Low-carbon heating at home with solar diversion and storage and heat pump. #heating #DHW #storage
A Dickens of a tale of the ghosts of water and space heating past, present and future... We continue to be frugal, and thus "low" users on the Ofgem TDCV scale. Notes for an online talk to Manchester Carbon Co-op EcoHome Lab, 12 December 2024, given over Meet sharing my original HTML script page and scrolling through it!
692s"EcoHomeLab talk A Home heat Carol" (captions) Uploaded . Downloads:
combi boiler and it wasn't condensing, we've had it since we moved into the house about 20 years ago.
[00:12]
Boring gas [] combi boiler, but we've got aerogel wall insulation and the superb Radbots,
[00:18]
and four of us at home and we consumed about 3.1 megawatt hours in gas that year.
[00:25]
For reference, the Ofgem 2017 low gas typical domestic consumption value, was 8 megawatt hours
[00:33]
a year. So we were low and we weren't cheating by heating on the slide with electricity. Our
[00:37]
electricity gross consumption if we had had no solar PV was 2 megawatt hours for that year
[00:42]
compared to the Ofgem 2017 low class 1, which means when you're not heating with electricity
[00:50]
of 3.1 megawatt hours per year. So we were low on both scales. Class 2 users may be all electric
[00:56]
heating with storage heaters or whatever and high is much higher for class 2 in that one. Net exports
[01:02]
were about 1.9 megawatt hours for that year. A little thing called the pandemic stopped me
[01:08]
putting in the Thermino. I was ready to go in early 2020 but kind of it didn't seem the to be the most important thing to be doing at that moment. But finally, it got round to 2022, and in March
[01:19]
the Thermino and the Eddi went in. So for reference in case any of you don't know,
[01:24]
when I started this thing this Sunamp product line was called Uniq. By the time I'd actually
[01:29]
got around to doing it in 2022 it was called the Thermino and I went for a size equivalent to
[01:34]
about 150 litres water tank and the Eddi, the myEnergi Eddi, which I wanted to use for a solar
[01:42]
diversion but also because it was highly controllable remotely and there's some nice stuff on the
[01:47]
Internet, people showing how to do it, so that I could choose to top up the Thermino when there were low grid [carbon-intensity] times. So the first curiosity when I got the Thermino was that when I filled it up
[01:58]
the first time nominally it holds seven kilowatt hours roughly but when you're but that's when it's
[02:03]
hot so you draw that seven kilowatt hours worth of heat and it's still just hot enough whatever
[02:08]
that means I think 45 degrees or something. When you're heating from cold mains water in March
[02:13]
actually it's more like 10 kilowatt hours and ours [DHW consumption] is about four kilowatt hours a day so that implies it's just short of two days' worth which is just about right. But anyway there's a very clever valve which we have in our system called the Intasol valve which is entirely mechanical -
[02:28]
there's no electronics, computing, Internet, anything. So it's designed to work with a
[02:32]
combi and a thermal water store and our Thermino looks like a solar thermal water store to it.
[02:38]
If the output of that when you turn on tap if the output is from that is hot enough it just uses
[02:42]
water straight from that although it then has a tempering valve to cap the temperature at 50
[02:47]
degrees to prevent scalding. If the water from the Thermino is not hot enough to use directly, ie
[02:52]
it's below about 50 degrees it routes it very cleverly it mixes it down to 28 so any
[02:58]
combi will accept it and feeds it into the Thermino so it bleeds residual heat into the
[03:03]
combi to reduce your demand so you really get that 10 kilowatt hours of storage. So in fact
[03:08]
therefore internally my measure that I'm not allowed to share with you because of an NDA with
[03:12]
Sunamp shows between zero and 100% [well] minus 30 and 100% well actually 110% because you can
[03:20]
overfill it with diversion a little bit get a little bit more in and I don't do that when
[03:24]
I'm topping up from the grid so it means that if there's ever some sunshine I don't expect
[03:29]
it can still be absorbed by the system. Our DHW demand about four kilowatt hours a day is
[03:34]
something like a megawatt hour a year and with diversion about half of that so basically all of
[03:39]
summer and some of the shoulder months is covered. It's 500 kilowatt hours a year. And if you look at
[03:45]
the chart you'll see the green bars are where it's diverting (this is for September in the year it
[03:50]
went in 2022) and you see the start of September it's basically all green and that shows we were
[03:55]
using water most days right so that it was topping it up but then towards the end of the month we're
[04:00]
starting to use boost.
The boost top up from the grid but it's complicated it says what's the
[04:06]
threshold. The black is what's the threshold for grid intensity I'm going to use for allowing top
[04:10]
up. The green and sort of yellow squiggly line near the bottom is grid intensity actual grid, so grid means it's 'super-green' which means it's in the bottom quartile
[04:22]
for the last week and there is no storage being drawn down into the grid which means probably
[04:28]
the storage is topping up so maybe we should too. And the little yellow plus signs are samples
[04:34]
where the Thermino is indeed being topped up. Interestingly the little black crosses are when I
[04:39]
respond to dips in grid frequency and stop topping up so I'm providing a tiny amount of grid
[04:43]
frequency support as well. And as I gained more confidence in this I made the boost algorithm
[04:49]
more aggressive so for example last year we basically drew almost no gas for hot water - only
[04:55]
for space heating - so only in the winter so we basically had zero effectively zero gas bills
[04:59]
other than standing charge most of the year and that's only done as I say oh there's a an actual
[05:04]
cap on this top up of being 170 something grams of CO2 per kilowatt hour so it will only top up
[05:13]
when it's less carbon to top up the Thermino and use it from there than it would be by burning gas
[05:18]
directly. So our footprint is lower than it would have been we used virtually no gas and it all got
[05:22]
turned off when we were capped at the end of November... So ...
The Present - Heat-pump and Eddi
so ...
[05:28]
At the end of November, well at the start of November middle of ... November, we disconnected the
[05:33]
Thermino from the combi so that when Octopus would come in they would just have a plain combi system
[05:39]
to upgrade but various things to try and make sure that it would all go back together again.
[05:44]
Well it hasn't it's not perfect we're going to have to do more work than I wanted to but it's OK.
[05:48]
But for the moment we have a relatively vanilla Daikin heat pump with water cylinder curiously
[05:55]
160 litres nominal so nominally the same capacity as the Thermino but three times the size and we
[06:02]
have DHW, domestic hot water, and space heating from that and it's running right now. The interesting
[06:09]
thing is that generally for electricity ignoring anything to do with PV or whatever and ignoring
[06:14]
heat we use about five kilowatt hours of electricity per day in the house and the space heating and
[06:20]
domestic hot water together via the heat pump is coming to about the same day so it's roughly
[06:24]
doubling our electricity consumption. Which is fine. It's probably going to be true for the whole of
[06:28]
this month it doesn't look like it's going to get much colder. This is a sort of mildish month. So
[06:33]
the Eddi was driving [the] Thermino via its heater output one and we diverted that instead to control the
[06:40]
immersion heater on the hot water tank. So normally once a week the heat pump system is programmed to
[06:49]
use the immersion heater to raise the whole hot water tank to about 60 degrees C to ensure that
[06:55]
any legionella is killed. Now, you don't need to do it once a week some people do once a day some
[07:00]
people keep it permanently at 60 - you don't need to do that. We probably turn over half the tank
[07:05]
volume every day and none of us except possibly me as the oldie in the house are in any way
[07:12]
vulnerable. I do baths rather than showers anyway and so I have now modified it to be a bit smarter
[07:17]
which says it waits eight days since the last time the tank was at maximum temperature and then in a
[07:23]
slot in the morning between midnight and 6am if we get to super green it will then run the
[07:31]
pasteurisation cycle then which takes about three kilowatt hours. If that hasn't happened by 14 days
[07:37]
in - and I could push that out if I was feeling like it - if that hasn't run by 14 days
[07:41]
it forces one the next morning. Now, that reduces the overhead. If I were running a pasteurisation
[07:47]
cycle every day that would be throwing away about well that would be heating putting in about three
[07:53]
kilowatt hours of heat at a CoP [Coefficient of Performance] of one. By doing it every sort of 10 days I expect this to be it's an
[07:59]
average of 0.3 kilowatt hours per day at a CoP of one and the rest of the CoP of I reckon I'm
[08:05]
estimating them between two and 2.5 or better so it's diluting that a lot less and it's doing when
[08:11]
when grid intensity is low it's nice on the grid demand is low. On my Eddi I have - I had - a spare
[08:19]
slot for a current clamp and so I put on a second one for that to monitor the entire supply to the
[08:26]
heat pump so the H4 boundary. And interestingly even [the] little OpenTRV box and so on are all supplied
[08:32]
via that even though it goes out the front of the house to the heat pump, through the heat pump, down
[08:37]
underground, through the wall, and all the way around the side of the kitchen, it's all supplied from
[08:41]
there. And when it's quiescent so when it's not doing any heating of any sort hot water or space heating
[08:46]
that's using about 10 or 11 watts which is not bad. Interestingly the circulation pump on its own
[08:51]
adds about another 50 watts. And I've set out with this heat pump that it is not [connected to the Internet] because at the moment
[08:56]
doesn't need to be connected I've set it up all sensibly statically so for example it keeps the
[09:01]
hot water at 45 degrees basically from 7-ish 7:30 in the morning till 4 pm and then from 7 pm
[09:10]
till 9:30 the rest of the time it will keep it to 42 degrees so we use less hot water at peak times
[09:17]
heating electricity for hot water. My son just ran a bath 10 minutes to four so it kind of nearly
[09:22]
defeated the mechanism just now but it still uses less. Radbots are controlling call for heat for the
[09:28]
space heating and it's an on-off thing and many installers will absolutely hate that. And I wrote
[09:33]
my first academic paper about why we should microzone and why TRVs should be fine but they
[09:38]
also would absolutely hate the fact that it's microzoned and not running all the time.
[09:42]
(I think I can get all these 3D graphs on at once.) The top one is temperatures as seen by the Radbots themselves
[09:48]
on the TRV so [] the temperature doesn't wave around quite as much as that. That's
[09:53]
a temperature [] right next to the radiator. Bottom trace is a spare Radbot I shoved
[09:58]
outside in our porch and interestingly it now gets the cold breeze from the from the heat pump
[10:04]
so it's a lot more wobbly temperature than it used to be. The next one down with a purple line shows
[10:09]
call for heat to the heat pump and you see it's only on about maybe a third of the time. And when
[10:14]
we get down to design temperature at minus two degrees outside which will be about three times
[10:21]
the heating degree days I think that won't quite get up to 100. But an interesting point running it
[10:27]
like this is if I ran it all the time the way the installers would like me to I'd be spending a
[10:32]
kilowatt hour a day just running the pump. I'm spending I'm saving two thirds of a kilowatt hour
[10:38]
just on not running the pump so I could have a slightly less good CoP and still be ahead by
[10:43]
by avoiding that parasitic load. There are options in the Daikin to only sample running and so on.
[10:49]
Anyway, next graph is percent open each of the TRVs and when it's over over 49 then it's calling for
[10:55]
heat. So this is in the week coming up to a few days ago.
What's in the Future?
I'll [reconnect] the
[11:01]
Thermino. I'll change the algorithm for top up to say that when the current grid intensity is about
National Grid [NESO] prediction for the next two days then I will top up the Thermino on the grounds
[11:17]
that although I'm topping them up with CoP one, immersion electricity, actually I'm gaining a
[11:24]
factor of two because I'm storing energy now that I do not have to get from the grid when
[11:29]
the intensity will be higher later.
Show Notes
Recorded on the Zoom H1n on the desk in front of me.
Original script
2019: Beforetimes
Before the pandemic, though I had been plotting a Sunamp (then "Uniq") it did not quite land.
With our old non-condensing gas combi, but with aerogel wall insulation and Radbots, and four of us at home, in 2019 we consumed about 3.1MWh gas.
For reference, the Ofgem 2017 "low" gas Typical Domestic Consumption Value (TDCV) was 8MWh/y.
We were not cheating by heating on the sly with electricity! Our electricity gross consumption as if we had no solar PV was about 2MWh, compared to the Ofgem 2017 "low" "class 1" TDCV of 3.1MWh/y. (Class 2 users may be all-electric, and "high" is 4.6 vs 7.1 for class 1 vs 2.) Net exports were about 1.9MWh.
2022-03: Past - Thermino and Eddi
In March 2022 the Eddi and Sunamp Thermino 150 ePV heat battery were in.
The first curiosity is that though nominally the Thermino holds about 7kWh, heating from cold ~10°C mains takes/holds more like 10kWh. Our DHW (Domestic Hot Water) demand is about 4kWh/d. In conjunction with the Intasol valve that residual heat can be bled into the combi input feed to reduce required temperature lift. My internal measure of how full the Thermino is therefore ranges from about -30% to +100%.
Well, ~110% at the top end, because it is possible to 'overfill' it a little, eg during solar diversion, until the Thermino disconnects the heating element. I avoid overfilling during "boost" (more in a moment), to always allow a little space for unanticipated diversion.
Our DHW demand is something like 1MWh/y. About half that is covered by solar PV diversion during the summer months (and shoulders).
In the chart below for September 2022 diversion starts to fade over to boost:
When solar diversion is not available, I have a script on my main Raspberry Pi to check when grid carbon intensity is low compared to recently, and in particular also lower than burning gas directly, and then "boost" (top up) from grid, as seen in the example from :
As I gained more confidence I made the boost algorithm more aggressive until by the time the heat pump went in almost no DHW was coming from gas, and our carbon footprint was lower than if it had been. 2024's gas consumption up until the supply was capped was about 0.8MWh.
2024-12: Present - Heat-pump and Eddi
As of the end of Octopus fitted a heat pump. That process took a week, or a year, or twelve years, by different perspectives. It should probably have taken two days!
To keep the installation as "normal" as possible for Octopus we disconnected the Thermino at the start of November. I aim to get it reconnected by the time solar diversion becomes a thing again, some time in .
Broadly, all heat (including DHW pasteurisation via Eddi and the immersion) has been consuming the same electricity as everything else in the house, ie ~5kWh/d for heat to a total of ~10kWh/d. The weather is reasonably mild, at ~4 heating degree days above our baseline of 12°C.
The heater output 1 from the Eddi that was driving the Thermino is now driving the immersion in the DHW cylinder. Rather than running oblivious every day or every week this waits until eight days since the tank was last at maximum temperature (ie with the immersion thermostat open) and then waits for a 'super-green' grid time to run — bottom quartile carbon intensity for the last week, and no grid storage draw. If one does not arrive for 14 days then the pasteurisation cycle is forced. This cycle will only run in the wee hours (midnight to 6am), so impact on the grid, carbon footprint, and CoP dilution, are all minimised.
I am using a second external current clamp on my Eddi to monitor entire heat-pump consumption, ie H4 boundary. I have no other automated monitoring or data collection from the heat pump yet.
Note that the heat-pump is not currently Internet connected. I am happy for time to be synced, but I am not expecting remote control. I intend to add a MODBUS card for additional local monitoring and control, eg for better grid interaction such as running the heat pump at some times ahead of or in parallel with diversion or boost to improve effective CoP.
I am running the heat pump in a way that many installers will not approve of: microzoned with (Radbot) TRVs, with an on/off call for heat from them; the heat pump is using weather compensation for the flow temperature [hart-davis2024zone]. Normal running of the heat pump over the week or so to as we emerged from storm Darragh, with external temperatures almost high enough for the heating to be off entirely ( ~14°C); (a) temperatures in the porch outside (4o) and at each TRV/radiator over ~7d, (b) boiler control and call for heat over ~36h, (c) per-TRV percentage open (50%+ calls for heat centrally).
2025-02: Future - Heat-pump and Thermino and Eddi
The Thermino will be on heater output 2, which the Eddi will move over to automatically during solar PV diversion once the DHW cylinder is maxed out.
The boost algorithm that I am working on will, in super-green times, look to see if current grid intensity is much lower than the mean over the last week or forecast for the next couple of days. If so then the Thermino will the boosted via the Eddi. The "much lower" factor that I have in mind is currently 2, ie there will be an effective CoP of about 2 from storing heat at that time.
[09:48]
