Measurement and Prediction of the Journal Circumferential Temperature Distribution for the Rotordynamic Morton Effect
Academic Article
Overview
Research
Identity
Additional Document Info
Other
View All
Overview
abstract
The journal is the part of a shaft that is inside a fluid film bearing and is usually assumed to be circumferentially isothermal. Recent work has shown that under certain vibration conditions, a significant temperature difference (T) can develop around the journal circumference. The T may cause the shaft to bend leading to a synchronous vibration instability problem, termed the Morton effect (ME). A test rig was developed to verify the asymmetric journal temperature of the ME and its prediction using a five-pad tilting pad journal bearing (TPJB) operating with an eccentric shaft to replicate a circular vibration orbit. The bearing is tested at various conditions including: supply oil temperature at 28C and 41C, bearing operating eccentricities of zero and 32%Cb, and rotor speed up to 5500rpm. The journal temperature distribution is recorded with 20 sensors located around the journal circumference, and the measurements provide a benchmark for predictions from a time transient model with the three-dimensional (3D) fluid and solid finite element method (FEM), and with a simplified ME prediction approach using only steady-state results. The test results follow the predictions exhibiting a sinusoidal-like temperature profile around the circumference with an angular lag of the hot spot location behind the high spot location (angular position on the rotor that arrives at the minimum film thickness condition each rotation) by a speed-dependent angle. Increasing the supply oil temperature reduced the journal T, while increasing the bearing operating eccentricity increased the journal T. The agreement between the test and predicted results is significantly better for the 3D FEM transient model than for the steady-state-based model in terms of journal T and hot spot position. An improved version of the latter approach is proposed and yields significantly better correlation with the test measurements.
