Metering Electrical Power

September 1, 2016

One of the invaluable lessons that I learned while a student was the dictum, "You can't improve something if you can't measure it." While this idea is most often applied to management, it's easy to see how it applies to science and technology, also.

A corollary concept is that you need to know how closely you must make a measurement for it to be valid. If you're measuring the oxidation rate of an alloy by examining the weight change of a specimen (as my thesis advisor did for his thesis, more than half a century ago), a lot depends on the accuracy and resolution of your balance.

Nobel Physics Laureate, Enrico Fermi, had an interesting observation on measurement,

"There are two possible outcomes: if the result confirms the hypothesis, then you've made a measurement. If the result is contrary to the hypothesis, then you've made a discovery."

Enrico Fermi at a blackboard. Fermi's smile might indicate that he was joking when he wrote the wrong expression for the fine structure constant, alpha, a fact easily discovered through dimensional analysis. (Image from the Smithsonian Institution, via Wikimedia Commons).

One measurement that impacts all homeowners is their electrical power usage. In Tikalon's Northern New Jersey region, the cost of electricity is about $0.095/kilowatt-hour, heaped atop a "delivery charge" of about $0.02/kilowatt-hour. This is slightly below the national average of about $0.105/kilowatt-hour. Electrical power of a simple series circuit, as shown in the figure, is easily calculated in combinations of the load resistance R, the current I, and the voltage E, as shown.

(Created using Inkscape.)

Since the load resistance of your home is variable and essentially unknown, the power can be calculated only as the product of current and voltage. Today's electronic and computer technology allows high precision measurement of current and voltage, their multiplication, and integration, so it's possible to measure power usage this way. However, residential power meters have existed for at least a century; so, how do power meters measure the electrical power consumed?

Power meters use electromechanics in the form of a motor whose speed is proportional to the product of voltage and current. The motor is a conducting disc, typically aluminum, that's driven as a two-phase induction motor by a voltage coil and a current coil. The driving force is an eddy current in the disc, and the spinning disc drives a mechanical counter. Since the current and voltage are alternating, the supplied and measured power are affected by the phase difference between the (roughly) sinusoidal voltage and the current waveforms. Only the in-phase components of voltage and current produce power.

The electrical power meter on the outside of my house. The motor disc is visible through the central slot. Newer meters have an error of just a fraction of a percent,[1] which is remarkable, considering their operating environment. (Photo by the author)

While residential electric meters accurately measure the total power used in a house, energy-conscious people would be interested in how much power is used individually by their home electrical items. Such items include light bulbs, refrigerators, televisions, computers, dishwashers, home heating systems, air conditioners, and water pumps. Newer models of many items are more energy-efficient, a prime example being LED light bulbs. A little analysis of power consumption data would indicate which energy-hogs should be replaced, but only if there were an easy way to monitor the power used by individual items.

That's the problem recently addressed by electrical engineers at the Massachusetts Institute of Technology (MIT, Cambridge, Massachusetts).[2-3] They've developed a non-contact sensor that monitors the power demand of devices through electromagnetic emissions from their electrical cables. Everyone who's tried to receive AM radio inside a house knows that such emissions exist. The advance made by the MIT engineers is to associate these emissions with particular devices and wirelessly transmit the power usage data.

The MIT electric power monitor at the surface of a multi-conductor cable. (MIT photo by Bryce Vickmark.)

Steven Leeb, a professor in MIT's Department of Electrical Engineering and Computer Science and two of his students, David Lawrence and John Donnal, have published the results of their research in a recent issue of the IEEE Sensors Journal.[2] Another paper on the topic including co-author James Paris is still in press.[3]

Their approach to power measurement has several advantages, the first being that no signal wiring is required. The postage stamp sized sensor is simply placed on the power line. Precise orientation is not important, since the sensor is self-calibrating. The device takes frequent data samples very quickly, which allows detailed identification of spikes and patterns in the voltage and current. The system software uses this information to identify the types of electrical devices connected to the power line when used with a calibration reference load.[3]

The research team was able to eliminate errors arising from mis-orientation of the sensor by having an array of five sensors in the power monitor. Each of these is offset slightly from the others, and the software determines which sensor produces the best signal.[3] Testing at Fort Devens Army base, and aboard the U.S. Coast Guard cutter, Spencer, helped to build a catalog of waveform signatures for different electrical devices.[3] A user interface was developed to present the power data in a human-readable format.[3]

One criticism of home automation products is that their cloud-based data interface might allow criminals to access power usage information to determine whether or not a home is occupied. The raw data from the MIT system is under a homeowner's control.[3] During testing, the MIT power sensor discovered a significant energy-saving opportunity, and also an anomalous voltage pattern that revealed a wiring flaw that diverted dangerous voltages to some copper plumbing.[3].

Says Leeb, "For a long time, the premise has been that if we could get access to better information [about energy use], we would be able to create some significant savings." [3] Leeb estimates that a commercial version of his power sensor would monitor an entire home for a cost about $20-$30. One help for home use is the ease of installation. Says co-author, John Donnal, "It just goes on with a zip tie."[3] This research was funded by the Grainger Foundation and the U.S. Office of Naval Research, among other agencies.[3]

References:

1. American National Standards Institute, Inc., "ANSI C12.20-2010, American National Standard for Electricity Meters - 0.2 and 0.5 Accuracy Classes," August 31, 2010, (1.2 MB PDF file).
2. David Lawrence, John S. Donnal, and Steven Leeb, "Current and Voltage Reconstruction From Non-Contact Field Measurements," IEEE Sensors Journal, vol. 16, no. 15 (August 1, 2016), pp. 6095 - 6103, DOI:10.1109/JSEN.2016.2574245.
3. David L. Chandler, "What's wasting power at home? Ask your pap! - New system from MIT can identify how much power is being used by each device in a household," MIT Press Release, August 1, 2016.