Quantifying the Reliability Modelsâ•… 135
9.2╇ Allowing for Diagnostic Intervals 9.2.1╇ Establishing Diagnostic Coverage We saw, in Section 8.1.3, how the down time of unrevealed failures could be assessed. Essentially it is obtained from a fraction of the proof-test interval (i.e. half, at the unit level) together with the MTTR (mean time to repair). Some databases include information about MTTRs and those that do have been mentioned in Section 4.2. In many cases there is both auto-test, whereby a programable element in the system carries out diagnostic checks to discover unrevealed failures, as well as a manual proof test. In practice the auto-test will take place at some relatively short interval (e.g. eight minutes) and the proof test at a longer interval (e.g. a year). The question arises as to how the reliability model can take account of the fact that failures revealed by the auto-test enjoy a shorter down time than those that remain unrevealed until the proof test. The ratio of the former to the total is a measure of the diagnostic coverage and is expressed as a percentage of failures revealed by the test. There are three ways in which diagnostic coverage can be assessed: 1. By test: in other words failures are simulated and the number of diagnosed failures counted. 2. From field data: as for test, but using maintenance records. 3. By FMEA: in other words the circuit is examined (by FMEA described in Section 9.3), ascertaining, for each potential component failure mode, whether it would be revealed by the diagnostic program. Clearly 60% diagnostic coverage could be demonstrated fairly easily by either method. Test (or field) data would require a sample of only a few failures to establish a ratio of 60%. For 90% coverage however, the data sample would now need to exceed 20 failures and the cost and time begin to become onerous. For 99% coverage the sample size would now need to exceed 200 failures and this is likely to be impracticable.
9.2.2╇ Modeling Consider a dual redundant configuration with components subject to 90% auto-test coverage. Let the auto-test interval be 4╛hours and the manual proof-test interval be 4380╛hours. Assume that the manual test reveals 100% of the remaining failures. The reliability block diagram needs to split the model into two parts in order to calculate separately in respect of the autodiagnosed and manually diagnosed failures.