Interpretation of Dynamic Data Plots for Troubleshooting and Resolving Vibration in Large Rotating Machinery

Abstract

Vibration in large turbine-generator rotor trains can be diagnosed by recognizing the specific correlated patterns visible in data presented in Bode, polar (Nyquist), shaft centerline, shaft orbit, and frequency spectrum plots, among others. Properly interpreting these data plots requires the recognition of true rotor behavior when affected by various faults. By the authors’ experience, the most prevalent faults seen in practice are the direct and indirect effects of distributed mass eccentricity within individual rotors or across an assembled rotor train. The presence of mass eccentricity leads to various effects of inertia during machine operation (often torque or load dependent), which create the principal graphical data signatures to look for and utilize in diagnostic assessment. The graphical signatures that result from the presence of imposed forces “fighting against” a rotor’s natural inertia are readily and independently recognizable, even when in combination with additional contributors of dynamic vibration such as that caused by resonant responses within critical speed regions, rubs, and bearing/fluid instability. This paper describes and presents typical data signatures in the various types of plots, including examples of shaft bows or other unresolved distributed mass eccentricity on individual rotors, plus examples of rotor misalignment on assembled multi-rotor machines caused from bent or off-square coupling faces and radially offset, misaligned bearings. Furthermore, understanding the most prevalent root causes underlying the diagnostics leads to the recognition of proactive shop strategies to prevent these faults and to substantially improve the probability of a successful startup following a turbine or generator rotor service outage.