Heat pump controls, we need better

These blogs are usually triggered when I come across issues (several times) that bother me… here goes.

Theory and practice show that the best performance for any (1) heat pump is achieved if the average water temperature that the heat pump ‘sees’ (the water temperature passing through it) is kept low (2), whilst still achieving the required level of comfort.

It is all very well installing underfloor heating or radiators that can give high heat-outputs with low circulating water temperatures.   However, the actual temperatures may deviate significantly from the theoretical, and this is highly dependent on the control system provided.  It is often assumed (by both installers and users) that the controls are capable of doing what you need… not always the case.

Whilst the SCOP rating and the predicted flow/return temperatures would seem to have the greatest bearing on energy-efficiency, it really hangs in how well it is controlled.

I am hopeful that over this decade we will see far better control systems. It strikes me that we are in a ‘middle’ phase with some claims of ‘intelligent’ control, but actually its not always helpful in practice. Furthermore, many programmers and controls are still user-unfriendly. No wonder that many are not programmed as they should be.

Whilst things have got better, I still come across installations that are clearly not ideal. E.g. inadequate controls, or sometimes too many and too complicated.  It is often very hard to estimate how worthwhile (cost-effective) any modifications would be. This would ideally involve some performance monitoring over time. So, annoyingly, many systems remain in a non-ideal state.

I don’t want to worry people unnecessarily here.  Most installers know what they are doing!  Many systems are well optimised, and do just what they say on the tin, but people still need to be mindful that it’s not always the case.

The below are some typical general examples, but there are many other offerings and options out there.

Some typical control methods

When things are simple, there are fewer potential pitfalls

Here we have an as-simple-as-it-gets system.  In this case the room control is the ‘target temperature’ or ‘auto adapt’ type. This modulates (variable inverter) the flow temperature according to room temperature so the actual water temperatures can adjust itself from anything below 30°C to above 40°C depending on the demand.  The average water temperature over the year can be surprisingly low, giving low running costs.   In this case, the Room control is supplied with the heat pump.

This is a good system for open-plan houses, and extremely well-insulated houses, but when we get into larger houses with different rooms used at different times, and radiators upstairs with underfloor downstairs, we can start running into trouble…. To be clear… It will heat adequately, but the running costs could become higher than they should.


The next level of complexity may involve a buffer cylinder.  This is simply a tank of water. It might store something in the order of up to 20 minutes worth of heat.   It is generally needed in buildings where there are many zones (many rooms), and where you may at times want only one zone on.
The simple principle is this –
* The heat pump heats the buffer cylinder to a certain temperature (variable over the year).
* Heat is pumped from this cylinder to rooms as required.

Unlike the first example where the controller ‘learns’ the lowest (and most efficient) water temperature possible, in the buffer tank case, a pre-determined temperature has to be set, and this usually uses a ‘weather compensation curve’ (put simply 50°C when very cold out, 30°C when mild, and a sliding scale between the two).
The ‘curve’ setting is usually found by trial and error, but as I am all too often finding, this is often set higher than it actually needs to be. E.g. in my example it is heated to 50°C.  Water at the radiator may (due to mixing) be at 45°C, and the thermostat in the room will ‘cycle’ the pump with on and off periods.   The radiator at a constant 35°C might suffice.

Compared to our first example, the heat pump ‘sees’ water at a considerably higher temperature, so the average COP is worse.

Ideally, the set point of the buffer would not only depend on the outside temperature, but it could reduce for certain periods of the day, and also reduce depending on the house temperature.  Many manufacturers have a facility for this, but I very rarely see it being used well.   Some controls allow two curve settings over the 24hr period.  Most don’t.

Underfloor and radiators on the same system

A next level of complexity is where mixing valves are used.  This is typical when both underfloor and radiators are used on one heat pump.

Here we can see an underfloor circuit that only needs water at the lukewarm temperature of 35°C.  Whilst our heat pump would heat this very efficiently, unfortunately the system takes all water up to 50° (less efficiently), only to mix some of it down to 35°C.   If our control was ‘cleverer’, it could drop the buffer set temperature for certain periods of time when only the floor needs heating. However, few currently available controls will cater for this.

These examples are only a few of many options and combinations. Of course, the manufacturers have thought this all through, however, it would be impractical to make controls for every circumstance.. it could be too complicated, so they tend to evolve a ‘system’ and hope it will suit most applications.

Intermittent and continuous heating

For houses that are unoccupied during the day, or offices that are vacated evening and night, there is no point keeping them warm all the time.  That said, if the heating is completely off for too many hours, then the required catch-up (heat-up) may be difficult for our heat pump.   In practice, a compromise is usually sought, with ‘occupied’ and ‘unoccupied’ temperatures.  This is of course dependent on the building, be it insulated or not, heavy stone or lightweight etc.

For constantly-heated homes, heat pump installations can be simpler.  But for intermittent heating, some extra complexity and extra capacity may be needed.

That said, for most of the year, there is spare capacity, so it may be perfectly acceptable in many situations to forego intermittency during the relatively brief coldest periods of the year. Again, good controls would be key to optimising this.     In the meantime, owners can set controls the best they can, but my feeling is that because programmers and controllers can be quite a chore to set correctly, many are not set well.

There seems still to be a lot to learn about the best way to operate heat pumps in different buildings with different situations, with respect to intermittency or constant.

A final gripe, .. for anyone trying to set a digital room thermostat to a frugal setting.  This could be in a workroom of maybe a big lofty room in an old house.  The ‘jump’ between settings of 16°, 17° or 18°C is great.  If energy-use is a minor issue, then you simply go 1° higher, but if you are keeping a tight rein on cost (e.g. a village hall), a finer control is desirable.    Manufacturers… please allow ½°C adjustment increments.  (It seems that 1°F increments are however about right). Gripe over.



  • (1)  The exception here could be a ‘transcritical’ heat pump using Refrigerant R744 CO2. These produce very hot water with relative efficiency, but don’t generally achieve the high COPs at low temperatures that other conventional types can.
  • (2)   How low can you go and still get an increase in COP? Whilst graphs show a trend of increased COP down to say 30°C, what happens below this?  I have done various ad-hoc test to the best I can, and found that some older fixed-speed designs tend to flatten off around 25-30°C.  But I have been surprised how efficient some inverter units work. This is could be to do with two things. i) better refrigerant control with an electronic expansion valve and ii) better motor power at very light loading.


Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.