00343 �2133ef95cf1f1920885bd442e7266a

347

A Monitoring Plan for Detecting Product Degradation

check. The generał philosophy is initially to test a string of consecutive units at the beginning of production to show that the design and manufacturing processes conform with the reliability requirement. The size of the string, designated as c and referred to as the clearance number, will be related to the reliability reąuirement. Once the product ^Mproves” itself by having c consecutive units pass the acceptance test, the sampling phase of acceptance is initiated and the product is monitored at an intermediate ratę, Level 1 and a reduced ratę, Level 2.

When a failure occurs during the initial testing, 100% testing is continued and c consecutive units must pass before sampling can be initiated. When a failure occurs during the Level 1 sampling, initial testing is reinstated to show conformance to the requirements of the design and manufacturing processes. When a failure occurs during Level 2 sampling, the estimated failure ratę is checked for concordance with the reliability requirement. This check is madę to allow for a continuation of sampling if the cumulative number of failures is consistent with the reliability requirement and if there is no other indication of a change from the reliability reąuirement at the time of the failure. If the estimated failure ratę is not concordant with the reąuirement, the initial testing of a string of consecutively produced units is reinstated. If the assessed failure ratę is concordant with the reąuirement, sampling, where sampling starts with the unit immediately succeeding the failed one, may continue on a probational basis until a string of c consecutive sampled units, is tested without failure. There is no disruption in the sampling process if concordance is determined. The sampling is probational in the sense that if a failure occurs during the probational period, initial testing, 100% inspection, is resumed. Given successful test results, it is assumed that the reliability reąuirement is being met.

Since in practice, testing of units designated for inspection will not be instantaneous, it is necessary that a time restriction be put on testing with regard to shipment of units to their destination. If the delay between test unit selection and test completion (which includes designation as a defective or non-defective) exceeds some value, all product acceptance stops. The time restriction can be clock time or number of produced units. If the test time is not being exceeded, product acceptance will continue.

The monitoring plans that are developed in this paper can be seen to be similar to the Dodge (1943, 1947, 1951) continuous sampling plans where 100% inspection is followed by sampling. However, the main objective of the Dodge plans is to control the Average Outgoing Quality (AOQ), where testing is non-destructive, by resorting to 100% inspection during periods of poor