Optimal Designs of Tilting-Pad Thrust Bearing operation with the combination of numerical and machine learning techniques

K Tserpes,Konstantinos P Katsaros,Pantelis G Nikolakopoulos,G Lampeas,S Pantelakis

doi:10.1051/matecconf/202134903003

Abstract

Hydrodynamic thrust bearings are machine elements used in many rotating machinery in order to support axial loads. The investigation of the lubrication in such mechanisms using numerical analysis methods has been the major subject of many studies over the years. Furthermore, the evolution of technology in the last decade brought the concept of industry 4.0 and machine learning techniques have started to play important role in the operational optimization of such assemblies. The aim of this study is to examine optimal designs of tilting pad thrust bearings by combining numerical analysis and machine learning techniques.

Highlights

Hydrodynamic thrust bearings are machine elements designed to carry axial loads in rotating machinery
Simulations were performed for rotating speeds from 2000 to 12000 rpm and applied external loads from 650 to 2300 N
The minimum film thickness was 4μm in all studied cases

Summary

Introduction

Hydrodynamic thrust bearings are machine elements designed to carry axial loads in rotating machinery. Studies of thrust bearing’s hydrodynamic lubrication with numerical analysis methods have been the subject of many works over the last years. In terms of the traditional numerical analysis and the contribution of machine learning techniques the target has been set to minimize friction losses and build hydrodynamic thrust bearing designs with a “greener” environmental effect [1,2]. It is well known that, hydrodynamic thrust bearings have a variety of commercial and industrial applications. A multi-grade SAE 10W40 and a bio-lubricant ASW 100 are used in the current investigation in a variety of loads and rotational speeds. A data lake is built and used as an input to a machine learning algorithm in order to investigate optimal combinations for the bearing’s operation. AWS100 is found to be the optimum lubrication solution for the examined cases

Basic Assumptions

Governing Equations

Viscosity Model

Numerical Analysis

Machine Learning Techniques

Results