Downloaded from https://www.everyspec.com ADS-79D-HDBK reliability. Several methodologies are available for such purpose. Among them, accept-reject testing has been widely used. In accept-reject testing, the null assumption is that reliability to be demonstrated meets the requirement. In addition, two types of risks are considered and previously discussed: Type I Error , such that: and Type II error such that: where and are usually chosen to be 1%, 5%, 10%, or 15%, depending on the maximum risk tolerated. The significance level and may be chosen to be equal or different.
The accept-reject test consists of running n single sample CI/HI readings from separate component tests at specified conditions and mission duration and accepting the reading if r or fewer sample readings out of n fail during the tests. The validation process is rejected if more than r samples fail during the test. Obviously, outcome of the accept-reject tests depends on the designed-in reliability of the validation process, the total sample size and maximum allowable number of failures, and the underlying risk tolerated. The sample size can be determined by:
once the reliability goal, R L1 , and allowable number of single sample CI/HI readings from separate component tests that may fail for an accept decision, R , are determined and confidence level (1- ) is specified. Known total sample size and the number of allowed number of failures, the reliability of the validation process corresponding to a risk of can be estimated by: The accept-reject test consists of following steps: a. Determining a reliability goal R L1 as appropriate for the product or process; b. Selecting a Confidence Level appropriate for the accuracy required from the results of the accept-reject testing; c.
Selecting the test duration and conditions to achieve desirable reliability level for demonstration; d. Calculating the CI/HI Sample Size for the testing; e. Performing the testing with calculated sample sizes to validate the required Reliability with given number of allowable failure. If additional failures are observed within the testing duration, the sample size needs to be recalculated or confidence level needs to be re-determined. 152