Definition
The bathtub curve is a reliability model describing how failure rate typically changes across an asset's life in three phases: a higher early-life rate (infant mortality, from manufacturing or installation defects), a roughly constant useful-life rate (random failures), and a rising wear-out rate near end of life (age-related degradation). Its shape explains why a single average failure rate can mislead.
Key facts
- Describes three distinct phases: early-life (infant mortality), useful life (roughly constant rate), and wear-out (rising rate).
- A conceptual model, not a guarantee: actual assets can deviate from this shape depending on design, environment, and maintenance.
- Explains why burn-in testing targets infant mortality and why age-based replacement targets the wear-out phase specifically.
- A single average failure rate calculated across an asset's full life can obscure which phase it is actually in.
What does Bathtub curve reveal about failures?
The bathtub curve describes the typical shape of failure rate over an asset's operating life: elevated shortly after commissioning as manufacturing or installation defects surface, settling to a roughly constant rate of random failures during useful life, then rising again as age-related wear accumulates near end of life.
How do teams use Bathtub curve in failure analysis?
Failure rate is the specific quantity the bathtub curve plots over time; the curve is the conceptual shape that quantity often follows. Weibull analysis is the statistical method used to fit actual failure data to a specific shape and determine which phase of the bathtub curve an asset population is actually in, rather than assuming the generic three-phase pattern applies.
FAQ
Frequently asked questions
How does failure rate typically change over the phases of an asset's age?
The bathtub curve describes it: a higher rate shortly after start-up (infant mortality), a roughly constant rate during useful life, and a rising rate again as the asset approaches end of life (wear-out).
Last reviewed: