Elastohydrodynamic Ball Bearing Optimization Using Genetic Algorithm and Heuristic Gradient Projection

Abstract

Rolling element bearings operation depends on some variables contributing to the machine element performance. The present work attempts to improve the performance of rolling element bearings through the increase of fatigue life and the reduction of bearing wear. The formulation is based on Elastohydrodynamic to maximize the realistically evaluated minimum film thickness without significant increase in viscous friction torque. The multiobjective problem can then be stated as maximization of minimum film thickness and minimization of total friction torque. Design vectors are reduced in the present study relative to previous studies as some variables are considered as dependent variables. A new important parameter is introduced in this study as a design variable, which is the viscosity of lubricant (η0 (Pa.s)). Lubricant viscosity contributes drastically in either increasing minimum film thickness separation or increasing the frictional torque arising in bearing. A multi-objective optimization using Genetic Algorithm is used in order to evaluate Pareto optimal solutions. Another multi-objective problem has been formulated such as a two objective problem involving maximizing minimum film thickness, and minimizing the bearing elements size (i.e.: ball diameter and mean diameter) through subjecting the bearing to the maximum allowable compressive stress of the elements. Heuristic gradient projection method is used in solving such a problem, as it can efficiently seek the optimum point in less than five iterations. In such a case, the design variables are reduced to two variables, which are the ball diameter and mean diameter. Full design vector consideration is also performed.