When people hear the word reliability, it’s often interpreted as meaning “zero failures”. While that’s understandable, it’s not what reliability engineering is really about.
The Certified Reliability Engineer (CRE) Body of Knowledge, produced by the American Society for Quality, defines reliability as:
“The probability that an item will perform a required function without failure under stated conditions for a specified period of time”
Crucially, each part of this definition needs to be clearly understood and agreed from the outset, particularly between customers and suppliers. Many reliability problems arise not because the definition itself is wrong, but because these elements are interpreted differently.
There are four key elements in that single statement worth unpacking:
Probability
Reliability is never 100% – everything fails eventually. Reliability engineering is about managing uncertainty based on evidence, not eliminating failure entirely. Acceptable levels of reliability therefore need to be explicitly agreed.
Required function
Failure is defined against what the system is meant to do, not whether it simply switches on or appears to work. For example, a pump that runs but cannot deliver the required flow rate has failed its function, even though it may still be operating mechanically.
Stated Conditions
Environment, operational usage, loads, and stresses all matter. Reliability only has meaning in context. If operating conditions are assumed rather than clearly defined, suppliers may demonstrate reliability under benign conditions while customers experience something very different in service.
Specified time
Reliability without a time basis, or an equivalent measure such as operating cycles or missions, is essentially meaningless. “Reliable” for one hour, one mission, or one year implies very different expectations.
A shared understanding of probability, function, conditions and time at the start of a programme is often more valuable than the most sophisticated reliability analysis applied later.
Reliability engineering isn’t just about calculating metrics like failure rates or availability. Those numbers are only useful when clearly tied back to function, context, and time. In practice, many reliability problems stem from poorly defined functions, unrealistic assumptions, or misalignment between stakeholders. Get those wrong, and even the best modelling tools won’t help.
This definition underpins everything that follows in this series, from leadership and ethics, through to modelling, testing, and lifecycle decision-making…
Next up…
Reliability Bites #2: The benefits of reliability engineering.
Leave a Reply