A quality control brainteaser from the Eng-Tips engineering forum.
Suppose you’re buying a billion pieces of a given product per year with break strength as your main quality criterion. You use Acceptable Quality Level (AQL) sampling and reject any lot with more than your threshold for failures.
What should you do if you want to reduce the defect rate?
This was the question posed by a member of the Eng-Tips engineering forum, and the resulting discussion touched on issues of root cause analysis, quality control metrics and supply chain management.
Reducing the Defect Rate in Destructive Testing
The most obvious answer to the question above is to lower the AQL threshold. However, as most engineers have learned one way or another, the most obvious answer isn’t always the best one. With so many statistical tools and metrics available to quality engineers, it’s worth considering other options.
But first, some additional context for the problem:
- Although the average value of break strength is 6 sigma better than the minimum specification, the defect rate is still high, with 20 percent of the shipments having a 25 percent failure rate, while the majority of shipments have a 0-10 percent failure rate.
- The cost of the product (a commodity fastener sold at retail) is low, but the cost of testing and rejected shipments is high.
- Visual inspections or other methods cannot be substituted for the destructive test.
The high cost of destructive testing and inability to substitute it with another inspection method is what drove the Eng-Tips member to seek further opinions. As it turned out, the other forum members unanimously agreed that focusing solely on quality control measures was not the ideal solution.
Instead, they suggested identifying and addressing the root cause of the high failure rate. The process allowing faulty parts to be made is what needs to be fixed. Hence, the initial advice was to work closely with the supplier and conduct a comprehensive root cause analysis to determine the underlying reasons for anomalous failures.
This is good advice, but it doesn’t answer the original question.
Exploring Different Quality Control Metrics
Supplier issues aside, the non-normal data distribution observed in the destructive testing calls the adequacy of the existing statistical metrics into question. One potential alternative is to use proof testing, where the criterion is whether the test article withstands a given load without failing. In this case, the test load would need to be high enough to break the unsatisfactory fasteners but low enough to leave the good ones intact.
Whether a proof test like this is possible depends on the details of the product’s design, but if it is, it could also be automated to test any arbitrary number of incoming parts, or even all of them. Since the issue seems to be with the supplier, this is the only way to reduce the defect rate that’s apparent to customers with any certainty, though if the ultimate issue is that the supplier’s process is out of control, there may not be enough fasters left to ship to the customer.
Incentives in Supply Chain Management
With the supplier’s own quality process identified as the root cause of this issue, the discussion turned to questions about incentivizing better supplier behavior. The original poster admitted that their company is using multiple suppliers, and expressed the desire to introduce a bonus system for quality improvements.
The response was that grading suppliers based on reject rates and returning defective parts to suppliers with higher failure rates was the better option, rather than trying to use incentives to coax better results from suppliers that are under-performing. Ultimately, what’s needed in cases like this are stronger contractual terms and consequences for consistently supplying defective parts.
A Holistic View of Quality
The real lesson here is that controlling failure rates in destructive testing requires a holistic approach that goes beyond traditional quality control methods. While AQL sampling has its place, it’s no substitute for process improvement, root cause analysis and supplier management.
Metrics like proof testing and tighter acceptance criteria can help reduce failure rates, while greater collaboration between the purchaser and supplier, coupled with effective incentives and consequences, should foster improvements in product quality.
By implementing these suggestions, organizations can strive for lower failure rates, reduced costs and improved customer satisfaction.
