One of the most neglected tools in the vibration analysis toolbox is phase. This neglect is partially due to the extra time and effort often necessary to collect the information. Many analysts are also not properly trained in phase measurement and analysis.
With the top-of-the-line data collectors now available, little time and effort are necessary to add this valuable data to route collection. Useful phase data can now be taken on a route, simply by carrying along an extra accelerometer.
Absolute Phase vs. Relative Phase and Cross Channel Phase
Absolute phase is the actual phase of the vibration at the frequency of interest, as the accelerometer placement relates to a reference point on the rotor. Absolute phase measurements must be taken with some type of tachometer or key phasor.
Relative phase is simply the phase difference of the vibration phase at the frequency of interest between a reference point and a second point. The analyst may choose any two points to apply the accelerometers.
In many cases, absolute phase is not necessary because many diagnoses are made by simply analyzing relative phase. Cross channel phase is a measurement of relative phase. Cross channel phase measurement requires a two-channel analyzer/data collector, but not all two-channel data collectors will record cross channel phase as route data.
Bent rotor, misalignment, looseness and other problems can be more reliably diagnosed when cross channel phase data is included with amplitude and frequency measurements.
Misalignment
For example, a cross channel phase point is built and two accelerometers are placed axially on the bearings on each side of a coupling. If the accelerometers are placed in the same quadrant (i.e. both on top, left side, right side or bottom) and same direction, a coupling with angular misalignment (Figure 2) will exhibit a delta phase of nearly1 180-deg (±30-deg) (see Figure 1).
Figure 1
If you cannot place the accelerometers in the same direction, and they must be placed opposed to each other (see Figure 4), a 180-deg phase error is introduced because of accelerometer orientation. This error must be accounted for by either adding or subtracting 180-deg from the result. Before making the correction, a delta phase of approximately 0-deg (± 30-deg) would indicate probable angular misalignment.
Taking radial cross channel phase measurements across a coupling will help diagnose parallel (offset) misalignment (Figure 2). As in the axial measurements, when the accelerometers are mounted in the same quadrant and same direction, misalignment is more likely the closer the delta phase approaches 180-deg. Another good indicator of misalignment is when the delta phase between the radial vertical and horizontal on the same bearing is nearly 0-deg or 180-deg.2
Figure 2
Figure 3. Cross channel phase screen
Notice the information available in the screen capture (Figure 3) taken as a routine route point in a polar plot showing phase and amplitude of the two channels. Notice A is the reference channel and is shown to be at zero on the polar plot.
The exact amplitude of both channels:
Channel A = 0.311 ips
Channel B = 0.295 ips
The relative phase angle between the two channels (Delta Phase) is in degrees. Notice the delta phase is 1730 which is essentially 1800.
The phase coherence value is 1.
The frequency of the vibration data represented is 3270-rpm.
Bent Rotor
In another example, the same type of axial cross channel phase measurement is set up as above, but on bearings that are on the same shaft with the rotor between the bearings. If the delta phase is 180-deg (± 30-deg) when both accelerometers are oriented in the same direction and same quadrant, a bent rotor or severely cocked bearing is indicated. If the analyst is unable to place the accelerometers in the same direction and they must be placed opposed to each other, the orientation error must be corrected.
If the accelerometers are opposed, and the orientation error is not corrected mathematically, a delta phase of 0-deg (± 30-deg) would indicate a bent rotor (see Figure 4) or severely cocked bearing. If this is indicated, and it is not possible to measure the physical run-out of the rotor, a run up or coast down should be performed to rule out the possibility that the rotor itself is resonating. This should be done because the mode shape of a resonating rotor can mimic a bent rotor.
Figure 4
If this shaft (Figure 4) were bent and monitored without phase, a high one-time vibration would be present and time would likely be wasted trying to balance the rotor. While you may be able to balance out some of the radial vibration, most of the axial will remain.
Looseness
Measurements can be taken on the foot of the machine (or bearing housing) and on the base of the machine, or on the base and the ground or floor. If everything is tight, the vibrations should be in phase. As looseness progresses, the delta phase will increase along with an increase in amplitude difference between accelerometers. Figure 5 shows two sets of accelerometers. The accelerometers of Set #1 are taking relative phase between the foot and the base, and the accelerometers of Set #2 are taking relative phase between the base and the floor.
Figure 5
Trending
Route phase data can be trended and compared to amplitude data (see Figure 6). In this way, a worsening situation can be evaluated over time. For trending phase, as with amplitude trends, repeatedly placing accelerometers in the same position is always necessary.
Figure 6. Trending software
Changing phase is a red flag. If the delta phase changes more than 30-deg, even if it stabilizes on the change, this clearly indicates a significant change on the machine. It may be a dirty rotor causing imbalance, or it could be a loose foot reducing the stiffness and bringing the resonant frequency of the machine or structure near operating speed. This latter condition will more likely be detected when the reference accelerometer is on a different machine in a machine train. If both accelerometers are on the same machine there may be no relative phase change, even when the absolute phase changes substantially.
When you have phase, amplitude and frequency, the vibration cause will be easier to diagnose accurately. As the tools available to the analyst improve, diagnoses should become easier and require less effort to reach the same or higher levels of confidence. In addition, the more the analyst uses each particular tool, the more competent he or she will become at analyzing the data, detecting machine faults and accurately determining their sources.
Notes and References:
- For practical purposes, a 7 ± 30-deg variation from a phase quantity is considered negligible.
- Avanti Notebook 2nd Edition, Ralph T. Buscarello, Update International.
Pumps and Systems, May 2009