Thirty Rejected Brackets and a Supervisor Who Wants a Warning Letter
At a sheet-metal unit near Aurangabad, a batch of 250 MS mounting brackets was drilled on Monday morning. Quality control rejected 30 pieces because the mounting holes were 1.5 mm off-centre, outside the plus or minus 0.5 mm tolerance. Every rejected piece came from the first three hours of the shift. The drill jig had been re-shimmed on Sunday evening by a newly joined trainee who was trying to remove play in the clamp. On Monday the operator used drawing revision B, which moved the hole centre by 2 mm; only two of the five drawing copies in the shop had been updated to revision B, and the trainee had used an older copy. The operator, Rafiq, has 11 years' experience and says the jig 'looked fine'. The production supervisor wants the 30 pieces scrapped and Rafiq given a warning letter so the batch can be closed before the customer's Friday dispatch.
1. Build a 5-Why chain to the root cause. Explain why stopping at 'operator error' would be wrong in this case.
Read the worked answer
5-Why chain
- Why were the brackets rejected? The mounting holes were 1.5 mm off-centre.
- Why were the holes off-centre? The jig was set to the old hole position, not to revision B.
- Why was the jig set to the old position? It was re-shimmed using an outdated drawing copy.
- Why was an outdated copy available and used? Drawing revision B had been issued to only two of five copies, old copies were never withdrawn, and the person setting the jig was an untrained trainee working unsupervised.
- Why did that go unnoticed before the batch ran? There was no first-piece inspection and approval before production, and no authorised jig-setting record.
Root cause: a drawing-control and first-piece-approval failure - not an individual's carelessness.
Why 'operator error' is the wrong stop If the analysis stops at Rafiq, the action is a warning letter, and nothing changes: three of the five drawing copies in the shop are still old, the jig can still be re-shimmed by anyone, and the next batch will repeat the defect with a different operator's name on it. 'Operator error' is a symptom that ends the investigation early. The test of a good root cause is simple: if this person were replaced tomorrow, would the problem disappear? Here it would not, because any competent operator would have used the paperwork available to him. Also note Rafiq's 11 years of experience - an experienced operator failing is a signal that the system gave him the wrong input, not that his skill vanished.
2. Distinguish containment actions from corrective actions for this case, giving two of each with timing.
Read the worked answer
The distinction Containment protects the customer and stops the defect spreading today. Corrective action stops it happening again. Containment without correction means you will contain the same problem next month; correction without containment means bad parts may already have shipped.
Containment (today, before the shift ends)
- Quarantine and segregate. Tag and physically separate the 30 rejected pieces and hold the rest of the batch. Check whether any of the 30, or any part drilled with the same jig setting, has already moved to the next operation or to another customer. If it has, that customer must be told.
- Stop the drill and verify. Halt drilling on that jig until a competent person has checked the setting against revision B. Then do 100% inspection of the 220 remaining pieces and of any other batch drilled since Sunday evening, and re-check the jig dimension with a gauge, not by eye.
Corrective action (within the week)
- Drawing control. Establish one master drawing per part, stamp the revision, withdraw and destroy the three old copies, and issue revisions against signature so no one can hold an uncontrolled copy on the shop floor.
- First-piece approval and authorised jig setting. No batch starts until the first piece is measured and signed off by a supervisor. Jig setting is done only by an authorised person, and the setting is recorded - who set it, to which revision, and when. Add training and supervision for the new trainee, and add a jig-check step to the machine's daily checklist.
Note the sequencing: the warning letter is neither of these. It changes no dimension and prevents no recurrence.
3. The Friday dispatch needs 220 good pieces and only 220 good pieces remain. Recommend what the supervisor should do and what he should tell the customer.
Read the worked answer
Recommendation
- Do not scrap the 30 yet. Measure them properly. An off-centre hole of 1.5 mm on a bracket is often recoverable - weld and re-drill, drill oversize and fit a bush, or use a slotted washer if the design allows. Any of these recovers 30 pieces at a fraction of the cost of 30 new ones, and it is a decision the customer's engineer can approve in a day.
- Get written approval for the rework method or a deviation before touching the pieces. A repaired part that the customer has not approved becomes a second rejection and a credibility problem.
- Protect the Friday dispatch with the 220 good pieces produced after verified first-piece approval, and 100% inspect them so the doubt does not travel to the customer.
- Tell the customer today, not on Friday. Waiting is the mistake that turns a quality problem into a trust problem.
What to say to the customer 'The first batch of mounting brackets had a hole-position deviation caused by an outdated drawing copy in our shop. We have quarantined the affected pieces and verified the jig against the current revision. 220 pieces will be inspected and dispatched for Friday. The remaining 30 can be recovered by re-drilling oversize and fitting a bush, or we can remake them; we are sending you the method for approval today and will confirm the quantity and date on Friday. We will also send you our first-piece inspection record with the dispatch.'
On the warning letter: it should not be issued. Rafiq used the paperwork he was given, the trainee was unsupervised, and the drawing control was defective - all management responsibilities. Disciplining him would also guarantee that the next operator who spots a problem keeps quiet about it, which is a far costlier outcome than 30 brackets.