Measuring real time gas consumption

[DBMandrake]

there are fitting instructions for each of the meter types in the pack, an uber-diyer such as yourself should have no issues . suspect I have same meter as you, 1970s job. It sticks (they provide mounting tape) right on to the glass face of the dial, so should be immune to your cupboard problem, and it watches the dial. I just recalibrated mine after a year of running (i.e. I checked what loop thought the gas meter reading should be, versus what it actually is) and it was only out by 20 units or so. happy with that.

Here is my meter:



The irony is that it looks like it has hidden away low voltage switch contacts that would allow direct connection to a monitoring system but only "approved" devices could be connected, no doubt only devices supplied by the gas supplier

It looks like light would get in through the side of the plastic casing over the readout, so hopefully that wouldn't be sensed erroneously by the sensor as the cupboard is dark until the door is opened.

the electric one is less accurate, due to inductive coupling it over-reads when the usage is low (I have tried to calibrate, but gave up, too damn hard).

Inductive coupling isn't inaccurate per se, I've used professional grade current clamps for measurement purposes and they can be very accurate, as can hall effect current clamps for DC.

I suspect one reason for inaccuracies is going to be that if it is only a simple current clamp and doesn't measure the voltage (and I don't see how it can without any direct electrical connection) then it must be assuming a nominal mains supply voltage. Power is Voltage x Current after all, so if you measure current but only assume voltage there will be errors that depend on voltage fluctuations for the time of the day and based on local voltage drop from items such as an electric shower.

They might have a voltage drop heuristic in the software that assumes a certain amount of voltage drop for a certain amount of current draw (for example if your shower draws 40 amps, the voltage might drop by 5 volts in a "typical" installation - that's a guess, I haven't measured it) but that wouldn't account for fluctuations caused by the trends of your neighbours etc.

So absolute accuracy is impossible with a simple current clamp but its still a damn sight better than a bill (and potentially bill shock) every 3 months with no way to go back and work out what the cause was. A relative reading is all you really need there as long as absolute accuracy is in the ball park. For me the real time feedback is much more important than absolute accuracy.

as others say, when it comes to switching your provider, take their monetary and tariff stuff on their site with a pinch of salt.
take your own meter readings periodically to check the accuracy, dump their data into your choice of format and do your own thing via all the comparison sites.

I'm unlikely to take their switching suggestions seriously, I'm not someone who chops and changes providers all the time to save a few p here or there, I want it purely for rapid feedback of gas and electricity use so I can find out what does and doesn't use a lot and take quick action to correct any problems rather than getting bill shock 3 months later...

[paulockenden]

The biggest problem with electricity monitoring is that most products completely ignore the power factor (which is often more important that measuring the voltage).

There's only two whole-house products that I know of that measure Amps, Volts, and factor in the power factor: Smappee and OpenEnergyMonitor (the latter might involve some assembly).

Eco-Eye Smart does measure the power factor (via a capacitive sensor - foil wrapped round the cable!) but doesn't directly measure voltage, so it assumes 240.

I could write loads about this... Oh look, I already have - see this month's issue of PC Pro magazine!!!!!

(Oh, and to bring this back on topic, Smappee also has a gas/water sensor).

[DBMandrake]

Ah good point!

I completely forgot about taking power factor into consideration - I should know better as I'm well aware of what AC power factor is. Whoops.

In that case its no wonder its not super accurate, especially if you had a lot of loads like fluorescent lights. (We do have one in the kitchen and two in the garage)

Still it's probably better than nothing (eg a bill every 3 months) and most of our energy use is gas so I think loop will still be a worthwhile source of information and feedback.

[bruce_miranda]

I have the OEM version built up. It's pretty darn good. The errors come from the fact that if you measure full home utilisation then when the utilisation is really low the errors creep in. It's a compromise. It's like having your multimeter set to 250V all the time and trying to measure a 1.5V battery.

[sandyman]

so, the reason I gave up on calibrating the low end of the electric range on loop was thus: having looked at the observed electric usage over the day in loop , done the maths on where the most significant cost segments of my electric bills were coming from, it seemed much more important to find the "on all the time" lower power things, as the 24hr multiplier really adds up (no sh*t sherlock!) than worry about the high current short usage stuff. I turned everything I can find off, and still showing usage on loop. I seem to remember it thinks we're at 0.3A when really sub 0.1A based on count-the-time-between-meter-flashes maths. hence why I'm convinced the electric version is crap at low usage. my house can use a lot of amps at max (2x elec ovens, 40amp induction) and not a lot at min (all lighting LED, AAA+ rated fridge/freezer) so the range is a tricky one, as someone else mentioned which range setting on the multimeter is appropriate...

if you know of a better way to do this cheaply and without soldering or otherwise hacking, i am all ears. otherwise I will just have to wait for a smartmeter.

[paulockenden]

That's definitely power factor effects you are seeing there. It's also why some monitors will tell you that solar panels are producing (or even consuming) a few hundred watts, even when it's totally dark out there.

OEM or Smappee are the best options, if you want to avoid this.

[DJBenson]

Here is my meter:



The irony is that it looks like it has hidden away low voltage switch contacts that would allow direct connection to a monitoring system but only "approved" devices could be connected, no doubt only devices supplied by the gas supplier

It looks like light would get in through the side of the plastic casing over the readout, so hopefully that wouldn't be sensed erroneously by the sensor as the cupboard is dark until the door is opened.

Inductive coupling isn't inaccurate per se, I've used professional grade current clamps for measurement purposes and they can be very accurate, as can hall effect current clamps for DC.

I suspect one reason for inaccuracies is going to be that if it is only a simple current clamp and doesn't measure the voltage (and I don't see how it can without any direct electrical connection) then it must be assuming a nominal mains supply voltage. Power is Voltage x Current after all, so if you measure current but only assume voltage there will be errors that depend on voltage fluctuations for the time of the day and based on local voltage drop from items such as an electric shower.

They might have a voltage drop heuristic in the software that assumes a certain amount of voltage drop for a certain amount of current draw (for example if your shower draws 40 amps, the voltage might drop by 5 volts in a "typical" installation - that's a guess, I haven't measured it) but that wouldn't account for fluctuations caused by the trends of your neighbours etc.

So absolute accuracy is impossible with a simple current clamp but its still a damn sight better than a bill (and potentially bill shock) every 3 months with no way to go back and work out what the cause was. A relative reading is all you really need there as long as absolute accuracy is in the ball park. For me the real time feedback is much more important than absolute accuracy.

I'm unlikely to take their switching suggestions seriously, I'm not someone who chops and changes providers all the time to save a few p here or there, I want it purely for rapid feedback of gas and electricity use so I can find out what does and doesn't use a lot and take quick action to correct any problems rather than getting bill shock 3 months later...

That's exactly the same as my meter and it works fine with Loop. I've had no problems at all with my device which has been running for over 12 months now. The lack of official API is annoying (as are the promises from Loop about one being forthcoming) but the PyLoopEnergy API works great and it's super simple to integrate it into Domoticz to capture realtime and historical usage/consumption trends.

The only down side is that you have to set the device up as an electricity sensor in Domoticz (for both gas and electricity) so you can't use the cost per unit functions to calculate periodic costs (if you try and use any of the other types of gas/electric meters in Domoticz I guarantee you will have a world of pain - such as readings showing in the UI that don't exist in the database, BIG readings too!!)

[DJBenson]

Here's my code required to get reads into Domoticz from Loop via PyLoopEnergy;

#!/usr/bin/python3
import pyloopenergy
import time
import urllib

def gas_trace():
  print("Gas =", le.gas_useage)
  f = "http://127.0.0.1/json.htm?type=command&param=udevice&idx=18&svalue=%s" % int(le.gas_useage*1000)
  urllib.request.urlopen(f).read

def elec_trace():
  print("Electricity =", le.electricity_useage)
  f = "http://127.0.0.1/json.htm?type=command&param=udevice&idx=17&svalue=%s" % int(le.electricity_useage*1000)
  urllib.request.urlopen(f).read

elec_serial = 'REMOVED';
elec_secret = 'REMOVED';

gas_serial = 'REMOVED';
gas_secret = 'REMOVED';

le = pyloopenergy.LoopEnergy(elec_serial, elec_secret, gas_serial, gas_secret, pyloopenergy.IMPERIAL, 39.1)
le.subscribe_gas(gas_trace)
le.subscribe_elecricity(elec_trace)

Just grab your serial and secrets from the Loop portal as described in the PyLoopEnergy instructions, create two dummy electricity sensors in Domoticz and update the IDX values in the script. You can also change from imperial to metric meter and alter the calorific value for the gas consumptions if required.

I have this running constantly using systemd on my Raspberry Pi Domoticz box and it feeds electricity reads in at a rate of one reading every 30 seconds and one every 15 minutes for gas which gives a fairly decent resolution to my graphs.

[DBMandrake]

So loop arrived yesterday and I installed it.

A few minor hiccups along the way - the first was trying to find the "product code" which it asked for but didn't tell me where to find it - I was checking on the electricity kit box and in the order details but it turned out to be on the gas kit box with no indication that this random looking code was something special. Silly me.

The next thing was that the setup process only walked me through setting up the electricity sensor and then told me I was all done, even though I'd entered a dual fuel product code from the gas kit box... so I had to manually go in to add the gas sensor afterwards.

I also had some signal issues with the gas sensor - at first I couldn't get any connection at all even though it is at most about 6 metres from the receiver. (30 metres range - ha right! ) It turns out that the orientation of the receiver and probably the gas sensor unit itself is very critical. There wasn't much choice about how to place or orient the gas sensor and still have the cable reach the meter, (with sitting on top of the meter the only choice) so I found I had to turn the receiver broadside on to the direction of the gas sensor - when it was edge wise there was no connection at all. Turning it broadside has given me a modest 2 bars. The receiver is kind of resting on the pile of cables at the back of the AV cabinet in the living room to get it to sit in a usable orientation.

I'm not happy with the signal level or the wobbly footing of the receiver at the moment so I think I'm going to have to get some doubled sided sticky foam and stick the unit to the AV cabinet vertically in the right orientation to the gas sensor to ensure a reliable signal. I realised after having the signal issue that unfortunately, a tall fridge is on the other side of the wall from the boiler closet so the line of sight from the gas sensor would pass directly through the fridge to the back of the AV cabinet where Ethernet is available. Bother.

The installation process recommended that I attach the sensor to the 1cu ft clock dial using bracket B (I think) so that the two small holes are beside the pivot - this seems to work well and after checking it this morning the reading is bang on to the last digit, as long as you turn the use of gas off and give it 15 minutes for the reading to catch up. (obviously it is always running a bit behind with 15 minute updates if gas is currently being used)

The next problem I had is that we are on SSE Hydroelectric - which is a choice in the list of providers but they have zero tariffs listed for SSE! C'mon loop, you can do better than that... So I had to use the manual tariff method by taking figures from a previous bill. It also just gave me a default British gas tariff during the setup and didn't ask me to go select the correct provider and tariff - I had to go searching to find that. (To discover they had no tariffs for SSE anyway)

The electricity reading seems reasonably good so far - at least for loads like the electric shower which (due to a power factor of 1) look pretty much spot on at 9kW, and the kettle is spot on at 3kW. Lighting loads around the house (which are also a power factor of 1 in most cases) also seem to agree quite well. So as a relative measure of power use it looks pretty good, albeit not accurate enough for true billing purposes due to the lack of power factor measurement.

Until more usage data is gathered its a bit hard to see what the data will look like - so far I've only noticed the live view and the daily view, I guess more will become available once there is more data history.

One thing I found quite peculiar was the tiny little 3" long Ethernet cable stub - which is far too short to plug into a switch/router and orient the receiver upright (as recommended) and keep it away from metal! (as recommended) It's not made very clear at all in the documentation that the RJ45 socket is not just a pass through, you can actually connect the device to your router via the socket and a normal Ethernet cable as well. Clearly there is a 3 port 100Mbit Ethernet switch inside so you can actually use either port as the router connection or passthrough, but C'mon loop, that 3" long Ethernet cable is just silly. It's not even CAT 5e/6 cable - it's flat non-twisted pair cable which is out of spec for Ethernet. Two RJ45 sockets side by side would have made so much more sense if you want to provide pass-through.

I'd recommend people just use a normal Ethernet cable as I ended up doing, so the receiver can be positioned upright and away from the switch/router and just ignore that silly little RJ45 lead sticking out... I would also ding Loop for recommending in the setup process that people can connect another device to the loop if they don't have enough Ethernet sockets without also warning them that it is only a 100Mbit port.

Edit: On re-reading my post it looks like I'm being very negative about loop - I'm not. I'm just someone that is picky and detail oriented so I do notice small niggles especially if they could have been avoided by better design or instructions. But so far it looks like it will do what I want quite nicely - the daily graph does look quite good, and my only real concern is making sure I can attach the receiver somewhere securely and get a reliable signal from the gas sensor. The electricity sensor is giving a solid 4 bar signal although it is somewhat closer to the receiver and doesn't have a fridge in the way...

Here's my code required to get reads into Domoticz from Loop via PyLoopEnergy;

Thanks for that - that's brilliant.

I already have a Raspberry Pi running Domoticz and Munin graphing the heating system and I'm reasonably good with Python so I should have no trouble setting this up.

[DJBenson]

systemd unit for anybody that is interested;

[Unit]
Description=Loop Energy Gas/Electricity Consumption
After=multi-user.target domoticz.service

[Service]
Type=idle
ExecStart=/usr/bin/python3 /home/pi/domoticz/scripts/pyloopenergy/pyloopenergy/loop.py
Restart=always

[Install]
WantedBy=multi-user.target

When paired with my code above, this will import gas readings every couple of seconds and a gas read (if there is one) every 15 minutes. You can slow down the read speed by adding a sleep command to the bottom of the python script.

Gotta say, now I've familiarised myself with systemd, I am systemd-ising everything