What does the Weibull shape parameter beta mean for maintenance?

What beta and eta describe

The Weibull distribution describes time to failure with two parameters. The shape parameter, beta, sets the form of the failure-rate curve. When beta is below 1, the risk falls with age, a pattern typical of early-life failures such as commissioning faults or manufacturing defects. When beta is close to 1, the risk stays roughly constant and failures look random. When beta is above 1, the risk rises with age, which is the pattern of wear-out. The scale parameter, eta, is the characteristic life: the age by which about 63.2 percent of units have failed. Read beta and eta together, because eta alone says nothing about the shape of the failure pattern.

Worked example: the same eta, two shapes (hypothetical)

Suppose a bearing type has eta = 5 000 hours. Compare the chance of failing by 2 500 hours under two shapes.

• Beta = 2.5: F = 1 - exp(-(2 500 / 5 000)^2.5) = 1 - exp(-0.1768) = about 0.162, or about 16 percent.
• Beta = 1: F = 1 - exp(-0.5) = about 0.393, or about 39 percent.

The wear-out shape gives a lower chance of failing early, but risk rises steeply with age after that. The same eta with a different beta tells a different maintenance story, so beta must be estimated from data before any replacement age is chosen.

Use and common mistakes

Beta drives the replacement policy. If it is above 1, a preventive replacement age can reduce wear-out failures, but the cost of early replacement must be weighed against the risk avoided. If it is near 1, scheduled replacement will not help, and the answer lies in condition monitoring or design changes. A frequent mistake is fitting beta from very few failures and treating the result as exact. Another is leaving out suspensions, the units still running at the time of analysis; excluding them can bias beta. For early-life failures with beta below 1, the fix is often in installation, commissioning or incoming quality control rather than in the replacement interval.

Follow-up questions