This is a filled-in three-legged 5 Why for one made-up complaint: weld nuts that spun loose at the customer. It runs one chain for why the defect happened, one for why it was not caught, and one for why the system allowed it. Every answer has evidence next to it, and each leg ends at a cause you can change.
The case: bracket 4410-B is a laser cut, press brake formed steel bracket with an M8 nut projection welded on. The customer bolts it into an assembly. Lot 2610-14, 1,200 pieces, shipped and then came back as a complaint: nuts spinning loose when the bolt was torqued.
Problem statement
37 brackets, part 4410-B, lot 2610-14, with the M8 weld nut spinning when torqued at the customer. Returned parts pushed out at 1.1 to 1.6 kN against a 4.0 kN minimum.
Leg 1: Why did it happen? (occurrence)
| Why | Answer | Evidence |
|---|---|---|
| 1. Why did the nut spin? | The projection welds were too weak to hold the torque. | Push-out on returned parts: 1.1 to 1.6 kN. |
| 2. Why were the welds weak? | Weld current density was too low, because the electrode tip face had grown. | Tip face measured at 9.4 mm against an 8.0 mm limit. |
| 3. Why had the tip face grown? | The tips had made far more welds than usual since their last dress. | Weld counter: about 1,100 welds that shift. |
| 4. Why so many welds between dresses? | Tips are dressed once per shift, by the clock. The run rate nearly doubled three weeks ago. | Work instruction and the new schedule. |
| 5. Why is dressing by the clock? | The interval was set when the cell made about 600 welds a shift and never tied to weld count. | Original setup sheet. |
Root cause: tip dress interval based on time, not weld count. Action: dress every 400 welds with a controller interlock.
Leg 2: Why did we not catch it? (detection)
| Why | Answer | Evidence |
|---|---|---|
| 1. Why did weak welds ship? | Nothing between the first piece and the box tested weld strength. | Control Plan for step 40. |
| 2. Why only the first piece? | The push-out test is done at shift start, right after the tips are dressed. | Test records, all time stamped at shift start. |
| 3. Why does that miss it? | A fresh dress makes the strongest welds of the shift. The test only ever checked the best case. | Trial: 0 of 40 failed after a dress, 9 of 40 failed at end of interval. |
Root cause: the check tests the best-case part, never the worst. Action: push-out test on the last part before each dress, and current monitoring on every weld.
Leg 3: Why did the system allow it? (system)
| Why | Answer | Evidence |
|---|---|---|
| 1. Why was the dress interval not reviewed when volume went up? | Nobody treated the run rate change as a process change. | No change request raised. |
| 2. Why not? | It was handled as a scheduling decision by production planning. | Schedule change email. |
| 3. Why did that not trigger a review? | The change request trigger list covers tooling, material and method, not volume or shift pattern. | Change procedure. |
Root cause: process change triggers do not include run rate. Action: add run rate and shift pattern to the trigger list, with a PFMEA review.
How we knew it was the root cause
We turned it on and off. 40 nuts welded with tips at the end of the old interval: 9 failed push-out. 40 welded right after a dress: 0 failed. The cause explains why earlier lots were fine (lower volume) and why only some parts in the lot failed (the ones welded late in each shift).
The whole investigation is written up as an 8D on the 8D report example.
Everything on this page is made up for the example: the part, the numbers, the customer and the team. Use your own forms, rating scales and customer requirements.
Use these next
- How to do a 5 Why, the method behind this example
- 8D report example
- PFMEA example
- Control Plan example
- Quality glossary, for any term on this page
- Free tool: 5 Why helper
- On Etsy: the 3-Legged 5 Why Analysis, the same form ready to fill in