Tong, Xiaomeng (2018-05). Nonlinear Prediction of Thermally Induced Rotor Instability (Morton Effect) and Experiment Verification. Doctoral Dissertation. Thesis uri icon

abstract

  • The Morton effect (ME) is a thermally induced rotor instability problem that most commonly appears in rotating shafts with large overhung masses and supported by fluid film bearings. The time-varying thermal bow, due to the asymmetric journal temperature distribution, may cause intolerable synchronous vibrations that exhibit a hysteresis behavior with respect to rotor speed. First discovered by Morton in the 1970s, and theoretically analyzed by Keogh and Morton in the 1990s, the ME is still not fully understood by industry and academia experts. Traditional rotordynamic analysis generally fails to predict the potential existence of ME induced instability in the design stage or troubleshooting process, and the induced excessive rotor vibrations cannot be effectively suppressed through conventional balancing, due to the continuous fluctuation of vibration amplitude and phase angle. The early ME prediction methodologies adopt three assumptions which simplify the ME analysis and meanwhile reduce its accuracy in predicting the ME response. (1) The ME is assumed to be static and speed-dependent and thus its stability can be analyzed with the classic control theories in the frequency (speed) domain. Nevertheless, the ME is proven to be highly transient and process-oriented, and not only the operating speed but also the transient speed acceleration can cause instability. (2) The thermal bow effect is only considered at the rotor overhung side and is simplified with thermal imbalance, which neglects the ME induced thermal bending moment. (3) Early ME prediction only focuses on the single rotor-bearing side and neglects the coupled effect on rotordynamics and bearing behavior from the other side. The assumptions above are abandoned in the current dissertation and replaced with the high-fidelity transient finite element analysis (FEA). The FEA is based on the recent thermo-elasto-hydro-dynamic bearing analysis and featured with (1) the nonlinear transient prediction of ME in the time domain through numerical integration, (2) analyzing both rotor-bearing sides to simultaneously account for the overall thermal bow effect, and thus the coupled rotordynamics effect from both rotor-bearing sides can be considered, and (3) the bowed rotor method, which also considers the thermal bow at the rotor midspan besides the rotor ends. The thermal bending moment is also included in the rotordynamics model, making this method more accurate than the traditional thermal imbalance method. Moreover, the ME testing rig is designed to measure the temperature distribution with 20 sensors across the journal circumference and record the peak-peak temperature difference (?T) with respect to various operating conditions. Supply oil temperatures of 28?, 41? and bearing eccentricity of 0, 32%Cb are tested with rotating speed up to 5500 rpm. The journal ?T is found to increase almost linearly with speed up to 5500 rpm, slightly increase with eccentricity from 0 to 32%Cb, and decrease by nearly 20% by raising the supply temperature from 28? to 41? due to reduced viscosity. The proposed high-fidelity ME analysis is applied to predict the journal temperature distribution against the measurements, and predictions are also compared with the simplified ME analysis proposed in earlier literature. The high-fidelity model demonstrates a better accuracy in predicting the maximum rotor temperature Max?(T), the peak-peak temperature difference Max?(?T), and the phase lag between the high spot and hot spot compared with the simplified ME analysis.

publication date

  • May 2018