ANSTO's research capabilities, led by the OPAL nuclear research reactor and associated instruments provide access to users investigating areas as diverse as materials, life sciences, climate change and mining/engineering.
Effects on Fracture Behaviour in a Welded Branch Connection
Many gas pipelines have connections made to allow a branch to feed new facilities, especially gas-fired power stations. The connections are often made on in-service pipelines by hot-tapping, welding directly onto a pressurised pipeline.
The branch is welded directly to the live pipeline and then reinforced with 2 half sleeves cut longitudinally from a section of pipe, with a hole made in the side of one half (figure 1). A special shutoff valve is attached to the branch, the branch is pressurised and then a cutter removes a coupon of metal from the main pipe.
The cutter is withdrawn through the valve and the valve is shut, and then attached to the new branch pipeline.
 |  |
| Figure 1 (left): Pipe, branch welded on, components of reinforcing sleeve, and completed branch connection, welds are shown with bold lines.Figure 2 (right): Samples for measurement were cut from this branch connection made during welding qualification trial. |
The welding process generates significant residual stresses which may contribute to premature failure by fatigue, stress corrosion cracking (SCC), hydrogen assisted cold cracking (HACC), fracture, or lead to unacceptable deformation.
Residual stress may alter the critical flaw size for crack growth. The branch pipe has a total thickness of 28.2mm, more details are available here.
Experiments were done on Kowari using a monochromatic neutron beam with Λ = 1.67 Å. This combination of wavelength and scattering planes resulted in a scattering angle of 90o. A nominal gauge volume of 2 x 2 x 2 mm3 was used to perform through-thickness scans at. A section through the weld and parent was cut into a 6 mm thick plate to derive do values across the weld, heat affected zone (HAZ) and parent.
Stresses were determined in the saddle, an area where those residual stresses will not be relaxed by any future hydrotesting. The maximum stresses (290 MPa) were along the weld direction (figure 4) with lower stresses both in the direction of the branch.
 |
| Figure 3: Stresses in direction of weld (out of page), the maximum measured stress is 290 MPa, the interpolation plot incorrectly shows a higher value. |
The branch can fail by plastic collapse or by brittle fracture, these failure mechanisms are modelled separately. The weld defect modelled is a flaw through the entire weld, aligned axially to the run pipe, situated at the weld crotch (figure 5).
 |
| Figure 4: Axially aligned full section weld defect in crotch |
If this vessel were to be hydrotested the residual stresses would be reduced. The reduction is more significant at the crotch, where the weld is parallel to the major hoop stress. Thus the residual stresses were measured at the saddle. The stresses are taken at the crotch, where pressure induced stresses are highest.
Combining these two stresses is a conservative method of assessment. A single fracture mechanics case is presented, that of a through wall cracking equivalent to the full section weld defect in crotch. This is a very conservative model as weld cracking will only penetrate the outer sleeve (a possible 14 mm deep crack, not a full thickness 28 mm crack).
The inner sleeve has higher toughness (80 J vs. 40J) and limited residual stresses as it has not been welded. Additionally, the stresses reduce away from the branch weld, while the analysis assumes they are constant. These conservative results are shown in figure 6.
 |
| Figure 5: Critical crack size with different residual stress values. |
The critical crack size allowable increases even with very conservative estimates of residual stresses based on neutron measurement. The branch was shown to be safe as measured by neutron diffraction using ANSTOs residual stress diffractometer, Kowari.
Reanalysing with the actual residual stresses significantly increased the critical crack size using very conservative estimates of residual stress based on these measured values. The accuracy and good spatial resolution of the neutron measurements makes this a useful integrity assessment and research tool.