Batch Rejects on a CNC Production Line
A manufacturer produces 2,000 aluminium brackets per week on a CNC machining centre, then deburrs and anodises them. Over three weeks the customer rejects 4% of brackets for hole position error exceeding 0.3 mm, against a tolerance of 0.15 mm. The quality officer's inspection records show rejects clustered at the start of each shift and after lunch breaks. Operators say the machine 'settles in' and that they check the first article only at the start of the day. Coolant concentration has not been tested for two months. The customer is now demanding a formal corrective action report and has threatened to move the contract, which is 30% of the plant's revenue.
1. Analyse the likely causes of the rejects and distinguish between special and common causes.
Read the worked answer
The clustering pattern is the key evidence. Rejects concentrated at shift start and after breaks point to special causes tied to machine state after idle periods: thermal drift as the machine warms up, coolant that has separated or changed concentration during downtime, or fixtures that shift when re-clamped. The untested coolant is a plausible contributor to dimensional variation (poor cooling changes thermal expansion and chip clearance).
Common-cause variation would show a stable scatter across the whole shift; this pattern does not.
A sound analysis would:
- Plot rejects against time of day and machine state to confirm the pattern.
- Measure coolant concentration and temperature at start, mid-shift and after breaks.
- Check whether first-article inspection is done only once daily (it is) and whether warm-up cycles are run before production.
- Verify fixture clamping force and tool wear records.
The distinction matters because the corrective action differs: special causes are fixed by changing the warm-up and inspection regime and restoring coolant control, whereas common causes would require process capability work. The 0.15 mm tolerance against a 0.3 mm observed error also suggests the process may not be capable even when stable, so capability should be checked after the special causes are removed.
2. Draft the substance of the corrective action response the plant should send the customer.
Read the worked answer
The response should be structured, evidence-based and committed to verification.
1. Containment (immediate): 100% inspection of current stock and work-in-progress; quarantine suspect batches; notify the customer of affected serial numbers.
2. Root cause: state the findings, e.g. thermal drift during idle periods combined with degraded coolant control, and the absence of warm-up cycles and start-of-shift first-article checks. Avoid blaming operators; describe the system failure.
3. Corrective actions:
- Mandatory warm-up cycle before production each shift and after breaks.
- First-article inspection at every shift start and after any stoppage over a set duration, with recorded results.
- Coolant concentration and temperature tested and logged daily, with a control chart.
- Fixture and tool-wear checks added to the schedule.
4. Verification: monitor rejects for a defined period (e.g. four weeks) and report the rate and capability data to the customer; audit the new checks.
5. Prevention: update the quality manual and train staff; add these checks to the standard operating procedure.
The judgement being tested is that a corrective action must address containment, cause, correction, verification and prevention, and must be credible to a customer whose contract is at stake. Promising 'greater care' without measurement would not satisfy the customer or fix the process.