Downloaded from https://www.everyspec.com ADS-79D-HDBK analysis should be the development of physical and functional models of the system, subsystem, and components as a means to determine the likely faults that may arise and their effect on the functions of the various elements of the system. Models of the fault modes, developed through either simulation and modeling or empirical measurement and analysis through testing, should be used to develop first estimates of the fault behavior as it progresses from initiation to failure. This is often described as ―Physics of Failure‖ modeling and analysis.
This modeling and analysis is accomplished with the scale and resolution acceptable to model the particular fault and item geometry. For example, to understand the presence and progression of a fault mode, the modeling of crack size propagation should be capable of representing crack geometries of the critical crack size as calculated by the analysis. Similarly, if pressure transients of 0.5 psi are important, the model is ineffective if it can only model transients of 2 psi. If a CBM system design is being undertaken, selecting the most effective faults for inclusion in the effort is normally done in a selection process.
From the total population of possible fault modes for all parts, components and subassemblies in the systems of the aircraft, the criticality analysis employed by RCM is used to determine which faults are important enough to justify sensors and data collection for monitoring. While fault modes which affect safety naturally rise toward the top priority for inclusion, fault modes which result in degraded availability and increased maintenance effort can also become high priority for development.
The same basis for criticality in RCM analysis applies to CBM, i.e., if RCM analysis has indicated that a particular failure mode requires inspection or remediation, those same modes can be investigated for feasibility analysis for CBM. Fault modes that represent single point failures that have led to the loss of aircraft, death, or major injury are obvious candidates for investigation. Other faults that drive significant costs or readiness degradation are also strongly acceptable for CBM feasibility analysis. This feasibility analysis should include trade studies which optimize the cost (example:
weight, system complexity, data collection, and processing infrastructure) for the benefit of being able to detect and diagnose the specific fault being considered. There are no fixed or rigid criteria that mandate a particular fault mode as requiring CBM application—the decision to sense and measure data to identify faults and base maintenance decisions on that information is like any other design decision that optimizes cost and risk with benefit. The results of FMECA and fault models should be used to develop a candidate group of faults with features or characteristics that can be obtained from signal processing of sensor data and used to accurately detect the presence of fault modes.
These ―features‖ are referred to as Condition Indicators throughout this ADS. This selection process, which is application dependent, establishes the domain of the feature (example: time, frequency, wavelet) and the property of the feature (example: energy, rms value, sideband ratios) that will be employed to develop the feature (or CI) for use in fault diagnosis. The FMECA results are also used to consider which faults require feature extraction and CI measurement in flight versus those that can be delayed until after flight. In general, the use of signal processing algorithms and software onboard the aircraft during flight should be prioritized such that: a.
Algorithms to compute CIs for faults on components which are flight critical. Any faults for which the progression could lead to loss of the aircraft are strong candidates for ―onboard‖ processing. Further ranking of the CIs can be done through risk analysis of the fault likelihood. For example, if one fault has an occurrence of 1 per 100,000 flight hours and another 1 per 10 million flight hours, inclusion of the former before the latter seems reasonable. b. Algorithms to compute CIs for faults which are combat mission critical. Again, ranking within this category by occurrence factors is the most reasonable approach. 73