On the <scp>tilting‐pad</scp> thrust bearings hydrodynamic lubrication under combined numerical and machine learning techniques
Konstantinos P. Katsaros, Pantelis G. Nikolakopoulos
Abstract
Abstract Thrust bearings are machine elements designed to support axial loads in rotating machinery. The hydrodynamic lubrication analysis of such bearings has been a major subject for many studies over the years, leading to important conclusions for design parameters that affect their optimal operating conditions. Furthermore, the last few years, the influence of the industry 4.0 concept has brought new tools that can revolutionise bearings' design. The aim of this study is to combine numerical analysis and machine learning techniques in order to identify optimal thrust bearing's hydrodynamic designs. For this purpose, the Reynolds equations are solved using the finite difference technique on a 2‐D grid of a tilting pivoted bearing's pad. The bearing pressure distribution; load carrying capacity and friction are calculated for multiple operating conditions. The data produced are used as input for the training of regression models that predict the behaviour of the thrust bearing for a wide range of loads and rotating speeds. Simple and multi‐variable, linear, polynomial and SVM regression models are compared for their accuracy to predicting the bearing's operating conditions. The major findings related with three different lubricants, a monograde SAE 30, a multi‐grade SAE 10W40 and a bio‐lubricant AWS 100 that are compared and their optimal operating conditions are suggested in terms of minimum friction force and maximum load carrying capacity for the bearing pad. AWS 100 is found to be the most suitable lubricant that provides the bearing with low operating friction and high load carrying capacity in all studied cases.