In this study, effects of the root fillet geometry and the tooth asymmetry on the tooth bending stresses and the fatigue lives of spur gears are investigated. For this purpose, an existing gear analysis model, Load Distribution Program (LDP), is employed to define four basic tooth geometry variations. These four variations are (i) symmetric tooth profiles (i.e. identical loaded and unloaded flanks) with full circular root geometry (at the maximum radius possible), (ii) symmetric tooth profiles with an elliptical root geometry, (iii) asymmetric tooth profiles (i.e. loaded and unloaded flanks at different pressure angles) with full circular root geometries, and (iv) asymmetric tooth profiles with an elliptical root geometry on the right (loaded) flank and a circular root geometry on the left flank. Under these conditions, variations (ii), (iii), and (iv) are predicted to have maximum root stresses that are 7.6%, 22.4%, and 24.3% less than that of the baseline case (i).
Actual test articles representing these four variations are procured and qualified through dimensional measurements of the profiles as well as the root fillet regions. Certain teeth of each test variation are instrumented by using multiple strain-gauges placed at different root fillet locations. The strain measurements under various tooth load levels are compared to predictions to verify their accuracy for all four variations considered.
Single tooth bending fatigue tests are also performed to obtain fatigue data for each variation of the test gears. The resultant tooth bending fatigue performance of each gear variation is shown to correlate with the level of root stress reduction achieved. Experiments indicate that the most significant life increases compared to the baseline conditions are achieved with the last variation (asymmetric toot profiles and an elliptical root shape), where the mean life is increased by more than 30 times. It is also shown through examination of the broken teeth that the critical locations where the cracks initiated agree well with the predicted locations of the maximum root stresses.