Scrap that nobody could attribute to a cause
Scrap was measured per shift but never linked to machine, material lot or operator. We joined the three and found two thirds of it came from one upstream setting.
- Scrap rate
- -27%
- Material saved
- $780k/yr
- Lots traceable
- 100%
- Time to first finding
- 4 wks
The situation
Bramwell measured scrap accurately at the end of each shift and could do nothing with the number. It was not linked to the machine that produced the part, the material lot it came from, or the settings in use at the time.
Every quality investigation therefore started from scratch.
Constraints
The quality team had been promised analytics twice before and had stopped engaging. Anything we built had to produce a usable finding within the first month or it would be ignored like the previous attempts.
What we did
We joined scrap records to machine telemetry and material lot data, which required fixing timestamps that three systems recorded in two different time zones.
The first finding arrived in week four: roughly two thirds of scrap on one line traced to a single upstream temperature setting that had been changed during a maintenance visit eleven months earlier and never changed back.
What came next
We expanded the reason code taxonomy from six to nineteen with the quality team, so future investigations start from something more specific than “material”.
The scrap reduction is real, but the durable change is that a defect can now be traced to a lot and a machine in minutes.
Results
| Measure | Before | After |
|---|---|---|
| Scrap rate | 5.2% | 3.8% |
| Lots with full traceability | 0% | 100% |
| Time to trace a defect | 2 days | Minutes |
| Reason codes in use | 6 | 19 |
What we left behind
- A project in the client's own repository, with tests and CI
- Documented definitions with named owners
- Runbooks for every scheduled job, written for a non-engineer
- Two weeks of training for the team taking it over
Other engagements you might want to read
Tell us which number you cannot trust
Send one paragraph about the report that always starts an argument. A data lead, not a salesperson, replies within one business day.