Run three separate chains. Why did it happen, why was it not caught, and why did the system let an unproven cause close? Write the evidence beside every answer. Prove the occurrence cause by turning it on and off: change one thing, run, and watch the defect follow it. Stopping at the first leg lets the next defect through.
Wednesday, 7:30, the die room. Die D-3082 is on the bench. A tooling lead and a quality technician stand at the bench with the old 8D, which says the cause was the operator. Nobody in the room believes that any more.
A repeat defect needs three questions, not one. Why did it happen. Why was it not caught. Why did the system let the first answer close. Run each as its own chain. Book 1 covers the basics in Find the Cause; this chapter is about the legs that were skipped.
Leg one: occurrence
- Why was the second hole too far from the first? The strip moved in the die before the hole was pierced.
- Why did it move? The pilot that locates the strip was worn. It measures 3.940 mm. New is 4.000 mm.
- Why was it worn that far? It stayed in the die past its life. Nobody could say when it was last changed.
- Why was there no record? The preventive maintenance list covers the press and the die as a whole. The pilot is not on it.
- Why is it not on it? Nobody owns wear parts inside a die. That is a cause you can control.
Prove it by turning it on and off
This is the test the old 8D never ran. Put three pilots in the same die, one at a time: 4.000 mm, 3.970 mm and the worn 3.940 mm. Run 2,400 pieces each. Read the hole spacing at pieces 1, 600, 1,200, 1,800, 2,400.
All three read 38.03 on the first piece. That is why the first piece check passes. By piece 2,400 the new pilot has moved 0.02 mm. The 3.970 mm pilot has moved 0.08 mm. The worn pilot has moved 0.21 mm and reads 38.24, which is 0.04 mm over the limit. Turn the cause on and the defect appears. Turn it off and it goes away. Now the cause is proven, not stated.
Leg two: escape
- Why did the defect reach the customer? Nobody checked spacing late in the run.
- Why not? The control plan has one spacing check, on the first piece. The final sample uses a plug gage on hole size.
- Why does that miss it? A plug gage reads the diameter of a hole. It cannot read how far apart two holes are. Both holes pass.
- Why was it chosen? It was already at the station. Nobody asked what it measured.
Leg three: system
Why did a root cause that was never proven get approved and closed? The 8D form had no field for a test. The closing review checked that every box was filled. It did not check that the cause was true. And the effectiveness check, no complaints for 30 days, was shorter than the 8 weeks between long runs, so it could never have seen the failure. That is a system cause, and it is the one that will let the next repeat through.
Read more: how to do a 5 Why, a worked 5 Why example and the free 5 Why helper.
The mistake to avoid
Stopping at the first leg. You can prove the occurrence cause perfectly and still ship the next defect, because the gage that missed it is still at the station. Run all three, even when the first one feels finished.
- Ask the escape question on a blank page.
- Write what each check can and cannot read.
- Add to your root cause step: how was it proven?
Open the printable Monday Board
Related on ShopFloorOps
The part, the die, the lots and every number in this book (sensor bracket BK-6139) are a made-up example. The tool sidebars show ShopFloorOps tools, which I make and sell. Every chapter teaches the step itself, so you can use your own forms. New to a term? See the quality glossary.