* For referenced figures, please refer to ELECTRICAL CONTRACTOR magazine.
If the data recorded appears to violate the laws of physics, apply the test of reasonableness to the data before attempting to fix something that isn't real.
I often have files of power quality data sent to me for an explanation of what happened, which is often accompanied by the question: "Is this real?" There are numerous reasons why the data recorded may not reflect what is actually going on in the electrical circuits. Rule 7 will take a look a few of the common problems.
The first step is to know what you are going to hook up to. Something as simple as whether you are monitoring a four-wire wye or crazy-leg delta circuit makes a big difference in what measurement leads have to be placed where on the circuit itself and connected to which monitoring input point. An open-ended delta or single-corner grounded delta is often connected in a different manner than a full delta. Connecting for power quality measurements rather than just energy measurements can also make a difference.
There are three-phase PQ monitors that have single-ended voltage-measuring inputs with a common that is to be connected to the neutral wire and a ground connection that must go to earth ground. (This is contrasted by differential input monitors where each phase has two input connections: Va+ and Va -, Vb+ and Vb-, etc). All of the inputs need to be connected in a wye circuit, or else you might read one-half the line voltage on the neutral input, which wouldn't be real. Similarly, connecting such a device to an ungrounded delta will now give the circuit a ground reference. The input impedance of the monitoring inputs should be high enough so as to not cause problems with any ground-fault detectors on the circuit.
Make sure the connections are secure, not just for correctness of the data but also for safety reasons. A loose clip lead can land in another location and cause a short between phases or ground. Figure 1 shows an example of the "unreal" results of such an improper connection. There are a series of transients going to zero volts within a cycle. At the same time, there is no change in the current waveforms (not shown here). If it were a real load-side event, there would normally have to be very large current transients, almost short-circuits to ground, to get those kind of transients similar to those shown in Figure 2. And if it were a source-side event, the reduction in the RMS equivalent of the voltage would most likely have an effect on the current being drawn by the loads. Instead, it turns out that vibration from a large pump turning on a floor below the monitoring location caused the alligator-clip lead to come loose on the connection, resulting in an open circuit to the instrument's monitoring input from the voltage connection point.
Another common problem is using current probes within their ratings. People often seem unsure of the current levels that they will be monitoring, so they opt for larger-than-needed ampere-rating current transformers. Most CTs should be used to measure currents that are at least 10 percent of the full-scale rating, which for the 1,000A CT used in Figure 3 would be 100A. The user was measuring a little more than 5A, which isn't even enough to provide adequate magnetization current for the CT, much less operate it in the linear region. The waveform recorded shows significant current distortion. Particularly disturbing, if this were "real," is the 12 percent even harmonic distortion? Having such distortion levels in an electrical system would normally cause much distress in transformers and other electromagnetic devices. In reality, the CT is being misused for the application, and the waveform is an artifact of inadequate current amplitude for the CT to function properly.
People try also to do the reverse. Some PQ monitors have measuring inputs that can take a current crest factor of three. People interpret this as meaning that they can put three times the full-scale range into the CT. The 1,000A CT is subjected to 2,500A for extended time periods, and they don't know why its plastic body and handles have melted to look like a fried egg on the skillet. Doing such not only voids the warranty, it is a safety hazard.
Remember, that it is very, very difficult to change the laws of physics. If you have recorded data that attempts to do such, be sure to apply the tests of reasonableness before submitting your finds to the Nobel Peace Prize committee. EC
BINGHAM, a contributing editor for power quality, can be reached at 732.287.3680.