Using historic test program data to improve the repair process

Data collected during an automatic test program execution can provide much insight into the operation of the unit being tested that was not available when the original test program was developed. When a test program is being formulated, the developer generally does not have any data on the precise failure modes of the unit. Typically, the predicted reliability analysis data along with the generally accepted failure mode assumptions are used to develop the requirements for testing the unit. If the assumptions made during the test requirements document (TRD) development turn out to be inaccurate, the test program will be suboptimal. This could result in a longer mean time to repair (MTTR) as well as a greater demand on the spare parts supply line and thereby result in a higher cost for repair. Efficient estimators of the repair process statistics must be derived from the test data. Test programs executed on automatic test equipment have test results available at execution time that can easily be saved to form a historical data warehouse. The analysis of this data will reveal how well the test program is performing as well as how the program can be improved. Improvements can result from more accurate diagnostic callouts or less time taken to obtain the correct callout. This paper will demonstrate a methodology that uses historic test program data to validate the TRD assumptions and improve the repair process where the original TRD assumptions are shown to be invalid.