A May 2026 analysis by Therness (process piping fabricator) tracked 19,000+ production welds across six months of ASME B31.3 shop fabrication. The numbers are worth reading:
**Overall rejection rate: 3.2%**
Of 19,000+ welds, 3.2% failed initial inspection (visual + RT). Not catastrophic, but in a 10,000-weld project, that is 320 rework cycles. At $150–$400 per rework weld (cut-out, re-bevelling, re-welding, re-NDT), the cost adds up fast.
**Rejection by joint type:**
- Girth welds (pipe-to-pipe): 1.8% failure
- Branch welds (stub-on, weldolet): 5.7% failure —> 3× higher than girth welds
- Nozzle / socket welds: 3.1% failure
Why branches fail 3× more? Two reasons the Therness analysis identifies:
1. Restricted access for welding and grinding of branch welds often sit in tight configurations
2. Fit-up variability of branch saddle prep is more sensitive to fit-up tolerance than simple butt joints
**Rejection by defect type (ASME B31.3):**
- Lack of fusion: 52% of all rejections — more than half
- Porosity: 18%
- Slag inclusion: 16%
- Undercut: 9%
- Cracks: 5%
The finding that matters: lack of fusion dominates regardless of welder skill level. This is a process parameter problem (travel speed, torch angle, interpass temperature), not a skill problem.
**From the Therness report:**
"The branch-to-run ratio of 3:1 rejection suggests that current WPS parameters developed for girth welds are being used on branch welds without adjustment. Branch geometry changes heat-sink behavior, weld pool fluid dynamics, and access. The WPS must account for joint geometry, not just material thickness."
Source: Therness — ASME B31.3 Process Piping Inspection Study, May 2026.