Author Topic: Current model smart meters give -32% to 528% variation in measured power  (Read 355 times)

0 Members and 1 Guest are viewing this topic.

Offline peter

  • Newbie
  • *
  • Posts: 19
  • Karma: +6/-0
Did anybody else see the IEEE published report on the accuracy of current model smart meters when subjected to modern load profiles generated by standby household equipment?

It makes you wonder how the bill from <insert: Power Company> is calculated since the power meters clearly are off the planet when measuring current waveforms with large amounts of high frequency components.

The University of Twente and the Amsterdam University of Applied Sciences have published a paper demonstrating gross overbillings by smart energy meters, ranging from -32% to +582% of actual power consumption.
They tested about 12 meters, including ones that look a bit like those found here in AUS.

"Experiments went on for six months, with individual tests lasting at least one week, and sometimes several weeks. Researchers tried to reproduce regular household energy consumption patterns and didn't focus on putting the smart meters to stressful conditions." according to the article.

I find it beyond belief how much error AC power measurements can have.

Since it's an IEEE paper, you need an account to download it from them.
Additionally, since it was published in IEEE Electromagnetic Compatibility Magazine this means it is peer reviewed
and so has a very low B$ content.

Look at this link where you can find a link within to Sci-Hub's copy of the paper. To download from there, you need to install the Tor browser.
for another summary.

Fair use quotes from the document:
"Controlled experiments performed on static energy meters confirm that they can present still faulty, and substantially higher, readings. The main cause of interference appears to be the current sensor.Meters with  a Rogowski coil current sensor showed a positive deviation of 276%, or an increased reading of 376%, using a controlled power supply with undistorted voltage and defined impedance, compared to the reading of a conventional electromechanical meter based on the Ferraris principle. Meters with a Hall sensor showed a deviation of registered energy of -46%, or a decrease in energy reading to 54%.
Using the mains supply in the laboratory, from 9 static meters 5 showed positive deviations of up to 582%, which is a higher energy reading of 682%, and 2 showed deviations of around -30%, equivalent to a reading of 68%."