The Recall That Needed Data: Traceability, Not a Quality Issue
A batch of parts came back from the field — a batch of 500 had a missing screw. The customer needed a recall. But the machine didn’t track which parts were in that batch. The operator knew the range (serial numbers 12000-12500), but the PLC didn’t log it. The recall was 5000 parts instead of 500. This is about traceability, not about the quality failure itself.
What traceability means
Every part that leaves the machine should have a record: which serial number, which batch of raw material, which operator, which machine settings, which cycle count. If a defect is found, you can trace it to the exact batch and recall only the affected parts. Without traceability, you recall everything.
The data to log
| Data point | Why it matters |
|---|---|
| Serial number | Identifies the part |
| Timestamp | When it was made |
| Raw material batch | Which lot of material |
| Machine settings | Pressure, temperature, speed |
| Cycle count | How many cycles since PM |
| Operator ID | Who ran the machine |
| Test results | Force, vision, leak test |
What we did
1. Added a serial number scanner. Every part has a barcode. The scanner reads it at the start of the cycle. The PLC logs it. The part’s record starts with its serial number.
2. Logged the test results. The press force, the vision pass/fail, the leak test — all logged against the serial number. If a part fails in the field, we look up its test data and see if it was within spec.
3. Logged the machine settings. The PLC logs the pressure, temperature, and speed at the start of each cycle. If a defect is found, we know the settings at the time. If the settings drifted, we can trace it.
4. Exported to a database. The PLC writes to a SQL database (via an OPC UA server). The data is queryable. We can look up serial number 12345 and see every parameter at the time it was made.
The recall cost
The original recall (without traceability) covered 5000 parts. With traceability, we found the defect was in serial numbers 12000-12500 (500 parts). The recall was 10x smaller. The traceability system paid for itself in one recall. The cost of the system (PLC logging, barcode scanner, database) was $3000. The cost of over-recalling 4500 extra parts was $45,000.
The traceability rule: Log every part: serial number, timestamp, material batch, machine settings, test results. Store in a queryable database. When a defect is found, trace it to the exact batch. Without traceability, every recall is a full recall. The system costs thousands; the over-recall costs tens of thousands.
