,
Abstract
Oscillatory transients are caused by switching operations, nearby lightning strikes, or other impulsive noise exciting electrical resonances present on the network. The most common in the distribution system is defined by IEEE 1159 as low frequency (under 5 kHz), and is often caused by capacitor bank switching. The peak magnitude of the voltage can rise to up to twice the normal value. Despite the high peak voltage, capturing a short-lived transient can be difficult with traditional RMS-based waveform capture triggering. Waveshape triggering is much more suited to capturing these transients; this methodology is presented here.
Oscillatory Transients
A typical oscillatory transient is shown in Figure 1. Likely a capacitor bank switch upstream has created an impulse that activated a system LC resonance, formed by the capacitor and local inductance from loads and transformers. This resonances causes the voltage to ring at the resonance frequency. This ringing persists for several of its periods, with the exact duration depending on the damping introduced by line and load resistance. A heavily damped resonance may not last for even a full period, while a very underdamped ringing can last for 10 milliseconds or more. In theory the peak voltage from a passive resonance circuit can reach twice the source peak voltage. In addition to overvoltage problems, these transients can cause complex loads such as variable frequency drive controllers to shut down as a preventative measure due to excessive waveform distortion. Thus, it’s important to capture and quantity these events.
Figure 1. Oscillatory transient from cap bank energizing
In Figure 2, the capture in Figure 1 is zoomed and analyzed in ProVision. The peak under and overvoltages are marked in per-unit (pu) values, that is, relative to the baseline voltage, as 0.33 pu and 1.32 pu. The voltages may be read directly off the point table in ProVision by moving the point marker to the lowest and highest peaks. The resonance frequency may be computed by measuring the time of one full oscillation period (again using the point table), and computing the reciprocal to get frequency. Here the period is 0.73 milliseconds, giving a frequency of roughly 1370 Hz. ProVision graph annotations were used to mark these measurements as shown in Figure 2.
Figure 2. Detailed breakdown of cap bank transient showing voltage excursions and resonance frequency
To perform this analysis, the event must be captured in the first place. Despite the 33% overvoltage on an instantaneous basis, the short duration (relative to a 60 Hz period) of the peak translates into a fairly small change in the RMS value during that cycle. The RMS Voltage stripchart (Figure 3) shows the event from that perspective. Here the one cycle min, max, and 1 minute average values are plotted. The event is marked with the orange arrow. The cap bank switch causes a steady-state step increase in RMS voltage of roughly 1V (from a 282V nominal). The extra voltage due to the transient itself adds another volt to the one cycle max reading, for a value of 284V. This small increase for that single-cycle max reading appears harmless in the RMS stripchart, and is too small to trigger with a traditional magnitude-based waveform trigger.
Figure 3. Peak transient voltage barely noticeable in RMS voltage stripchart with one cycle resolution
The key to capturing disturbances which change the waveform shape, but not the RMS value, is to use a metric that quantifies distortion from the ideal 60 Hz sine wave. There are several metrics that could be used, but Total Harmonic Distortion (THD) is well suited due to the sinusoidal shape of the ideal waveform. It’s especially appropriate in this case since the distortion itself is also sinusoidal (although not generally 60 Hz harmonic). The THD is a percentage comparison of the 60 Hz fundamental magnitude relative to the combination of all the other harmonics:
where Vn is the magnitude of the nth harmonic. A value of 0% indicates a pure 60Hz sine wave with no distortion. Typical steady-state voltage THD levels are 0.5 to 5%.
PMI recorders with advanced waveform capture measure THD every 60 Hz cycle. If the THD measurement changes by more than the threshold from one cycle to the next, the capture trigger is fired, and a waveform is recorded. One key aspect is that a high THD isn’t the criterion, rather it’s a change in THD from one cycle to the following cycle - this change indicates a change in the waveform shape, and thus some sort of event. With an oscillatory transient, the THD will increase for just a single cycle, then return to the previous value, before the capture has ended. If the voltage THD changes in a steady-state fashion (e.g. a nonlinear load turns on, bumping up the voltage THD), a capture will trigger during that transition (possibly an interesting waveform that may need examination), but since the new, elevated THD is no longer changing, no further captures from that change will happen.
The THD values for the event in Figure 1 are shown in Figures 4 and 5. Here the ProVision harmonic analysis tool is used to compute THD for various single cycles within the capture. In Figure 4, the grey selector window is over a “normal” cycle before the disturbance. The THD is displayed as 0.8%. In Figure 5, the grey selector window is centered around the ringing transient. Here the THD for this cycle is 10.4%, a very large change. The THD in the following cycles (not shown) is back to the baseline 0.8% again. Clearly THD can be used to distinguish the disturbance cycle from the rest, given a 10x change in value (compared to a 1% change in RMS value). The trigger value is shown in the waveform capture list report, as seen in Figure 6, with this event circled in orange.
Figure 4. THD before and after the transient is low, at 0.8% (above)
Figure 5. THD during the transient is high, at 10.4% (above)
Figure 6. THD trigger value shown in the waveform capture list report (above)
The THD change threshold is adjustable separately for each voltage channel (Figure 7, outlined in orange). The default is 5%, and a suggested range is from 2% to 10%. A lower value can be used to trigger on smaller disturbances such as zero-crossing distortion, commutation notches, etc., and a larger value may be needed on the neutral, or in situations with excessive baseline voltage distortion (such as from a generator or inverter). Note that this value is the amount the THD must change from one cycle to the very next (either higher or lower) to trigger. The baseline voltage THD is not relevant, just the change in THD. Because of this, even a value of 2%, which seems low as a steady-state THD, is actually still fairly significant.
Figure 7. THD trigger thresholds in the waveform capture configuration dialog
Conclusion
Oscillatory transients can produce large peak voltages or notches in a cycle, degrading insulation and possibly tripping off VFDs and other sensitive loads. Capturing these events, which often produce very little change in the overall RMS voltage, requires a trigger based on changes in the waveform shape itself. The voltage THD is such a measurement, and is adjustable to cover a range of triggering situations.
Chris Mullins
VP of Engineering and Operations
cmullins@powermonitors.com
http://www.powermonitors.com
(800) 296-4120