The customer asks about a serial number
A customer gets in touch: a part with a certain serial number has failed in the field. The customer wants to know whether it is a single failure or whether more parts may be affected, and wants an answer within a couple of days.
To answer, you need to know which order the part was made in, which material batch it came from, in which machine and with which tool it was machined, which measurements were made and what they showed, who measured and with which gauge. If all of that is linked to the serial number, you can say within a few hours which other parts share the same conditions. If it is not, you have to guess, or treat everything made over a long period as suspect.
What traceability covers
Traceability is usually divided into three parts that belong together:
- Product traceability: which part it is, and which part number, drawing revision, order, batch or serial number it belongs to. It lets you follow a part forward to the customer and back to the material.
- Process traceability: how the part was made. Which machine, which tool, which program or setting, which shift, and whether anything deviated during the run.
- Measurement traceability: how the part was inspected. Which characteristic was measured, against which requirement, with which gauge, by whom and when, and what the result was.
The word traceability is also used in metrology, for the unbroken chain of calibrations from a gauge to national standards. That is a different thing, but the two meet here: for a measurement to be credible, you need to be able to show that the gauge was calibrated and approved when the measurement was made.
What must be recorded when it happens
Most of the context only exists in the moment. The operator knows who she is, which gauge she has in her hand and which order is running. A week later nobody knows for sure. So the context must be captured when the measurement is made, not when someone asks.
A measured value without context can only be compared with its limit. With the context you can answer questions nobody had thought of asking when the measurement was made. Is it one particular cavity that deviates? Does the fault only concern one material batch? Were all the parts measured with the same gauge?
A few things make it easier to get the recording right:
- Let the context come along automatically where possible, for example order and part number from planning and the gauge from its identity, instead of being typed in by hand.
- Record deviations and actions against the measured value they concern, with time and name. A note in a separate binder can rarely be linked to the right part afterwards.
- Keep the original value. If a value is corrected, both the original and the new value should remain, with who changed it and why.
- Decide how long the data should be kept. Customers in the automotive industry, for example, often specify long retention periods.
How fine should the traceability be?
Traceability costs. Recording a serial number for every part is reasonable for a safety-critical component, but may be unnecessary for a simple washer where a batch is enough. The question to ask is what happens if something goes wrong: how small a quantity do you want to be able to contain the problem to? The finer the traceability, the less has to be blocked and sorted when there is a fault, but the more has to be recorded every day.
A good starting point is to go through the customer’s requirements and the characteristics that are especially important in the FMEA and the control plan, and decide the level of traceability from them.
Requirements from customers and standards
ISO 9001:2015 (clause 8.5.2) requires the organisation to identify its products and their status with respect to inspections, and to control the unique identification and retain the documented information needed when traceability is a requirement. Clause 7.1.5 also requires that the calibration status of measuring equipment can be determined.
IATF 16949:2016 (clause 8.5.2.1) goes further. The organisation shall analyse the internal, customer and regulatory requirements for traceability, and develop a plan for how they will be met, including for parts with safety or regulatory requirements.
The customer’s own requirements may be stricter than the standards, both on what is to be traced and how long it is to be kept. Always read the customer’s requirements.
Summary
Traceability is not something you build when the auditor or the customer asks. It is the result of the context being recorded every time a part is made and measured. When it is in place, a complaint becomes a contained investigation instead of a broad block, and a measured value becomes evidence instead of just a number.
