Beyond understanding what will fail (failure mechanisms) we need to know when a system will fail. Both kinds of information can be found in a variety of sources.
Unlike the stock market where we are regularly warned that past performance does not indicate future performance; in reliability engineering past performance is the best predictor of future performance.
Looking for failure data
Sure historical data is not perfect and as customer expectations, use, and environmental stresses changes so will product failure rates.
The absolute best reliability data is the lifetime and failure mechanism information for the product in customer applications and use. I’ve heard that the best way to estimate reliability is to ship and track all your products till they fail, then you will know their reliability performance.
Unfortunately, we nor our customers have the patience to experiment with reliability performance in this manner. We have to make decisions today about reliability performance in the future.
Field Data
This is by far the most valuable data.
While not generally available for a startup creating their first product, most organizations are working on variations of existing products. Even if you are working on your first product getting it into the hands of customers and tracking the reliability performance will provide useful information.
Warranty returns may provide time to failure information, along with failure analysis provides failure mechanisms.
Customer complaints may provide failure mode information and with a little work may provide information on failure mechanisms. Customer calls may include questions on the operation or maintenance of a system, which may not appear with direct product returns. A customer that is unable to use or is frustrated by a product even though the product operates as designed is not a trouble found issue, it is a human factor reliability failure.
Field representative information may provide insights on the full range of failures of a product. Not all product have field representatives, yet those that have, have the unique ability to gather use, environment, expectations, and detailed feedback about a product’s performance. It may take some training and encouragement to enable the representatives to collect and report this information, yet having their insights may provide insights for reliability improvements before field failures occur.
Distributor/Dealer information may provide similar information as a field representative. Similar to field representatives they can provide useful information, yet they may not have the same focus as personal within the organization. As with representatives, we may have to provide training and encouragement to gather and report reliability pertinent information.
Industry Data
Some industries share reliability information. Most do not. One example is the Telecordia electronics failure rate database. While focused on the telecommunications industry, the same basic electronic components are widely used in similar environments.
In recent years the gathering and reporting of component-level failure information have diminished as the cost of component failure analysis has become less cost effective in relation to simply replacing items at the circuit board level.
Public Data
GIDEP (Government-Industry Data Exchange) is a program to gather and share reliability information. Participating companies provide information and in return have access to the databases of failure information. (During my work within a corporation, our lawyers advised against participating as it may lead to ‘issues’ with vendors of components that we report with unfavorable failure rates. Therefore, I do not have direct experience with these databases.)
GIDEP has four data interchanges:
Engineering: Including engineering reports, testing results, specifications, methodologies and techniques.
Reliability-maintainability: Including failure rate, failure modes, replacement rates, and demonstration & field performance reports.
Metrology: Including technical data on test systems, calibration, test equipment procedures, and measurement technologies (part of NIST – National Institute of Standards and Technology)
Failure experience: Including information on significant problems regarding parts, components, processes, fluids, material, safety, and fire hazards.
You may also find reliability related public information with other government agencies including NASA, Departing of Commerce, and the Department of Energy. RIAC (Reliability Information Analysis Center) which just recently merged with other information analysis centers and is known by DSIAC (Defense Systems Information Analysis Center)
Related:
Sources of Reliability Data (article)
Reliability Testing Considerations (article)
Alex Miura says
Very good article! When we choose the source of warranty returns only, we might consider that typically the failure behavior is premature or constant failure rate. If we have to choose wear out failure behavior, I do believe that testing is the best way to figure out.