I was watching this video https://youtu.be/zPhlm8afr9o?si=K1_HLTAJZcRfV7Je&t=982 by Mars of Renewable Heating Hub the other day, where he is questioning a claim that it is possible to achieve a system efficiency / SCOP / SPF of 340% from a system that needs to run at 55C on the coldest days.
It got me thinking, what can we say about this from the data on HeatpumpMonitor?
1) Weighted average flow temperatures on the coldest day vs SPF:
At a high level we could look at the following chart that plots weighted average flow temperature on the coldest day vs SPF: https://heatpumpmonitor.org/?chart=1&filter=query:cooling_heat_kwh:1:lte:t,hp_type:Air:eq:t,boundary:H4:eq:t&period=custom&selected_xaxis=measured_mean_flow_temp_coldest_day
First we can see we don’t actually have any systems that run at a weighted average flow temperature of 55C on the coldest day (weighted average meaning that most heat is delivered at that flow temperature), the highest we actually have is system 444 in London with a weighted average flow temperature of 49C and a SPF of 2.4. System 170 in Devon meanwhile has a weighted average flow temp of 48C and achieves an SPF of 3.8, highlighting how wide the range of performance can be!
The central line of best fit gives the following equation:
SPF = -0.0688 x Weighted average flow temp + 6.4303 +- 0.61
We would therefore expect the average system that actually delivers most of it’s heat at 55C in practice on the coldest days to achieve:
-0.0688 × 55 + 6.4303 = 2.6 +- 0.6
The higher end 90% PI performance systems at 55C might achieve around SPF 3.2 and the lower end 2.0.
2) Simulated results using our Dynamic Heat Pump Simulator
I’ve been slowly improving our dynamic heat pump simulator over the last year, the latest versions includes the ability to simulate space heating performance over a full year and interpolated COP model based on the detailed Vaillant datasheets. Here’s an example run with a 5 kW vaillant continuously heating a house to 19.5C, running at near 55C on the coldest day achieving an SPF of 3.34 over the year (space heating only). This is without modelling defrosts and so I would expect perhaps a further 0.1-0.2 reduction in COP in reality which could bring us to a result that might be quite close to the top 90% PI of the HeatpumpMonitor data above. Note that this is space heating only, I would again expect a reduction in the combined SPF once DHW demand is included - unless of course a well optimised heat pump cylinder is added to a high temperature emitter system - perhaps not that unlikely?
The Ecodan COP model suggests 3.44 is possible, the simple carnot with variable offsets 3.53. Interestingly if you double the heat loss, switch to a 12 kW Vaillant COP model with it’s higher datasheet performance and adjust the emitters accordingly to again target 55C the simulated performance rises to 4.2 on space heating! The on paper performance for the 12 kW Vaillant really is a class above!
3) Steady state design flow temperature vs actual required flow temperatures
Playing around with a number of simulation runs I noticed something that makes a lot of sense conceptually but is really quite interesting to quantify!
-
First: steady state heat loss calculations and design flow temperature calculations do not take into account internal gains such as body heat and heat given off by electrical appliances in a house. Internal gains in a typical house may be around 300-400W. For a house with a real world heat loss of 3.4 kW and a low temperature radiator system with a design flow temperature of 40C, this makes the difference between needing to run at 40C without gains and only needing 38.3C with 330W of gains included.. For a high temperature system with a design temperature of 50C including 330W of gains drops the required flow temperature 47.4C. The impact of gains increases as the design flow temperature increases, in our example here the difference increases from 1.7K @ 40C to 2.6K @ 50C.
Modelling a system that requires 55C not including gains results in 52C after including 330W of gains (same 3.4 kW heat loss above).
-
Second: Taking this same 55C without gains 52C with gains steady state system and simulating it in the dynamic model we only see sustained periods at ~49.8C (max 50.5C). The design temperature of ~-1.1C (calculated based on the 99.6% percentile of the outside temperature dataset used for the simulation - important!) was not quite experienced for long enough, the warmer outside temperatures either side results in average heat demands that are lower than what we would experience if we had multiple days in a row at design temperature.
The result of this example is that a system with a design temperature of 55C on paper based on an accurate heat loss calculation and steady state assumptions may well result in significantly lower flow temperatures in practice due to: internal gains, short duration of design temperatures and also higher 99.6% actual experienced design temperatures than assumed. The performance of this simulated steady state 55C system, ~50C when dynamically simulated results in a space heating SPF of 3.62. If this simulation is realistic we could feasibly imagine an SPF of 3.4 being achieved once defrosts and DHW demand are included - assuming a high performance heat pump cylinder. Allocating 2x 1 hour DHW periods from which the space heating needs to recover by running at slightly warmer flow temperatures only has a marginal effect reducing SPF from 3.62 to 3.59.
Conclusions
As ever it’s a bit of a complex picture.
-
It’s probably the case that if we monitored 100s of systems that actually needed to run at 55C all day, on the coldest days, we would find average SPF’s that center around 2.6±0.6 (90% PI)
-
It’s likely that the better designed, installed and optimised systems, especially those with heat pump hot water tanks and optimised heat up DHW schedules might come in around SPF 3.0-3.2.
-
The 12 kW vaillant simulation suggests higher SPF’s could be possible for larger systems perhaps these could even hit SPF’s of as high as 3.6-3.8?
-
Systems with accurate heat loss calculations and steady state design temps of 55C may well do a better as they may run as low as 50C in practice with gains and dynamic effects taken into account. (-0.0688 × 50 + 6.4303 = 3.0±0.6).
-
Systems with oversized heat loss calculations and 55C design temperatures could do even better of course!
-
It is possible to hit higher SPF’s from systems that run at 55C for short periods in a day e.g to boost temperatures but then fall back to lower temperatures such as 45-50C for longer periods. I will do another post with an example of this another time.
Would love to hear peoples thoughts on the above!
.