Smart Meter Vs. Standard Meter (with Photos!)

Michael Giberson

Oncor has been testing smart meters in part by monitoring several several homes with side-by-side smart meter/standard meter pairs.  (See week four results from Killeen, Texas; week four results from Temple, Texas; week four results from the DFW area will be posted tomorrow, April 7, assuming Oncor sticks with their schedule.)  So far it looks like the smart meters and standard meters are keeping it close, with most of the meters no more than 4 or 5 kwh’s apart after four weeks.

One pair of meters in the Temple test is reported at 12 kwh’s apart, a difference of about $1.60 on a monthly bill, but the older meter is the one with the higher reading.

What? You don’t trust Oncor? Well they are posting photos online each week from the side-by-side meters. (Which won’t convince any skeptics, but will serve to document the historical moment.  Years from now some digital artist will be inspired to remix these images into a work of art.)


7 thoughts on “Smart Meter Vs. Standard Meter (with Photos!)

  1. The problem may not be due to the meters themselves but due to the transmission of the data from the Smart meter to wherever its sent and recorded.

    For example, the consumption level reported in the couple’s lawsuit shows that they were billed for 11,000 kWh or more per month for three months, which is almost impossible to imagine. This consumption level means that the couple have a load of more than 15 kw that is on for 24 hours every day for 3 months, in order to consume a maximum of 33,000 kWh for those 3 months. How could that be possible?

  2. Great links! The Oncor press releases include all their meter data, suitable for some casual statistical analysis. This is especially timely for me and some colleagues, since we’re just putting together some statistics homework problems for the paired t-test.

    Using the Temple and Killeen data, I estimate an average difference of 1.5 kWh between meter types, with the mechanical ones giving the higher readings. With 24 meters reported, this difference is statistically significant with a p-value of 0.026 for a one-sided test. If the Dallas data is consistent with these, Oncor stands to lose a few bucks in direct billing once the smart meters are installed.

  3. Re: “Oncor stands to lose a few bucks in direct billing once the smart meters are installed.”

    * Actually, since Oncor is only the wires company, it is some energy retailer in Texas’s restructured retail market that is losing a few bucks in direct billing.
    * Which do we trust: old meter or new meter? If the new meters are right, that implies the old meters were exaggerating consumption slightly. If the old meters are right, that implies the new meters undercount consumption slightly.

    Re: “How could that be possible?”

    Obviously looks like an error. My guess is that the old ways of collecting and processing meter data had numerous checks built in to reduce the common sources of errors; with a new way of collecting meter data perhaps they hadn’t worked out and fixed all the different ways things could go wrong.

  4. The following information is of the “I read something about this somewhere” variety, so take it with a few grains of salt, but I’m guessing that the reason that the two meters have different readings has to do with corrections for phase shifts between current and voltage. One will probably be measuring volt-amps and the other will be measuring watts.

    I believe I read that old meters were designed with the intent of measuring power usage on resistive loads, which do not cause phase shifting. Modern electronics do, and the difference could lead to differences in power measurement.

  5. Respectfully, I think probably not, Tom. That wouldn’t necessarily be a minor difference, and I would hope that new meters wouldn’t be designed on the sly to measure volt-amps instead of kWh. Let me explain.

    The current entering most complex devices (i.e., more than resistive) is not strictly in phase with the voltage across its terminals. Mathematically, we can break that current into two components: one in phase with the voltage, and another 90 degrees out of phase with the voltage.

    The component that is in phase with the voltage transfers energy from the power system into the device. Somebody has to burn fuel and inject that energy into the system somewhere in order for your device to consume it. As consumer of the energy (measured as kWh), you have to pay a share of the cost of the energy input.

    The component that is 90 degrees out of phase with the voltage doesn’t transfer any *net* energy from the system into the device. Rather, the energy carried by that component flows into the device during one half-cycle and flows right back out in the next half-cycle. Your device doesn’t consume it, and nobody is out there burning fuel to supply it. That said, there are losses and other details back on the system that imply a cost for this “reactive” power, but it’s generally not as expensive as “real” power, and you shouldn’t be paying real-power prices for it.

    You’re right that a conventional meter isn’t sensitive to the reactive component, as it was meant to measure the net energy passing through it. The new meters should be designed to be able to measure both real and reactive power, but it would be a scam to measure kVA and bill for kWh. It would be like selling a can of peas that’s mostly water. Some water is expected in a can of peas, and you know that when you buy the can. That’s normal and part of the price, right? But you’d feel ripped off if you opened the can and it was barely more than pea-water. Reactive power is like that water. It has to be there as part of the product, but it’s not something you can really do anything with.

    I’ll bet somebody was expecting me to talk about beer and foam…

  6. It always pays to get more data. Looking at the 4-week cumulative consumption data for Killeen, Temple, and Oak Cliff, I get an average difference of only 0.93 kWh between meter types. A paired t-test on these 30 meters gives a t-statistic of 1.34 for a (two-sided) p-value of 0.1905. This is insufficient evidence to demonstrate a difference in the two types of power meters.

Comments are closed.