U.S. Army ADS-79D-HDBK — ADS-79D-HDBK.pdf

ADS-79D-HDBK Downloaded from https://www.everyspec.com condition maintenance. This component was selected for our case study because a large amount of CBM data is available for it. The fault detected involves pitting damage on gear teeth of the NGB spiral bevel gears. The CI used is Input DA1. This CI was recorded from the time when damage was thought to have occurred until the suspect gears were removed. An increase in the magnitude of the CI value indicates a fault. After TDA to confirm damage, new gears were installed and the CI was again recorded to confirm a healthy component. A total of 16 NGBs are included in this study.

The TDA results on suspect gears for these NGBs are 4 green, 2 yellow, and 10 red. In our analysis, CI values are categorized according to the color of the NGB from which they were recorded. CI values recorded after gear replacement are all considered green. Three separate analyses were performed with different subsets of the CI data. Dataset 1 is comprised of the last recorded CI value before gear replacement and the first recorded CI value after gear replacement, for each NGB. After recognizing that multiple CI recordings may be taken on a single day, the maximum CI values recorded the day before gear replacement and the day after gear replacement were identified for each NGB.

This is Dataset 2. Finally, the maximum CI value recorded over the entire suspect gear interval was identified for each NGB, along with the maximum CI value recorded the day after gear replacement. This comprises Dataset 3. Figure I-16 , below, provides the Weibull plots for both the green and red thresholds using Dataset 1. As it turns out, these thresholds are essentially identical, and the same occurrence is observed using Dataset 2 and Dataset 3. A yellow category does not exist for Input DA1 CI data for the Apache NGB. Therefore, only the green threshold will be considered for the remainder of this case study.

Summary statistics for both thresholds are given in Table I-IV at the end of this section. Figures 17, 18, and 19 provide the Weibull plots for Datasets 1, 2, and 3. For each figure, faulted CI values (yellow and red) are depicted on the plot, but right-censored CI values (green) are not. The straight line is the Weibull fit, and the curved line to the left is the 90% lower confidence bound. The CI value that provides 90% reliability with 90% confidence is found by locating the point where the line reliability = 90 ( y -axis) intersects the lower confidence bound line, then reading the point‘s value from the x -axis (Input DA1). 170