Repair Rate Triples in Week Three
Repair rate on a structural job has risen from a 4% baseline to 18% in week three. The work is 600 ft per week of 5/16 in fillet, FCAW-G, 0.052 in E70T-1, with three experienced welders plus two new hires who started in week three. Rejects are mostly elongated slag and lack of fusion on the second pass of three-pass fillets. They cluster between 1 pm and 4 pm, and disproportionately on machine #3, a 12-year-old 350 A unit with a 100 ft #2 cable coiled on a steel hook and a work lead clamped to a painted surface. The new hires were cleared by the lead hand after he watched one bead. Wire is a new lot. The shop reaches 95 °F in the afternoon, doors are open, and fans blow across the work. Rework runs about $22/ft including NDT.
1. What do you do in the first hour, and what data do you collect?
Read the worked answer
First hour
- Hold the suspect work and tag it; do not let it move to NDT as if it were normal production.
- Take machine #3 out of service until it is checked. Fix the obvious: clamp the work lead to clean bare metal, uncoil and inspect the cable, check the lugs and ground.
- Check both new hires' WPQs against the WPS and the code. A lead hand watching one bead is not a qualification; if they are not qualified for the process and position, that is a stop-work finding independent of the repair rate.
- Do not open a disciplinary process. Nothing here yet points at attitude.
Data
- Reject Pareto by welder, by machine, by hour of shift, by pass number, for the two weeks before week three and for week three.
- Measured current and voltage at the stinger under load on each machine, with a calibrated meter, not the dial.
- Gas flow at the nozzle, and whether fans or open doors disturb the shielding.
- Wire lot certificate, storage condition, and whether the lot changed on all machines or only some.
- WPS parameters actually in use at each station.
2. Rank the probable causes and explain how you would separate them.
Read the worked answer
Candidate causes and how they would show up in the data
- Machine and cable: a painted clamp plus a coiled cable raises resistance and inductance, so the meter can read in range while the arc current is low. Expect rejects on machine #3 across all welders and all hours. That is a cold, poorly penetrating second pass, producing exactly the lack of fusion and slag seen.
- Shielding disturbed by the fans and open doors in the afternoon: expect rejects across all machines after about 1 pm, not tied to one machine. This also fits the time-of-day cluster.
- New-hire technique: expect rejects concentrated on the two new hires, on any machine.
- Wire lot: expect a step change on the day the lot changed, on every machine.
- Heat and fatigue at 95 °F: expect a general afternoon drift, and it compounds the others.
Separating them The machine and the time of day both point to the afternoon window, and the new hires started the same week, so the variables are confounded. Stratify: if rejects on machine #3 come from experienced welders too, the machine is the driver. If new-hire rejects appear on every machine, technique is a second, independent cause. Only after stratification can you allocate effort.
3. Quantify the cost of the problem and set corrective actions with exit criteria.
Read the worked answer
Cost
- 18% of 600 ft = 108 ft/week rejected
- Direct rework: 108 x $22 = $2,376/week, before lost arc time and schedule effects and before the risk of an escaped defect reaching the owner's inspector. At a 4% baseline the same figure would be 24 ft, about $528/week; the excess is roughly $1,850/week.
Actions
- Machine #3: replace cable and lugs, clean grounding point, calibrate output, return to service only after a measured check on a test coupon. Owner: maintenance, 48 h.
- All machines: measure output at the stinger on a weekly schedule.
- Fans and doors: reposition or screen so shielding is not cross-dragged; add hydrating breaks and cooling in the afternoon.
- New hires: verify qualifications and pair each with an experienced welder; monthly parameter audits.
- Production control: parameter log per shift per machine, and a weld map so a defect can be traced to a machine and an hour.
Exit criteria 100% visual plus 10% UT hold for two weeks, released only when the weekly repair rate falls below 5% for two consecutive weeks with the machine and hour stratification flat.