Enter your criteria for Estimation Test Plan

Stat > Reliability/Survival > Test Plans > Estimation

Enter planning information to determine the sample size to estimate percentiles or reliabilities with a specified degree of precision. Planning information may come from design specifications, expert opinions, historical estimates, small pilot studies, or other sources.

Note

The data you collect for an estimation test can be censored or uncensored. A time-censored test plan or failure-censored test plan often provides precise results while minimizing test costs. For more information, go to Data censoring.

In Distribution, select a distribution to model your data.
Base your decision about the distribution on process knowledge or an evaluation of the distribution fit. For more information, go to Distribution fit for reliability analysis.
From Specify planning values for parameters, select what values to use to describe the distribution from the previous step.
For the distributions except for the exponential distribution, provide one of the following.
- Both distribution parameters
- Either distribution parameter and one percentile point
- Two percentile points
The exponential distribution has only 1 parameter. Provide one of the following:
- The scale parameter
- One percentile point
If you select an option with a percentile point, the specifications include the percentile value and the percent. For example, if you expect 15% of units to fail by 100,000 cycles, enter 100000 in Percentile value and 15 in Percent.

Optionally, select that a planning value for a distribution parameter is a fixed, known value. Do not use this option unless there is a strong physical or historical reason.

Assume shape is known

The option available for a Weibull distribution.

Assume scale is known

The option available for a distribution besides the Weibull distribution or the exponential distribution.

If you select this option, Minitab assumes that you do not need to estimate this parameter. Therefore, a smaller sample size is required.
Under Parameter to estimate, select an option to indicate what you want to determine.
- Percentile for percent: Estimate the time at which a certain percentage of the units will fail. This value is useful for determining a warranty period for a product. Enter a value between 0 and 100 to indicate the percentage of units that will fail. The nth percentile has n% of the observations below it, and (100–n)% of observations above it. To see the percent values that correspond with parameters of different distributions, go to Determine sample size requirements for estimating parameters.
- Reliability at time: Estimate the percentage of units that will survive a certain length of time. This value is useful for estimating how many units will survive through a warranty period. The time could be a standard time measurement, such as number of hours, or another measurement that indicates a certain amount of usage, such as number of cycles, number of weight units, and so on.
Specify how to define precision for the test. Then, specify the Precision values.
1. From Precisions as distances, select whether to measure the precision as a distance from the reliability estimate to either the lower or upper bound of the confidence interval (CI). Selecting the lower bound is common because many reliability standards are defined in terms of the worst-case scenario.
2. Enter one or more precision values to indicate the maximum distance from the confidence bound to the estimate. Enter values greater than 0. For example, suppose you want to estimate the 10th percentile of your failure time distribution, and the lower bound is to be no more than 25 hours less than your estimate. In Precision values, enter 25.
Note
The precision of an estimation test plan is based on the width of the confidence interval for the parameter that you are estimating. A wider confidence interval provides a less precise estimate, but requires fewer units to test.