The traction behavior of lubricant films forms the foundation of dynamic modeling for aeroengine mainshaft ball bearings. Its accuracy directly determines the reliability of predicted dynamic responses and the available design safety margins. Existing traction models produce artificial friction in the zero slip region and exhibit strong sensitivity to ball size effects, which leads to significant deviations from experimental observations. These limitations make them unsuitable for high-fidelity analyses of aeroengine mainshaft bearings. In this study, a self-developed high-speed traction test rig was used to systematically measure the traction–slip responses of three aviation lubricants, including the newly developed 4102 (7 cSt) and the inservice 4050 (5 cSt) and 4010 (3 cSt). The tests covered a wide range of operating conditions, including maximum Hertzian pressures of 1.0 to 1.5 GPa, oil supply temperatures of 25 to 120 °C, entrainment speeds of 25 to 40 m/s, and slide–roll ratios (SRR) of 0 to 0.3. The evolution of lubricant traction characteristics was examined in detail. Based on the experimental data, a four-parameter and three-coefficient traction model was proposed. This model eliminates the non-physical traction outputs at zero slip observed in previous formulations. When embedded into the bearing dynamic simulations, the maximum deviation between the predicted friction torque and the measured values is only 3.79%. On the basis of typical operating conditions of aeroengine bearings, lubricant selection guidelines were established. Under combined high-speed, light-load, and high-temperature conditions, the high-viscosity lubricant 4102 is preferred because it suppresses cage sliding and enhances film stiffness. When the cage slip ratio is below 15% and lubrication is sufficient, the low-viscosity lubricant 4010 is recommended, followed by 4050, in order to reduce frictional heating. This study provides a theoretical basis for high-accuracy dynamic design and lubricant selection for aeroengine ball bearings.
Deng et al. (Sat,) studied this question.