The latest release of SignalCalc Turbo, version 2.0.198, includes a valuable new feature called Virtual Probe Rotation. Frequently requested by turbomachinery manufactures and rotating machinery engineers, Virtual Probe Rotation greatly improves analysis capability on certain types of large rotating equipment, especially in the power generation industry.
Shaft vibration response due to forces such as unbalance usually occurs in the vibration plane with the most compliance. With large steam and gas turbine generators, this vibration plane is often fairly well aligned with the true horizontal plane. Support stiffness for bearing pedestals, for example, is generally going to be more compliant in the true horizontal plane due to construction. To the contrary, structural stiffness from the supporting foundation and bearing pedestal, combined with hydrodynamic restraining forces from the bearing itself as a result of gravity load of the rotor, produce a net stiffness which is often greatest in the true vertical plane as shown in Figure 1.
For these reasons, it is desirable to install proximity probe pairs in the true vertical and horizontal orientations. As illustrated in Figure 1, however, it is much more convenient to install pairs of radially mounted proximity sensors at 45°L and 45°R orientations because of the horizontal split line and construction of the bearing housing. Proximity sensors at 45°L and 45°R orientations provide adequate vibration measurement for a machinery monitoring and protection standpoint. However, for diagnostics purposes, important information can be missed with this arrangement.
Machine response in the true vertical and horizontal planes due to resonances, for example, can be quite different. When it is possible to install proximity probe pairs in the vertical and horizontal planes, differences such as resonant frequency, peak amplitude, and phase response are quite apparent in measured data. But when these sensors are installed at 45°L and 45°R, the observed response is effectively a composite of two separate responses and often difficult to interpret.
For example, the 1X filtered shaft orbit of a generator shaft in Figure 2 displays the highest displacement in the horizontal plane. This orbit is drawn by combining the 1X filtered signals from two separate sensors at 45°L and 45°R orientations in X-Y coordinates. While the orbit provides an accurate view of shaft vibration in two dimensions, neither sensor is seeing the full magnitude. In fact, with direct analysis of 1X amplitude and phase response of the signals from each of the two individual sensors, it is not possible to distinguish machine response in the true horizontal plane from the true vertical plane.
Virtual Probe Rotation solves this problem. This feature enables the user to virtually rotate orthogonal sensor pairs to a different angular orientation. The resulting vector data graph, such as a filtered 1X bode or polar, displays the result of a vector transform calculated based on rotation of the sensor pairs through a user-specified angle. This allows the user to view the machine response in the true vertical and horizontal planes. The formulas for the vector transforms in rectangular format, yielding the vibration vectors as seen by the virtual probes, is as follows:
Figure 3 shows Bode and Polar graphs of 1X running speed vector data from the same generator shaft, using a sensor mounted at 45°R, before the vector transform is applied. As can be seen, machine vibration shows some increase above 3,000 rpm with flat phase response, so it is difficult to determine if this is due to effects of a machine resonance.
In Figure 4, Virtual Probe Rotation has been used to apply the vector transform to the same Bode and Polar graphs by rotating the probes through an angle of 45° in the clockwise direction. As can be seen, the probe pair for channels 11 and 12 has been rotated to align with the true vertical and horizontal planes. The Bode and Polar graphs for channel 11 now show the actual machine response in the true horizontal plane. Machine response above 3,000 rpm is much more pronounced, and a corresponding change in phase can now also be observed. Together, this data points strongly toward the effects of a machine resonance.
Without virtual probe rotation, separate resonances in the true vertical and horizontal planes can be masked or misinterpreted. Sometimes machine response can be interpreted as a “split resonance”, with partial response being observed for both resonances by both sensors. With the transformed data obtained from virtual probe rotation, however, the diagnostics engineer is now able to cleanly separate machine response in the true vertical and horizontal planes. This enables more accurate identification of machine and structural resonances, and also provides valuable insight into understanding of machine unbalance response when balancing large rotors in the field.
If you have any comments or questions regarding SignalCalc Turbo, please contact Rob Bloomquist, Product Manager, at firstname.lastname@example.org or +1 (408) 216-8405.