Downloaded from https://www.everyspec.com ADS-79D-HDBK D.7.6 Normal and Faulted CI levels. The purpose of every CI is to distinguish between faulted and healthy components, so the effectiveness of a CI is based on its ability to separate these two populations. To determine how well a CI separates faulted cases from the healthy ones, one must first identify these two data sets. Two methods are used to make this identification: teardowns and seeded fault testing. Teardowns are used to determine the actual condition of components for which values of a CI have been calculated, usually components that are suspected of being faulted.
In seeded fault testing, a component with a known fault is placed on a test stand to determine how its CI values differ from healthy components. Since it is impractical to tear down every component, the going assumption is that the vast majority of components are healthy. A good CI should provide enough separation between known faulted components and the rest of the fleet that a threshold can be selected such that the known faulted components are above it and that the vast majority of the rest of the fleet is below it. Two thresholds are commonly established for each CI. The lower threshold, or caution threshold, indicates that component‘s behavior is anomalous.
Maintainers should inspect such a component and order a replacement. The higher threshold, or exceedance threshold, indicates that the component has a significant fault. Maintainers should replace such a component. The initial thresholds are set using engineering judgment and statistical analysis. As more data is collected and faults are found, the thresholds are revised to more accurately convey the condition of the component. Figure D-11 shows a portion of a spectrum from a faulted AH-64D aft hanger bearing and an average of spectra from the fleet. This is the section of the spectrum that is used for the Aft Hanger Bearing Energy CI.
Note that the largest peaks in the faulted spectra correspond to the fundamental BSF of 182.9 Hz and its harmonics. The average spectrum was calculated using 10 spectra (or the maximum number available) from each monitored tail number. FIGURE D-11. Comparison of AH-64D aft hanger bearing faulted spectra The faulted bearing that produced this data was sent to Corpus Christi Army Depot for teardown. It found that the grease was contaminated with dirt, and that spalling and corrosion pitting of one single ball initiated failure and caused secondary damage to the other balls and the races (Figure D-12). 96