Sometimes making an assumption is a good thing. You can achieve more with less. A well-placed assumption saves you time, work, and worry. The right assumption may even be left unstated: It’s so good that no one questions it.

Have you ever assumed that the failures for a system follow an exponential distribution? Did you assume that tallying up the total hours and dividing by the number of failures was appropriate? Did you even check?

Assuming an exponential allows you to quickly and easily state reliability. Just mention the mean time between failures (MTBF) and you’re done. Better yet, you can run a test or two and calculate the MTBF. Maybe you can even use an old parts count prediction method to predict the MTBF. This entails just making a few simple assumptions here and there. When assuming the failures are random in nature, the reliability statistics become very easy.

My favorite use of a well-timed assumption involves designing an accelerated life test. Let’s say you have 20 samples and about 3 months to run a test. If all goes well you can test for 2,000 hours. For a product expected to survive 10 years, you will have to rely on some form of acceleration. This is where accelerated life testing comes into play.

You know that you can use higher temperatures and achieve an acceleration factor of 10 or more. But your temperature is limited to a certain value or you’ll risk melting some key elements of the product. Although 20,000 hours is not the 5 years you had hoped to simulate, you’re limited by the physics of the materials and failure mechanism.

Wait a minute: If you assume that the failures over the 5 years occur in line with an exponential distribution, you can use the memoryless aspect of the probability of failure, and simply tally the total hours run during the test, including the acceleration factor. Thus you have 20 samples, each run for 2,000 hours with a 10´ acceleration factor, or 400,000 hours. If you do not have any failures or just one failure, this becomes an MTBF of 400,000 hours or approximately 45 years.

What a great assumption! It has saved the day and you may even be able to shorten the test or reduce the number of samples. Of course, you will just have to remain quiet about the nature of the bearing wear-out mechanisms expected to cause failures. And, you are relieved that you do not have to run the test for, say, 4,000 hours, as you know from field returns that this is about the time at which the bearings start to fail. It’s OK: It was just an assumption.

If you or those you know are making similar assumptions, stop them. They are most likely misleading themselves and others based on the use of a faulty assumption. Always check your assumptions carefully. Understand the failure mechanisms. It is important to conduct the right test with as few assumptions as necessary, and use only assumptions that are true, not merely convenient.

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 Musings on Reliability and Maintenance Topics.