Try it while you read
Open the free Weibull Analysis tool and click Load sample: it reproduces the case study below.
Theory: one distribution, three failure modes
The Weibull distribution describes time to failure very flexibly with two parameters: the shape β and the scale η.
- β < 1: failure rate decreases, early failures from manufacturing defects.
- β ≈ 1: constant failure rate, random failures.
- β > 1: failure rate increases, wear-out (fatigue, abrasion, corrosion).
η is the characteristic life: 63.2 % of the units have failed by then, whatever β is. On a Weibull plot (ln t against ln(−ln(1 − F))) the data form a straight line with slope β.
Key point
Step by step
- Collect the times to failure from a test or field data.
- Paste them and choose the unit.
- Check the plot: points close to the line mean the model fits.
- Read β, η, B10 and MTTF; enter the mission time for its reliability.
Case study: life test of 8 actuators
Eight actuators are run to failure on a test bench: 120, 190, 240, 310, 365, 420, 500 and 610 hours (the Load sample data).
- β = 2.45: wear-out.
- η = 390 hours, MTTF 345 hours.
- B10 = 155 hours, B1 = 59 hours.
- Reliability at the 200-hour mission: 82 %.
The customer requires B10 ≥ 300 hours, so the design fails the requirement. Because the failures are wear-out, the team looks at the bearing material and lubrication rather than at assembly quality.
Common mistakes
- Mixing failure modes (early and wear-out) in one fit: the plot bends.
- Ignoring units that did not fail; a censored analysis is needed for those.
- Drawing conclusions from 3 or 4 failures.
Frequently asked questions
Is MTTF the same as MTBF?
For non-repairable items the mean life is MTTF. MTBF is used for repairable systems. For a Weibull with β = 1 (exponential) both equal η.