Implementation of the Average-Log-Ratio ALR gear-damage detection algorithm on gear-fatigue-test recordings

•Early remote detection of gear tooth bending-fatigue and pitting damage.•Remotely distinguishes tooth bending-fatigue from tooth pitting damage.•Up to ten opportunities for detection from same working-surface damage. The Average-Log-Ratio, ALR, gear-damage detection algorithm is exercised on accelerometer recordings made during earlier-performed accelerated gear testing. Results of ALR computations from tooth bending-fatigue failure and pitting failure are displayed and discussed. The periodic behavior of the rotational-harmonic frequency spectra of tooth working-surface damage is verified and utilized in ALR detection of both tooth bending-fatigue and pitting damage. Rotational-harmonic frequency spectra out to the tenth tooth-meshing harmonic of damaged gears are utilized. ALR computations of gears failing in tooth-bending fatigue are shown to provide strong periodic contributions out to (and beyond) the tenth tooth-meshing harmonic, whereas tooth pitting damage is shown to generate rotational harmonic spectra with strongest contributions determined by the fractional size of the pitting damage on tooth working surfaces. This differing character of ALR rotational harmonic spectra between bending-fatigue and pitting damage allows remote detection of damage that can distinguish between these two classifications of gear damage.