Dynamic Identification of 280mm Diameter Tilting Pad Journal Bearings: Test Results and Measurement Uncertainties Assessment of Different Designs

Tilting pad journal bearings (TPJB) are crucial elements in turbomachinery applications providing stiffness and damping characteristics that determine rotor system dynamic behaviour. Hence, a correct design and an accurate dynamic properties prediction is fundamental for the successful industrial operation of rotating machinery. Current design trends in turbomachinery aiming at higher efficiency and power through weight optimisation and higher operating speeds determine, in centrifugal compressor applications, the development of large flexible rotors that are particularly critical from the rotordynamic standpoint. The dynamic feasibility of this type of machine is based on bearing solution and its stiffness and damping characteristics that must be predicted with a certain level of confidence that must be known to increase the accuracy of rotordynamic behaviour and avoid unpredicted vibration issues when rotors are operated. Furthermore, large centrifugal compressors make the bearings operate at very high peripheral speed where the transition from laminar lubrication to the turbulent regime occurs, leading to increasing predictions accuracy.

To this purpose a test campaign on different large TPJB solutions operating in turbulent lubrication regime has been performed to evaluate pro and cons of each bearing configuration. The tested bearings differ each other in terms of bearing technology, pivot design and pad micro-geometry, lubrication-layout and manufacturing process.

The experimental activity has been carried on a dedicated test rig [1] specifically designed for investigations on large size high-performance bearings for turbomachinery. In the present work both static performance and dynamic identification of the tested TPJB solutions are presented and compared to numerical model predictions.

The test configuration is characterised by the test article (TA) floating at the center of a rotor supported by two rolling bearings. The TA is statically loaded by a hydraulic actuator and excited by two orthogonal hydraulic actuators with a single tone or multitone dynamic load. The linear coefficient computation method is based on the dynamic measurement of forces, accelerations and relative displacements of rotor and bearing. In the experimental setup, an actuator applies a static force to which the dynamic loads are superimposed.

An uncertainty analysis has been performed to validate the experimental results and increase the validity of results comparisons from different bearing technologies. In the literature, the experimentally determined bearing dynamic characteristics are reported with their uncertainty interval, usually with 95% confidence. Few papers give details about the adopted technique to perform such an analysis on the results of the journal bearing dynamic coefficient identification. In the present work a statistical approach [2] developed to estimate the systematic error related to the computation TPJBs dynamic coefficients is presented and applied to tested configurations.

P.Forte, M. Libraschi and E. Ciulli, "Set-up of a novel test plant for high power turbomachinery tilting pad journal bearings," Tribology International, vol. 127, pp. 276-287, 2018.

M. Barsanti, E. Ciulli and P. Forte, "Random error propagation and uncertainty analysis in the dynamic characterization of Tilting Pad Journal," in Journal of Physics: Conference Proceedings, vol.1264, 2019.