|
Related Products: |
EGSA-013, EGSA-015, EGSA-210, EGSA-220, EGSA-310, EGSA-320, EAMA-001, EAMA-0003, EAMA-005 |
|
Related Documents: |
EZDP-2100, EZDP-2104 |
|
Type: |
Operation |
|
Keywords/Labels: |
rotor flux, InsightCM, RFM baseline waveforms, signal processing, operating state, alarm management |
|
Summary: |
This topic describes how to set up the Rotor Flux Monitoring System. |
The data and waveforms collected by the Cutsforth Generator Field Monitoring System can be used to determine the level of deterioration in the insulation of the rotor windings that result in shorted winding turns.
Damage to the winding insulation leads to:
-
Increased demand on excitation systems
-
Imbalances in the rotor field
-
Increased vibration
-
Degradation of flux field integrity
-
Reduced generator capacity
-
Increased risk of rotor ground fault
-
Potential unexpected generator failure or forced outage
Advance knowledge of insulation failure and real-time feedback of the extent of the damage helps the plant determine the best course of action for potential continued operations and planned vs. unplanned maintenance activities.
The Cutsforth Generator Field Monitoring System acquires signals from the generator’s existing rotor flux sensor at over 50kHz. The rotor flux waveform is then analyzed to determine the following:
-
The beginning and end of each rotation cycle
-
Which coil is closest to flux density zero cross
-
The flux density peak amplitude value of each coil
-
The percent deviation between the leading and lagging peak amplitude value of each coil
This section offers a brief technical overview of the rotor flux capabilities of the Cutsforth Generator Field Monitoring System:
-
Rotor flux signals processed in real time.
-
Peak voltage amplitude of each coil is identified as waveforms are processed.
-
Monitoring system plots the Flux Density Curve (FDC) and identifies the Flux Density Zero Crossing (FDZC).
-
The FDZC is the point at which the integral of the flux waveform is zero volts, which affords the greatest sensitivity to detecting a shorted turn in a coil.
-
-
Deviation in voltage between poles is calculated in real time.
-
Deviation in voltage between poles is calculated for coil nearest the FDZC and reports the numbers of shorted turns on that coil.
-
A high-speed waveform and full feature set is captured each time the FDZC aligns with a different coil.
-
Total flux in Volts RMS is calculated and displayed.
-
Offers multi-feature alarming.
-
e.g. Total flux and number of shorted turns on FDZC coil
-
This section covers the viewing of rotor flux data in InsightCM™. For assistance with getting your Generator Field Monitoring System connected to and communicating with InsightCM™, refer to these documents, or contact Cutsforth Support if further assistance is needed:
-
EZDP-2092 – Installing InsightCM™
-
EZDP-2093 – Installing InsightCM™ Images from Windows 10 Server
-
EZDP-2095 – InsightCM™ Configuration for Generator Field Monitoring
-
EZDP-2097 – Activating InsightCM™ Server License
Monitoring for RFM conditions and alarms can only be operational when the generator is online or false faults will be registered. To do this, you must collect baseline waveforms according to these instructions to be able to accurately determine how to set the generator ON operating state.
-
From the InsightCM Data Viewer page, click Force Trigger to collect the following baseline waveform sets:
-
Force Trigger just prior to the generator spinning up on turning gear
-
Force Trigger during spin up just prior to the generator coming online
-
Force Trigger just after the generator comes online and is at lowest load
-
Force Trigger when the generator is at normal operating load
-
-
Annotate the waveforms by clicking Add Comment.
-
Save these waveforms so they are never deleted by selecting Data Event > Retain Data Event.
Note that these four waveforms will possibly NOT be properly analyzed or display properly in the system at this time from the default waveform processing settings. Export these waveforms and send them to support@cutsforth.com for analysis at which time we will advise on each individual setting required on the Signal Processing Properties tab of InsightCM.
Cutsforth offers many user-definable RFM parameters so that analytics and calculations are specifically tailored to the unique flux signal that each generator produces. It is difficult to advise what the settings might be for any given unit. On first power up, it is advised to perform the waveform collection steps referred to in Collect RFM Baseline Waveforms, and send the resulting TDMS files to Cutsforth for analysis. For proper interpretation, Cutsforth must have:
-
The TDMS exports of the four waveforms collected
-
The number of Poles
-
The number of Slots per Pole
-
The number of Windings per Coil
-
The MW's capability of the unit
|
Number of Poles |
The number entered in this field indicates the quantity of poles in the generator. A typical generator has 2 poles. Some have 4 poles. |
|---|---|
|
Smooth |
If the signal is excessively noisy, check this flag, and smoothing will be applied to the measured flux signal to reduce the noise. |
|
Smoothing Length |
If the Smooth flag is checked, the signal is smoothed with a filter that is Smooth Length points long. |
|
HP FFT Index |
To help separate the flux wiggles from the lower frequency power cycle signal, the ‘HP FFT Index’ specifies the frequency above which a highpass filter is applied to the flux signal. The frequency is specified as the number of cycles per power period. For example, a value of 20 means that frequencies below 20 cycles per 1/60 of a second (for 60 Hz generation) are removed by this highpass filter. As another example, a value of 1 means one cycle in 1/60 second, which is the power frequency of the generator. |
|
Cycle Detection Tolerance Angle |
After the cycle is detected, the positive-going flux wiggles should reside between the ‘Start Angle’ and ‘End Angle’ of the power cycle. (The negative-going flux wiggles must fit between this range after adding 180 degrees.) Any wiggles outside this region are rejected. Since the cycle detection start index has some “jitter”, this tolerance angle is used to allow that angle window to expand a little on both sides by this tolerance angle. This tolerance angle should never be more than about half the angular cycle length of a flux wiggle. For example, if 12 wiggles (due to 6 coils) occur between the ‘Start Angle’ and ‘End Angle’, which for example might span 150 degrees, then each wiggle spans about 13.636 degrees (from 150/11), so this tolerance angle should not be more than 13.636/2 or about 6.8 degrees. |
|
FDZC Coil Tolerance Angle |
Capturing the flux signal when the FDZC is near a coil position implies that a tolerance of that position is necessary, since the chance of the FDZC being exactly on a coil position is small. When the FDZC is at a coil position +/- this tolerance angle, the flux signal is captured. |
|
FDZC Number of Inaccessible Coils |
Since the FDZC position may never reach coils 1 and 2 at full load, this number allows the flux signal to capture the coil amplitude deviations when the FDZC is at this number plus 1. For example, if the number of inaccessible coils is 2, then when the FDZC is at coil 3, the results for coil 2 and 1 are also captured. |
|
LPF Cycle Detection |
If this flag is set to true (checked), then the LPF is applied. |
|
LPF Passbands |
As mentioned above, some flux waveforms have structure that can confuse MJCD method. If the frequency of the wiggles is faster than any other structure, a lowpass filter can remove the wiggles and focus on the rest of the structure. The utility of the MJCD is to find a maximum of the ΔY/ΔX values that is close to the “between” region as this parameter walks along the waveform so that the cycle start can be placed in this region. The following figures show the effect of this lowpass filter. The first example is from an LGE coal-fired generator. Note that the amplitude of the wiggles in the “between” region are about the same as the flux wiggles. |
Applying the LPF to the signal with a passband of 0.022 results in the following signal.
Note that the flux wiggles are essentially eliminated and the overall background structure of the wiggles is preserved. The ΔY/ΔX value has a very distinctive maximum at the center of the “between” region.
A good starting value of the LPF Passband is computed as follows.
-
LPFPassband = (AngRange/360)/NumWiggles
For the signal above, with NumWiggles = 2*NumCoils = 2*8 = 16, and AngRange = 150–32 = 118, the LPFPassband is (118/360)/16 = 0.0205.
Adjusting this value slightly up or down can help improve the detection of the wiggle extrema.
-
Remove the instructions for the system to collect waveforms when in the Default operating state.
-
Create an ON operating state based on the Flux Probe RMS level as recommended by Cutsforth as a result of analyzing the TDMS waveforms collected on first power.
When the generator is offline, the Flux RMS amplitude will be low, and when the generator is online, this level will rise abruptly. The level for the ON state should be such that the RMS signal is produced by an online generator and that it is stable for at least a minute before collecting and analyzing waveforms.
It is initially recommended that trend lines be set to collect every 15 minutes and that waveform sets be analyzed every 4 hours.
If the Cutsforth RFM template was used to create the asset, then the default alarms will be set such that when the number of shorted turns on any coil becomes greater than one, an alarm is triggered. It may be desired to change this such that an alert occurs when less than one turn is shorted, or more if a unit already has an existing set of shorted turns present on any given coil.
The InsightCM™ Data Viewer screen layout is fully customizable according to the user’s preference and also allows the viewer to observe multiple data points in both graphical and numerical form at the same time.