,
Abstract
A little-known graph in ProVision, the RMS Waveform Capture graph, can actually be one of the most illuminating. In cases where interval graphs don’t have enough detail, and raw waveform capture plots “hide the forest for the trees,” the RMS Capture graph is a good intermediate representation of complex RMS change events.
RMS Interval Graphs
The RMS Voltage and Current Interval Graph (Graph, RMS Interval, RMS Voltage and Current from the main ProVision menu) is a good starting point for analyzing a data file, and usually the first graph where a sag or other event is noticed. Figure 1 shows an interval graph on a 277V wye service to a gym. The graph has been zoomed to show a voltage event (circled in red). The voltage rises on phase A, and sags on phase B, and the event merits further analysis.
Figure 1. Voltage event that merits further analysis. Voltage rises on phase A, and sags on phase B.
We can zoom in even further in the interval graph (see Figure 2), and turn on the Toggle Point Table to get the min/max voltages, but the data in the interval graph is still limited. The stripchart interval was set to 1 minute in this recording, and thus only a one cycle max, min, and average are present at each data point. From the interval graph, all we can tell is that phase went from roughly 281V to 291V, and phases B and C sagged to around 252V and 255V, respectively. Those are one-cycle readings, but we don’t know where they happened during the one minute interval, or how long those values lasted. The current readings don’t change drastically, possibly a sign that this event was not caused by the monitored load.
Figure 2. Toggle Point Table displays min/max voltages
Waveform Capture Graphs
The next stop in the analysis is often the Waveform Capture graph. The waveform for this event is shown in Figure 3. It’s clear that the current on phase B was involved, but the current decreased rather than increased, which again points to the load not being the source of the event. The voltage waveforms look relatively normal though, and it’s not obvious why the phase B current was reduced for two cycles. The 10 volt swell on phase A voltage is not visible, and the 28V sag on phase B is difficult to spot. The phase B current looks much different than the phase B voltage shape.
Figure 3. Waveform Capture graph
In this case, the raw sinusoidal waveforms are showing too much detail to be useful. An RMS voltage change is difficult to spot in the sine wave, due to the scaling involved. In this example, phase A’s RMS value rose 10V, from a starting value of 280V. On a 280V RMS sine wave, the peak instantaneous voltage is 1.41 (square root of 2) times the RMS value, or 395V. The full sine wave goes from +395V to -395V, so the full span is 790V. This is displayed on the waveform graph as +/-400V, or an 800V span. Given this very large span, it’s easy to see why a 10V swell, or even a 30V sag would be difficult to spot. A 10V RMS change in sine amplitude translates to just a few pixels on the graph, when the scaling is 800V on the Y-axis. In Figure 3, the swell is just barely visible on the monitor, and the sag in phase B voltage is easy to miss. Since the RMS values are “hidden” in the scaling, it’s difficult to tell how long the sags or swells last.
RMS Capture
The same event is much more easily analyzed with the RMS Capture graph (Graph, Waveform Capture, RMS Capture in the ProVision menu). This graph is computed by ProVision from the raw waveform data. A sliding cycle RMS window is applied to the data points. The window slides point-by-point, and the RMS value of the data in the window is plotted as a function of time. Since at least one cycle’s worth of data is needed to compute the window value, the overall length of the RMS graph is one cycle shorter than the original waveform. The resulting graph is a “continuous” RMS graph, showing the RMS voltage and current point-by-point. Because the graph is autoscaling to RMS values instead of peak sinewave values, the graph scale works in our favor, instead of working against us, in terms of visualizing RMS changes.
The RMS Capture graph for this event is shown in Figure 4. The nature of the event is much more apparent. The RMS values here are even more precise than the interval graph, since the sliding window has more time resolution than the one-cycle stripchart max/min values. We see that the RMS value for phase A rises from 281V to over 293V, and the RMS current rises roughly in step (the time separation between the voltage and current plots are due the phase shift, from the delta connection, and load phase angles). The 30V sag on phase B is also obvious, and in this view, the phase B current shape closely follows the voltage shape, indicating that the current was merely following the voltage. In this graph, the phase C voltage and current appears more out of step than phase B, unlike the waveform graph, but the scaling shows that the change in phase C current was not nearly as large.
Figure 4. RMS Capture graph
Two key pieces of information can be seen from the RMS Capture graph. First, the entire event only lasted six cycles. The RMS values are back to their nominal values by 140 ms in the graph, and the worst-case excursion was closer to three cycles long. Second, given the way the current follows the voltage, this appears to be a voltage event from upstream, and not caused by the load. Possibilities include a transmission or subtransmission line-ground fault (which could look like this after the delta transformation), or a recloser operation on a neighboring circuit.
This recording contains another good example of the power of the RMS Capture graph. Again, starting with the RMS Voltage and Current interval graph, we see a sag on phase A, in Figure 5. We can’t tell how long the sag lasted, but we can tell that it’s around 10V on phase A, and the other voltage and current channels seem relatively unaffected. In this graph, the waveform annotation marker is enabled (the vertical line down the graph), indicating a waveform capture was triggered at this time, in this case two closely spaced waveforms. Clicking on the marker launches the waveform graph, Figure 6. There are actually two waveforms triggered, roughly a second apart, both are shown in Figure 6. It’s not obvious at all what happened, and at first glance, it appears that nothing changed in the waveforms. This was purely an RMS change event, and the raw waveforms are not useful at all (except to exclude other event types).
Figure 5. Starting with the RMS Voltage and Current interval graph, a sag is visible on phase A
Figure 6. Two waveforms triggered, roughly a second apart
The same two events are shown as RMS Capture graphs in Figure 7. Here we easily see that the RMS voltage for phase A sagged once in each waveform, for roughly 3 cycles in the first, and over 6 cycles in the second (still recovering by the end of the waveform). There are changes in the other voltage phases, including increases in RMS voltage in the first event. These two waveforms are around 70 cycles apart. Given the RMS changes, and the timing, two recloser operations are likely the cause.
Figure 7. RMS Capture graphs
It’s important to keep in mind that the RMS Capture graph is not the best view for all event types. Although excellent for RMS changes, true waveshape disturbances are best analyzed with traditional waveform capture, and the RMS graph can actually be the misleading view. In Figure 8, a likely line-line fault is shown. The disturbance and ringing are clearly shown in the current traces, and zooming in reveals corresponding effects on the line voltage. The original fault (possibly arcing or failing insulation) only lasted a few milliseconds, with the rest of the disturbance consisting of ringing for another half-cycle. The RMS Capture graph for this event (Figure 9) looks much different – in this graph, it appears that the voltage was almost constant, and the RMS current had a sudden one cycle step increase. This shape is purely due to the sliding, box nature of the RMS window calculation – the narrow surge in current during the fault magnifies the RMS value while the sliding window covers it, and as soon as the window passes the fault time, the RMS value output falls back to the nominal value. With a short event like this, the raw waveform data is much more useful, and the RMS capture graph shouldn’t be used.
Figure 8. Likely line-line fault
Figure 9. In this graph, it appears that the voltage was almost constant, and the RMS current had a sudden one cycle step increase
Conclusion
The RMS Capture graph is a powerful method to visualize RMS changes in voltage and current, giving waveform-like detail without the inherent scaling issues of looking at raw sine waves. Residing mid-way between coarse interval graphs and ultra-fine raw waveforms, the RMS Capture graph is perfect for voltage magnitude analysis. Since the RMS graph is computed by ProVision from the raw waveform capture data, any PMI recorder than can record waveforms can provvide RMS Capture graphs. ProVision is required (Winscan does not have this feature), and ProVision 1.50 or later is highly recommended, due to very much improved graphing speed.
Chris Mullins
VP of Engineering and Operations
cmullins@powermonitors.com
http://www.powermonitors.com
(800) 296-4120