Earth Notes: 10 Years at Home with GSHP, Western France: Case Study

Updated 2024-02-01 21:47 GMT.
By Damon Hart-Davis ORCID logo.
Learn from a home ground-source heat-pump experience in France. #GSHP #heatpump
We have to install millions of heat pumps across Europe to meet climate targets, some of them GSHPs for even higher efficiency than ASHPs. What are they like to live with in reality? This decade-long case study from a home in France gives an idea, and reveals a CoP of over 400%!

Many thanks to "Topher" (Chris S), who has provided reports and images for many years!

Executive Summary

Topher says: In all, I am really pleased with the results over the 10 years...

Points of note (far more detail below):

A Real System in France, Since 2010

"Topher" (Chris S) in western France has very kindly allowed me to record some details in his words lightly edited (and show some of his photos) for his GSHP installation in August 2010 (pompe à chaleur).

May 2010: interesting boring machine and compressor

See more detail in this Green Building Forum thread.

It is a 15 KW 3 phase heat pump made by Nibe in Sweden, but badged as Atlantic - the distributor in France. It will be supplied by two 100 meter vertical bore holes drilled through water courses, but will use a closed circuit supply of two pipe loops per bore hole.


Topher supplied some log file outputs for about 3 days from 2011-02-23.

Topher's note on the meaning of logging parameters (from the Nibe Manual).


Topher notes that:

The boreholes were done long before the heat pump installation. [May 2010]

2010-08-22: report 1

May 2010: huge amount of spoil and water from boring second hole
Of my two bore holes one is 102 meters the other is 95 meters deep. Both will be connected to a manifold, then from the manifold to the heat pump in the house. I am keen that the pipe lengths to both bore holes are equal, so that the flow to both bores are equal to get the best possible heat transfer. So the horizontal trench for the 95 meter deep bore hole needs to be 7 meters longer (102 - 95 = 7 meters). I am hoping to get this by having a zigzag trench from the 95 meter bore hole. Just one zig, not too difficult.

At this point Topher made the terrible error of confusing me with someone who knows anything on the topic and wondered if it was important to have the same length pipe runs of it 7% was ignorable: ah, if only I knew!

2010-08-23: report 2

Work started today. A huge amount of stuff arrived. Fittings, inertial cylinder, many containers of antifreeze, and so on. The trench route question has been solved. The lengths of the pipe runs to the bore holes will be be the same, there will be just a short length of trench where all 8 pipes from the boreholes will be together. I have told my French Heating engineer about this blog and he is going to read it. Who knows he might even contribute a bit too.

2010-08-24: report 3

Due to the house layout, the best position for the heat pump and 200 litre inertia tank is in a loft space on the first floor. There is plenty of room in the loft space, but it needed 5 people to lift the heavy heat pump up the stairs. This was done with efficiency and good nature by the team.

Today, Dominique and Pierre are connecting lots of pipework. Tomorrow a new person will be digging the trenches for the borehole pipes to be connected to the manifolds and then through the wall, into the house and up into the loft space. I will be connecting a 'Current Cost' energy monitor (Model CC128 ENVI) to the three phase supply to the heat pump. I hope to record every single watt that it consumes.

I am very aware of the huge loudspeaker that the floor of the roof space makes to the visitor's bedroom below. Previously I was planning to have a heat pump made by a French Company called GéoThermic. I visited the factory and listened to the noise it made, it was significant. I was planning to have it mounted on hangers on the very solid wall (1 meter thick) with insulating pads. I also visited a neighbour who had a very noisy heat pump (much louder than a fridge) in their kitchen. They were annoyed about the noise, but had grown used to it.

Then I discovered Atlantic - they import Nibe heat pumps from Sweden. I was able to visit the classroom where they teach installers about the product. Using the inaccurate measurement of my ears, they are as quiet as a fridge. The compressor and heat exchangers are inside a sealed metal box. It must have a great deal of sound insulation. The heat pump will be on additional sound insulating feet on a thick base of high density chipboard which will be on the floor of the roof space. The Atlantic spec says the noise level is 21 dB(A).

2010-08-26: report 4

In the garage are eight 20 litre containers of antifreeze for the borehole circuits. The heat pump supplier recommends protection to minus 15 degrees Celsius. Some of the containers are pre mixed at 33%, others are neat glycol. The glycol manufacturer says that a 30% mix gives protection to - 15 Degrees C.

I took a sample of the 33% mix and put it in the freezer. My freezer has -16, -18, -20, -22, and max settings. I do not know how accurate the control is. At -16 and -18 it stayed as a bright pink liquid. At -20 it has now become a slush.

Normal winters here get to -4 or -5 degrees, but mostly hover around zero. My neighbour who has been here a long time, said that once it got to -15, but that was very rare. So I think it will be OK.

The borehole installation is covered by AXA insurance for 10 years. I was surprised to see that they do not demand a certain antifreeze concentration.

Trenches were supposed to be done yesterday but the digger driver was not available. He is here today, and my rather badly maintained garden is being destroyed. It will re grow.

Topher also did some handwaving recalculation of his system parameters for comfort:

The heat pump COP is quoted as 4.25 at 35 deg Celsius. I have underfloor heat so I will run at 35 deg or less. At 35 deg, the heat output is quoted at 15.42 kW. So if I understand correctly, I will take 15.42/4.25 = 3.63 kW from the electrical supply and 15.42 - 3.63 = 11.79 KW from the 95 and 102 meter bore holes.


I need 11.79 kW from the bore holes. Bore hole 1 (95 meters) and bore hole 2 (102 meters) give a total length of 197 meters. This does not include about 25 meters of trench at a depth of about 1 meter, which will provide a little more, but I have ignored it. So I need 11,790.00 watts from 197 meters, this comes to 60 watts per meter.

I have 4 pipes per hole, no spacers between pipes, no grout in gray shale, and some layers of quartz, and clay. I think that both bores will be full of water.

Topher also bemoaned his lack of spacers (see pic):

I was unaware of the availability of spacers, Had I known, I would have insisted on their use. They are only simple plastic things, which can't cost much. They will increase the time taken to lower the pipes into the hole, but not by much, say 15 minutes max. If you don't use spacers pipes will probably touch each other, reducing their ability to abstract heat from the surrounding ground. If you have a lot of horizontal trench from the bore hole/s to the heat pump, then put spacers on the horizontal pipes too. The trench will need to be a bit wider, hence more damage to your garden. But your garden will be wrecked anyway, so no problem.

2010-08-27: report 5

GSHP borehole and pipework and a lack of spacers

The garden is now totally destroyed - just the rather poor quality grass bits, the flower beds remain intact. I had expected it, but when you actually see it happen it is a surprise. Trenches have been dug that slope downwards towards the bore holes; at the deepest they are 2 meters deep rising to say 0.5 [metres] as they enter the house. Today the 8 pipes from the bore holes will be laid in the trenches, they will be filled with antifreeze, then connected to the heat pump via the manifolds. I have pics if anyone is interested.

They will be topped up with antifreeze and pressurised with a pump as the circuit is closed. Dominique the boss will not allow a connection to the bore hole circuit for refilling with mains water in case of diluting the antifreeze. It makes sense. If I need to top up, I will have to buy a pump and get some correctly diluted antifreeze.

The Nibe heat pump has a control that allows the primary (bore hole) circulation pump to run 24/7. This will be useful to ensure all the air has been bled from the circuit. Later on it can be set so that it starts 20 seconds before the heat pump.

The heating circuit is almost complete. They are going to flush it out with mains pressure before filling with water and inhibitor.

It is 9.15 am and Dominique is hopeful that the pump will be switched on today. I am doubtful, we will see.

2010-08-28: report 6

I was right. Installation was not completed on Friday evening as planned. However, Dominique worked his socks off pushing himself very hard all day long. There was power cut in the afternoon, heavy rain in the morning, and a broken underground rainwater pipe to be repaired - not helpful.

Each bore hole loop is filled with the pink antifreeze mixture. When the pipe loops were put down the bore holes, the drillers put water in the pipes to make them [heavier] and assist their descent. At this stage of the installation, a small electric pump is used to force the pink mixture into the pipe. When clear water coming out the other end, turns pink, both ends are sealed. The next step will be connection to the manifold. Fortunately each pair of pipes has one plain and one with a longitudinal stripe. In this way, as long as all the stripes are connected to one side of the manifold, and the non-stripped to the other side, all will be well. Finally antifreeze will be pumped by hand into the closed system and the circulation pump started. Air will be bled out of the system and the correct pressure set.

The neat glycol has to be premixed with water before being put in the pipes. Apparently if neat antifreeze is put in the pipes and expected to mix with the existing water, it does not. Slugs of antifreeze circulate with slugs of water. This is not like adding inhibitor to a radiator circuit.

Work will re-start on Monday and take 2 or 3 days I expect. I am concerned about the garden being left in the best possible state. The trenches have taken up lots of space and clearly they will have the bright red clay surface. I would like the non trench areas restored as well as possible. There are some piles of soil - for example near the bore holes where the fine material was pushed out from the drilling process. I am hoping I can get Dominique to encourage the digger driver to do the best he can to get a good result.

When the installation is finished and the heat pump commissioned, a stage payment is due. The final payment will be made on the 1st December when the system is working correctly and keeping the house warm.

Warning: 2010-08-30

I have learned something today that I wish I had known before. If you are having ground source put in an existing house, plan very carefully the clearing up of the site afterwards.

If one contractor is doing both the borehole and the heat pump installation, it will be easier. The trap I fell into was that the hole boring people said that all the material that came out of the bore hole would be used by the people that dug the trenches as it was a fine sandy consistency and would go well with the pipes. Now the people doing the trench say that the hole boring people should have removed the material from the bore holes.

I am trying to get Dominique to give what was once a poor grass area the best possible tidy up before leaving. We will see the result in a few days.

2010-08-31: report 7

We are at the beginning of day 7. Virtually everything is connected up. The primary (bore hole) and secondary (underfloor) circuits are pressurized to just over 1 bar. Outside the inspection chambers for the two bore holes and the manifold remain to be completed. Dominique worked hard on the digger yesterday to try and make the site as good as possible. There is lots for us to do to get it flat and in a state for the grass to regrow.

The heat pump (made by Nibe, Sweden) is not yet connected to the supply. It has a very super little facility - three current transformers for the three phases, they go over the incoming supply wires to the house. The heat pump is set with a max incoming current value, which is the same as the rating for the supply. If the heat pump load were to cause an over current, hence a blackout in the house, it reduces and redistributes supply across the phases to ensure that the max current is not exceeded. Cool eh? Only on the 3 phase machines.

In France, one factor that affects the tariff is the max kW of the supply. If you exceed the max, it trips, and you have to switch off the offending device before re-closing the circuit breaker. Some people have relatively low supply to save money and do their load regulation manually. When we bought the house we had a 12 kW (3 phase) supply and would often exceed the max of 4 kW per phase. When we switched to the Tempo we upped it to 18kW and now have no problems. It is another example of the way the French control their maximum demand.

2010-09-01: report 8

Day 8, bright and early (8.15 am) Dominique and Pierre are working again. Both primary and secondary circuits are connected and have held their pressure. Final electrical connections are being made. The Nibe 3 phase models have current sensors for the incoming mains, so that if the potential consumption of the heat pump would exceed the max of the supply, the heat pump changes its phase distribution or waits till the house load has fallen. The current transformers have to [be] fitted to the three incoming power cables, in my case a very full consumer unit (old style with fuses) does not have enough space. I am considering having it replaced which would give me circuit breakers and space for the current transformers. I expect it will be very expensive. I may not do it.

Now it is lunch time and the heat pump has been powered up. Everything seems to be working except the connection to the room thermostat. The control interface is really good, a full colour display and intuitive controls. Dominique says it is far better than another well known make. At first power up, the system takes the installer through a commissioning procedure. The compressor has run briefly and in the limited time available to listen, I expect no problem with noise. The two circulating pumps are proportionally controlled, so no adjustments are needed. As I type the primary (bore hole) pump is running to get rid of all the air. My current monitor is showing 155 watts.

So far I am very impressed with the Nibe product, although the manual does not correspond with the screen images - usual software documentation problem.


2010-09-03: report 9

After nine and a half days work, it is done. Dominique and Pierre have left. You may think that 9.5 days is a long time, it is, but it was all solid work. Installing in the attic of an old (200 years) farmhouse was difficult. A new build would be much easier.

The Nibe heat pump is really excellent. It has many features, all easily controlled from a user-friendly colour display. Each screen has its own help screens. One very minor negative point - I had asked for a room thermostat. Normally the user adjusts the heap pump controls to get the desired temperature. The room thermostat with colour screen, would not communicate with the heat pump. Reason - down level firmware in the heat pump. Very quickly the Atlantic (French distributor of Nibe), sent me the correct level programme on the internet. It was transferred to the heat pump with a USB key. Cool eh? Perhaps the next generation will be connected to the internet, it would give remote update, remote control, and error reporting.

Last night I set the thermostat to 30 degrees to try and get the heat pump to start. It did. We had a hot night as underfloor heating has huge thermal inertia. It is very quiet. My concern about the floor of the attic space where it is installed, acting as a sounding board to the rooms below is totally dismissed. You have to stand in front of it to know it is working.

Both circulation pumps are software controlled. At the moment both are running 24/7 for 10 days to get all the air out of the circulating water. After 10 days they will return to automatic operation.

In late November, when it will be colder outside, Dominique will return to check that the system is working correctly under heating conditions. At present with outside temperatures of 17 to 26 degrees, it is doing nothing.

My plan to pay 40% at the start of work, 50% at installation complete, and 10% at the beginning of winter when correct working can be proved, seems to be working well.

It is really dead obvious, but when you experience something directly, it reinforces the well known rule - "The more preparation you do, the better will be the result." I wish I had learned more about the bore hole technology and discovered spacers. I also wish I had thought more about making the ground good after the trenches had been filled. If I had, I could have been more precise with agreements with the hole driller and the heat pump installer.

2010-09-08: report 10

The weather is cooling here and I have found the heat pump has been coming on at night. I have reduced the thermostat setting to try and prevent this.

Using power monitors is of no use in measuring real power consumption as people have told me. So I am planning to buy a PRO 1250D 3 phase kW Hour meter to measure the true consumption. Cost including VAT and delivery is £90.65 (GB Pounds). If anyone can suggest a cheaper alternative, I would be grateful.

In some places on the primary circuit there are unlagged parts, for example the filter and valve unit. When it is humid I get condensation which drips on the floor. I have to make up a little device to catch the drips.

Both primary and secondary pumps are still doing their 10 days 24/7 continuous running to get rid of the air. The temperature of the primary water is 14 degrees, so in the depths of winter the heat pump will be starting to raise the temperature from 14 degrees up to 35 which is the max set for the underfloor heating. I am hopeful of low energy costs.

2010-09-11: report 11

The 10 day continuous running of both circulating pumps has now ended. They were totally silent to my old ears. I thought I might switch it off for the first time and see what the 'CurrentCost' consumption monitor (not an accurate measurement) showed.

Power disconnected from Nibe unit by external circuit breaker - 0 watts. At least it is accurate at zero. Power to Nibe, but Nibe switched off by front panel switch - 14 watts Nibe unit switched on, but not operating - 110 reducing to 66 watts

So in the non heating season, switch it off at the circuit breaker.

Meter and Leak

2010-10-23: report 12

On Wednesday 20 October I installed a [kWh] meter and put the heat pump on. It was the first time it had been used to heat the house. It worked fine and brought the house up to 19 as set on the thermostat. I spent some time trying to understand all the different parameters that could be adjusted. I have a lot more to learn.

Everything was fine until Friday, when I noticed that the pressure in the primary (bore hole) circuit had dropped to about 0.2 bar. Previously it had been solid at about 1.5 bar. I opened the inspection cover where all the bore hole pipes are joined to a manifold. If you look back, you will see the picture before the cover was constructed. There were two small pools of red liquid - the manifold was leaking. I told the installer and he said that it was probably caused by the low temperature on the return side. When the heat pump is running the supply side is 9 to 11 degrees, and the return can be minus 2 degrees. It makes me concerned about the leak security of the manifold. In normal operation the temperature will vary from about +11 to - 2 degrees, and if this change is sufficient to provoke a leak, then is this going to be a continual problem?

I have seen a special manifold designed for this application. It has plastic pipe parts with electric resistance heaters which are used to melt the female socket to the inserted pipe. It would seem to give a much more secure connection. With 20:20 hindsight perhaps I should have insisted on one of these. Any one got information to add? It would be very interesting to hear about your experience.

This Saturday morning the heat pump had shut down by itself. The alarm help screen said that insufficient primary circulation was available to work the heat pump. Last night I could hear slight noise of air in the system. I await the arrival of Dominique on Monday to fix the problem.

2010-10-25: report 13

Yes it really is unlucky 13. Dominique came today and re-pressurised the primary circuit. It was hard to get the pressure above 1.3 bar. The frequent drips coming from one of the manifolds changed to a small stream. Apparently it is a defect in the brass casting of a 'T' fitting. The frequent temperature changes, +11 to -2 degrees every time the heat pump cycles on and off, causes repeated expansion and contraction and has moved the defect in the casting enough to create a pin hole. Repair coming soon.

2010-10-27: report 14

cracked union nut

The leak was caused by a brass adapter. It has an axial crack about 2/3 of its length along the larger female threads. I have it as a souvenir. Dominique says it is a very rare occurrence, but he has seen it before.

The heat pump is back on and working fine. Two items remain -

  1. I want to get a better understanding of all the parameters that can be adjusted and tweak the system to its ideal settings.
  2. I am concerned that the pump is short cycling. I want to increase the hysteresis so that it runs for longer periods at less frequent intervals.

2010-11-02: report 15

The short cycling was 5 to 10 minutes running only. This was reducing the life of the compressor so I turned it off. I contacted Nibe UK who were brilliantly helpful.

Apparently I should have a return water temperature sensor about half way up the buffer cylinder. The 200 litre buffer cylinder acts an additional load, if I just had the relatively small amount of water in the underfloor pipes, the heat pump would rapidly reach the set temperature, causing short cycling.

As a temporary fix I strapped a temperature sensor on the return pipe going the heat pump. It works fine and now the heat pump runs for about 30 minutes before shutting down. I await Nibe UK to say if it can be left as a permanent solution, or if I need to fit a tube to the buffer cylinder to take the sensor.

When the pump has been running for some time, the water going into the under floor pipes is around 40 degrees. I think this is too high. My next task is to find out how to reduce this.

2011-01-11: report 16

I had to return to the UK in October 2010 for a family emergency. I am still in the UK, but will return to the heat pump installation in France at the end of January. While I have been away, it has been holding the house at 16 degrees as I had set on the thermostat with outside temperatures as low as minus 5 degrees. I don't know if it has been short cycling - I hope not - I will find out.

On the question of how to connect the buffer cylinder, I have been struggling to get an answer from Nibe. At first I spoke to Nibe UK but could not get a clear response, so I was referred to Nibe Sweden. Finally today I made contact with someone in Sweden who really knows, and I am much clearer about what is needed.

Series connection is OK if the secondary load is fixed as mine is. The reason for parallel connection would be if there were individual circuits which had control valves that might shut them down and set the flow to near zero, or zero. Clearly this would be bad for the heat pump. In a parallel connection a second circulator would be needed. There is a circulator inside the heat pump which would pump into the buffer cylinder, the second circulator takes water from the buffer cylinder to the heat loads.

The most important point I have learned is that the temperature differential on the secondary should be 5 to 10 degrees, with max of 12. When I get back, I will see what differential I have, and take steps to get it to the ideal.

It is possible to connect an external sensor to measure water temperature. At present it is on the return to the heat pump. The technical expert said it could go on the supply side, and that I could experiment by first disconnecting it, and then trying it in different places. I will do this.

2011-02-13: report 17

After an absence of 3 months we have returned to France. During our absence the heat pump has been holding the house at 16 degrees, and I don't think it has been short cycling. I had hoped the temperature would be more like 12, but as I had to leave in a hurry, I had no time to check it. During our absence, EDF have had lots of red and white days (expensive units) but I console myself with the lower house temperature using less electricity.

I have been studying the installation and discovered two problems. First the temperature difference (supply temperature minus return temperature) on the underfloor is too high. It runs at about 12 to 13 degrees. For a concrete floor it should be 5 degrees (7 degrees for wood). The way to solve this is to increase the flow. The secondary circulator in the heat pump is set to max, so an additional pump is needed.

This high temperature difference causes another problem - the warm water output from the heat pump can get too hot for underfloor if the outside temperature is cold the heat pump has to run for a long time. I have measured 47 degrees - the ideal is 30 to 35. I am concerned that my very old underfloor pipework, and floor concrete could be damaged.

The second problem is the connection of the buffer cylinder and the and the underfloor circuit. My underfloor circuit has no control valves that open and close, the flow is fixed. So the cylinder and floor are connected in series with the floor coming first then the cylinder, and finally returning to the heat pump.

Nibe (heat pump supplier) say the cylinder should come first. Nibe use the supply temperature as the control to the heat pump, most other manufacturers use the return.

So on St Valentine's day the pipework is to be modified and a pump added. Then on Wednesday an expert from Atlantic (heat pump distributor) will come and set up all the parameters. I will have some questions for him as you can imagine.

If anyone is interested, I think I now understand the reasons for parallel and series connection of the buffer cylinder. Also, I think I understand the degree-minute system that Nibe use to control the start/stop of the compressor.

2011-02-16: report 18

This morning, the man from Atlantic arrived with my installer to carry out a commissioning process. I decided to pay for this (EUR350 more), because it increased the guarantee on the heat pump from 1 year to 3 years, with the manpower guaranteed for 1 year, plus I would be confident that it was installed and adjusted correctly.

The Swedish Nibe heat pump (distributed by Atlantic in France) has a set up procedure which gets all the values set up according to the individual environment. For example there is a parameter which can be changed to reflect how good the house insulation is. I thought that my house, renovated 25 years ago, was well insulated---apparently it is not that good compared to modern materials.

Everything was found to be satisfactory except for the temperature difference on the secondary heating circuit which supplies the underfloor pipes. The delta-T is 11 to 12 degrees, it should be 7 maximum. There are a number of solutions:

  1. Change the circulating pump inside the heat pump to a more powerful version. I don't yet know if this is possible. In my view this would be the ideal solution.
  2. There is already a second pump in the circuit which was added yesterday to boost the circulation, but it is less powerful that the pump inside the heat pump. It is best if they both have the same performance. However the result may still not be enough, but it would be relatively easy to do.
  3. Change the connection to the buffer cylinder so that it is in parallel to the heat pump, with a more powerful circulator to pump from the cylinder to the underfloor. I think this would be certain to solve the delta-T problem, but it has probably has the most work to achieve. Below I have supplied an explanation of series and parallel connection of the buffer cylinder.

This excessive delta-T is not harmful; it causes less than optimum performance, reducing the efficiency or COP of the heat pump.

The man from Atlantic is responsible for advising the best solution. I am keen to hear what he says.

Series/parallel explanation: geek alert!

OK. But a warning first - I am a retired electronics engineer, not a heating engineer, but I am fairly confident of my facts. It will be interesting to see if people disagree. I will do the buffer cylinder first, then the Nibe degree minutes separately.

The buffer cylinder (BC) is used to give a sufficient quantity of water as a heat load for the heat pump (HP). My heat pump needs a minimum of 300 litres of water. The BC has 200 litres and the underfloor about 160, giving a total of 360.

The BC can be connected in series or parallel with the HP. When in series, the warm water from the HP flows through the BC and underfloor or radiators sequentially. When connected in parallel the warm water from the HP flows through the BC and back to the HP. There is another circuit from the top of the BC through the underfloor/radiators and back to the bottom of the BC. This needs an additional circulating pump.

In series connection the flow from the HP can be through the BC first or the heat load first. There is a lot of debate about which is best. Nibe is clear in drawings that BC is first then the heat load. Nibe controls the start/stop of the HP based on the supply temperature, I believe that most other manufacturers use the return temperature. If the heat load is underfloor this stops the temperature getting too high. If you have radiators, then perhaps they should be first as a higher temperature is desirable.

If you have a heating load which varies, for example motorised valves which open and close then a parallel connection is needed. If it were to be connected in series and the heating load was suddenly shut down the heat pump could be damaged.

In a parallel connection the BC can have a '3 pipe' or a '4 pipe' connection. The warmer top connections are always separate - one to the HP and the other to the heating load. In the '4 pipe' system the return side on the bottom of the BC are separate, in the '3 pipe' the returns are joined with a T connection to a single point on the bottom of the BC.

Nibe Degree Minutes: more geekery!

Same health warning---I am not a heating engineer. It might be helpful to take a piece of paper and draw the 'heating curve' and the 'degree minutes' formula described below to help understanding. My apologies if this is too long, but I am trying to be clear.

Nibe---the Swedish manufacturer of my heat pump uses a number of parameters to control the compressor. Supply and return water temperature is measured with sensors. External and internal temperatures are measured in the same way.

There is a graph of supply water temperature on the vertical axis and external temperature getting colder on the horizontal axis. This is the heating curve and represents the heat needed to keep the house at the correct temperature as the outside ambient varies. Looking at the curve, as the outside temperature falls, the supply temperature rises. The slope and offset of the heating curve can be varied; it can also have fixed upper and lower limits. With this flexibility, the heating curve can be adjusted to suit the heating needed by the house.

Degree minutes are calculated every minute by the control circuit. The formula is actual flow water temperature minus the calculated flow water temperature taken from the heating curve. As each minute passes, the new value is added to the previous. Suppose the actual flow temperature is 25 degrees and the calculated temperature is 30 degrees. At the start, the degree minute will be -5. Suppose one minute passes and things remain the same, a new -5 is added to the existing, and we get -10. Things continue in this way until the degree minute value arrives at a preset point. Let's say the preset is -60. As you will have guessed this preset can be changed too.

At -60 the compressor starts, and the flow water temperature starts to rise. When it gets to 29, a degree minute is -1. When it gets to 35, a degree minute is +5. These positive degree minutes eventually reduce the negative value to zero. At zero, the compressor stops.

While the flow temperature minus the calculated temperature remains positive, the degree minutes become a larger and larger positive value. On my machine when it reaches +100 it does not get any bigger. I don't know why, maybe it is a Nibe fixed value. However as the flow temperature falls, because the compressor is now off, it eventually gets below the calculated value and now the degree minutes become negative. The +100 reduces and eventually gets to -60 which is the preset compressor start value and the cycle repeats.

If you change the degree minute setting from -60 to say -100, then the hysteresis increases, that is the temperature has to fall further for the pump to start, and the temperature rises higher before it shuts off. It is not necessarily best to have low hysteresis because the pump will cycle on and off more frequently and compressor life will be reduced. Nibe say 20 minutes on, and 20 minutes off is ideal. As in life, the ideal is hard to obtain, as the on and off time depend on the external temperature and the sizing of the heat pump with respect to the heat load.

2011-02-23: report 19

Dangerous to put this in print, but I think it is complete. There are a few odds and ends to sort out, but it is practically done. Whew, it was started in August last year.

The buffer cylinder is now connected in parallel with a three pipe system, and a more powerful circulator added. Nibe Sweden advised 3 pipe in place of 4 pipe as it would give the best isolation from the fairly powerful underfloor pump from the less powerful circulating pump inside the heat pump unit.

I have an old gas boiler which I was using on EDF red days - red days are very expensive. I was advised to check the costs, and I found that the heat pump was cheaper (just a bit) than the gas using last years prices. I have to do an update on current prices, but I expect the same result. Amazing eh? I think so.

I am recording consumption data, and learning about the logging that the heat pump provides. If you put the logging data on a spreadsheet you can plot graphs of temperature and similar things.


2011-03-19: report 20

I got back to France yesterday. House temperature was 14 degrees. I had set the thermostat to 12; I need to find out what parameter to change to get it right. With the low temperature setting and the warmish days, the heat pump has not been running much. If my calculations are right, the average run time was 14 minutes, and the average starts, 39 per week. I don't know if this is good, bad, or unimportant. What do you think?

After having watched a compressor cycle at the unchanged thermostat setting of 12 degrees, (it ran for 11 minutes), I increased the thermostat to 19 degrees. Due to the sudden large demand, the degree minutes became a very large negative number (measured return temperature minus calculated return temperature from the heating curve). I had set -100 for pump start. It started, and as the difference was so large, the number got more and more negative while the heat pump output warmed up. When it got to -400 the auxiliary resistance heating cut in and the temperature rise got faster. Eventually the underfloor temperature got too high in my view, and I switched off by lowering the thermostat. I subsequently found that the resistance heating was set to come on at -400; I have left it there for the moment.

There is a setting for max flow temperature and it was at 48 degrees, much too high in my view. I have lowered it to 32, but I still get high temperatures. It does not control the pump directly as you might expect. I need to study the operation to get it right.

As the house is now close to 19 degrees inside, the temperatures are lower at pump switch off. I now see the reason for resistance heating. If you have a sudden demand for a big temperature difference, adding on peak electric heat will get there faster, but will cost more of course.

2011-04-09: report 21

Now that it is much warmer, the heat pump runs very occasionally, and I have discovered a new problem. Due to the high ambient, say 13 degrees, the heating curve will calculate a low temperature for the heating flow. This may be just a little above the actual temperature in the buffer cylinder. The result is a short run time for the compressor. Not good.

I have found these ways of increasing the run time.

  1. Increase the degree minute setting at which the compressor starts. During the winter it was minus 60, I have now set it to minus 120. This means that the temperature in the buffer cylinder will have to get colder before the compressor starts, so it will have to run longer before switch off.
  2. Increase the offset allowed on the max flow temperature. Previously I had 31 degrees max on the heating curve and an off set of +10. This gives a max temperature of 41 degrees at which point the compressor stops regardless of degree minutes. I changed the heating curve to a max of 29 and the offset to +12. This still gives a max of 41 degrees, but allows the temperature to get higher by 2 degrees than the heating curve calculation or the max of 29 degrees.

This seems to work. I now get a run time of about 20 minutes, I got 10 or less beforehand. I will have to wait until next heating season, to see the effect of this when it gets seriously cold outside. Always a compromise as we all know.

I am still recording the consumption, but I am not sure how best to use the numbers. Next week I will switch off the system at the main circuit breaker.

2011-06-15: report 22

Just got back to the house after a period away. The pressure on the secondary underfloor heating circuit was unchanged. Good.

The pressure on the borehole circuits was zero. Bad. This is the second time there has been a problem with this circuit. Previously the problem was a cracked brass pipe fitting in the manifold where 8 pipes from the two boreholes (4 pipes per borehole) are joined to the large diameter supply and return pipes to the heat pump. Dominique the installer said it was a defective part. Everyone else who has looked it it says it was overtightened. The high density black polythene pipes are very stiff and hard to manipulate. They are joined to the brass manifold with a threaded fitting which holds the pipe and compresses a gland. I understand this is an old technology used for the cold water supply into the house.

It is possible to obtain plastic fittings which are melted to the pipe with a heating element enclosed in the fitting. As I understand it, the fitting has a bar code which is scanned, then a special piece of equipment passes the right current for the right period of time to form a good melted joint. This appears to be a better method of obtaining a secure joint. However if a problem does occur the repair is much more expensive.

I have assumed that the leak is in this manifold. I could be wrong. It might be underground. I really hope not.

2011-06-23: report 23

Dominique came this morning to find out why I had zero pressure in the borehole circuits. He could not find any leaks. He re-pressurised the system to 1.7 bar and asked me to monitor it for any pressure drop.

The explanation is that there was a pocket of air trapped in the system which had been there since installation, or had formed with air coming out of the water/antifreeze mixture. Eventually it moved to one of the automatic air vents and escaped. This caused the pressure drop.

Let's hope this is the end of all the interventions and the future will just be using and tweaking.


No the ground loops were not pressure tested. You have pointed out another lessons-learned item. With hindsight they should have been. The borehole company has a 10 year insurance with Axa so I may be OK from a cost of repair point of view, but I will have to suffer the upheaval if the leak is underground.

The 8 pipes from the 4 borehole loops go into a manifold made from brass fittings (not welded - another lessons learned perhaps). One of these leaked previously but after repair, and for the rest of the winter heating season the pressure has been constant. It only dropped after the heat pump had been switched off - very strange.

Yesterday and today is much hotter and the pressure which had dropped to 0.5 bar is now 0.6 bar. There are lots of things above ground which could be causing the loss of pressure.

I think a pressure gauge would be a good idea, should not cost much. Get one with a pointer that you can set to the actual reading - much easier to spot a drop. You should also have a filter on the flow into the heat exchanger.

When I had the first leak the heat-pump shutdown automatically as it could not get enough heat from the primary circuit.

Third Leak

2011-08-30: report 24

I am now on my third leak in the primary circuit. The first was some time ago, during the heating season, and was caused by a broken pipe fitting in the manifold connecting the black high density polythene pipes from the bore holes to the main flow and return to the heat pump. The broken fitting was caused by over tightening or defective brass in the fitting. The installer claimed the latter, everyone else said it was the former!

The second was during the summer - there was a sudden loss of pressure. It was thought to be an air bubble because when the pressure was restored and the circulation pump run for 10 days, the pressure stayed up.

About 8 weeks later, I noticed that the pressure was zero again. This time I could not get more than 1.0 bar because I ran out of the pink antifreeze. When I looked at the manifold, it was surrounded by the pink coloured water. I am now waiting for Dominique to find the leak and repair it.

The inspection chamber is about 60 centimetres deep and has a small cover. It is very difficult to work on the manifolds because the space is so restricted and the pipes are very stiff. Four smaller pipes are connected to one big one and one manifold lies on top of the other. I am going to have a much larger inspection pit made, since it seems to need frequent visits.

In hindsight I should have insisted on a plastic manifold with the pipes heat welded together. It might have been more expensive, but it would have been much more reliable. I am not sure about a pressure test, because the pressure has held up for weeks between failures.

If you are having pipes jointed underground, I would strongly recommend all plastic fittings with the pipes heat welded. It would be interesting to read about other's experiences.

2011-09-06: report 25

Dominique came today to investigate the third leak. It was found to be a brass fitting. What I would call a straight one inch coupling, male to female. The flow and return manifolds are made up with one of these on each side. If you look at picture 10, dated 5 October 2010 you will see it. It is on the right hand side next to the elbow. So now both the flow and return have had the same fault. You can clearly see the crack on the fitting. I hope to post a picture in a few days. Both fittings have had the same longitudinal crack. So I seem to be the unfortunate recipient of two defective machined brass castings.

Dominique has taken the fitting to his supplier to complain. He has borne the cost and time of making two repairs because of these defective parts.

The system now has the primary pump running 24/7 for a few days at 1.5 bar. Fortunately there is a builder on site doing various projects. Tomorrow he will construct a new inspection pit, much larger than the old one which will make any future leaks much easier to fix.

2011-11-24: report 26

I got back to the house a few days ago. The heat pump was off and the interior temperature was 14 ºC. I put the heat pump on, and in the usual 3 or 4 days, it climbed to the set figure of 19 ºC. If you have been reading this, you will know that I have had a lot of leaks in the primary. So far the pressure is holding at one bar - hooray. Lets hope the leaks are finally fixed.

I noticed that the temperature difference on the primary was higher than usual. I took out the filter on the primary circuit that stops the crud going into the heat exchanger. It was 50% or more blocked. I cleaned it and the temperature difference dropped to the usual 5 ºC. I believe that this has made the heat pump more efficient as it can now abstract more heat from the ground.

If you have a system with a filter in the primary, it would be interesting to know how clean it is, and if after cleaning, the temperature difference is less.

2012-02-06: report 27

Its all obvious and predictable now, but I was surprised. When the house was first refurbished 25 years ago, it had a 9 kW resistance heater for the under floor heat. I found that it could just maintain 19°C inside with 0°C outside. As my heat pump was 15 kW, I calculated it could get 19°C inside with -10°C outside. I am going to find this is wrong.

First it is not possible to get 15 kW continuously. The pump is an on/off device and continuously raises the temperature until is is shut off by the heating curve calculation, or a temperature limit. So I am getting less than 15 kW. I have tweaked the settings to get it to stay on as long as possible and be off for the least time. The external temperature has varied from -1 to -5°C and the inside is gradually reaching 19°C. I have recently returned to France and the internal temperature was 14°C. With the huge thermal mass of concrete and the low ambient it takes a long time to heat up.

Second I am now not sure if the underfloor system is able to transfer enough heat into the building when it is very cold outside. The water temperature in the underfloor pipes varies from 32 to 42°C. If I could keep it at a constant 42°C, I would get a better heat transfer.

EDF the electricity supplier are having a bonanza period. With the cold weather they are making every day allowable a 'red day' with the very high tariff. Grrrr.

Bores OK

2012-02-05: report 28

In the early part of this story, there were questions about the bore holes Would they provide enough heat to the heat pump. Well, for that bit at least I think it is OK, just as Paul in Montreal had said it would be. I have 2 x 100 meter boreholes with two loops in each borehole, this feeds a 15 kW heat pump.

Following continual use over one week, at worst the supply is +7.1°C and the return +2.6°C. I believe Nibe's specification for COP measurement requires a primary supply of 0°C and an output of 35°C.

2014-02-15: report 29

The heat pump had been working without problems until 5 Jan 2014. I was in the UK and a friend who checked the house regularly told me that it had stopped with a temperature alarm. I contacted the installer, and after a few days he was able to investigate. He found the return temperature from the primary circulator to be minus 8 degrees; In my experience, it had never been lower than plus 2 degrees - clearly a wrong value. This circuit pumps the glycol around the closed loop that takes the heat from the two 100 meter deep boreholes. This value caused the system to shut down. He thought (wrongly) that the sensor was defective and ordered a new one. After 11 days it finally arrived.

The result was the same. He found the circulating pump for the primary had stopped working, hence the minus 8 degrees. The pump is a specially made product by Grundfos for Nibe the heat pump manufacturer. A new one off the shelf, can only be purchased from Atlantic, the agent for Nibe in France. Its retail cost is EUR1,572. The pump caused a 4.5 amp fuse to blow and gave off that tell-tale smell of burning electronics.

When the system was installed I paid EUR466 to have it set up by an 'expert'. I was not impressed by his 'expertise' but I was also buying 3 years insurance, so fortunately, I will not have to pay the huge cost of the new pump.

Finally on 13 February (39 days after the fault was first reported) the new pump arrived. It was smaller than the old one and is marked efficiency 'A' like a fridge. Apparently there have been three different designs of this pump over the years. This must be in my view, because of reliability issues. The pump has a permanent magnet and I assume it is a DC device. The control electronics is in a plastic container mounted on the side of the pump. Its speed can be varied remotely. I have set it to 85% of full speed, this gives a supply/return temperature difference of 5 to 6 degrees.

I expect to be presented with a bill for manpower. If/when it arrives, I will ask for a contribution from Atlantic on the grounds of the product having an inherent design defect. As I am in France, I will probably lose.

Let me know if anyone is reading this and wants to hear the end of the story. OK thanks for your interest. Things happen slowly here, when something happens, I will let you know.

2014-02-21: report 30

The bill arrived in the post today. 40.00 euros per hour, 7 hours work, plus 20% TVA or VAT. Total 336.00 Euros. I suppose I should be grateful, a heating technician in the UK, would probably be twice the price.

I have decided to pay up and keep quiet. The only good thing is that the invoice includes a detailed list of the work done.

I wanted to keep the old pump to dismantle it, but it had to go back under the terms of the insurance.

Four Leaks

2015-01-14: report 31

In total I believe I have had 4 leaks from brass fittings. This could have been due to defective fittings or plumber over tightening. I suspect the latter. With the constantly fluctuating temperatures, any weakness will be assisted to fail.

The most recent a year ago was in the primary. The antifreeze mix leaked out, and the heat pump shut down due to poor circulation. The liquid escape was not excessive but it made the chipboard flooring become like soggy cardboard. Finally this spring time I am going to disconnect the heat pump, move it, fix the soggy floor, and re-install it; rather a plumber and a carpenter are going to do the work.

Apart from this, it continues to work well. I find I have to adjust the settings when the ambient is mild and adjust again when it gets cold. I do this to achieve at least 18 minutes of 'on' time for the compressor. I have not found compromise settings that would give 18 minutes minimum in both mild and cold conditions. If anyone has suggestions on this I would be pleased to hear them. Nibe heat pumps have a huge variety of settings - probably too many. For example the heating curve slope can be varied, max and min can be set, overshoot set, a kink can be put in the slope, and more. . . .

2015-05-29: report 32

electricity meter 3 phase and heat meter
Two meters. The small one on the left measures three phase input to the heat pump in kWh. The one on the right measures heat output in MWh.

In previous reports I have complained about 3 or 4 leaks caused by brass fittings cracking and leaking primary or secondary fluid. I believe it was caused by over-enthuistic tightening, the installer says, as you might expect, that the fittings were defective. The most recent leak in the primary caused the pump to stop because of low flow, but more significantly the liquid damaged the chipboard flooring and the heat pump began to list to one side.

I had the heat pump and inertial cylinder disconnected, the floor repaired with a better quality product, and the heat pump and cylinder re-installed. One extra bit of equipment was added - a heat meter. It has an ultrasonic device to measure flow, and two sensors that measure flow and return temperature. I gave the system a quick test and everything works fine and I can see that it had produced one or two kW of heat output.

In the winter I will note the readings and hopefully be able to measure the COP of the heat pump. I also have a data logger, and hope to use that to give a graphical display of heat over time.

2015-10-06: report 33

As it was a bit chilly, I put the heat pump on for two days. It consumed 29.7 kWh of electricity. It gave 117 kWh of heat output as measured with my new heat meter. This gives a COP or efficiency of 3.94. Or I could say, I bought 29.7 kWh from EDF. I got 87.3 kWh free from the ground.

2016-10-23: report 34

This is probably the last report []. Some facts may be of interest...

  • Installed and running for 6 years
  • Some leaking pipes caused by installer over-tightening joints
  • One failed primary circulating pump replaced under extended warranty
  • Compressor 9429 starts
  • Compressor hours 2946
  • 18.7 minutes average run time -- too short in my view
  • Using input electrical power versus heat output COP = 3.86

2018-11-22: report 35

Just in case anyone is still there...

It is now 7 years since my GSHP was installed.

In 2015 I added a heat output measuring device. It measures the flow rate of the water by ultrasound and taking the supply and return temperatures calculates the heat output produced.

I measure the energy input with a standard kWh meter. I divide the output by the input to get the CoP (coefficient of performance).

Here are recent readings with their dates...

  • - 3.74
  • - 3.74 [date was given as 20/4/19 but likely a typo]
  • . - 4.23
  • . - 4.27

I am constantly tweaking the parameters to get the compressor run time to be as long as possible. With the mild weather this is difficult.


2021-04-19: bonus and final update

GSHP 15kW and inertia cylinder
General view of 15 kW heat pump with inertia cylinder to the right.

Topher said to me in an exchange of emails today:

The Nibe GSHP has been keeping the house at 15 degrees while we have been locked down. I ... have just [re]calculated the [] COP. It is 4.35 and I am confident it is a true figure. The run time average has been 16 minutes. Hard to make it better with such a light load.

Here is a longer report that he has posted:

It is the end of an era. I am selling the French house. A lucky Frenchman will have the benefit of the Nibe GSHP. A few final comments.

  • It has been installed for 10 years and is still giving a COP of 4.35 with average run time of 16 minutes.
  • Only one problem with the Nibe equipment - the high power primary circulating pump failed during the warranty period and was replaced FOC.
  • All the other problems have been to do with plumbing, an enthusiastic young apprentice who overtightened union nuts that later cracked, expansion vessels that need a charge of nitrogen, air bleed fittings that need replacement. The latest repairs will cost me about EUR160.
  • Some time ago I installed an output measuring device. Secondary water flow measured with ultrasound, supply and return water temperature measured with thermistors, a display unit measuring MWh. The incoming 3 phase power is measured with a kWh meter. Values from these meters produce the COP given above.

In all, I am really pleased with the results over the 10 years use.


2021-09-16: follow-up question

On Twitter, @TrystanLea asked:

Is the COP of 4.35 since they installed the heat meter in 2015? I'm not sure if I'm reading that correctly?

To which Topher answered:

Oct 2010 GSHP Installed, followed by a number of months of piping and sensor modifications. See above for details.

Nov 2015 put in system to give measure of heat output, details above in blog. I estimate the accuracy of the measurements to be not worse than 1%. I assume the accuracy of the kWh heat input meter to be 1% too.

COP measured from Nov 2015 to April 2021 when house was sold.
Worst COP 3.43
Best 4.55
Average 3.99

I reckon that is pretty good. For every 1 kWh of heat I bought from EDF, I got 4 kWh of heat into the house.

Mitsubishi made the innards of the heat pump. Nibe (Swedish) made the control electronics, sensors, valves, and box. Atlantic (French) put their badge on it and a local plumber (French) installed it. And I paid for it.

~10521 words.