What is mean time to failure (MTTF)?

In reliability analysis, MTTF is the average time that an item will function before it fails. It is the mean lifetime of the item.

With censored data, the arithmetic average of the data does not provide a good measure of the center because at least some of the failure times are unknown. The MTTF is an estimate of the theoretical center of the distribution that considers censored observations.

The MTTF can be used in several ways; for example:

To determine whether a redesigned system is better than the previous system in demonstration test plans.
As a measure of the center of the distribution when the distribution fits the data adequately.
To compare selected distributions with a distribution ID plot.

For example, you are studying how many miles automobile tires survive. You create a distribution ID plot of the results, and get the following table of MTTF:

Table of MTTF

		Standard Error	95% Normal CI
Distribution	Mean	Standard Error	Lower	Upper
Weibull	69545.4	629.34	68322.8	70789.9
Lognormal	72248.6	1066.42	70188.4	74369.3
Exponential	75858.8	2865.18	70446.0	81687.6
Smallest Extreme Value	69473.1	646.64	68205.7	70740.5

The Weibull and smallest extreme value distributions have similar MTTFs of around 69500 (Mean column on the table). The exponential distribution has the largest MTTF at 75858.8.

If all the distributions provide an adequate fit, you may want to select the lognormal or exponential distribution because of their slightly better MTTF.