Porosity Spike After a Shielding-Gas Changeover
A GMAW cell runs 1/4-in. A36 brackets with 0.035-in. ER70S-6 at 26 V, 260 A, and 35 cfh of 75% Ar / 25% CO2. The reject rate from visual plus 10% PT sampling held at 1.4% for eight months. In the three days since the gas supplier changed out the cylinder manifold, it has jumped to 11.3% (27 of 239 parts). Every reject is clustered in the first inch of the weld or at a restart; mid-bead porosity is rare. Two nozzles in service are heavily spatter-caked, one gun liner is kinked, and the new manifolds arrived without a dew-point certificate. The scheduler has 640 brackets due Friday and proposes raising flow to 50 cfh and "running through it." The welders were performance-qualified on this WPS 14 months ago.
1. What is the defect distribution telling you, and in what order do you run the checks?
Read the worked answer
The distribution is the strongest clue. Porosity confined to arc starts and restarts points to a gas-coverage transient at initiation, not to moisture in the base metal or electrode - that would scatter porosity along the whole bead.
Rank the causes accordingly:
- Delivery system - a kinked liner or a spatter-blocked nozzle chokes flow at the cup even when the regulator reads 35 cfh.
- Gas supply - an unpurged or leaking manifold, or off-spec dew point, adds moisture right at the changeover point where the timing fits.
- Parameters - pre-flow time too short relative to wire run-in.
Test sequence: measure flow at the nozzle with a calibrated flowmeter (not at the regulator), pressure-test and purge the manifold, replace liner and nozzle, then check pre-flow and run-in settings. Change one variable at a time and run a 20-part check after each, so you don't credit the fix to the wrong change.
2. The scheduler wants to ship. How do you disposition the 239 parts already made?
Read the worked answer
Refuse the run-through. Everything made since the changeover is a defined suspect lot, and the 1.4% historical baseline is not evidence that an 11.3% lot is acceptable.
Disposition:
- 100% visual on all 239, with PT concentrated on the start/restart zones where the defect lives.
- For surface-breaking porosity, remove to sound metal and repair under a qualified repair procedure, then re-inspect; budget for distortion and the repair hours, not just the weld time.
- Hold the 640 until a 20-part verification run returns to baseline, then release under normal sampling.
Document the hold with dates, cylinder and manifold IDs, and part serials. That record is what makes a supplier claim defensible and what lets you prove the correction worked.
3. Evaluate the 50 cfh proposal on technical and cost grounds.
Read the worked answer
Reject it on both counts.
Technical: 50 cfh is outside the flow range the WPS was qualified at. Exceeding a qualified range makes the procedure unqualified, and higher flow at the nozzle exit promotes turbulence that can aspirate air and worsen porosity. Flow is a diagnostic, not a cure.
Cost: moving 35 → 50 cfh is 43% more shielding gas on the cell for zero benefit, and it buys you a hidden root cause that will reappear at the next cylinder change.
The honest answer to the scheduler is a schedule, not a shortcut: 24 hours to isolate the cause, a verification lot, then a decision on Friday's 640 based on data.