In this study, an experimental methodology is presented to separate various components of the power loss of a gearbox. The methodology relies on two separate measurements. One is designed to measure total power loss of a gearbox housing a single spur gear pair under both loaded and unloaded conditions such that load-independent (spin) and load-dependent (mechanical) components can be separated. With the assumption that gear pair and rolling element bearings constitute the bulk of the gearbox power loss, a second measurement system designed to quantify rolling element bearing losses is proposed. With this setup, spin and mechanical power losses of rolling element bearings used in the gearbox experiments are measured. Combining the sets of gearbox and bearing data, power loss components attributable to the gear pair and rolling element bearings are quantified as a function of speed and torque. The results indicate that all gear and bearing related components are significant and a methodology such as the one proposed in this study is warranted.

A tribo-dynamic model of ball bearings is proposed to predict their load-dependent (mechanical) power losses. The model combines (i) a transient, point contact mixed elastohydrodynamic lubrication (EHL) formulation to simulate the mechanics of the load carrying lubricated ball-race interfaces and (ii) a singularity-free dynamics model and establishes the two-way coupling between them that dictates power losses. The dynamic model uses a vectoral formulation with Euler parameters. The EHL model is capable of capturing two-dimensional contact kinematics, velocity variations across the contact as well as asperity interactions of rough contact surfaces. Resultant contact surface shear distributions are processed to predict mechanical power losses of example ball bearings operating under combined radial and axial forces. An experimental setup is introduced for measurement of the power losses of rolling-element bearings. Sets of measurements taken by using the same example ball bearings are compared with those predicted by the model to assess its accuracy in predicting mechanical power loss of a ball bearing within wide ranges of axial and radial forces.