Earth Notes: Milk Tanker Thermal Store with Heat Pump (2007)

It seems that ground-source heap-pump (GSHP) might not have a high enough CoP (Coefficient of Performance) to make our London home carbon-neutral or net-zero-site-energy over a year for space heating and domestic hot water (DHW).

Brian's suggestion was that a second-hand milk tanker's stainless-steel tank might be about the right size. He is not responsible for the unverified and indeed slightly deranged hand-waving below...

Milk Tanker Going Down!

Bury a 20,000l-ish (ie 20m^3, 700cuft) second-hand stainless steel milk road tank (well insulated) and full of water (plus maybe corrosion inhibitors) in the garden. In the summer, energy from solar thermal panels is pumped in, and in the winter pumped out again for space heating and DHW, and much less energy is required to pump the energy this way. Indeed, any excess RE year round, such as the dump load for a wind-turbine or even solar PV, that would otherwise be discarded, could be captured in the thermal store. This is much more efficient then trying to generate and store energy as electricity, for example, when all that is needed is heat. I have to do the sums for the energy capacity vs our energy usage and so on, but this looks like an interesting alternative to GSHP. Milk is good for you!

Here's an initial sketch, validated by any sane person...

Thermal Store Capacity of 20kl Tank

Assuming that the water in the thermal store neither freezes nor boils, and that a range of 10°C (mean ground temperature) to (say) 90°C (to avoid building up any vapour pressure) is possible, then our store can retain at most 80°C * 4.18J (heat capacity) per gram of water, ie ~334J/g. 20kl of water is approx 20Mg ie 20Mega-grams (and about 10% of our house's heated volume as it happens), and thus our store has a capacity of maybe 6.7GJ (Giga-Joules).

(This corresponds to a storage capacity of ~90kWh/t for a delta-T of 80K, or ~116kWh/t for a delta-T of 100K maybe with the help of antifreeze (etc) so practically ~10t/MWh at maximum with simple and safe technology, cf ~60kWh/t for a delta-T of 50K such as from 90°C down to 40°C, something like 6kWh/t for ultracapacitors, well under 30kWh/t for lead-acid batteries (16kWh/t for AGM @ typical 50% DoD), 100kWh--200kWh/t for Lithium-chemistry batteries of various types, and up to ~200kWh/t for Zinc-air batteries. So hot water has good energy density if all that is needed is heat. For comparison, the energy density of jet fuel is ~43MJ/kg, ie ~12MWh/t, or about 100x that of Li batteries by weight, thus no battery jumbo jets!)

Note that to maximise the efficiency of the heat-pump we should probably try to keep the thermal store's temperature range as close as possible to the temperature that we want to get our house and/DHW to, to minimise the thermal gradient that the pump is working against. That temperature would be somewhere between (say) 30°C+ for underfloor heating up to 70°C for DHW. Keeping the thermal store too warm (too far above the underground mean ~11°C) will increase heat loss into the surrounding ground, though that surrounding ground might then actually become part of the store. But my initial guess is that keeping the thermal store temperature between (say) 10°C and 30°C might be best, though that range reduces the capacity to 25% compared to an 80°C range.

Another way to increase overall efficiency might be to have the heat-pump heat water just enough that each day's new solar input (even mid-winter) is enough to bring water up to the (DHW) temperature required, minimising the thermal gradient that the heat-pump has to work against. Proper thermal stratification in the store(s) may help with this. When solar input has not been sufficient, which it won't be mid-winter, the heat-pump will have to overcome more of the temperature gradient itself, resulting in a lower CoP.

Our Energy Demand (Circa 2007)

We don't need any significant space heating for about half the year (we do use about 10kWh/day for DHW year-round though). We use a mean of maybe 20kWh/day for space heating for six months of the year, so, roughly, 180days * 20kWh/day = 3.6MWh = ~13GJ for winter space heating.

Our 2007 space-heating and DHW energy consumption was ~26GJ.

Even if our 20kl thermal store was 'full' at the start of the 'winter' period, at ~90°C, it would hold only ~50% of the energy needed for space heating, ignoring heat-pump costs and losses through imperfect insulation, etc. Suppose then that we buried two ex-tankers rather than one, and heat losses and heat-pump energy costs proved not to be huge, maybe that would cover our space-heating requirements. (And in summer we can reverse the pump to cool the house and add to the store.)

To cover DHW in the six months 'winter' period would require a third tanker!

(As of the fairly mild 2012, space heating gas demand was down to ~2MWh during the six winter and shoulder months, so just about do-able with one 20kl tank and extreme insulation! DHW for that six months would be another ~0.7MWh.)

Filling the Energy Store

This initial analysis ignores two issues entirely:

• How do we put energy into the thermal store renewably?
• RE does not just stop in the winter in the UK.

Assume the bulk of the energy is to be collected with solar thermal panels unshaded, south-facing, at a fixed angle of ~36° to the horizontal. In London, mean insolation per square metre year round would be ~3.1kWh/day (see PVGIS), maybe 4.8kW/day/m^2 in July down to 0.89kWh/day/m^2 in December. Insolation of about 2/3rds of that is available with panels at 90° (ie vertical, eg on a wall), but with more of the output in mid-winter, peaking at 2.8kWh/day/m^2 in August and down to 0.95kW/day/m^2 in December. At ~50%-efficient energy capture and the ~36° incline, (and assuming that we could use solar PV to drive any pumps, etc, needed) given that our year-round load averages 20kWh/day we'd need 13m^2 of panels. Probably not impossible since our (horizontal) roof area is ~40m^2, even though it is half east-facing and half west-facing and is not entirely unshaded.

With a much smaller energy store (to cover a few very cloudy days for example) then we'd have to meet heating/DHW demand (peaking at 30kWh/day in winter) at the seasonal level, say 0.95kWh/day/m^2 for vertical panels in December. That implies ~35m^2 of panels (at 90% overall efficiency allowing for much less storage loss) to cover winter demand (ie roughly 2.5 times the capture area of a system with year-round storage) which might still be physically possible, though competing for space with possible solar PV, and possibly too large a mass of water to get hot enough in poor weather, and with a huge and possibly dangerous excess of unused heat energy in summer. Overheating problems can probably be mitigated with:

• Vertical (or near-vertical) collection panels to minimise summer energy absorption cf winter.
• An overhang can help protect against overheating in high summer sun.
• A 'fail-safe' [archive] drain-back system to empty the collectors when the system is 'full' eg the tank is over ~80°C (and in very cold weather to prevent freezing). This could partially filled, not just all-or-nothing.
• A heat sink (dissipation) system, eg reverse heat pump into ground or drain/greywater, or a heated outdoor pool, or an oversize/seasonal thermal store.
• Adjustable panel shades/covers.

Thus the tank/store is worthwhile if its capital and running costs are less than the extra capture area otherwise needed. And clearly, this need not be 'all or nothing': 3 months' storage rather than 6 months' might still reduce the necessary solar thermal panel capture area significantly, though I have not yet attempted to compute the trade-off numerically.

Compared With Solar PV

I have previously estimated that as of 2007 I could just about eliminate our net household electricity use over the year with 4kWp of grid-tie solar PV to cover our ~7kWh/day consumption. Very roughly solar PV requires 6m^2 per kWp, so the 4kWp would require ~24m^2, ie roughly double the solar thermal collector size. (Note that some of the 13m^2 thermal collector panel space budget might in fact be used for PV to drive circulation pumps so as to eliminate the external (grid) energy draw.)

One big advantage of solar PV over solar thermal is that 'excess' energy can simply and safely be ignored, ie not used, without any damage being done. Although PV is much less efficient than thermal at capturing energy, in theory coupling it with a heat-pump can recoup a lot of the difference (it might be just possible to have 20% efficient PVs and a heat-pump CoP of 5).

In principle any excess can be sold (or given!) back to the grid, which while not currently financially very rewarding, does displace CO2 emissions from generation elsewhere.

Adding a House Thermal Store

It would be useful to have an energy store (an oversize hot-water tank) in the house to cover a day or more of DHW and heating demand. A (tall, stratifiable) tank such as the TORRENT 450 'RE' (450l) if allowed to vary between (say) 60°C and 95°C can store ~70MJ which is somewhat less than our peak 30kWh/100MJ peak daily winter demand so probably represents a minimum usable size and would mean little scope for my favoured morning hot bath! The Solus II 2200L (2200l) might store ~90kWh/320MJ which would maximise the chance of using a heat-pump efficiently to warm the water and solar thermal being able to top it up to (higher than) DHW temperature even mid-winter. The Solus documentation even describes paralleling two tanks if necessary. A TiSun PC5000 holds 4950l which might store a week's worth of heat.

Water should be stored at over 60°C to kill bacteria such as Legionella, but water this hot represents a scalding risk and should be distributed at or below 50°C: regulation of storage/outlet temperatures is important for health and safety.

Heat can be pumped into the house store from our seasonal store as needed, for example when we have available solar PV so as to minimise grid energy draw, and the house store will ensure that there is hot water available all day/night even though the solar PV is clearly only available for a small part of the day, especially in winter.

(Where available, solar PV could be replaced or supplemented with other RE such as wind; but PV is likely to be the best RE source in urban environments.)

A larger in-house thermal store would ensure that more hot water and heating would be available on a cold winter morning before the sun came up, and if expanded to cover several days, eg maybe to a 2kl/2m^3 store, would allow the system to ride out several cloudy days (ie where little solar PV is available to run the heat-pump) without needing to draw grid power to run the heat-pump.

If not enough solar PV is going to be available to cover heat-pump use (in addition to other electrical loads driven from the same PV panels) then off-peak or 'green'/renewable electricity could be purchased, and/or the heat-pump could be kept off during grid peak load where more carbon-intensive fuels are being burnt, ie ~4pm to ~8pm in the UK, to minimise the CO2 contribution from running it.

The toughest strain on the system is mid-winter when there is least insolation. In England on average in mid-winter 1kWp of solar PV should produce about 1kWh of energy. If the heat-pump from our thermal store has an overall CoP of ~5 which seems plausible given that our store should be well above the 11°C ground temperature, ie closer to our house-store/DHW temperature, then to support 30kWh/day heat demand mid-winter we'd need ~6kWh/day of PV, from ~6kWp of panels, on top of covering other loads.

In general it would probably be best to have the solar thermal panels heat the house store directly until it is at its maximum temperature (which is probably ~20% of total demand on a mid-winter day), and then divert it to heat the seasonal store (eg mainly in the summer). This pumping can be driven directly by solar PV if need be.

Keeping Warm in a Grid Power Cut

If heating circulation pumps can be kept running on battery power then the house can kept warm from the house store even when the grid fails, and DHW would also be available. Circulation pumps might likely consume much less than 100W, and DHW can be driven by water mains pressure with mechanical thermostatic mixer values (TMV) to limit DHW temperature so requiring no energy input while the tank remains hot. (Note that the TORRENT RE tanks are reported to be supplied with a TMV, and the Solus II can be supplied with one.)

If the heat-pump can be operated directly from solar PV when the grid is down then the house store can be kept topped up indefinitely, and thus the heating remains available indefinitely.

This is more robust than a typical gas combi boiler which will stop if either gas or grid electricity is unavailable.

Some Costs

An engineer from a UK solar-thermal systems supplier kindly humoured me with some outline component costs while pointing out that this is not the same as full install costs for a well engineered and reliable system that he'd be prepared to build or install! He suggested that 13m^2 of evacuated tube solar hot water might cost something under £9k (November 2007). For a small house the bulk of the replumbing (etc) might be in the region of £2k-£3k, excluding any of the complications of heat-pump interaction with a huge thermal store as proposed above.

(Normal practice is to size a domestic solar thermal system to avoid venting even in mid-summer to minimise maintenance, etc, which means that most of the year some or most (maybe 80% in winter) of the energy will need to come from another source such as gas. The thermal store that this supplier might add at maybe 2 to 3 times the volume/capacity of the central heating system itself, and not much over 300l, but getting up to 90°C, is to carry the system's use beyond sundown and through short cloudy intervals, etc.)

The TORRENT 450 'RE' (T450RE) house thermal store is listed at ~£2400 as of Nov 2007.

Perry Process also humoured me, and suggested £5k ex-VAT, ex-craneage, etc, for a 20,000l stainless-steel second-hand tank (316L) as at November 2007, so we might hope that, excluding works and insulation, for a total capacity of ~60kl (60m^3) and ~20GJ, the core thermal store might cost well under £20k, ie < £1/MJ. (Compare: ~£50+/MJ, ie ~£200/kWh and over 70kg to 50% discharge at 2007 prices, for lead-acid batteries, the RE electricity staple.)

Combining With an Existing System

For reference, our current CH/DHW is provided by a [archive] "Potterton Performa 24" 'instant' mains gas combi boiler rated at 24kW, SEDBUK band D, corresponding to a DHW flow of 9.8l/min raised through 35°C. The manufacturer tells me that this cannot be used to heat DHW fully or partly 'pre-heated' in a house thermal store by solar/GSHP.

Some Important Unknowns

Various parts of this scheme are not clear to me yet:

• Details of the plumbing.
• Whether 'standard' heat-pump units should/could be used to transfer energy to and from the thermal store.
• What CoP might be reasonable to expect: I has to be much better than 3 or 4 round-trip to make this scheme better than 'conventional' GSHP. An ATES system in London achieves a heating CoP of 5.
• The cost and disruption and any planning consents needed to install these vast water stores. (Maybe a University high-energy physical department would pay to put cosmic-ray detectors in for their studies!)

Other Thoughts/Ideas

• Doing everything to improve insulation first, ie reduce the need for heating, is likely to be by far the best thing to do in terms of cost and CO2 savings. Eg, 350mm+ in loft space.
• Rather than buring a tank outside, a cellar of 60m^3+ capacity could contain the store.
• Recover drainwater/greywater/graywater/sullage heat overnight, and/or to prewarm incoming water for heating and/or heat-exchanged with or heat-pumped to the lowest levels of the house or seasonal thermal stores. Possibly 70% of heat could be usefully recovered from (say) bath/shower waste.
• Bury the (top of the) storage tanks more than about 6m deep (since the temperature gradient moves about 1m/month) and let the surrounding soil add to the thermal mass of the store providing that there is nothing to carry excess heat away such as moving groundwater.
• Keep the water in the seasonal and house stores thermally stratified.
• Spread the thermal store over several separate tanks, both for engineering robustness in case one springs a leak or whatever, and also to potentially keep one for space heating nearer 30°C and for DHW nearer 60°C to reduce the thermal gradients that the heat-pumps have to work against to improve efficiency. That would mean added complexity, and possibly needing two heat pumps, so might not be worthwhile.
• Maybe use rainwater run-off to fill the tanks rather than potable mains water.
• Possibly as much as 3.5kWh/day of our DHW energy demand is my morning hot bath (40°C water temperature rise (~10°C main to ~50°C bath) times 75l water (150l capacity half full of me) times 4.2kJ/l = ~13MJ = ~3.5kWh). If I dodged the soap every other day or showered some of the time (neither of which helps my aches as much as a bath), then we could trim our demand a little. Shifting bathtime to the evening might help me use 'residual' heat a little better, without the noticeable heat/temperature loss from a smaller house tank/store through the night.
• Driving underfloor heating from lower down a stratified house store/tank at (say) 30°C might be more efficient than normal high-temperature radiators competing with DHW for the hottest water at the top of the tank.
• As suggested by Ungrounded Lightning Rod at [archive] OtherPower/Fieldlines, for better value possibly I should use glazed and vacuum/evacuated tubes in series to maximise bang-per-buck: glazed first to get the temperature up cheaply, then vacuum to really peak up the temperature just before the water goes into storage.