Multi-objective optimization of spherical roller bearings based on fatigue and wear using evolutionary algorithm

In Spherical Roller Bearings (SRBs) design the fatigue and wear lives are the most important factors. The fatigue life of bearing is connected to dynamic capacity (Cd) and wear life of bearing is linked with the elasto-hydrodynamic minimum film thickness (hmin). Multi-objective optimization (MOO) of SRBs has been considered in the present study. For SRBs optimization problem, two objectives (Cd and hmin), eight design variables, and twenty-two constraints have been considered. Bearing pitch diameter, roller diameter, number of rollers, effective roller length and the contact angle are five design geometrical variables and other three are constraint parameters. Objective functions have been optimized individually as well as simultaneously. Elitist Non Dominated Sorting Genetic Algorithm (NSGA-II) is used to solve a non-linear constrained optimization problem of the SRB design. A convergence methodology is performed to the bearing design for global optimum results. Results obtained from NSGA-II runs of MOO have been used to draw Pareto-optimal fronts (POFs). Optimum bearing dimensions are selected by considering the knee-point solution on the POFs. Results indicate that the dynamic capacity of optimized bearing got enhanced thus increase in life of the bearing. A sensitivity analysis is conducted to identify the sensitivity of objective functions with design variables. The sensitivity analysis plays an important part in deciding the tolerances, which can be provided to design variables for the manufacturing of optimized bearings. The results obtained from MOO problem have been compared with available literature and are found to be better.