legacy version of the manufacturing specification did not require. This paper provides a non-destructive evaluation (NDE) approach to pipe seam toughness that can reduce the need for these disruptive pipe cutouts and increase the available dataset for the pipe toughness of specific assets. The NDE approach is currently specific to electric resistance welded (ERW) pipes, but data is provided from a limited proof-of-concept for Electro Flash Welded (EFW) pipes. The current process first estimates the 85% shear transition temperature of the seam in the CVN impact energy vs. temperature curve using the machine learning (ML) model. The predicted transition temperatures are shifted based on American Petroleum Institute (API) API 1176 to account for the difference in strain rate between a CVN test (impact) and fracture initiation (quasi-static). This shifted transition temperature is then compared to the minimum operating temperature. If the operating temperature is higher than the transition temperatures, the pipe material is confirmed to be in the upper shelf region of the CVN impact energy vs. temperature curve. If upper shelf values are applicable, a distribution of CVN toughness values for like samples is used to determine a conservative toughness estimate. A secondary machine learning model can provide an actual estimate of the toughness value.



