Here’s a question for you: Do you ask your suppliers the right questions concerning reliability?
If you are getting the right information from your suppliers, then you would enjoy few supplier related field issues, or as little downtime or low warranty costs.
Asking the right set of questions will help you gain the understanding you need to improve your reliability performance.
The 3 Initial Reliability Questions
What will fail?
Rather than ask, “what is the reliability?”, ask if they know how their product will fail.
Capacitors, gearboxes, or anything you purchase for use in your product will fail.
Does your supplier know what will fail?
The intent is to determine the specific failure mechanisms affecting the component or subsystem.
If you know how something will fail, you are then able to:
- Design additional margin into the product to delay or avoid failures
- Focus on critical to reliability process and assembly steps to minimize failures
- Evaluate the impact of the use environment set of stresses on initiation of failures
When will it fail?
Some failure mechanisms lead to failure only when in an over-stress situation.
Other mechanisms accumulate damage over time. Still, others mix the two paths to failure.
The intent here is to understand how the failure mechanism occurs given your design and environmental conditions.
The supplier may have a time to failure model (maybe even a detailed physics of failure model) that allows you to estimate the time to failure for your situation.
Another outcome from this question is knowing when to look for the failure or when to detect precursors to a failure.
If the failures occur early in life for a subset of parts, it may be possible to improve upstream process controls to eliminate or identify faulty parts.
If the failures occur over time dependent on experienced loads, then monitoring loads may forecast expected time to failure for specific units.
This question also helps you evaluate how well the supplier knows how their product will function or fail when in use.
What is your evidence?
Assuming the supplier provides information answering the first two questions, ask for the supporting evidence.
How do they know what they are claiming concerning reliability performance?
If you receive vague responses to the first two questions, you will likely find there is little evidence supporting reliability claims.
If the supplier engages in a frank discussion of failure mechanisms and how they may reveal in your situation, the supplier likely has an extensive body of work to support their claims.
If, as one supplier told me, the reliability claim on the data sheet is pure marketing as all the competitors make the same claim, you may need to explore other options.
Or, you are talking to the wrong person.
If the evidence is a Mil Hdbk 217 prediction or any other generic failure rate database, it’s time to ask them about any real data or evidence they may have instead.
If the evidence includes detailed descriptions of failure mechanisms including elements of material science, assembly and process controls, extensive and long-term life testing, and time to failure models, you have the necessary information to move forward.
The follow-up set of questions
With the initial discussion based on the first three questions, you will likely have more questions for your supplier.
You may want to explore the different classes of failures or the scope of what the supplier evaluated to support their claims.
What kinds of failures should we expect in early life?
Some consider out of box defective parts found during assembly or similar failures to be quality issues and not reliability related.
If it fails for any reasons it is a reliability issue.
The underlying cause may be process control related, yet it led to a failure.
Call it what you may, a 10% defect rate in your assembly process is still a problem. (Of course, your threshold to define a problem may differ.)
The tricky bit here is understanding the source of the early failures.
If the supplier is screening to minimize such failures, you can count of experiencing the failures given no inspection process is perfect.
If they have design and process controls that are stable and capable, that is better.
- Early life failures typically occur when there are:
- Design errors that significantly over stress the part.
- Part, assembly, installation elements that are unstable or incapable of meeting requirements.
- Supplier’s supply chain issues leading to unacceptable variation or latent defects.
- Not all early life failures are due to supplier mistakes, get the understanding to make that decision correctly.
What will cause a premature failure?
This is a follow-up to the question about early life failures.
Get the information on the chain of events that will lead to the expected early life failures.
If the assembly and design are good, what will eventually fail?
The intent here is to explore how the part will fail over time.
Keep in mind that everything fails eventually. If the supplier claims it will not fail, as for the evidence (and check if they have a patent on the never-fail technology).
Everything fails so if they do not know how they are not thinking about the reliability of their product.
If they focus on the early life and design related issues, this question shifts the focus to the longer term failure mechanisms.
This should evoke a discussion that may include:
Expand this discussion to include the expected effects of storage, standby or quiescent modes, full-time active use, or cyclic use patterns.
Get the information you need
Often just asking for failure rate information does not provide suitable information to make informed decisions.
FIT, MTBF, Useful Life and similar single point estimates related to reliability are actually of little use.
Get the details you need to determine if the supplier’s parts are what you need in your design.
Understand how the items will fail both early life and over time.
Gain the knowledge you and your team need to select the right suppliers and to design and build reliable products.
Supplier Management Fundamentals (article)