Failure statistics can be presented in various forms, showing various content. A failure rate may show the rate of defective pieces after a process step. But also, the proportion of reported defects at our end customer.
Depending on the industrial field, organizations need to register both internal and external failures, and to draw an inference from the registered data. Failure statistics (or otherwise called failure rate) are a general wording with multiform meanings. Having the failures and defects properly registered, traced and reported, the company can have many benefits, such as:
- Knowing where to take actions, where to intervene.
- Seeing where we are compared to our goals and what our real quality performance is.
- Transparent information to the management for further decisions.
The ISO/TS 16949:2009 (and the newer IATF 16949:2016 require not only to have management reviews (based on ISO 9001), but also to trace the Cost of Poor Quality (CoPQ), which is coming mainly from warranty claims and internal rejects. Without tracing failure statistics, it is impossible to get information about CoPQ.
Examples for failure statistics and failure rates:
- Internal Reject Rate (IRR) of a manufacturing line.
- Fall-off Rate (FOR) of a manufacturing process.
- 0km failure rate of a product family.
- Field failure rate of a given model year.
- Share of defect causing field (Pareto).
It depends on the company (and its customer) how the available data is used, traced and reported. Some examples from the automotive industry:
Example for various failure statistics | ||
| Rate | Formulae | Example |
| Internal Reject Rate (IRR) | Rejected parts / All initiated parts * 100% | 300 parts was initiated on the manufacturing line, but 15 were rejected for various reasons at various stations, which means an IRR of 5%. |
| 0km failure rate | Claimed parts / Shipped parts * 1 000 000 | We shipped 1 200 000 products this year, and 12 pieces were claimed with defect. All defects were accepted, and our 0km failure rate is now 10 ppm. |
| Pareto of field claims by causing fields | – | 37% – supplier related 32% – manufacturing related 17% – development related 8% – no failure (product conforming to the spec.) 5% – failure caused by the customer (rejected claim) 1% – unknown root cause (under analysis) |
Tracing the same quality metrics as the customer does is key. You can save a lot of time, by being on common ground with the customer.
Example (see table below): both of the supplier and the customer calculates 0km defects in ppm.
The frame of reference differs, as the customer calculates based on the failure date (observation), while the supplier calculates based on the manufacturing date. On top of that, they use different basis for their calculations (quantity produced vs quantity received).
In this case, the numbers will differ, generating misunderstandings. You should be on common ground with your customer!
Supplier’s vs Customer’s failure calculation | |||
| January | February | March | |
| Supplier’s calculation | |||
| 0km defects based on manufacturing date | 2 pcs | 1 pc | 0 pc |
| Quantity (produced) | 200 000 pcs | 240 000 pcs | 200 000 pcs |
| Defect ratio | 10 ppm | 4.2 ppm | 0 ppm |
| Customer’s calculation | |||
| 0km defects based on failure date (observation) | 0 pc | 2 pcs | 1 pc |
| Quantity (received) | 180 000 pcs | 250 000 pcs | 210 000 pcs |
| Defect ratio (ppm) | 5.5 ppm | 8 ppm | 4.8 ppm |
Conclusion
- Failure statistics (or otherwise called failure rate) are a general wording with multiform meanings.
- A failure statistic can be presented in various forms, showing various content. A failure rate may show the rate of defective pieces after a process step, but also the proportion of reported defects at our end customer.
- Failure statistics are usable for knowing where to intervene, seeing where we are compared to our goals and providing transparent information to the management for further decisions.