This paper describes an initial feasibility effort completed in a Small Business Innovative Research (SBIR) Phase I project for DOT-PHMSA to develop a novel approach for the assessment of the remaining strength of corroded pipe. The goal was to assess what failure criterion might be for modern line-pipe steels (e.g., X80) compared to pre-1970 pipes that were used to validate the B31G and RSTRENG methodologies. In the process, the methodology will be able to handle many situations that B31G cannot handle, i.e., high longitudinal stress, corrosion that is more circumferentially oriented than axially oriented, interactions of longitudinal and hoop stresses, and multiple adjacent corrosion patches without empirical criteria.
For the assessment of improved failure criteria, two series of SENT specimens were tested with a systematic variation in the bluntness of the notch from sharp crack to large radii. This was done for a 1960 vintage X52 steel, and a modern X80 steel. The sensitivity to notch acuity was much lower in the modern X80 steel, but both steels had ductile fracture start before maximum load even with the bluntest flaw.
To better assess the driving force, an automated meshing system was pursued, where field laser scans of corroded pipe surfaces were used to created detailed 3D FE meshes of the corrosion profile. One example case was used to assess all the nuances of creating the FE mesh in an automated system, which is quite complicated. The results were used to compare predictions for this case to B31G, Modified B31G, and RSTRENG predictions using the X80 steel pipe properties. In the process of these initial FE analyses, it was found that the assumption used in the B31G type analyses (including RSTRENG) used an equivalent flaw depth with the flaw actual length, whereas the more accurate procedure would be to use the maximum flaw depth with an equivalent length.
The system if developed is intended to be Cloud-based on the Ohio State Supercomputer Center, so that anyone world-wide could conduct such FE analyses without having to have a license for the FE solver software or a supercomputer. Such a validated FE analysis would also a B31G Level 3 analysis.