I was recently given a copy of this fascinating report describing a 2-year study of a district heat network serving 40 highly efficient homes.
Using district heat in this way flies in the face of the prevailing view in the UK, which is that DH is incompatible with low energy housing. In this country we assume that low heat demand in homes means that heat losses from the network will always outweigh useful heat delivered.
But as you’ll see in this post, this view isn’t correct. The Lystrup Danish network serving near-Passivhaus-standard terraced homes has just 17% losses. Achieving this required a very low temperature heat network as well as careful design, commissioning and monitoring of the resulting system.
For engineers and district heat geeks, the report is packed with fabulous nuggets of information, the most interesting of which I’ve pulled out in the following post. If you’re not a DH geek, I imagine you’ve already stopped reading and are now watching Strictly.
Still here? Strap in and let’s GEEK OUT!
To start, it’s worth highlighting a couple of things about the homes themselves. First of all, the network spans 40 houses spread across 7 terraces, so it’s a small, low density network. This is much more challenging than, say, a big block of flats because of the additional pipe required to reach each house. This means a denser development should have lower capital cost and lower losses than seen at Lystrup.
According to the design, the primary energy for heat in these homes (as set by Danish regs for a “low energy class 1” house) is 30kWh/m2.yr. How does this standard compare to a Passivhaus in terms of heat demand? A direct comparison isn’t straightforward, but let’s give it a try.
Passivhaus says your heat demand must not exceed 15kWh/m2.yr. That’s final demand, rather than primary energy, which is what the Danish standard specifies. Converting from one to the other requires a primary energy (PE) factor, which depends on the fuel. As a rough estimate let’s say there’s a 90% efficient boiler feeding the network, with network losses of 17%. So that gives us a PE factor of 1.34 [calculated like this: 1/(0.9 * 0.83)] and therefore an annual heat demand in these Danish houses of around 22kWh/m2 (compared with 15kWh in a Passivhaus).
So we can reasonably assume that these Danish houses aren’t quite Passivhaus standard, but pretty damn close. Viewed another way, these houses require less than half the heating of typical new-build housing in the UK (around 45kWh/m2.yr).*
In real life (as opposed to on paper), the Danish occupants used primary energy of 51kWh/m2.yr, significantly higher than the design demand of 30, which the report puts down to comfort take. That’s quite a gap between design and practice, and it would be easier to accept the comfort take theory if they’d checked the completed homes to rule out poor construction. For example, I’d have liked to see an air-tightness test and maybe some thermal imaging. But I won’t get hung up about it here.
The design flow and return on the network is 55/30, which is incredibly low. These low temperatures greatly reduce the heat losses from the pipework and are what makes it viable to supply heat to homes with very low demand.
In the first year the actual return temps were 34°C – higher than expected because of defective control valves. They brought this under better control in the second year.
The first thing you might think is, what about Legionella risk? The designers addressed this risk by ensuring there’s no more than 3 litres of domestic hot water (DHW) standing in the dead legs on the secondary side of the heat exchanger. That gets you off the hook in Denmark.
The next thing you might think is, how the heck do they get sufficiently hot water off the DHW plates? The report explains that they’re only getting a 3° drop across that plate, so the DHW temperature is 52°C, still hot enough to cause scalding.
And the last thing you might think is: that temperature drop between flow and return might be achievable across a DHW plate but you’ll never get that return temperature from a radiator circuit! But surprisingly you’d be wrong – the return temps off the radiators were between 28 and 33°. It’s just that the engineers have to be very careful when they commission the radiators.
Another interesting thing about the low temperature network: it’s fed from a high temperature network, with flow temps of 80°C in winter and 60°C in summer. So you could, in theory, plug in a low temperature extension to any existing network.
In the first year of operation, heat losses were 19.9%, which was higher than the design losses of 17.7%. I contacted the engineers who carried out the study and they provided more information showing that in the second year the heat losses were 17%. When you consider how little heat the homes actually use, this is an impressive number – i.e. that’s a 17% piece of a very small pie.
Nevertheless, the engineers still reckon the network is oversized and, if they were to do it again, the heat loss could be “significantly decreased” even further. Nice to see that even the Danes fall into the trap of oversizing networks!
And they’d already used very liberal diversity figures. For the homes that were not fitted with experimental thermal stores called DHSUs, there was additional load for the first two houses but after that the slope of the diversity curve was roughly -1/n. In other words, after the first two homes, additional users did not add simultaneous load to the network. That’s almost unbelievable, especially in the UK market where engineers desperate to protect their PI routinely apply diversity factors of 0.7 or even no diversity at all! And remember, they still reckon it’s oversized!
Total cost for supply and installation of the new network and HIUs came to €8,160 (about £6,400) per terraced house. Or rather, that would be the average cost per house if the DHSUs were omitted. Adding these thermal stores resulted in higher cost and higher standing losses, which outweighed the intended benefit of lower peak demand. So I’m omitting them from the cost calc.
In summary, (low temperature) district heating can work with very thermally efficient homes, even on low density networks. But it can only be delivered with very careful design and close monitoring of the resulting system. I suspect that in the UK, we’re not quite there yet. But it can be done.
*Big thanks to Nick Devlin of Fabric Building Physics for advising on all things Passivhaus.
If I’ve read table 4 in the paper right, the losses were measured at 13 kWh/(m2.a) (that’s delivered heat and m2 of dwelling) – so about the same as the notional passivhaus heating demand.
So really good, but not quite quite as low as expected from the 17% losses figure – the actual delivered heat was around 53kWh/(m2.a) including DHW. And only 0.6kWh/(m2.a) pump power despite pipework being on small size even for Denmark.
Hi Alan,
Thanks for the comment. That’s the way I read table 4 as well (though there seem to be two table 4s!
The low pumping energy (despite higher than projected demand) looks like an indicator that the network is oversized and so lower losses might have been achieved through design.
The Danes (Dansk Industri) came to London back in October (2nd/3rd) to teach the UK how they do District Heating. (how to do District Heating properly)
There were a few local authorities in the room but not a single UK engineer part from myself. Tragic for the UK but great for me: I got the whole day with Marek Brand, system architect for that Lystrup scheme. 🙂
The Danish ownership model allows their engineers and operators to do their job. “I kept turning the flow temperature down until people complained” says the engineer. “How on earth do you get away with that?” I asked. “In the UK the lawyers would be all over you if you tweaked the design post commissioning…” He answer was simple. “My customers own my company. If I save them money they’re happy. If they’re not happy they fire me. Low temperatures save my customers money. They understand why I constantly work to improve the system. It works very well.”
On some schemes there’s a financial penalty for returning water over 30C. The customers ask for it, because overall it saves them money. Where a scheme wants to turn the flow temperature down but some houses can’t handle it because their radiators are too small, the scheme will actually buy them new one. Can you imagine your neighbours chipping in to improve your radiator performance so that they can reduce the flow temperature? Yes actually, in Denmark. “70/40” is the 1970s standard. 60/30 is commonplace and 50/30 is the new target. The town of Bjerringbro (home of Grundfos) runs a few experimental streets at 46C flow temperature with very small pipework indeed, though this is now pushing the limits of acceptability for domestic hot water. (delivered at 43C and requires insulated distribution pipework within the home)
The equipment do build 4G networks (50/30) exists. The Danfoss substations aren’t silly money: An Evoflat FSS will do your 47C DHW from 50C flow temperatures and list price is £950+VAT. You can pick up an Akva Lux II that does the same thing from a number of retailers for £400 + VAT. The right pipework is inexpensive too: a single 100 metre roll of Logstor PEXFlex twinpipe with the thickest insulation is £1600 in sizes up to DN32. £16/metre for a pair of insulated pipes. It’s hardly worth the trip to Wickes for some speedfit and a pack of Climaflex. That fancy high pressure pump? They’re of the shelf and only €2,000. 10 bar rated radiators and internal pipework? Even the cheapest stuff at Wickes is 10 bar rated these days, courtesy of the Eastern Bloc joining the EU and Soviet heating systems having been designed for 10 bar. (direct connected 17 storey towers)
Why don’t UK M&E consultants design heat networks properly? The business model is screwed up. Developers go to the lowest bidder because they couldn’t care less what the operational performance is. Every M&E consultancy with a copy of the “1970 Ladybird book of heating system design” (every standard referencing the 70/40 design temperatures or DS439 hot water diversity factors, or all of CIBSE’s output to date) thinks they can do it. Eventually a heat network operator buys the concession to operate the worthless scheme and bills the consumer for the mistakes made during it’s implementation. Oh, and the contract is locked down for 25 years, so no continuous improvement either.
The consumer pays the price in the short term. The industry will pay the price in the long term.
Back to Lystrup:
Remember the occupancy: this is housing for old people.
-The internal temperatures were at 25C 24/7/365, which goes some way to explaining higher than expected space heat load and return temperatures.
-Nobody in these houses woke up to go to school or work on a regular basis, which does influence the hot water diversity factors. The recorded curves are very much on the low side compared with say a block of one bed flats in London that’s used as a glorified hotel for city workers who are all taking the same tube.
-Keeping the potable water volume under 3 litres “gets you off the hook” for legionella in the UK too. Even at 45C delivered hot water temperatures. Not only does it get you off the hook but it works well too: 10 million gas combi boilers can’t be wrong.
-The houses were passive houses but they were LARGE by UK standards, and Denmark has twice the degree days of the UK. The heat load per service pipe was therefore still quite high. Halve the space heat demand for a UK size house and halve the space heat load and relative losses on that network will increase.
-SAP and UK custom/practice screw with heat network control. If you implemented time clock operation on this heat network (to get the SAP points) then you would introduce massive morning reheat peak loads on the network. Bigger pumps, bigger pipes, bigger losses.
-Direct connection with TRVs (what the damn things were designed for in the first place) eliminates most of the stupidity that can occur with space heating return temperatures compared with the UK custom/practice of indirect space heating plus independent circulating pump. (even if commissioned correctly the consumers will always whack up the flow rate to increase responsiveness)
-Beware poor UK water pressure. The Danes get something like 3 bar @ 18L/minute. We’re guaranteed just 0.7 bar @ 9L/minute @ the stop tap. Not all the equipment on the market (including the Danfoss substations I referred to earlier) can cope with a non pressure boosted UK supply.
I’ve picked Marek’s brains extensively and encourage you to do the same. Also strongly recommend any of the Dansk Industri events: they bring wealth of operating expertise and are here on trade missions to give it away. These Danes do have kit to sell and will bend over backwards to help you understand how to use it, but the events aren’t the high pressure sales events you might first think.
There’s another opportunity to help each other here: real UK hot water data. What are the observed load profiles/diversity factors for UK housing developments? On a retrofit we log it first – 10 sec intervals in every home – before designing the network. If you want the equipment you can have it at cost (£150/home) provided that you share the data afterwards.
We’re working with the Energy Saving Trust to unlock the datasets they’ve already recorded (10 sec interval data as far back as 07/8) to generate load profiles/diversity factors for the UK housing that they analysed. I’m hopeful that this will be made public in due course: early days yet.
Do you operate schemes? Do you have high resolution heat meter data? Do you want to collaborate to develop some proper curves to design to? Give us a shout. 🙂
Useful references:
Guidelines for Low Temperature District Heating
Peter Kaarup Olsen (COWI A/S), Christian Holm Christiansen (Danish Technological Institute), Morten Hofmeister (Danish District Heating Association), Svend Svendsen/Alessandro Dalla Rossa (Technical University Denmark), Jan-Eric Thorsen/Oddgeir Gudmundsson/Marek Brand (Danfoss District Energy)
EUDP 2010-II: Full-Scale Demonstration of Low-Temperature
District Heating in Existing Buildings, April 2014
Click to access guidelines%20for%20ltdh-final_rev1.pdf
Demonstration af lavenergifjernvarme til lavenergibyggeri i Boligforeningen Ringgårdens afd 34 i Lystrup
Lead author Christian Holm Christiansen
Report for EUDP 2008-II, May 2011
http://www.byg.dtu.dk/~/media/Institutter/Byg/publikationer/byg_rapporter/byg_r250.ashx
[Danish and English language – the full publication that the paper linked in the blog post draws from – phone a friend, ask Marek, or ask Google translate…]
Low Temperature District Heating Network Serving Experimental Zero Carbon Homes in Slough, UK
R. Burzynski, M. Crane, R. Yao, V.M. Becerra
DHC13, 13th International Symposium on District Heating and Cooling, September 2012
4DH – 4G District Heating
Chair: Prof. Henrik Lund
http://www.4dh.dk/
Marko, that wasn’t a comment – it was an essay! Great feedback though, and plenty of it.
Too lazy to write a short version. 😉
Let’s be in the same room next time Marek is in the UK and continue the discussion. You’ll love the guy. “Ah yes, Lystrup. It’s not bad but…”
To give credit where due, there’s nothing wrong with ladybird 70/40 if heat is genuinely free at source and was definitely going to be wasted anyway. (Battersea Pimlico; 1970s Danish coal plants; perhaps nuclear CHP today)
If you’re having to create it (net new gas fired CHP; perhaps heat pumps, boilers) this is where you really ought to exercise more care and thought, particularly in new build. I wonder when SAP might re-write Table 12 with the *real* emissions factors for heat networks, net of the distribution losses. That’d be a cat amongst the pigeons!