Jump to content

Diy GSHP


RobLe

Recommended Posts

Attached is a rather long description of the ground source heatpump that I built at the end of 2021.  It's completely diy - two 2nd hand fridge compressors(one for vacuuming down, one for the HP itself), bits of bent copper pipe, etc.  I've encapsulated it as a pdf so that it hangs together in one thread, and should be an easier read.  Thanks buildhub for a big filesize limit!

I started out knowing vaguely what I wanted to do, then watching every youtube on R290 compressor systems several times, googling it, reading stuff on "Ecorenovator" etc.  The youtubes are especially useful, watching people do things right - or wrong!  I've been quite nervous about the safety aspect of R290 (propane) as the working fluid, and anybody who tries anything like this should be too.  Propane is a great working fluid - it's not an F-gas, it has a GBW of 3, while most F-gasses are 1000+.  But it burns well, so must be treated with great respect - used outside and kept away from flames and electrics.

It was quite a steep learning curve - I'm ok at soldering copper pipes for plumbing, and regular electrics, but there was a lot to learn to make it work.  Which it does, and we are happy with it - Mrs RobL doesn't accept any low temperatures in this house!  There have been minor niggles (a fuse blew, lots of weeping compression joints in the water circuit before I watched a youtube on how to do them right).  All the issues have been easy to fix, and I'm comfortable with the idea that I can fix them rather than wait for somebody else.

 

DIY-GSHP-RobL-June2022.pdf

  • Like 4
Link to comment
Share on other sites

Thanks for posting your write-up - like others here I'm working my way through it. I can already see plenty of good engineering practice - what other projects have you done before this one? I'm guessing this isn't your first major build.

Link to comment
Share on other sites

Very interesting read, not all the way yet but you have spotted the vampire warming load issue (P6) which we have been discussing / investigating here: 

others will be interested in your observations / highlighting of the issue.

Link to comment
Share on other sites

You’re too kind everyone, thanks!  I’ve diy installed most green things - solar thermal, mvhr, ewi, and made lots of smaller stuff - hot wire eps cutters, ach testing, but this is the biggest home build.  
The electrics is close to my day job, it could have been done much simpler with just ‘relay logic’, but I’d not get the logging or the COP display.  I like the cop especially - it’s an easy ‘everything is ok’ indicator, no need to look at all the different flows, temps etc if the cop is normal.

I’ll go take a look at the standby thing!

Link to comment
Share on other sites

Thanks.  I will be interested in how you get on with building control? 

 

My only comment (and this applies to all GSHP's) is you spent £1500 on the pipes in the ground (which is less than i would have expected)  I paid less than that for my ASHP and did not have the work of laying the ground pipes.

Link to comment
Share on other sites

3 hours ago, RobLe said:

You’re too kind everyone, thanks!  I’ve diy installed most green things - solar thermal, mvhr, ewi, and made lots of smaller stuff - hot wire eps cutters, ach testing, but this is the biggest home build.  
The electrics is close to my day job, it could have been done much simpler with just ‘relay logic’, but I’d not get the logging or the COP display.  I like the cop especially - it’s an easy ‘everything is ok’ indicator, no need to look at all the different flows, temps etc if the cop is normal.

I’ll go take a look at the standby thing!

 

Great stuff.

 

I think the question I would ask if whether at the end of that your house can be easily operated by a random person you sell it to in the future?

 

We have various people around (or no longer around) who have doo-dah-diddled wonderful things that work great, but only potential with themselves as embedded maintenance engineer. One of our members looked at the great thing he had created, and then wrote a manual for his partner for when he popped his clogs.

 

I wrestle with this as a small LL making houses as efficient as possible but still suitable for tenants, and needing to be "tenant-proof".

 

Do you have strategies for *that*?

 

My strategy is to keep it very simple, work to long time lines, make sure I have enough money to employ engineer or handymen, and to have done most things myself at least once so I hope I have a clue.

  • Thanks 1
Link to comment
Share on other sites

Operating it is fine, it uses the same controls we had before for the gas boiler - a honeywell timer in the airing cupboard and honeywell thermostat in the hall.  Not that anybody other than me ever set the timer, but it's standard.  We have an mvhr too, nobody else would bother changing filters, we have a water filter, same again, solar water - it would just be decommissioned without me if it broke, PV - nobody else would notice if it bust(it hasn't).  Green tech needs maintenance or at least mechanical sympathy, it's not as obvious as a telly if it breaks.  Maybe without me, Mrs RobL would notice these things.... I think different people have different priorities, I never remember birthdays:-)  Insulation is great in comparison, it doesn't need maintenance.  

 

If the heatpump itself got bust, and I wasn't there, that is awkward.  If we keep the gas boiler forever, that could be used instead, Mrs RobL has been taught how to swap back - it's 2 water valves and a 3 way rotary switch.  Of course to keep this backup we'd be paying a connection charge for gas, maybe £150/year just in case.  Maybe we should, it's a simple backup that's already there!  

I doubt a refrigeration engineer would fix it if it were bust, do they generically exist?  Or is it a specialist repairing/replacing one specific manufacturers kit.  My experience of "product" is generally I buy it, and after 2 years I'm on my own repairing it.  If I get in touch with the manufacturer they really just want to sell another.  Ranty point about our Nissan Leaf - it has the oldest battery firmware, which incorrectly reports low SOH and is allegedly a free firmware change to fix.  After getting in touch, they have sent several texts, emails, even a card... to get a new car.... no firmware update is in sight yet though:-(

 

 

 

Link to comment
Share on other sites

19 minutes ago, RobLe said:

Insulation is great in comparison, it doesn't need maintenance.  

Mine did.  We had some big SW storms a few years back.  Noticed a damp patch in the bedroom ceiling.  Seems the wind had got under the loft insulation, then, when it had flopped down, had moved forward a little, just enough to allow it to wick up water.

Easy fix, just pull in back 2 inches.

Neighbour has the same problem, told him how to sort it, 4 years on and he is still complaining about the 'leaky roof', (expletive deleted)ing nob.

 

I have tried, on two PCs to print your PDF off.  Can't print it.

Edited by SteamyTea
Link to comment
Share on other sites

Cracking project! Wookey told me about this one at the pub. Reading your writeup I can't believe that he didn't "just" copy rather than buying a little Vaillant unit off fleabay... 😂

 

 

Thoughts:

 

- Serious props for JFDI-ing it. Was it as much fun as it looks? 🙂

 

- Can I come take a look one day? (I live 50/50 Cambridge/Vilnius)

 

- Agree on W2W units being far simpler operationally

 

- I think the reason that ground source units tend(ed) not to bother with inverter drives is their operating envelope. Unlike air sourced units our input temperatures are far more stable. And unlike air to air units (which must modulate else hot/cold draughts) you tend to be coupled to some thermal mass (so can cycle without adverse effect).  Also unlike air source units you have to keep the water going through the ground loop fast enough to achieve turbulence / mixing (else heat transfer falls away) yet not so fast that the pressure drop is excessive (and you burn too much power pumping the water). This extra constraint means that their sweetspot (in terms of power output / sizing) is narrower; and given that cycling is ok; why bother to introduce a costly element with a design life of less than 30 years (high power inverter drives) and suffer all the hassle of ensuring that your oil distribution / refrigerant distribution still works at a variety of flowrates and your noise isolation works at a variety of operating frequencies etc? 

 

- MCS actually has some good docs on ground loop sizing for turbuleant flow etc. Glycol will make a material difference.

 

- 5C and 40C are perhaps a bit tight for a ground source heat pump. I'd expect refrigerant at -5C (for brine at 0C)

 

- Running raw water was brave. Yes Ethylene Glycol is poisonous but it breaks down quickly in the environment. Propylene glycol is too sticky / too harmful for operating efficiency to be worth considering IMO. Alcohol is also an option...but difficult to source cheaply by the barrel for obvious reasons!

 

- Good idea allowing the ground loop to equalise for a while then kicking the pump and measuring the temperature in order to get an idea of ground conditions. Like the modelling.

 

- In brazing pub CO2 is often nitrokeg (mostly nitrogen, a little CO2) and dirt cheap. Rubbish for MIG welding but might be ok for your use. Could also use welding argon. Readily available. If anything like soldering then it's better to get hot fast and braze fast than it is to heat slowly. Less time for things to oxidise. So may be easier work with oxy. Or just a big mapp torch. There's a torch and then there's a torch.

 

- Compressions on aliexpress heat exchangers at 13 bar working pressure. Yikes thinks me. (both the compressions and the aliexpress) Good job it's outside!

 

- For practicing this kid of thing is useful for testing joints:

https://www.toolstation.com/testing-pump/p36395

 

- Getting even liquid distribution after the throttle and into the plate is supposedly quite difficult. If you're boiling in the tube entering the heat exchanger who knows what the distribution within that is like. 

 

- Pukka digital gauges / temperature clamps are not too dear these days. Pair with an app to do all sorts of lookup. £200 new; £170 used. I have a set (used for district heating actually - measuring differential pressure on water...) that's borrowable in Cambridge fwiw. Wookey has a matching airflow sensor.

https://www.amazon.co.uk/dp/B07W62L1DR?ref_=pe_27063361_487360311_302_E_DDE_dt_1

 

- Like the perf drain over the ground look. I'll pinch that and run the guttering into it. Perhaps greywater even. Warm in addition to wet.

 

- Like the automotive header tank. I'll pinch that idea too.

 

- I like the control logic on the (otherwise dumb as a rock) Danfoss unit that I need to finally fit this year. Rather than using an air stat and a fixed flow temperature it's targeting a flow temperature based on the weather and using integral control to decide when to run the compressor. That should drop your flow temperature at conditions other than design condition and bump your COP vs running at full temperature on a hysteresis stat. 

https://manualzilla.com/doc/5637761/danfoss-general-user-manual?page=13

https://www.manualslib.com/manual/903511/Danfoss-Dhp-H.html?page=50#manual

 

- I'll definitely pinch the mesh+basecoat+mineral wool construction method for an insulated box. Not for a heat pump but for the services entering our cabin build. (need to stop the potable water freezing between the 2 metre buried depth (where it won't freeze) and the ground level (where it will freeze) and the floor (a metre above that)

 

 

 

Ideas:

 

- Drop the flow temps by running longer at a lower temperature. Air temperature by hysteresis isn't ideal.

 

- How are your radiators balanced / how do you manages flowrates in those? Any additional total output (at a given flow temperature) to come from revising this?

 

- Suspect you may have liquid distribution issues in the evap. Ideally this wants a straight shot from your throttle device into the plate; perhaps with additional distribution device poking into the plate:

https://www.swep.net/refrigerant-handbook/6.-evaporators/asas5/

 

- How much superheat do you actually have and how much does it vary? Is it worth trying to do an electronic expansion valve at the same time as tweaking the distribution into that evap; or is the operating window so small that it doesn't much matter anyway?

https://www.hvacknowitall.com/blogs/blog/371661-adaptive-vs-fixed-expansion-valves

https://www.danfoss.com/en-gb/about-danfoss/articles/dcs/evaporator-injection-algorithm/

 

 

Link to comment
Share on other sites

@Marcocosic

Some insightful points there, I think you know a lot more about this than I do!  There's many things that I did wrong in retrospect, down to basic not knowing how to use a water:water compression joint.  Only after making 12 or so of them for the groundloop distribution manifold, did I realise that I needed jointing compound for a leak free fit.  And you shouldn't fit the pipe all the way in, or you can't keep tightening the nut later.  It's like I'm making it up as I go along...Gah.

 

Anyway - yes, please come and visit, cast an eye over it sometime!  address pm-ed.

Yes, it was great fun to do!  So many new things to learn how to do... it's hard to find technically interesting and useful projects, and this ticked both boxes.

 

Raw water worked a lot better than the glycol, I wish it didn't need the glycol.  After you mentioned alcohol, I looked it up - you can buy denatured(undrinkable) alcohol a lot cheaper than the glycol I got.  Hmm, maybe I should've done that.  If it has a major overhaul one day I will replace the heat exchangers with bigger ones, I think I can scrape back ~4C between them if I increased from the current 24plate to 60 plate.  The ones I got allegedly are working pressure 16bar, I've seen some that say 30bar.  They're £30-£70 each, for basic ones.

I followed that link for the swep throttle connection thank you! - I think the suggestion is that the tiny cap tube to BPHE joint is critical, and the internal shape should be always increasing in volume.  Hmm, you may be right, I'd not heard about that, but it makes sense.   Doubt mine is perfect there:-(.  I don't know how much a better throttle(EEV/PRV) would improve things - I don't think they're common for low power compressors, but maybe it's a future improvement?

I did see that mcs doc, it was very helpful.  I sized our system from it, so it should be no surprise we need glycol I guess.  Somewhere I saw a paper on water v glycol in gshp systems - I think the conclusion was that in countries with higher mean temps than the UK it was worth adding a bit extra groundloop pipe so that water could be used.  In colder climates there's hardly any headroom before freezing, so the extra pipe length needed becomes longer and longer and so uneconomic.

 

Our rads are certainly not perfect - I know the system works better if I increase the internal pump speed to full, but I've not characterised "better".  Most of the rads are in a 1-pipe arrangement, some in a 2, bit of a hodge podge really.  The 1-pipe means that they will never have been usefully flushed since being fitted in 1963 (1 pipe means every rad has a permanent bypass pipe, and the flow is only convection in the rad).  I could probably improve them quite a bit by doing that one day - simplest I think would be to add 22mm wide bore valves to every bypass pipe, then I can do a conventional powerflush one rad at a time.  We have powerflush fittings already I think - I remember it being done 10+ years ago by a plumber, but I know now it cannot have done much due to the 1-pipe.  It all "just works", so not a lot of incentive to rip it out and replace with 2 pipe everywhere, mess and pipes chased all over, no ta.

Our heat measurement system uses a flow meter, which I think is accurate, and two temp sensors on the water pipes inside the house.  The temp sensors seem significantly better than the spec.... but the spec is +/-2C over their range!  I've ordered some significantly better ones, and that should help enormously when they arrive - when being the operative word, as I ordered 4 months back, and they're still waiting - so many IC's are late/unavailable these days.  If I can measure the COP accurately, then I can see exactly what improvement changing pump speed has, or insulating the pipes in the HP itself (not yet done all), etc.  And what improvement flushing rads has too.  It's satisfying to have verifyable improvements.

 

From memory the superheat was always in the 4-10C range, the 4 times I measured it, but I am at the mercy of the accuracy of my gauge set, and how well the temp probe is connected.  If I get better COP measurements going, another technique is to slowly increase charge until the COP stops improving at the operating point?  Below is a graph (random googled graph) - I've seen this sort of thing before.  The graph does seem to imply that as the cold side gets colder, the optimum charge is lower - I should bear that in mind.  Would a TXV/EEV avoid this optimal charge varying? 

image.png.21bb6626d52a74b46c340979ecab5dc8.png

Link to comment
Share on other sites

  • 4 months later...

Updates - the heatpumps on again, as it's getting colder!

 

I added a bypass heat exchanger to the heatpump, which could allow some summer cooling - ie. groundloop bypassed via the extra heatexchanger to radiators.  TBH it's a bit rubbish, it can pull out 500W average over a 48 hour period when I tried - barely worth the complexity.  The issues I had with it are that by the time it's needed the ground has heated up quite a bit as my coils aren't that deep.  Also our radiators are 1-pipe so only any good at convecting hot water, not cold, from the supply pipe that runs around beneath them.  Not that we are really in need of summer cooling, it just seemed like a freebie.  It uses the same thermostat, inverted to NC rather than NO as a control for the system.  I guess the "COP" is good, it's 60W of pumps and controls, extracting 500W out, cooling COP of 8.

 

I improved the heatpump startup considerably - originally as a simple on/off unit it used an internal NTC element to give it a soft start, now it has a step-start with a relay instead.  I had my 5A heatpump fuse "wear out" after ~2000 startups previously, and I hope that dropping the startup current will prevent this occurring again.  The I2T drops from 60A^2s down to 20A^2s with this change, it makes a huge difference.  Both the NTC and step-start are options for the compressor - I'm not inventing stuff, just reading the datasheet!

 

Increased the maximum run temperature, needed for DHW.  It was 50C, now it's 55C, and slowly gets the top of our tank to 46C.  The 200litre tank isn't "heatpump ready", and only has a 0.42m^2 coil - I think this is just about ok as the heatpump is only 2.5kW - if it was more powerful, this small coil would be a significant issue.  There's an immersion to get it to 65C once a week, timed just after the heatpump has finished.

 

Added a run cap to the startup coil, in addition to the normal run coil.  The compressor datasheet says this improves COP by 5%.  It's a bit marginal a difference to be sure if it's true of not, but I'm a believer!

 

Total electrical energy for heating+DHW our 4 bed home 6Dec2021-10Oct2022 is 970kWh, which I'm very pleased with.

 

Added some silent 2W danfoss valves to stop a few rads heating up when the DHW is being heated:

image.thumb.png.663d9d24a353b454a7843edf2871db12.png

 

Link to comment
Share on other sites

  • 3 months later...

The compressor didn’t turn on a few times recently - everything else worked, the relays clicked, but no compressor, COP=0.0 reported, system power draw of 60W - just pumps. Gahhhhhh.    
The compressor is powered via the contacts of 3 relays in series - call4heat, pumps, and compressor itself.  It’s done like that to be sure the pump must be on if the compressor is, even if the controls go wrong - it’s all controlled by my rubbish software after all.  Anyway, after about a year of switching, the call4heat one had failed - that’s the one that usually does the turning off and gets the spark.  Turns out I should have fitted an RC series snubber around the contacts to save them from the spark every time the relay switched off.  Here’s a pic of the old one.  It still works 50% of the time in that state!  I replaced relay+socket, and put a series combination of 47ohms+100nF across the compressor and a MOV across it too.  It seems to work, I can’t see a spark now when it switches off. 

I’m glad I caught it early, the house temperature didn’t really suffer.  Obsessively looking at the COP is helpful after all 🙂

FC91335E-D0A7-4E7E-A1EB-42C2C2E233CE.jpeg

Link to comment
Share on other sites

3 minutes ago, RobLe said:

Turns out I should have fitted an RC series snubber around the contacts to save them from the spark every time the relay switched off.

 

Yes, any inductive load will just want to keep the current flowing and generate a huge back EMF in an attempt to do so. Perhaps you could use a SSR with built-in snubber for higher reliability?

Link to comment
Share on other sites

Yes, maybe a good idea.  That relay serves two purposes:  one pair of contacts is used at low voltage, it tells the software ‘Call4heat’, the software checks there’s flow, nothings too hot or cold, confirms it’s not short cycling.  The other contact has the compressor routed through it.  I could just omit routing power through it, use an ssd to switch the compressor on/off directly via software….rely on my software being ok 🤣  It lasted a year sparking, I reckon it’s fine now there’s no spark.

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...