Glossary term
Reliability
The probability that a system performs its required function for a specified time under stated conditions.
Definition
metricThe probability that a system performs its required function for a specified time under stated conditions.
Reliability is the probability that an item performs its required function for a specified duration under stated operating and environmental conditions. It combines functional definition, time or mission profile, failure criteria, usage severity, maintenance assumptions, and statistical uncertainty.
Reliability is a conditional probability. A statement such as “reliability is 0.98” is incomplete unless it says for what function, over what time, under which environment, at what load, with what maintenance, and using which failure definition. A pump, aircraft actuator, software service, retaining structure, and electronic sensor all require different mission definitions.
For a lifetime random variable T, reliability is commonly written as:
If the failure rate is constant, a simplified exponential model gives:
This model is useful in some electronic or random-failure contexts, but it is not appropriate for every mechanism. Wear-out, fatigue, corrosion, infant mortality, and maintenance-induced failures often require Weibull, lognormal, physics-of-failure, or empirical field models.
Engineering use
Reliability engineering connects design, testing, maintenance, spares, warranty, safety, and operations. It uses failure-mode analysis, reliability block diagrams, accelerated testing, life data analysis, field return data, inspection results, redundancy modelling, and uncertainty analysis. Reliability is not the same as availability: a system can fail often but be highly available if repairs are rapid, or fail rarely but be unavailable for long periods after one event.
Designing for reliability means reducing failure mechanisms, derating components, controlling environments, adding diagnostics, improving maintainability, validating assumptions, and monitoring field performance. For safety-critical systems, reliability may be tied to interlocks, proof testing, fault tolerance, and independent protection layers.
Common mistakes
A common mistake is using MTBF as a synonym for reliability without a time horizon or distribution. Another is combining laboratory prediction data, supplier claims, and field failures without reconciling duty cycle and environment. A strong reliability review states mission profile, failure definition, population, data source, censoring treatment, confidence bounds, dominant failure modes, and whether maintenance restores the item to as-new or only to a degraded serviceable state.