8.2.2. How do you plot reliability data?

8. Assessing Product Reliability
8.2. Assumptions/Prerequisites

8.2.2. How do you plot reliability data?

Create a probability plot and if the points line up approximately on a straight line, the assumed model is a reasonable fit

Graphical plots of reliability data are quick, useful visual tests of whether a particular model is consistent with the observed data. The basic idea behind virtually all graphical plotting techniques is the following:

Points calculated from the data are plotted on a log-log scale and, as long as they line up approximately on a straight line, the analyst can conclude that the data are consistent with the assumed model.

If the reliability data consist of (possibly multicensored) failure data from a non repairable population (or a repairable population for which only time to the first failure is considered) then the models are life distribution models such as the exponential, Weibull or lognormal. If the data consist of repair times for a repairable system, then the model might be the NHPP Power Law and the plot would be a Duane Plot.

The kinds of plots we will consider for failure data from non-repairable populations are:

For repairable populations we have

Trend plots (to check whether an HPP or exponential model applies)
Duane plots (to check whether the NHPP Power Law applies)

Later on (Section 8.4.2.1) we will also look at plots that can be used to check acceleration model assumptions.

Note: Many of the plots discussed in this section can also be used to obtain quick estimates of model parameters. This will be covered in later sections. While there may be other, more accurate ways of estimating parameters, simple graphical estimates can be very handy, especially when other techniques require software programs that are not readily available.