Any smart meter experts in?

I’m hearing things about LED lighting affecting smart meter accuracy possibly due to the harmonics generated – but I can’t find any studies / papers.

Anyone know anything?

I guess you are talking about this?

https://www.telegraph.co.uk/news/2017/03/06/smart-energy-meters-giving-readings-seven-times-high-study-finds/

Haven#t found the study online yet but it says it was conducted by University of Twente Enschede in the Netherlands.

However it also says
Since the publication of this article BEAMA, the trade association which represents manufacturers of smart meters, has been in touch to say that none the smart meters tested in the study are part of the UK’s official smart meter roll out.

That’s the one.

I’d seen somewhere that 5 of the 9 most prevalent UK smart meters display some of the same behaviour though.

And I’m looking at a house where two different measurement systems (one control and one billing) are giving answers where one is 60% of the other.  So I’m exploring answers on both measurement systems

If it was 5-10% lower on one I’d ignore it but at 40% lower… well it’s difficult to make a measurement system that’s that far out these days due to component tolerance, so one of the systems is either incorrectly installed (probably not the SM) or there is an issue with the assumptions in the algorithms and how it responds to the actual load presented.

I’ve been 30 years a power systems engineer and it wouldn’t be the first time I’d seen digital systems assumptions fail to cope with analogue reality.

PS the 7 times will be misleading I think. I suspect 7 times the loads that cause high harmonics might be more reasonable, noting that higher quality power electronics “should” produce less harmonic content.

@igm, what’s the control meter you’re looking at?

Smart meters (I assuming we’re talking MID billing meters?) have to go through a fairly rigorous process to get accredited, and then each one has to be individually calibrated and tested to a set spec. That’s not to say they are infallible…

But the control meter you have, might not even be measuring true power. By way of example, many (non-MID) meters will just guess/assume the relationship between the Voltage and current. Come to think of it, many don’t even bother measuring Voltage. Give us some more details.

Regarding harmonics. These (usually) have smaller and smaller effect, as they are (usually) a lower and lower proportion of the total system work. So, if a 0.1% harmonic, is measure at x7, that only makes it 0.7%. I’ll guess that LED lights do generate unusual harmonics due to to high frequency switching, but it’s unlikely to be a large proportion of the power draw.

Agree. The control measurement is more likely to be the culprit.   The 7 times though, as I understood it was 7 times the actual load causing the anomaly, not 7 times the harmonic content.

The meter is an EDMI though if need to check the model.
The standards, as is common place in parts of the energy industry, are written extremely well on the basis of assumptions about load (demand or generation).  They will be tested in line with those assumptions too

The control is associated with a pair of myenergi Zappi chargers – PV and battery follow. it won’t be as accurate as the SM ought to be.

The most likely error is the wrong CT has been placed on the incomer for the control measurement, or that it’s been put in the wrong place – that’s getting checked.

Next up the control measurement really is 40% low. That’s an issue once the PV, EV, battery and immersion heater are trying to balance to zero at the incomer.

Then there are the SM assumptions and their implementation.  I have worked with tapchange control relays in the past that had some very sensible assumptions in them that didn’t make sense once you had distributed generation. I know some of the deliberate decisions that were made around SM specifications (I work closely with some of the folk involved with that from a DNO perspective) and what they do to the capabilities on SMs.

I don’t know the assumed loads or if those SM implementations have been tested with potential modern loads outside their assumed loads.

I’m just going through eliminating things that might cause that discrepancy.

I assume in the study they were forcing the issue, and getting truly unrealistic results. But don’t know the details. As the telegraph pointed out, the meters in the study are not used in the UK.

Also, worth noting that although “smart meters” have only been around for a relatively short while, electronic meters have been around for decades. A smart meter is only an electronic meter with some build in communications.

CTs very often the cause. Double check them. Also, with import/export need to make sure CT orientation is correct.

The other thing that springs to mind is whether the smart meter will cope with import and export? Not really my field, so don’t know. But wouldn’t surprise me if it failed entirely to deal with exported energy.

Sounds like a complex system. But my guess is everything is working right, just not “wired” up correctly.

But wouldn’t surprise me if it failed entirely to deal with exported energy

Even if dealing with export as well as import is one of their main selling points.

As I say, not my field. I thought the main selling point was they they wouldn’t have to send someone round to read the meter. Happily stand corrected.

thought the main selling point was they they wouldn’t have to send someone round to read the meter.

I for one wouldn’t have bothered with one as I’m more than capable of looking at my meter and updating an app once a month.

How ever want your SEG payment for export from solar and your forced to get one.

@UrbanHiker given the state of the northern DCC there’s very little point to smart meters at all.
SMs have become semi-obsolete before they’ve been properly rolled out.

CT orientation is correct. There is no export until next week anyway.  CT location might be an issue, but shouldn’t be looking at the values I’m seeing and where the CTs might be misplaced.

Voltage is available to the control algorithm but whether it’s used I know not. But that wouldn’t account for the discrepancy unless I’m so far over nominal that appliances were popping regularly.

Possible that a single phase voltage has been indicated to the measure algorithm as a three phase voltage (which it might multiply by 0.58 and then multiply each phase’s current individually then sum). If there’s only one phase current (there is) that would given about the right discrepancy.

As for electronic meters having been around for a bit, firstly I agree, and secondly that might be the problem. They’d have been designed and spec’d prior to the prevalence of LED lights etc, and if they didn’t redo the spec when they went to smart meters then there’s your problem (ie it’s not a problem with the meter exactly, it was just designed for an old load scenario – which doesn’t sound unlikely).

I’ll see what turns up.