Development of a predicted value for maintainability, whether it be a mean corrective maintenance time, mean downtime, or other parameter, is based on time estimates of individual tasks associated with maintenance of a system and its individual parts. The reliability data, such as failure frequency and failure mode and effects, play a role in maintainability prediction as well, since it is the failure modes, their effects and frequency of occurrence that ultimately define individual maintenance tasks to be performed. In order to develop a maintainability prediction, the predictor must assume that a person with a particular skill level will be assigned to maintain a particular subsystem within the system of interest. Further, estimates of fault isolation time and fault detection capability must also be developed based on the testability design and assumed effectiveness of any built-in-test (BIT) features incorporated at various system levels. Often, time estimates derived in this manner are based on experience and knowledge of the individual performing the estimate. Therefore, the chances for bias in a prediction exist, especially if such factors as maintenance environment are not well understood early in the design phase. It is important then to verify the maintenance task database that is used to obtain the predicted values as a means of improving the maintainability design aspects of the system, and to improve the chances of performing a meaningful and successful maintainability demonstration.
In addition to the preceding discussion, testability
aspects, such as BIT effectiveness, are not easily verified in a formal
demonstration test. This is due in part to the fact that such failure
mechanisms that cause transient or intermittent behavior are not easily
simulated in a laboratory environment (where many demonstrations take place).
Further, the number of failures induced in a demonstration represent such a
small part of the overall number of failures that may occur during fielded
operation and are therefore too small to really demonstrate the diagnostic
capabilities of a system design. Because of this, a well planned verification
program, that optimizes naturally occurring failures during development and
subsequent testing is needed to assess the diagnostic characteristics of the
design. In short, demonstration techniques and methods that are documented in
this appendix (and previously in MIL-HDBK-471A) are inadequate for testability
demonstration. Further discussion of testability verification is provided in
Section B.7.0 of this appendix.