Electricity usage forensics

[MikeFromLFE]

Okay, I switched off the flat this morning between 9:30 and about 2:30. Have also been periodically peering at the monitor box. I wasn't expecting this, but on the otherhand it does fit the problem rather nicely...

4:30 am (I'm a crap sleeper) the monitor briefly showed a usage from 08/03/2018 of £73.62. Seconds later it reverted back to £120.03. I was taking photos rather than writing down the figures, so I have proof I wasn't dreaming!

9:30 am, just before The Big Switch Off, it showed £74.47 and £120.89, bouncing between the two several times. Also photographed.

2:25 pm, just after switching everything back on again, it showed £74.47 and £120.89 again. No bouncing this time though.

Sooooooo, no power leechers this morning, but WTF with the meter readings?!
Possibly just an artefact of the power cycling?


Thanks for the extra info and numbers for comparison, folks.
Quisling: there's an immersion heater, but I keep it switched off at the wall except for maybe once or twice a year when it might get 30 mins or so of use.

I checked it yesterday: switch is off and the water's definitely cold.

Are the two readings, readings from different rates - ie one is day rate use, and the other is economy 7. Do you have a dual rate meter, and not know it? My meter (dual rate) exhibits similar behaviour, but slower.

Sent from my Moto E (4) Plus using Tapatalk

[Kim]

Could the display wossname be receiving data from two meters?  I thought there was encryption involved, but nothing would surprise me.

[Canardly]

Aha!

https://newstarget.com/2017-03-29-smart-meters-could-be-over-billing-you-by-a-whopping-582.html

[Polar Bear]

That's an interesting thought Kim.  If there are a number of smart meters there are they cross-chatting?

[nikki]

Are the two readings, readings from different rates

As I understand it (based on the back of the bill and conversations with npower on Saturday) there's just one rate per kWh and then a standing charge per day. The standing charge should be in the region of about a fiver...

Could the display wossname be receiving data from two meters?  I thought there was encryption involved, but nothing would surprise me.

That would be a simple explanation for the doubling, wouldn't it.

To get them to investigate that, I'm going to have to put forward a convincing argument as to why it was happening with the old style meter too.

This is what we're dealing with:



I mean, what could possibly go wrong?

I've just had a very long chat with the chap with the same meter as mine (although supplied by British Gas). He's super on top of his daily usage and it's minuscule, so I don't think I'm getting his added onto mine. I've told him if I discover his bills are so low because he's using my electrons I'm coming around to get my money back! 

Interferencey noise stuff could be an option maybe. Or just the shonky electrics all through the building.

I've left a note in the hallway to see if any of the other residents have interesting stories to tell...

[nikki]

Does anyone else have one of these Duet II smart meter monitors?



If so, could you please do me a massive favour, so I can identify if the two readings thing is an anomaly or not?

1. Unplug it so it's without power.
2. Plug it back in again and go to more > menu > meter balance so you get the screen in the above photo
3. You'll have empty spaces whilst the meter boots up and establishes comms with the meter (I expect)
4. Watch it as the fields start to get populated and note what you get for "cost"

I consistently get a lower figure for 30 seconds or so before it then settles at the higher figure.

What do you see?

[Quisling]

I take it you have a physical meter, plus an in-home display?  If so, are they giving different readings?  If you know the model of the physical meter then can you post the details and I'll do some digging, although the fact that the old meter also started recording higher usage seems to rule out a meter issue specifically.

Also, you mention the standing charge is about a fiver - hopefully that's not PER DAY as that would be extortionate for a domestic bill.

[nikki]

I take it you have a physical meter, plus an in-home display?  If so, are they giving different readings?

Yes, there are both. I don't get a meter reading per se on the in-home display, but I'm starting to take daily readings which I can cross-reference after a bit of sums. Not a lot to go on yet though - will report back in a few days.

If you know the model of the physical meter then can you post the details and I'll do some digging, although the fact that the old meter also started recording higher usage seems to rule out a meter issue specifically.

https://www.landisgyr.eu/product/landisgyr-e470-zigbee-smart-electricity-meter/ it seems.

Also, you mention the standing charge is about a fiver - hopefully that's not PER DAY as that would be extortionate for a domestic bill.

That would hurt, wouldn't it! 26.030 pence per day since the 8th of March.

[Kim]

Does anyone else have one of these Duet II smart meter monitors?

We used to, when we had an nPower smart meter like yours (Robin Hood have since replaced it with a different smart meter with a much less useful remote display unit).  I don't remember behaviour that you describe, but I tended to switch it straight to proper human-readable power units, rather than the cumulative bill display that didn't really tell you anything about power consumption.  What it did do was sometimes fail to realise there was a gas meter present, but that's academic for you.

I take it you have a physical meter, plus an in-home display?  If so, are they giving different readings?

Yes, there are both. I don't get a meter reading per se on the in-home display...

Are you sure?  It was a menu option on ours.  Extremely useful, and something that's conspicuously missing from the PipIt 500 we now don't use.

[nikki]

Are you sure?  It was a menu option on ours.  Extremely useful, and something that's conspicuously missing from the PipIt 500 we now don't use.

I thought it was ludicrous when the installation chap said it wasn't a function!

Have dug deeper and found it under 'advanced'. 

They're currently in agreement with each other.


I switched the flat off again for 8 hours, doing meter readings before and after. Didn't catch an electron thief...


Now trying to get a measure of a typical night's heating. Left it on as per usual last night and did before and after readings as well as some very generous back of envelope sums:

Racked up 4.19 kWh between midnight and 8:30am. Call it 4kWh between 12am and 8am.

Assuming that was all from the 2kW heater (which maybe runs more like 2.2kW) with no other appliances or standing charge, rough calculation of heater on for a total of 2 hours out of 8. Scaling that up to 24 hours would be around 6 hours of heater time and 12kWh over a day.

Met Office says it was 5-6.5 Centigrade between midnight and 8am. We've had some flippin freezing day's but I suppose this would be fair enough to take as a happy medium. There'll have been plenty of days between August and February when the heating wouldn't have been on at all. I was in Japan for over 30 of them!

I know the heater hasn't really started becoming active until about 10pm-ish for the last couple of nights. So I don't think I'm running at that level all through the day, but even with that worst case scenario I think I'd be well short of a 25kWh per day average. Yesterday's total was ~12kWh including heating and kettles and showers etc.

Heating's off for the night to get a baseline for background activity. Taking the opportunity to test camping sleep systems. I'll be underneath all this down if anyone needs me... 

[Kim]

Can you program your heating-control arduino to log the duty cycle?  Or, even simpler, just keep a running total of time spent in the on condition since bootup.

Or, in the absence of easy ways to sense it's drawing current, log the temperature of the heater directly...

[nikki]

I tried an LDR over the indicator LED on the remote control bit that intercepts the mains to the heater. I've cocked up with the code or wiring somewhere though and it's going a bit bonkers.

Will probably have another go stripping out the GPS stuff to make it simpler and see what happens from there.


0.39kWh used between midnight and 8:30 without any heating.

[Quisling]

0.39 kWh overnight equates to about a 50 watt constant load which is realistic if you just have a small fridge and a couple of small bits of DC equipment charging, without any heaters on.
To get a sense of how hard your space heating needs to work (assuming you don't adjust the desired room temperature) you can download free heating degree day data from www.degreedays.net .  Use 15.5C base temperature (good enough for this exercise) for a local reliable weather station and select "daily" data.  The number of degree days each day indicates the extent to which you had to provide heating, i.e. when the temperature falls below 15.5C.
I note your meter can record reactive power (kvArh) units.  Has it been set to record this?  If so, has it recorded much?  It shouldn't do, as reactive occurs when you have poor power factor, i.e. you're pulling inductive loads (like motors, or lots of old magnetic ballast fluorescent lighting).

A plug in energy logger like this might be useful https://www.amazon.co.uk/dp/B071NZ53WR/ref=sspa_dk_detail_0?psc=1
It gives instantaneous power (kW) and energy over time (kWh) recording and would allow you to understand better the kWh being used by each appliance in turn.

To put your load into context, we average 13-14 kWh/day over the year (with a little bit of electric heating in the dead of winter, and part electric domestic hot water in the summer months when the boiler is switched off) in a sizeable house with 2 kids, a large fridge freezer and another chest freezer (lighting is all LED or CFL so not much load there).  25kWh/day is a lot so the culprit is likely to be a heating appliance of some sort.

[ian]

Meter incompetence seems (to be the only) standard. My last water meter reading was about 70 cubic metres out. Since we only routinely use 35-40, that's quite a bit. They've re-read and now have a correct reading but are still sending us reminders three weeks later for the old bill.

BG on the other hand, took two months to be convinced that our gas meter was metric and not imperial.

By-the-by, with gas-fired heating, my last bill for dual fuel even with BG was about £370 for the winter quarter, and that's for a large detached four-bedroom house (so about £4/day).

[Kim]

I tried an LDR over the indicator LED on the remote control bit that intercepts the mains to the heater. I've cocked up with the code or wiring somewhere though and it's going a bit bonkers.

It's probably flashing at 100Hz...

[nikki]

It's probably flashing at 100Hz...

Current hypothesis is blutak drift causing a threshold value to Be In The Wrong Place

The number of degree days each day indicates the extent to which you had to provide heating

Interesting tool, thanks!

Here's what I got for 3 years of 12C and the airport's weather data:



So, much of a muchness except for Beastliness in February.

I note your meter can record reactive power (kvArh) units.  Has it been set to record this? 

Haven't been able to find it on the monitor.
Apparently I can request a copy of the 30-minute interval data log when I phone them next week. Given it takes them a fortnight to send an email I'm not sure how long we'll have to wait...

Will maybe get one of the energy loggers if it looks like there's a specific appliance that looks problematic. Everything I've kept an eye on so far seems to be in the right sort of ball park.

Certainly the £6 a day average they're wanting to charge me since the 8th of this month seems a little harsh. £2.89 is my highest daily spend recorded so far this week...

[Kim]

Reactive power's probably in the menu on the meter itself.  Not that it should matter, as domestic customers are only billed for real power, and it would surely take a quite spectacular cockup to start billing you for that by mistake.  Sounds like the load is predominantly resistive anyway.


Anyway, have we explored the 'are they reading the wrong meter' hypothesis?  How do meter point access numbers work?  How does nPower's billing department work?  Is there life on mars?

[Polar Bear]

I recalled this thread whilst taking tea with a friend this morning.  He recounted to me how he has an 'estimated' gas bill for two days, yes, two days which he has been asked to settle.

The amount?  £235.

His normal spend is less than that in a year!

His strategy is to simply ignore their demand and await a proper reading.  He has of course read his meter and knows exactly how much he owes and it is in the region of less than 1% of the estimate.

[Quisling]

Reactive power's probably in the menu on the meter itself.  Not that it should matter, as domestic customers are only billed for real power, and it would surely take a quite spectacular cock-up to start billing you for that by mistake.  Sounds like the load is predominantly resistive anyway.

Agree: reactive isn't specifically charged except on commercial bills where the PF drops below (typically) 0.7.  That said, if it's recording high kVArh the meter may be incorrectly wired or programmed.  We had that when our PV array was installed.  MInd you, Nikki's old meter was also recording increased usage too so I doubt we have a meter issue.

If you can obtain the HH interval data (and check whether it's kW or kWh!) then sum each 48 periods to give daily total kWh then run a regression analysis against the daily degree day data (possible in Excel, or PM me and I'll help).  The regression equation, in the form y=mx+c, where c is the intercept with the y axis - i.e. the baseload energy use unaffected by weather, and mx (the slope of the line) describes the correlation between the degree days (the independent variable) and the kWh (the dependent variable).  You should be able to spot how well grouped the data is to the line and see if there is a lot of random variability or not. 
Also, if the HH data is in a row format (i.e. 48 columns with a day per row) then it's useful to use the Excel colour code formatting to create a "heat-map" (using red = high/green=low) which (if you zoom out enough!) allows you to quickly spot anomalous days or intervals, or patterns such as a particular load occurring at the same time each day.
It'll only be any good though if the HH data is in fractions of a kWh otherwise all you'll get is a load of zero's and the occasional 1, but hopefully it'll be to 2 or 3 dp.

[Kim]

This is where the blinkenlight detector and RRDtool come in handy...

[nikki]

Anyway, have we explored the 'are they reading the wrong meter' hypothesis?

If I were super keen I could do readings from all the meters in the cupboard...
Not yet that keen though - it's a bit tricky doing even the one 'cos of the way the light reflects off the screens.

I really want to know what's going on with that flash of the lower cost value when I switch the monitor on. The internet hasn't yet provided anyone else with a similar meter set-up willing to see what happens with theirs though, so looks like I have to wait for the tech dept and hope I get someone vaguely competent.

When I've got a few more days' readings I'll see if I can spot any obvious patterns or relationships.


Regression analysis is new to me, but I'll wait and see what they send me and take it from there. Have figured out what HH stands for, at least 
Colour-coded spreadsheets should be within my skillset.

£235.

Impressive!

[Kim]

I really want to know what's going on with that flash of the lower cost value when I switch the monitor on. The internet hasn't yet provided anyone else with a similar meter set-up willing to see what happens with theirs though, so looks like I have to wait for the tech dept and hope I get someone vaguely competent.

My guess would be that there's a lag while it updates the cost-per-unit from the default value hard-coded in its firmware with the correct value obtained from the meter, so the effect is harmless and unrelated to your billing situation.

But that's only a guess.

[ian]

I recalled this thread whilst taking tea with a friend this morning.  He recounted to me how he has an 'estimated' gas bill for two days, yes, two days which he has been asked to settle.

The amount?  £235.

His normal spend is less than that in a year!

His strategy is to simply ignore their demand and await a proper reading.  He has of course read his meter and knows exactly how much he owes and it is in the region of less than 1% of the estimate.

Wait till the muddle imperial and metric gas. I wasn't expecting our first sub-quarter bill to be >£1000. I may have uttered a bad word and dangerously snortled my cup of tea. It makes me wonder if they always had the meter misclassified. The previous owner was a bit fuzzy around the edges and I assume there's some database of meters (leastways, I seemed to remember some mess-up with the previous house – a new build – and EDF refusing point-blank to believe it wasn't a flat because some database said so*). It was also EDF that literally couldn't stop itself sending our cycle/utility shed electricity bill to the meter itself, c/o the wall. It's not like they let these people run nuclear power plants or anything.

*part the developers fault I think, part they didn't have an actual street address because the council was dithering.

[nikki]

My guess would be that there's a lag while it updates the cost-per-unit from the default value hard-coded in its firmware with the correct value obtained from the meter, so the effect is harmless and unrelated to your billing situation.

But that's only a guess.

Yeah, maybe. The timing smells a bit wrong though, unless it's a bit like GPS sentences and the price per kWh isn't broadcast at as high a frequency as the other data. *shrug*

Today's experiment has been to find out what sort of level of usage would be needed to hit the 25 units per day they're claiming for me.

Heating went on at midnight and took about 3 hours to get up to temperature, I think. Had hit 9.48kWh by the time I got up at 8:30am.
2.5 kettles, some hob action, an hour of roasting potatoes, half an hour of shower, 10 mins of hoovering, 10-15 minutes of microwaving 2 loads of laundry, an assortment of battery and device charging ... and I'm up to 23.88kWh.

Averaging that much every day, including outside of Winter and outside of the country? Really?!

[Polar Bear]

Are you collecting data for an artwork here?