What it measures
The failure rate expresses how likely a component is to fail in the next time interval, given that it has worked so far. It is usually represented by the Greek letter lambda and measured in failures per unit of operating time.
The bathtub curve
The failure rate of many products follows a typical shape with three stages:
- Infant mortality: a high, decreasing rate. Units with manufacturing or installation defects fail first.
- Useful life: an approximately constant rate, with random failures.
- Wear-out: the rate rises again because of aging, fatigue or material degradation.
Not all components follow that shape. Some show only the constant phase and others wear from the start.
Relation to mean time between failures
If the rate is constant, the mean time between failures is its inverse. This is the case of the exponential model, widely used for its simplicity. A frequent mistake is to read the mean time as the expected life of a single unit: it indicates the average across a population in the constant-rate phase, not how long each part will last.
Where the data come from
They come from accelerated life tests, field history or reliability databases. Accelerated tests apply higher stresses than normal and extrapolate; the acceleration factor depends on a model of the failure mechanism, so the extrapolation carries uncertainty.
How to interpret it
The failure rate makes it possible to compare design alternatives and estimate a system’s reliability by combining those of its parts. But it depends on the conditions of use: temperature, voltage and vibration all change it, so a catalog value should not be applied without considering the real environment.