Tilt Pad Bearing Distributed Pad Inlet Temperature With Machine Learning—Part I: Static and Dynamic Characteristics

Abstract

AbstractUncertainty in mixing coefficients (MCs) for estimating pad leading-edge film temperature in tilt pad journal bearings reduces the reliability of predicted characteristics. A three-dimensional hybrid between pad (HBP) model, utilizing computational fluid dynamics (CFD) and machine learning (ML), is developed to provide the radial and axial temperature distributions at the leading edge. This provides an ML derived, two-dimensional film temperature distribution in place of a single uniform temperature. This has a significant influence on predicted journal temperature, dynamic coefficients, and Morton effect response. An innovative finite volume method (FVM) solver significantly increases computational speed, while maintaining comparable accuracy with CFD. Part I provides methodology and simulation results for static and dynamic characteristics, while Part II applies this to Morton effect response.