No abstract

Accurate prediction of cavitation is an important feature in hydrodynamic bearing modeling. Especially for thermo-hydrodynamic modeling, it is crucial to use a mass-conservative cavitation algorithm. This paper introduces a new mass-conserving Reynolds cavitation algorithm, which provides fast convergence and easy implementation in finite element models. The proposed algorithm is based on a variable transformation for both the pressure and mass fraction, which is presented in the form of a complementary condition. Stabilization in the streamline and crosswind direction is provided by artificial diffusion. The model is completed by including a simple and efficient thermal model and is validated using the numerical values of a reference plain journal bearing experiment under steady-state conditions. In addition, a transient analysis is performed of a journal bearing subjected to a harmonic load. It is shown that the proposed cavitation algorithm results are in good agreement with the reference measurement results. Moreover, the algorithm proves to be stable and requires only a small number of iterations to convergence in the Reynolds-based finite element model.

Journal bearings are used to support rotors in a wide range of applications. In order to ensure reliable operation, accurate analyses of these rotor-bearing systems are crucial. Coupled analysis of the rotor and the journal bearing is essential in the case that the rotor is flexible. The accuracy of prediction of the model at hand depends on its comprehensiveness. In this study, we construct three bearing models of increasing modeling comprehensiveness and use these to predict the response of two different rotor-bearing systems. The main goal is to evaluate the correlation with measurement data as a function of modeling comprehensiveness: 1D versus 2D pressure prediction, distributed versus lumped thermal model, Newtonian versus non-Newtonian fluid description and non-mass-conservative versus mass-conservative cavitation description. We conclude that all three models predict the existence of critical speeds and whirl for both rotor-bearing systems. However, the two more comprehensive models in general show better correlation with measurement data in terms of frequency and amplitude. Furthermore, we conclude that a thermal network model comprising temperature predictions of the bearing surroundings is essential to obtain accurate predictions. The results of this study aid in developing accurate and computationally-efficient models of flexible rotors supported by plain journal bearings.