Heat Pump Experiment: First Results

Friends, it’s been a couple of weeks now since I had the Low-Loss Header (LLH) removed from my heat pump installation (description here) and I thought you might like to know how the installation was performing.

In short, it’s performing very similarly to how it was performing with the LLH. The longer version follows:

Performance: Temperatures

Below are various graphs showing 3½ days of data from midnight on 15th October.

The first graph shows the internal temperature through the period. The set temperature was 20 °C and heating control was entirely by weather compensation i.e. the heating power is determined entirely by the outside temperature.

Click on graph for a larger version. The internal temperature of the house slowly approaching the set temperature of 20 °C. The heating is controlled entirely by weather compensation i.e. the heating power is set solely by the outside temperature.

The graph below shows how shows various system temperatures varied during the monitoring period.

Click on graph for a larger version. The graph shows how various system temperatures measured during the days following 15th October. The blue line is the outside temperature.  The orange line is the internal temperature also shown in detail in the graph above. The grey dots show the flow temperature in the radiators measured every 2 minutes, and the red line shows the same data averaged hourly. The purple dots show the temperature of the water used to heat the domestic hot water tank. The cycle in the early hours of the 18th is the weekly anti-legionella cycle.

Concentrating on the space-heating performance alone we can simplify the graph above by just using hourly averages and omitting the hot-water heating cycle.

Click on graph for a larger version. Simplified version of the previous graph. The blue line is the outside temperature.  The orange line is the internal temperature and the red line shows the flow temperature.

This allows us to see much more clearly the astonishing perfection of weather compensation: the flow temperature is increased (and decreased) in response to the decreases (and increases) in the external temperature: no internal thermostat is used. The setting to achieve this result was as follows:

• The ‘heating curve’ was set to 0.45. This number is a label for one of a set of curves which describe the heat loss of a dwelling as the external temperature falls.
• The ‘O/T threshold’ (O/T is short for outside) was set to 14 °C. This is the temperature above which the heating circuit will be inactivated.
• R/T modulation (R/T is short for room temperature) was set to ‘inactive’ which corresponds to just using the outside temperature to set the heat pump power.

There are a couple of interesting features of this data.

• Firstly, the outside temperature shown is captured by the monitoring system from a weather station a few miles from my home. In operation, the heat pump uses a temperature sensor place outside the house. The two sensors give similar results, but heat pump operation is based only on the local sensor. I think this is why there is a lag between the flow and outside temperatures which is particularly clear on the morning of the 16th.
• Secondly, since the LLH removal (on 7th October 2023) the main problem had been that the house had stayed too hot – typically 22 °C – and didn’t seem to cool. But finally, on the night of 16th October 2023, the outside temperature fell to 3 °C – passably cold – which helped the heating system to come into equilibrium.

So regarding the system performance, it is clear that the heat pump can keep the house warm when it gets cold outside. In some sense, this is all that matters. But how was the efficiency of the heat pump during these few days?

Performance: Heating and Electrical Power

Below are graphs showing the electrical power consumed and the thermal power delivered. The first graph shows data taken every two minutes, and it is clear that as the weather gets colder on the night of the 15th, the heat pump operates on longer and longer cycles. These increase in length from about 30 minute on-off cycles at midday on the 15th, to operating continuously once the outside temperature falls below roughly 7 °C.

Click on graph for a larger version. Top: Graph showing electrical power consumption (kW) and thermal output (kW) over the monitoring period. Data are taken every 2 minutes. Bottom: details of the cold night from midday on the 15th October to midday on the 16th.

We can see that in the early hours of the 16th October, the electrical power is around 600 W, and the heating power is around 1.8 kW corresponding to a COP of about 3. This is not bad, but not great either.

This view is interesting – showing the dynamics of the heat pump – and indicating that it needs a heating demand of approximately 1.8 kW – approximately 36% of its nominal output – in order to operate continuously.

We can get a more general idea of what is happening if we average the power measurements over an hour. This data is shown below.

Click on graph for a larger version: Graph showing electrical power consumption (kW) and thermal output (kW) over the monitoring period. This is the same data shown in the previous graph but now averaged over 1 hour.

Based on these averaged power measurements we can calculate the COP by dividing the thermal output data by the electrical consumption data. This is shown below.

Click on graph for a larger version: Graph showing Coefficient of Performance (COP) over the monitoring period obtained by dividing the thermal output data by the electrical consumption data. Also shown against the right-hand axis in blue is the external temperature (°C)

We see that the COP falls to around 2.5 to 3.0 during DHW or Anti-legionella cycles. When the heat pump is used for space-heating, the COP varies between ~3.0 when the outside temperature is 3 °C, and ~3.9 when the outside temperature is 12 °C. I don’t have exactly comparable data, but looking back at data from December 2022 , this data looks similar.

Click on graph for a larger version. Graph showing COP versus external temperature when the internal temperature is 20 °C. The blue dots show daily averages of COP (including DHW cycles) from December 2022. The green circles show rough estimates of COP (excluding DHW cycles) from October 2023.

Conclusion

The good news from this preliminary examination is that the installation does not appear to perform obviously worse than it did with the Low-Loss Header.

However the bad news is exactly the same: the installation does not appear to perform noticeable better than it did with the Low-Loss Header.

As you might expect, I will be gathering data as the winter progresses.