DIY W2W heatpump

[Robl]

We now have a relatively well insulated house in the UK, needing a peak of 1.5kW over the last year to keep it warm in the coldest of weather. That would be 1.5kW of heat continuously on for 24hours on the coldest day, in addition to the 500W of people and dog and standard electrical use. Our house warms up by 10degC per kW of continuous heating, and the existing water radiator circuit couples to internal air with 5degC per kW of heating, as measured at the boiler output. That simple statement took me a while to get the data for it - we have a "smart" gas meter, which logs daily measurements, which makes it a lot easier, and I've got a diy logger which gives nice graphs. From this, if we heat with 2kW continuously, the water circuit would operate 10C above the target room temperature of 21C, and the house could warm up even in the coldest of weather we get, to above 21C. The radiators are oversized - they're mostly original from 1963, but they could could keep up back before the house had external wall insulation fitted (big recommend!), and now the heatload is a lot lower.

I would like to replace our existing old gas boiler with a small capillary tube metered heatpump, and I'm musing with building my own gshp unit using an R290 compressor. AC Hacker and others on this forum that have persuaded me to consider it - so many impressive projects! If I did it, the unit would likely live outdoors next to our brick garage resting on a concrete block, the exposed sides having rendered rockwool insulation and a little airflow. The insulation means the air temperature in the box will be closely related to the garage next to the house, say 10-20C temperature all year round unless selfheated, and the render will keep it all watertight. It would be a lot simpler to put it indoors, but I think it will be ok outside, and it removes the propane worry.

To get the low grade heat in the first place, we'd need at least 100m of hdpe pipe buried in the back garden. This is another tricky part, but my hope is that a trencher would make it relatively painless to make a series of 1.2m deep 200mm wide cuts that pipe would be pushed into, giving 3 off 40m parallel loops. For the moment though I'm focussing on the heatpump itself, probably as it's most fun:-)

I picked up a 2nd hand compressor, NLE12.6MN, by Secom, 530W in and 2.3kW heat out from 5C up to 35C. I've hopefully attached a picture of the setup I'm thinking of. It's a bit complicated, as there's a direct "free" cooling path that I intend as well as the heatpump circuit itself.
I've plenty of diy plumbing experience, lots of electronics soldering, and I've watched lots of youtube vids on heatpumps with R290, some pro, some amateur.

Below are some details I've discovered randomly on the internet - please say if you think I've got stuff wrong!

Get the heatexchangers the right way round. The top should be the hot gas end, the bottom the higher density liquid. This doesn't matter so much when there's no phase change as the density doesn't alter much, but any refrigerant gas will have a density change. The left side ports (when "upright") should be the R290 side, so it's encompassed by water both sides.
An accumulator on the suction line reduces the likelyhood of liquid getting to the compressor, which is bad for it
Silver solder together the suction line and the capillary tube - Secop Capsel shows this, as do other places, but plenty of capillary systems don't seem to do it. I think it maybe helps nudge the operating conditions a little towards optimum keeping things more efficient when not quite right. Say there's too much subcooling, not enough superheat so the compressor might flood with liquid, this detail might prevent that.... I'm way into bluffing it territory now mind, I'd love somebody to point at something authorative on this:-)
Twist the remaining capillary tube up by bending it round something neatly, so it doesn't mind the compressor vibration.
Braze (15% silver) pipes to the to compressor, less critical joints elsewehere could be lower percentage silver soldered. The compressor joints must be brazed so that they are stronger and take longer to fatigue fail than silver soldered joints when subject to the continuous vibration stresses.
Bleed nitrogen or CO2 through the system while brazing, otherwise copper oxide flakes form on the inside and might damage the compressor.
Ideally all R290 joints to be brazed, possibly silver solder away from compressor - there seem to be strong opinions both ways about this - as I understand it brazing is so hot that the copper is weakened, but then silver soldered joints are not as strong. Least favourite are compression joints - all pipes and joints leak a little, but compression leaks most.
Secop Capsel is a great free tool that will calculate the optimum capillary tube size.
Danfoss Coolselector2 knows all the specs of Secop compressors (Secop=Danfoss), and is very powerful. I haven't mastered it - I don't know if it can simulate capillary systems or only more complex ones; I'm just using it to suggest appropriate pipe diameters and pick a compressor!

A few questions -
I'm assuming for a diy system, I need schrader ports on the discharge and suction lines so that I can connect up a guage manifold set and choose how much refrigerant to put in by matching running pressures to what Coolselector2 says. Do these have to be near the compressor, or actually better at the heat exchanger end of the pipes, so they're less affected by vibration? I'm just intending copper tee pieces for these connections.
I don't think I'll need the service valve on the compressor - I think that's just for when a cheap system is being comissioned, and then there's no need for 2 valves as no pressure measurements are made. I think I need the two service ports, as I won't know how much R290 to put in without measuring the running pressures. Is there a reason to braze on a schrader to the service port anyway?

Love the site, any advice taken with thanks!

[jeff5may]

Just got a free minute to check the forums. Life's been hectic for the last few months and all the side project stuff got put on hold.

It looks as if you are overestimating your compressor mass flow. Looks like a little pumpkin to me. If it's drawing around 500 to 600 watts, I would guess that it might move 1500 watts+ during ramp up, then taper off to somewhere around 1000 or so at equilibrium.

If you're using all copper fittings and tubing, you can use silfos rod with no flux. Once you start using silver, you definitely want to use flux. With the small stuff under about half inch (12mm), it's easy peasy. Just use an oxyfuel torch and the bare copper phosphorus rod. On the window unit and dehumidifier size systems, I use the little scuba tanks. They're super common, so it's not a big deal to exchange them.

[Robl]

I finished the heatpump build, pressure tested with CO2, vacuumed it & filled it with propane, then turned on the heatpump for the first time yesterday!
I charged it with the pump on, aiming for pressures I'd found with Coolselector2, which used maybe 200g of propane to give ~ pressures of 5bar(low)/13bar(high). Sadly I wrote the "before" gas can weight down on the can, which got all frosty and the number rubbed off.... and I can't remember it. The can originally had 360g in, I wasted a load last week hamfistedly putting the regulator on, and there's still some left hence ~200g. It was supplied in the "on" state, and like the CO2 one I have, I thought I had to tighten it on to stop leaks - but no, that just made it worse.. Uhh, amateur! I am doing all this outside with gloves and safety glasses on at least.

There's a picture attached - all the heatpump parts are on the outdoor grade ply, the compressor next to it is used as a vac pump. I've attached hoses to circulate water for now.
I ran it with a 3 thermocouple logger and an elec power meter manually read, with "cold" water from the tap at 8lpm being chilled down by a heatexchanger and dumped in the garden. The hot loop ran from a 27.4litre tank of water, pumped at ~8lpm through the "hot" heatexchanger, and back into the tank. I pushed the data through excel and I get some graphs, attached too. I got excel to do a linear fit on elec power in and delta(output temp), rather than use the actual data to calculate the COP, just so it's a noise free result, which I think is fine.

[Robl]

I've run the rig a few times now logging data, and analysed the data a bit better now, taking into account losses of the 27litre water tank to air - I did this by only running the circulation pump allowing the 27l tank to cool down, and then curve fitting to get the losses. I find 15W/degC relative to air, so when running hottest I get 300W extra loss, which is significant given the system is small.

I tried working out the superheat, which started at around +20C, then added some more R290, then I got +10C superheat. I added some more again - I think too much, not sure if I had any superheat at all at this point, and the efficiency dropped. I've added a picture of the calculated COP versus DT, and all the runs are similar except for the one with the highest charge, which gave to worst result.
Today I planned letting out some gas, to get back to the better performance - only to discover some must have leaked out - my pressures were a bit lower again, and the efficiency was good again. Hmm, clearly something isn't quite gastight. I have 4 heatexchanger connections that rely on nitrile rubber washers, and I left the manifold guage set on, I think it will be one of those things.

There's some vibration on the thin capillary tube, at the end, after the section soldered to the liquid line. It makes a white noise sound which I can deaden by touching it - I presume there's bubbles of gas in here? Is this a sign of something I can improve somehow? Should more of the cap tube be soldered to the liquid line, or the length be more/less? All the 4 charge levels I used made this happen, I think in a similar way. There's a pic of where this is coming from attached, I circled the pipe in red.

[Robl]

I've just about finished my ground-source heatpump project, it's installed and been running well for the last 3 weeks here - it provides all the heating for our house now, and the gas boiler has been switched off. If it goes wrong, there's a changeover switch so that the gas boiler can be swapped back in - success will be judged in a few months by that switch being left alone by my wife!

Many thanks to all on this website that have provided such great information, I'm not sure I would have started it without that.

I've written it up into a long pdf that's attached - I thought it would keep all the information together for an easier read for anybody else who was interested similarly.

[TobyB]

I've got several 'little' gshp projects like this in various states of
'not really started yet' so it's good motivation to hear that yours is
working, and doing what you predicted.

What did you end up doing out in the garden for a heat field?
edit- holy cats, I found it- and more- in the PDF! That's very detailed!

I'll be curious how your thermal reserve holds up to the cold days this spring.
Like you, we live in a pretty moderate climate, and DIY digging the field much deeper
than a a meter or so is almost impossible due to the geology. The colder climates
need to go deeper- so while 2m is well- known, I've not found a lot of data
about how well a shallower depth works when it's +5c in the winter.

Brilliant!

t

[Robl]

Hi Toby! Carnation WA seems a very similar climate to Cambridge Uk, but with a lot more rain in winter which is good for gshp as it makes the ground more conductive. I recommend looking at the doc below - it takes a bit of getting your head around, but will predict minimum pipe length at a certain spacing and depth, with borehole/linear/slinky all covered. Ground conductivity seems to be the most crucial aspect of it all.

https://www.gshp.org.uk/pdf/MIS_3005...ing_tables.pdf

All the temperatures here are being logged, I like graphs, I’ll add one in later to show behaviour of the heat field over time.

[Robl]

6 week update:
We have had a colder spell here, and the heatpump has been working for longer periods. After running constantly for about 10 hours in the morning a week ago, it tripped out on undertemperature, stayed off for 20mins, then back on again, tripped out again 30mins later - cycling like that. Prior to this, we left the house unoccupied and unheated for 3 days - this probably exasperated the freeze issue.

I suspect that once the heatexchanger ices up, just turning the system off doesn't properly de-ice it, so I changed the software fault condition so that on undertemperature it switched off the compressor and runs the pumps for 10mins - that seems to help a little. Ultimately though, the ground is getting colder, and the heatexchanger has been showing signs of freezing up and restricting the flow - which I expect is a horrible death spiral, so I decided to put glycol in the system.
I bought 15litres of "Coolflow DTX" glycol, to get a 25% glycol: water mix, and added it to the groundloop manually - while the system was off I blocked off the groundpump and put it in a header tank, draining away water, so all the glycol went into the groundloop, then released the pump and turned it back on.
It's definitely not as good with it in - I think I loose an extra 3C on the evaporator due to a combination of lower flow rate and poorer heat capacity.
The groundloop flow was 19litres/minute with tapwater, now it is generally only 12litres/minute. Originally I considered measuring entrence and exit brine temps - but these are difficult to measure as I have used flexi hoses, and I found if the heat ex freezes up, the out temp is irrelevent. There's a pic attached showing evaporator heat ex temp, also air temp , over the last 5 weeks or so.

I guess it is ultimately better now with glycol; I have dropped the trip temperature to -10C, and I doubt this will trip.

I measured the superheat of the heatexchanger, after it has been running and installed for a while. I calculated the evaporation point (from the suction line pressure) is 3degC below the measured gas suction line at the compressor, which is I understand about right.

So far total elec in of 304kWh used, and the calculated heat into the house is 1365kWh. I have ordered more accurate temperature sensors, to improve the accuracy of the heat output (DeltaT=5degC * flow rate = 7litre/min). I consistently measure ~2.4kW output, while the "coolselector" software predicts only 2.2kW. While it's nice appearing to have an efficient heatpump, I strive for accuracy:-)

[Robl]

I found an online manufacturers BPHE calculator tool, here:

https://www.heat-exchangers.uk/online-calculator/

The way you use it is to enter the inlet and outlet temps that you want (all 4 temps), and it then calculates which of their BPHE products could be used. It's a bit frustrating, as clearly they can calculate everything - but only give their parts as an answer. Still, they allow water, and a mix of water&glycol to be used; there's no propane option. When glycol is used, they typically swap their BPHE recommended to a larger one

For example, for water:30% glycol/ 70%water, with a 4C deltaT and 1.8kW transfer, they recommended B5THx16, similar area to the part I've used. Swapping to water:water allows a 3C temp drop to be entered, while they still picked the same part. Their heat exchangers might be better than my cheapie ones, but it holds that using a glycol/water mix means the thermal transfer is worse than with water.

I'll try and measure the deltaT I get for both heat exchangers, and decide if I care enough to upgrade them at some point - there's no rush, it's all working well still.

[jeff5may]

Been busy lately with life and such...

Good job, man! Haven't had time to read your pdf yet. Couple of answers to your questions, though.

Yes, the whooshing noise is normal for a cap tube system. It means you have refrigerant mass flow.

Regarding superheat: it's not a bad thing. You have to have at least a little bit of margin to make sure that the compressor is not ingesting any liquid. Propane has high heat capacity compared to most other refrigerants, so it actually helps your approach temperatures in the heat exchanger to have extra. 10 to 20 deg F isn't too much for a cap tube system.

For trenched ground loops using poly pipe, the goals are 5 watts per meter degC for 3/4 (19mm) pipe and 10 watts per meter degC for 1 inch (25mm) pipe. These goals are beatable nearly everywhere if the balance of the heat pump system is operating correctly. Just like the statement "everything worth insulating is worth super insulating", every ground loop worth sizing and installing is worth oversizing and installing.