How do you talk about reliability?
“The language we use matters.” Wayne Nelson
When we talk about our products or equipment, we may refer to the expected durability of the system.
- How long it will work before failure?
- How long before we have to make repairs?
- Will it work when we need it to work?
Our customers and investors also want to know how long will it last.
The concept of reliability is muddled with dependability, durability, availability, and trustworthiness. The term, reliable, in part means one can count on the product working as expected.
Our common language use makes the term unclear. What’s worse is we think we know what we’re talking about when talking about reliability.
Definition of RELIABILITY
1: the quality or state of being reliable
2: the extent to which an experiment, test, or measuring procedure yields the same results on repeated trials
“Reliability.” Merriam-Webster.com. Merriam-Webster, n.d. Web. 13 May 2014. http://www.merriam-webster.com/dictionary/reliability.
Full Definition of RELIABLE
1: suitable or fit to be relied on : dependable
2: giving the same result on successive trials
— re·li·able·ness noun
— re·li·ably adverb
“Reliable.” Merriam-Webster.com. Merriam-Webster, n.d. Web. 13 May 2014. http://www.merriam-webster.com/dictionary/reliable.
The common understanding isn’t measurable as stated.
Engineering Definition of Reliability
From an engineering perspective, we need a clear and common understanding of the terms we use when specifying, designing, and measuring product reliability. We need something we can specify and measure.
The common metric that many use is MTBF. Simply said it is the expected value of time between failures, or average. It’s easy to calculate and to specify. It also has the feature that a larger MTBF value generally means fewer failures over specific time period.
Yet, MTBF is a point estimate for what could be a complex behavior over time. MTBF is the inverse of the average failure rate. It doesn’t include other elements that we need to fully understand product reliability.
Stating the MTBF for a device, does not include over what time period the probability of failure applies. If you don’t think MTBF is a probability of failure, that highlights another issue with the metric.
Let’s consider the engineering definition of reliability.
“The probability that an item will perform a required function without failure under stated conditions for a stated period of time.”
O’Connor, P. D. T. And Kleyner, A. Practical Reliability Engineering, Fifth Edition. 2012. John Wiley & Sons. p. 1.
There are four elements:
- Environment (conditions)
- Probability of success
So, how does this work in practice. Let’s consider a cell phone. A possible reliability statement may be:
“The cell phone will make/receive phone call over the cellular network in North America with 99.9% probability of successfully operating over two years without failure.”
We may have to further define what a successful call entails. And, the range of environmental stresses and use conditions that may befall a cell phone in North America, yet we certainly can measure the probability of operating over 2 years.
You could also add other probability and duration couplets to include initial use (early life) and a warranty period, desired.
Setting the probability of success and duration connects the reliability statement we can use in the design, manufacture and maintenance to both the business and customer expectations.
In short when we talk about reliability, use the definition of reliability and include all four elements. A shorthand when all concerned understand the function and environmental elements, we can simply say 99% should last 1 year (or similar couplet).
We owe it to our selves and to our customers to avoid any confusion concerning reliability. The language we use can help provide a common understanding, or not.