Reliability or life testing involves estimating the expected durability over time of an item. This may be an entire system, a product, or an individual component. We may also focus on an element of a component, such as a material property. At the end of the testing we want to say something meaningful about the expected performance over time.

A ‘good’ life test focuses on the failure mechanisms in question. For example, if we know the polymer’s elasticity will degrade by chain scissioning, then a test that includes stress that causes chain scissioning to occur in a similar fashion as expected in use can provide results that reflect the actual performance. The trick is to understand the failure mechanisms first, then select the test stresses.

If the development time is long enough to encompass the expected operating time of the item, then we can simply use the item as it is expected to be used. The test in this case really just entails data collection and analysis. However, we rarely have ample time so require some form of acceleration.

Accelerated life testing (ALT) significantly highlights the need to apply the appropriate stress in order to age the item in a known fashion. We are attempting to cheat time. When the testing is done well, we gain a glimpse into the future. If the test is conducted poorly, then we witness something that will not come to pass.

Focusing on the Failure Mechanism

We can do a simple test of a life test design by asking “What is the expected failure mechanism and how does the test stress encourage that mechanism to occur?” If the answer is a failure mode such as “the product ceases to function,” then the design needs more work. What causes the product to fail? What fundamental mechanism leads to the loss of the function?

For example, if we test a new product with elevated temperature and humidity, because it is something we have always done, or a customer has requested it, or it is in an industry standard, we may or may not learn anything about the product’s expected lifetime. If the product is a handheld portable product, a high-temperature, high-humidity test will not serve to help us evaluate the very likely failure mechanism of shock stress resulting from dropping the product: We ‘pass’ the temperature and humidity test but learn nothing about the expected drop stress failures.

We should first sort out what is likely to fail, understand the failure mechanisms, then apply the appropriate stresses to excite those specific failure mechanisms. Science and understanding should support starting any life test.

Replicating the Failures as Expected in the Field

Test should be run failure. Did the test sample fail exactly or similarly enough (i.e., have the same failure mechanism) or did the test reveal a different path to failure? Running to failure allows us to check the validity of test design assumptions and avoid unpleasant surprises when the customers begin using the product.

Only once we have a well-characterized failure mechanism does testing to show a minimum reliability with a success test (a test designed to have no failures) become feasible. The key for any life testing is to replicate the failure mechanisms that will occur in actual use.

Using or Building an Acceleration Model

For specific failure mechanisms that have a time to failure versus applied stress relationship, we can use that information to create an acceleration model. The model provides an acceleration factor that translates the elevated time to failure test results to use condition expected time to failure performance.

Consider the failure of a hinge. If the failure mechanism is related to the flexing motion that occurs once a day in normal use and we replicate that motion in the lab 24 times a day, then assuming that the increase in flex motion rate does not introduce spurious failure mechanisms, we have a 24´ acceleration factor. This means in one day of lab testing we replicate 24 days of use. We have that glimpse into the future.

Some failure mechanisms have complicated relationships with stress. If the mechanism is chemical in nature, then the Arrhenius rate reaction formula connects temperature to the reaction rate if we know that specific chemical reaction’s activation energy (and we cannot guess or use a standard-based value here!).

The acceleration factor equations are based on either empirical evidence or detailed characterization of the failure mechanism. The field of physics of failure has catalogs of detailed formulas for specific failure mechanisms, plus the methods used to develop the formulas. These models may provide a means to directly estimate the reliability of an item based on the specific failure mechanism or a method to design our own life test.

The best way to design a life test is to let the customers use the product. Then, based on the understanding of how they use the product and what fails and when, we can design an adequate life test to predict what actually happens. This is not practical, so we make assumptions, sometimes confirmed by experimentation and characterization work, and attempt to peer into the future.

Bio:

Fred Schenkelberg is an experienced reliability engineering and management consultant with his firm FMS Reliability. His passion is working with teams to create cost-effective reliability programs that solve problems, create durable and reliable products, increase customer satisfaction, and reduce warranty costs. If you enjoyed this articles consider subscribing to the ongoing series at Accendo Reliability.

He can be reached at:
fms@fmsreliability.com or
(408) 710-8248