Related Topics and Links:

Test Design Assistant

ALTA Test Plan Utility

Reliability Demonstration Test Design

In a zero-failure reliability demonstration test (RDT), the engineer aims to demonstrate a specified target metric (e.g., reliability at a specific time with a given confidence level) by testing a specified number of units for a predetermined time. If no failures occur, then the target metric is demonstrated. This method has been adapted for scenarios where the target metric can be demonstrated even if some failures occur, as long as a specified number of allowable failures is not exceeded. For example, in a demonstration test where the number of allowable failures is 2, the target metric is demonstrated if no more than 2 failures occur during the test.[1]

ReliaSoft's RDT tool can assist the user in designing a demonstration test by solving for various values related to the test, such as sample size, required test time, the demonstrated reliability and the confidence level at which the target reliability will be demonstrated.

Tip: This tool could determine, for example, that if you assume your units follow a Weibull distribution with a shape parameter of 2 and you tested 10 units for 150 hours and none failed during the test, then you would demonstrate a 90% reliability at 100 hours with 90% confidence. It does not, however, calculate the probability that one or more units might actually fail during the test. If you wish to estimate whether a failure is likely to occur, one option is to use the Expected Failure Time Plot to estimate the first failure time based on the assumed distribution and parameters.

Control Panel

The following chart shows the information that each method can provide. The ReliaWiki resource portal has more information on these methods at: http://www.reliawiki.org/index.php/Reliability_Test_Design.

 

Metric to demonstrate:

Can solve for:

Parametric Binomial

The reliability at a specific time
or
Mean time to failure (MTTF)

The test time for a specified sample size
or
The sample size for a specified test time

Non-Parametric Binomial

Reliability at test time
(if acceleration factor = 1)

The demonstrated reliability
or
The associated confidence level
or
The necessary sample size

Exponential Chi-Squared

The reliability at a specific time
or
Mean time to failure (MTTF)

The total accumulated test time

Non-Parametric Bayesian

Reliability at test time
(if acceleration factor = 1)

The demonstrated reliability
or
The associated confidence level
or
The necessary sample size

 

 

© 1992-2017. HBM Prenscia Inc. ALL RIGHTS RESERVED.

E-mail Link