Earth Notes: On Solar DHW For Our Home: thinking allowed (2008--2019)

Updated 2022-09-30 12:30 GMT.
By Damon Hart-Davis.
Ruminations on Domestic Hot Water (DHW) from solar for our home, 2008 and on, and on... #solar #DHW #microgen
under combi 20191215
I had been through many iterations in my head on how we might make use of solar thermal to reduce our carbon footprint, learning quite a lot on each round. But at least a decade of rumination had not yielded a viable solution!

As of 2010 we covered most of the remaining available roof space with solar PV, on the assumption that we'll be preferring 'instant' ASHP in any upgrade of our central-heating and domestic hot-water system, and overall the reduction in carbon footprint with thermal or PV is similar especially given that excess PV output can be pushed to the grid to displace others' energy demands.

2008: First Time Around

Possibly the biggest insight is that our property simply does not collect enough sunlight to cover DHW (Domestic Hot Water) and CH (Central/space Heating) in mid-winter. At least in part because our roof surfaces face east and west, not south. Our whole ~40+m^2 roof will only receive an average ~20kWh/day mid-winter, when DHW and central-heating demand can average ~40kWh/day, though down from ~60kWh/day before conservation measures.

In 2008 I estimated our DHW usage to be ~6--8kWh/day (~2.5+MWh/y). We actually consumed at least about 8--10kWh/day in gas most of the year, some of which was/is for cooking, and our non-condensing gas combi boiler (GCNo 47 393 06) is 78.6% efficient and with a nominal maximum heat output of 24kW. Our DHW boiler-exit temperature had been ~50°C for a while, eg 51°C at the bath tap full open on a warm August day but I dropped it to ~45°C at boiler-exit with no complaints and (say) a bath at 40°C is pleasant for me, 37°C for our toddler. The cold water mains input temperature of 19°C 2008-08-30 mid-morning indicates that a temperature lift of ~20K--25K is required in warm weather, and maybe 30K+ in cold weather, ie a ~50% increase in energy demand in winter. (See other mains sample temperatures.)

It might be possible to cover most of of our DHW requirements for most of the year from the 6m^2+ of space remaining unused on our west roof (incident insolation >2kWh/m^2/day from March to September but only 0.5kWh/m^2/day in December cf ~4.7kWh/m^2/day June and July, mean 2.6kWh/m^2/day, ie a total of 960kWh/m^2/year isolation, and with an assumed collection efficiency of ~50% maximum). The advantages of using the space on the west roof are that it is completely unshaded, and facing away from the road minimises planning issues.

After our second round of grid-tie PV in 2009, the final roof layout leaves 2m from the north edge of the roof to the edge of the PV support frame.

Given that we have now dedicated the east roof to PV (~18% efficiency), consider that in conjunction with a top-efficiency heat-pump (CoP ~5) then we could convert the collected insolation to hot water with similar or better efficiency (>~50% overall) using the grid and/or batteries as our short-term energy store and exporting summer excess energy to the grid rather than wasting it. We could also change to a time-of-day or 'smart' electricity tariff and boost water temperature when grid electricity is cheap/abundant, which is probably increasingly 'green' as the UK's grid gets 'greener'.

Our current (Potterton Performa 24) gas 'combi' is fairly new but because it is only in efficiency band D (more than 10% below the best condensing models) and is not able to accept solar preheated water (though that might not be true given a device such as the Grant CombiSol for supplying DHW either direct from a solar-heated hot water tank (or thermal store) else diverted via an 'instant' combi with input from the hot tank mixed down to 25°C with a TMV as pre-heated water) then I am prepared to replace it. (This would also reduce our dependency on mains gas which is getting to be a more scarce and expensive commodity by the day!) Any heat-pump replacement would have to be a 'combi' too, ie providing hot water and space heating (using our existing radiators if possible). We don't really have ground collector space for a ground-source heat-pump (GSHP), so we'd want a high-CoP low-GHG-refrigerant air-source heat-pump (ASHP) such as an 'Eco Cute" CO2/R744-refrigerant model, which should be OK as London outside temperatures are rarely <0°C. The heat-pump should ideally be an 'instant'/'tankless' design to avoid "standby" (tank) losses from stored water heated with imported energy; holding solar pre-heated water in a tank with its standby losses is OK.) I would not want the maximum usable heat output in winter much below (say) 20kW so that, for example, a bath does not take much longer to run than now. But I also would not want the maximum electric input to be much over 10kW (~43A) since our main supply panel is only fused for 60A. Both might be achievable simultaneously if CoP never falls below ~2.

Note that the highest efficiency (CoP) would be gained for an ASHP by running it during the day when outside air temperatures are highest. But the cheapest and most plentiful electricity is usually to be had at night when neither space- nor water- heating are needed. Note that ~1t of water in the core of the house (in an IBC?) could act as a heat store for one day's combined near-maximum mid-winter heat requirement for us (~30kWh) with maybe a δt of 25K (eg heat stored at 65°C+ cooling to 40°C) preserving a reasonable CoP even at the highest store temperature. A tariff allowing scavenging of 'excess'/'green' electricity whenever available, primed with a forecast of the next day's temperatures to choose how much heat to top up the tank with, and conversely using 'dynamic demand' features to switch off at moments of grid strain and demand peaks, plus a day's thermal store for all heating (but using only a small hot-water tank in summer), could work very well in money and CO2(e) terms, and be friendly to a grid with lots of intermittent generation such as wind. Also, local microgenerated electricity (such as from PV) can be used rather than spilled to the grid for minimum cost/CO2. A smaller store, a grid-friendly (and heat-demand) modulated heat output, and a more airtight property, could deliver a similar CO2/cost performance.

Basic System Specification

This initial system should support the following:

DHW demand is estimated to be ~6kWh/day as of the end of 2009 (down from about 8kWh/d a year or so before, from behavioural adjustments).

Available initial collector area is unshaded ~4m^2--6m^2 on west-facing 23° pitched (concrete tiled) roof.

Note that the house used to have a hot-water tank and still has parts of the old system in situ though not connected, such as the airing cupboard with hot-water pipes routed through it, roof venting, loft-tanks, and so on, which may make refitting a tank-based system relatively easy. The old tank cupboard raw capacity is ~500l though the old tank was maybe ~100l. The interior dimensions of the cupboard itself (there is also more space in a larger immediately-adjacent storage area) is 0.5m x 0.5m x ~2.15m.

The house is a 3-bedroom end-of-terrace wood-frame construction in London.

I would like to maximise mid-winter energy capture/efficiency when we otherwise consume most gas/energy, at the expense of collection efficiency at other times of year if need be, to try to do our best to minimise our winter carbon footprint, and to minimise summer overheating/stagnation. (This might suggest drainback and evacuated tube or plate collectors, for example.)

Heating with Radiators

Note that to drive the radiators (to replace the other combi function) generally the radiator flow water temperature should reach ~55°C, though when the outside temperature is below 0°C a flow temperature over 60°C seems necessary (as of 2009-12-19) to deliver heat to the living room fast enough to keep it above 15°C.

2010: Revisiting Instant Solar DHW with PV + ASHP?

With high-efficiency PV and a heat-pump with a CoP around 3 suggested above then the overall capture of heat with a PV+ASHP system is similar to a simple solar-thermal system (though much more complex and expensive), though summer excess is shipped to the grid rather than wasted...

Could it match the instantaneous performance of our current 'instant' gas combi? For example, as of 2010-01-21 in the evening I can run a bath or a shower at about 1l/7s (~9l/min or ~140ml/s) at ~38°C (from a cold mains inlet temperature ~8°C, ie a 'lift' of 30K/30°C). At about 600W/K (4.2J/K/ml * 140ml/s) that implies about 18kW effective heat output from our ~24kW, ~80%-efficient boiler.

Move to summer, and run a shower or bath at the same rate, but now with the cold mains inlet at a little under 20°C, ie a "lift" of only 20K, implying a heat requirement of 12kW.

With a heat-pump CoP of (say) 3 which should be easy to achieve in summer, that implies an electrical input of 4kW, which indeed is something that we hope to expand our PV beyond in 2010 (a max PV capacity of over 5kWp and inverter output over 4kWp). Given that we might generally hope for reduced demand in summer, plus better CoPs and long sunny evenings for our west roof for example, we might well be able to cover much 'instant' DHW demand much of the time.

Back in winter, to support our 18kW heat demand, even if the CoP was only 2 (and generally we'd want it over 2.5 to ensure overall carbon-footprint reduction compared to simply keeping the gas combi) then we'd need ~9kW electrical input to match current combi performance, which is certainly do-able (40A vs our 60A connection) though clearly not likely to be covered from PV at the time!

In the case that our PV system was backed by batteries of course, we might still reduce peak grid demand and keep it away from peak (CO2/kWh) intensity, or could achieve the same with a hot-water store instead.

2011-04: Thoughts

IBC 1t tank

As of 2011 we seem to have DHW demand down to ~4kWh/d (ie about the equivalent of one hot bath) with maybe another 2kWh/d mean gas consumption for cooking.

If we aimed to get energy imports for DHW down to zero in high summer (without too much boiling/wastage other than holiday week) that implies about 1--2m^2 of collectors at between 50%--100% efficiency, or maybe 1m^2 on each of the east and west roof surfaces, which should still be possible with all our PV installed, with care.

Excess not needed for DHW in a 'combi' style tanked system can nominally help with space/central heating (CH) outside summer.

2m^2 should gather over 4kWh/day in June/July but likely under 0.5kWh/d in December when total heat demand (including CH) has been over 30kWh/d, and indeed there may simply be so little available insolation that a likely flat-panel collection system may deliver no support at all.

Although electricity generation with export of excess remains possibly best, some simple solar thermal on the side seems like a good idea; if mains failed our solar PV would not work without an expensive battery system and in any case could not export excess for others to use, whereas a little battery power for circulation could possibly keep us supplied with warm water for washing. (With battery support for the PV we could possibly keep the heat-pump running for more DHW/CH, using the storage of electricity and heat together.)

2011-09: Thoughts

I have just run across the interesting Volther Hybrid PV/T from Newform Energy.

The PV/T idea has interested me for a while (eg Zen PVTwin), but this is starting to look practical, with the added twist of using a heat-pump with a decent CoP, though it may only be feasible for systems larger than the 4 panels (~8m^2) that we could accommodate, and may exhibit new exciting problems such as grinding to a halt for the few days each of the last few year that we have had snow covering our roof, plus tank losses that we are otherwise immune to currently.

It appears that a 4-panel PowerTherm system coupled to a 200l thermal store (with no heat pump, just a solar pump), and (say) the CombiSol to work with our existing combi, could capture a little over 800kWh of heat and ~500kWh of electricity in a year, corresponding to up to 0.37tCO2/y saved.

A ~2t thermal store inside or under the house might reduce the risk of mid-summer overheating (acting as a heat dump) and extend the DHW season by a month or so, since there is about that much 'excess' heat available in the middle two months.

The cost per tCO2/y saved looks similar to our PV.

That should take us firmly carbon-negative and remove all gas consumption for DHW (leaving only cooking) in the summer. This arrangement could allow us to eventually swap out the gas combi for an ASHP relatively painlessly. Given our current plumbing layout we could also end up with hot water at our taps much quicker than now, at least in summer.

2012-01: Thoughts

Justin Broadbent of Isoenergy visited (2012-01-23) he pointed out that a heat-pump would be unlikely to save money (but we ought to wait until June to see the RHI details), that an ASHP might get us an SPF of 3 if done right but that a GSHP with borehole(s) might cost the same and have an SPF of 4 (combi style with ~200l thermal store with direct solar inputs), and that it might cost us something like £18k all in.

(JB also thought that our ex-airing cupboard could probably accommodate a ~200l tank/store if I wanted to put one there, and rightly reminded me that we also ought to consider upgrading remaining radiators to more-efficient newer ones and lagging all visible pipework if making this sort of expenditure.)

2015-09: SunampPV

Following a chat with Sunamp I found its SunampPV "Heat Battery" PV-driven DHW system that works with a combi, similar to the CombiSol arrangements above, and using a phase-change storage medium, to be seriously interesting. That is even getting past my normal worries about wasting the exergy of PV electricity in producing low-grade heat. Perhaps the low standing losses (~0.7kWh/d for 4--5kWh of heat storage cf twice that loss from a hot water tank) are a reflection of use of some of that exergy. (2016-03-03 note: standing losses appear to be <600Wh/d.)

See also blog and heat loss thread by JSH.

Given the nature of the SunampPV system to be able to mix down for my old combi or bypass the combi entirely when it can produce hot enough water directly, and its capacity being somewhat over our daily DHW demand, and the fall back to gas as needed, I think that this unit could eliminate most of our DHW gas demand (say 1MWh out of 1.5MWh/y) for the equivalent consumption of otherwise-exported electricity which we have available 9 months of the year. So this does present a significant grid balancing and gas demand reduction opportunity, especially if I do things such as preventing this unit from 'charging' from PV at times of peak grid demand 16:00--20:00 and at times of low grid frequency.

In principle future Sunamp systems charged by heat pump would retain more exergy but would likely be larger and less nimble in terms of responsiveness to fluctuations in PV generation, but those are only outlined on their site.

Transferring 1MWh of DHW consumption from gas to electricity would nominally worsen our footprint by (0.43-0.19)g/kWh*1000kWh = 240kg (CO2e per year, and using probably-conservative 0.43g/kWh electricity intensity), but would reduce flows by 1MWh/y in (gas) and 1MWh/y out (electricity) which has some benefits especially if (for example) some SunampPV charging is done at time of peak grid-wide PV (or wind) generation (esp curtailment) when non-load-following generation has trouble reducing far enough. Ignoring CapEx (at list prices, probably £2000 for supply and fit) the SunampPV would be broadly cost neutral since each kWh of gas avoided is approximately the same cost as income foregone from each unit of electricity not exported under the FiT.

2015-09-19: this morning, unusually, there were two showers and a bath in quick succession, totalling about 15kWh of DHW demand, thus exceeding the capacity of a reasonable daily store. There was somewhat over 2kWh of PV generation over the same time which could have been fed in to cover maybe half the DHW demand without gas. That would be useful in tempering energy flows into and out of the house.


Supposing it was possible to use the SunampPV as a ~24--48h thermal store (~5kWh heat), a ~24--72h electrical/battery store (4kW--12kW electricity), and a seasonal (ie ~6M) thermal store (~1.5MWh heat); what order should they be topped up in from PV that would otherwise be exported to:

  1. minimise gas consumption
  2. minimise carbon footprint and cost
  3. minimise grid imports especially in winter and at peak-demand times
  4. grid flows generally?

(With direct at-the-time consumption by appliances in the house implicitly taking top priority.)

I suspect that the battery store should always be topped up first since that probably preserves the most exergy and is the most flexible, ie can run household appliances as well as just providing heat. This could be as simple as being fastest and most aggressive to soak up otherwise-exported energy, though deliberately leaving some capacity to fill at highest generation (typically around solar noon, or when grid voltage or frequency are high) and some to discharge at peak grid demand or intensity (or low voltage or frequency) may maximise overall gains.

A possibility is to only allow the daily thermal store to recharge from 'excess' PV from (say) 11:00 (solar) to 16:00 (local) time, ie when south-facing PV systems are most likely to be dumping to grid (in a way that the grid may find uncomfortable to absorb) and ending before peak demand.

The seasonal/daily thermal store issue may be more subtle, ie the daily store should probably take priority (to allow minimising/displacing gas use for DHW in spring) but only if the seasonal store is meeting a pre-planned recharge schedule. It may be that the daily store can always be given priority, and providing that it is not too large nor too lossy, the seasonal store should have plenty of residual PV generation to recharge from relatively early in summer as daily insolation climbs quickly. This assumes that either DHW or space heating can easily fall back to gas if the thermal stores are exhausted.

The seasonal store (re)charge schedule could be as simple as a target 100% full at end September and 0% (ie empty) at end of March, with linear slopes on both sides, or with some modifications such as holding the target to 100% Oct--Jan to keep the seasonal store as full as possible at the expense of spilling to grid (and possibly DHW) in case of cold snaps. (If excess heat could be moved from the seasonal store to DHW, or in some other way they could be partly or fully unified, so much the better.)

If however the seasonal store is the only source of space heat other than falling back to (say) expensive electrical resistance heating, then it should probably fill much more aggressively. It should still wait significantly longer than the battery store (eg 60s) to allow the battery to to start filling first (giving it effective priority), and back off quickly if energy starts being imported.

If at all possible the seasonal store should preserve exergy, eg by being filled via a heat pump rather than electric resistance heat. (See also Milk Tanker Thermal Store with Heat Pump with magnetite.)

There are possible optimisations such as avoiding topping up the daily store if no DHW has been drawn for a couple of days, to avoid the standing losses and let the energy go to another store or spill (be exported) to the grid.

Looking at recent electricity and gas/DHW consumption figures (~4.5+4.5kWh/d), and 2014 PV generation (Mar--Sept inclusive > 9kWh/d) it looks like nominally all DHW could be satisfied with excess spilled to grid (or pushed into another store) for those 7 months out of the 12.

2016-04: Sunamp Stack

A more radical thought is to skip the SunampPV, rip out the gas combi, and get a Stack instead, initially with purely resistive heating for simplicity, with a heat-pump retrofit possible to maximise exergy and carbon savings.

Yes, get rid of gas.

The Stack is nominally a 60kWh heat store (1t, and the size of a fridge-freezer), for DHW and radiators, but since our highest mean daily demand mid-winter has been below 30kWh over several years, and more like 20kWh usually, maybe 40kWh max in one day, and that with a band-D boiler so probably 25% losses, a 20--30kWh store is likely enough heat for a day and a bit in winter. (Sample daily gas consumption data for Feb 2016.)

Afterwards I would add electric battery storage, probably "AC-coupled", with higher priority, so that there are little to no exports to the grid from October to March inclusive except possibly at peak demand/intensity time when I might spill excess to grid as a public service! Also, in the exporting months I'd still like to minimise exports around solar noon to be kind to the grid, and top-up imports would be in the low demand/intensity hours. I would get rid of gas entirely.

Yes, it'll cost more money and a bit more carbon annually, though should not be higher than a typical house without PV even then. (Total gas use was under 3MWh last year, for DHW and space heat, whereas our gross electricity exports were 2.8MWh, though not seasonally matching.)

(I suspect that ~2kWh battery and ~3kWp inverter throughput would prevent a large fraction of our residual electricity exports and re-imports. Curiously I already have about 2kWh of usable storage in my off-grid system, though dribbled out at ~3W not 3kW!)

Because this is no longer a backup or adjunct to a main source of heat as the SunampPV would be, I'd want to keep a minimum of a decent bath's worth in there, and top up immediately (eg even importing from grid) if the heat store went below that ~3kWh reserve, and indeed one should be able to have a succession of baths or showers out of it with intervening pauses with a top-up rate/element of (say) 2--3kW. (If more than one heat element is possible, the 'top up from mains immediately' below the low-water mark could be 1--2kW, to do so reluctantly as it were, with a larger element for diversion, allowing a high peak rate of fill with a secondary simple backup.)

In the heating season, given that there'd (almost) never be enough PV during the day to fill the store I'd want to nearly fill it overnight or when grid demand/intensity is low, opportunistically leaving a space for any PV energy that did arrive, so (say) to 3kWh below full capacity.

Outside the heating season I should let the store get to (say) a decent bath or two and then some (10kWh?), to avoid the risk of having to import during a dull day, but hoping to use a smaller part of the store to reduce unwanted losses. Any non-emergency top-up from imports should be as before overnight or when grid demand/intensity is low.

In any case I might choose to slightly overfill before peak hours (4pm--9pm) to leave any solar at that point to spill to grid, and might tolerate slightly lower minimum store levels during those peak hours, or high-intensity hours, to avoid importing from the grid before them.

2016-05-15: measurements from my Loop gas meter suggest that my relaxing hot bath consumes 5.4+kWh, even with a mains inlet temperature of ~16°C.

2018-08: Sunamp UniQ

Some AC-coupled PV electrical storage is imminent.

I still object to recipients of PV generation FiTs dumping power into exergy-wasting electric immersion heating with the implication that allowing the power to spill to grid would be 'wasting' it.

It is however of note that much of the time now the marginal grid CO2 intensity is not far off (and sometimes below) the CO2 intensity of burning gas per kWh of heat, ie electric immersion vs gas boiler, in the summer months at least.

What the grid would probably prefer is that not all the available PV generation were spilled to grid around solar noon. (EG Comparison of the Location and Rating of Energy Storage for Renewables Integration in Residential Low Voltage Networks with Overvoltage Constraints: "Storage in homes reduces overvoltage, offering clear benefits to the network operator...")

On my Raspberry Pi I already have PV generation figures for every minute. With the Enphase Envoy-S I may get not only that but also net import/export figures.

I think that the Enphase AC Battery 4h charge/discharge time is quite sensible to gently smooth grid flows.

Consider a Sunamp UniQ 3 (3.5kWh) with a 1kWh heater, only enabled to heat/charge when (say) both generation and exports have been consistently well above 1kW for several minutes on the trot, and only between 10am solar/UTC and 4pm local. This would prevent some PV spill in a period where demand is relatively low.

Per-minute monitoring is almost certainly enough to minimise accidental imports for DHW, especially with the electrical storage doing fine trimming in real time.

(The UniQ would work with, not replace, our ageing gas combi, and be just for DHW.)

That would likely cover most DHW demand from March to September, ie more than half the year, when PV generation is more than ~10kWh/d. Maybe that could reduce gas use by ~600kWh/y (16%), almost eliminating use for that half year. (Typical summer electricity demand ~5--6kWh/d, gas/DHW 3--4kWh/d.)

(The remote control could almost certainly be managed with a variant of the OpenTRV 'boiler control' with relay or SSR, with the stats hub being persuaded to TX instead once per minute to keep said remote device on...)

2019-09: Sunamp UniQ Again

It's that time of year again, when a man's fancy turns to PV diversion.

Here I look at what is exported after the Enphase AC Battery has stored what it can, ie that might be diverted to DHW. This is for winter months, so is going to be the worst available. But it gives an idea. Figures are in Wh. In September there's 9kWh/d available, Oct 4kWh/d, Nov 1kWh/d, Dec 0.2kWh/d (ie nowt!), Jan 0.5kWh/d, Feb 3kWh/d, Mar 5kWh/d. Computed with the following command:

% foreach m ( 1809 1810 1811 1812 1901 1902 1903 )
foreach? gzip -d < data/16WWHiRes/Enphase/adhoc/net_energy_20$m.csv.gz | awk
-F, '{sum+=$4}END{print sum}'
foreach? end

With DHW demand ~4kWh/d currently (gas used, nominal 80% boiler efficiency but possibly more like 50%), Nov/Dec/Jan are fairly useless. But other than those, most or all consumption may be coverable. Especially if we adjust the timing of our baths and showers a bit depending on sunshine.

Repeating for a minimum export threshold/margin of 100W (25Wh per 15m interval) shows possible diversion per day to be approx 8, 3, 1, 0.1, 0.4, 3, 5. So I can easily be nice to the grid spilling some after topping up the Enphase without materially hurting DHW availability:

% foreach m ( 1809 1810 1811 1812 1901 1902 1903 )
foreach? gzip -d < data/16WWHiRes/Enphase/adhoc/net_energy_20$m.csv.gz | awk -F, '$4>25{sum+=($4-25)}END{print sum}'
foreach? end

That might yield PV generation power use roughly in the following priority and amounts, eg on a bright morning with space free in all storage:

  1. 100W appliances
  2. 250W Enphase charging
  3. 100W spill to grid
  4. 2800W UniQ charging
  5. remainder (from generation over 3250W) spill to grid

Every electricity unit not exported would avoid a similar priced gas unit being imported. Assuming 1:1 which would be 100% (or at least, the same) efficiency conversion to usable DHW for both fuels.

There's very little in these months that a 3kWh element could fail to catch. Ie PV generation is rarely if ever going to be over that. I would however set the system to spill a small amount to export to reduce the chance of accidental import. And because it would be nice to the grid to do so.

(Ideally, to be nice to the grid, I'd suspend diverting when the mains frequency drops significantly, but I'm not going to get that this time around!)

The space under our combi is approx 100cm (high) x 60cm (deep) x 57cm, within the existing cupboard carcass.

In a conversation on Twitter: ...I have convinced myself that now it is legit to divert some PV to heat water with very low storage losses, since the nimble reduction of flows on both electricity and gas grids is some of that exergy at work...

(If actual gas DHW efficiency is anything like the 50% implied then switching to very nimble efficient resistance heating may give a near 2:1 gain normally expected of a low-end heat-pump.)

I have now engaged someone to do the system design.

~6045 words.