Minitab®  Support

Enter your criteria for Demonstration Test Plan

Learn more about Minitab
Stat > Reliability/Survival > Test Plans > Demonstration

Enter information to develop plans for a test to demonstrate that only a certain number of failures occur in a set amount of test time. Planning information may come from design specifications, expert opinions, small pilot studies, or other sources.

  1. Select the value for the analysis to calculate. If the analysis calculates sample sizes, then the required information for the plan includes how long to test each unit. If the analysis calculates the testing times, then the required information for the plan includes how many units to test.
    • Sample sizes
    • Testing times for each unit
  2. Enter the time limit for the test that applies to each unit or enter the number of units available to test.
    Note

    Each combination of sample size or testing time with a maximum number of failures allowed creates one test plan. You may want to create several test plans by entering multiple values for each property and then compare the results.
    • Testing times for each unit
      Enter one or more values to indicate the time limit for the test.
    • Sample sizes
      Enter one or more values to indicate the number of units available for testing.
  3. In Maximum number of failures allowed, enter one or more values to indicate the maximum number of failures that your test allows.

    Entering 0 results in the smallest sample size and test time. However, if just one failure occurs, the demonstration test fails and you cannot conclude that you have achieved the desired reliability. For more information, go to What is the difference between an m-failure test and a 0-failure test?.
  4. In Distribution, select a distribution to model your data. Then, enter the parameter to assume for the test plan.

    Base your decision abut the distribution on process knowledge or an evaluation of the distribution fit. For more information, go to Distribution fit for reliability analysis.

    When testing a redesigned system, you can often use a historical value to estimate the shape or scale parameter of the distribution. The historical value is usually a good estimate because the redesign does not affect this parameter.

    CAUTION

    The parameter value is an approximate estimate. If, however, the estimate of the parameter is too far from the true value of the parameter, then the sample size or the testing time for the test plan may not be accurate. If you are unsure of the parameter value, consider repeating the test plan analysis using a range of likely values for the assumed parameter. The results from the different analyses can help you to determine how sensitive the sample size or testing time is to the assumed value of the parameter.

    Distribution Parameter
    Weibull Shape: Enter a value greater than 0 for the shape parameter. A value of 1 makes the Weibull distribution equivalent to the exponential distribution.
    Exponential N/A
    Other distributions Scale value: Enter a value greater than 0 for the scale parameter.
  5. In Demonstration test goal, select the criterion to use for the test. Then, enter the minimum values for the test to pass.
    Because the goals all have equivalent expressions, the choice usually relates to the most convenient way to express the requirements. For example, For a Weibull distribution with a shape parameter of 4, the following produce the same test plan:
    • A true 5th percentile that exceeds 4,759
    • A scale parameter that exceeds 10,000
    The ranges of the scales on the probability of passing graph differ for specifications that produce the same test plan.
    The true percentile for a given percent exceeds this time
    Specify the maximum percent of units that fail before the specified time to meet requirements. For example, the requirement for a turbine engine combustor is for 2,000 cycles to lead to a maximum of 1 percent of failures.
    The true reliability for a given time exceeds this proportion
    Specify the proportion of units that survive and how long they survive to meet requirements. For example, the requirement for a turbine engine combustor is for a proportion of 0.99 to survive for longer than 2,000 cycles.
    The true mean time to failure (MTTF) exceeds this value
    Specify the minimum MTTF that meets requirements. For example, the requirement for a turbine engine combustor is that the MTTF is greater than 10,000 cycles.
    The true scale parameter exceeds this value
    This option is available when the distribution is Weibull or Exponential.
    Specify the minimum parameter value that meets requirements. For example, the requirement for a turbine engine combustor is that the scale parameter from an exponential distribution is greater than 10,000.
    For a Weibull distribution, the scale parameter is approximately the 63.2 percentile. For the exponential distribution, the scale parameter is the mean time to failure (MTTF) as well as the 63.2 percentile.
    The true location parameter exceeds this value
    This option is available when the distribution is not Weibull or Exponential.
    Specify the minimum parameter value that meets requirements. For example, the requirement for a turbine engine combustor is that the location parameter for a lognormal distribution is greater than 9.21
    For the normal distribution, the location parameter is equal to the mean time to failure (MTTF). The interpretation of the location parameter depends on the distribution. A normal distribution with a location parameter of 10,000 and a lognormal distribution with a location parameter of 9.21 have similar 50th percentiles.
  6. In Confidence level, enter a confidence level between 0 and 100. Usually a confidence level of 95% works well. The confidence level affects the risk of reaching an incorrect conclusion.
    • A lower confidence level, such as 90%, may reduce the sample size or testing time that is required for the test. However, a lower level increases the risk of incorrectly concluding that the system meets the reliability goal. For a 90% confidence level, the risk of this error is 10%.
    • A higher confidence level, such as 99%, decreases the risk of incorrectly concluding that the system meets the reliability goal. However, the test may require a larger sample size or longer testing time. For a 99% confidence level, the risk of this error is 1%.
Copyright © 2026 Minitab, LLC. All rights Reserved.