Research question: Do we see a clear relationship between measured average flow temperature when the heat pump is running on the coldest day and annual SCOP in the HeatpumpMonitor.org data?
This is going to be a bit of a long thread that I will add to over time, looking first at aggregate stats before then looking closer at particular systems.
Theory: Heat pump systems on weather comp should show quite a clear relationship between design flow temp and annual SCOP for space heating. The SCOP calculator tool that I built here is a an example of how this can be calculated theoretically Heat pump SCOP calculator. It’s not the only factor in the calculator that affects the SCOP, the other factors in that calculator are the practical efficiency of the heat pump and the minimum modulation level which determines how low the flow temperature can go in mild weather. These would typically be 50% and 40% respectively for systems without sophisticated control of mild weather cycling.
Figure 1: All systems with valid data (minimum 290 days).
As is often the case with HeatpumpMonitor.org real world data, we see a wide spread of values. Clearly there is more to annual SCOP than the average flow temperature when the heat pump was running on the coldest day. See Heat pump oversizing data analysis thread for another example of wide spread of results.
All systems but one have average flow temperature when running that are below 45C, most are significantly below, the average flow temperature of all systems on the coldest day is 36C.
On average, systems had a measured average flow temperature when running of ~7C below the stated design flow temperature entered on the form. There are 2 systems with average flow temperature 20C or more below those entered on the form. 12 out of 59 systems were within 2C of the form design temperature. Actual experienced outside temperatures where on average 0.6C higher than form entered design outside temperatures.
System 65, next highest SCOP (H2 boundary) seemed to have a bit of data loss on the coldest day, though it was also off for that middle part of the day as can be seen from the flow and return temperatures.
Average flow temperature when running 33.1C including DHW cycle.
Average outside temperature -1.7C
We can see a little missing/incorrect DHW heat data probably due to an air issue earlier in the day, the carnot efficiency to reproduce the space heating part was ~45%.
Weighted average flow temperature from histogram tool for whole year: 31.20C
Coldest day:
Average flow temperature when running: 38.9C
Peak flow temperature on space heating: 42.0C
Average outside temperature: -2.5C
Carnot efficiency on coldest day 44.7% (fairly standard)
This is a good example of a heat pump running around 40C rather than mid 30s but still one of the highest performing systems - this is likely due to average flow temperatures over whole year being relatively low.
I’ve made a couple of improvements to the HeatpumpMonitor.org daily data tool. It’s now easier to access e.g mean flow temperature when running vs outside temperature for each system e.g Heat Geek Vaillant:
Perhaps the more informative temperature to look at would be average flow temperature when it’s around 5C outside, or average flow temperature for the outside temperature at which the most heat was delivered … which could be drawn as a histogram…
Added an option to show weighted average flow temperature to the histogram tool which plots flow temperature vs heat delivered. This might be the better measure as it should capture the average flow rate at which heat is delivered over the year, which should arguably correlate better than the coldest day… now to automate this calculation for all systems
Sheffield Viessmann: 30.08C, SCOP: 4.97 (H4)
Bristol Nibe: 32.77 C, SCOP: 4.66 (H2)
Heat Geek Vaillant: 31.20 C, SCOP: 4.62 (H4)
Glyn’s Samsung: 33.05 C, SCOP: 4.41 (H4)
System 170, 11 kW Ecodan, 41.67C, SCOP 3.6
System 77, Woodam surrey, Vaillant, 38.27C, SCOP: 3.7
For systems with DHW mode detection with the ability to filter out space heating only, the correlation between average flow temperature on the coldest day when space heating vs annual space heating SCOP is not much better.
Thought I’d add a few comments on the above as a starting point.
Looking at both the first chart (SCOP vs coldest day flow temperature)
and the 3rd from last: SCOP vs weighted average flow temperature for the whole year:
The weighted average flow temperature is showing a clearer relationship as we might expect but it’s still interesting how wide the variance is at a given average flow temperature. At face value a system averaging a weighted average flow temperature of 33C could have a SCOP of anything from 3.5 to 4.5, that’s quite a wide range!
Likewise you can select a SCOP such as 3.5 and see systems with weighted average flow temperatures all the way from 33 through to 43C.
Other factors are clearly having a significant impact on performance. It would be interesting to list different design factors and work out statistically how significant each of these factors are in determining SCOP::
Design flow temperature & average flow temperatures
Make and model of heat pump (and related effectiveness of control algorithms e.g for mild weather cycling)
Oversizing factor
Outside temperature differences
System volume
Hydraulic separation, type and configuration
Indoor temperatures
Type of emitter system
Monitoring error?
Someone with a data analysis background and with more skill than I could probably unpick the impact of these factors in a clearer way.
I think I will try and pick out specific systems that span a range at a given flow temp and try to and see if I can make any sense of why these are giving such different results.
I’ve added weighted by kWh delivered, average flow - outside temperature to the measured params in the system list table.
It’s now possible to plot this using the plotting tool on the site, which also allows for colour coding the points by % carnot when running.
I think this result is actually quite strong:
Annual performance is strongly correlated to weighted by kWh heat delivered flow - outside temperature as we would expect.
We can see that higher % Carnot, yellower colours are above the line and lower % carnot bluer points are below the line. % Carnot being how effective the heat pump + system is at any given flow & outside temperature, what % of ideal carnot COP it’s getting.
Damon’s Viessmann Vitocal 150A scores the highest % carnot of all systems available on the annual view, though it’s fairly close to others 55% vs 54% carnot.
We dont have any other Viessmann Vitocal 150A’s with a full year of data, but we do have quite a few with a month of data, and the % carnot from these span a wide range, it’s clearly not as simple as buying a top end heat pump.
Damon’s Vitocal is 2nd from the top here and Toms install that required long primaries is the lower end of the % carnot range (Tom is a very experienced and good installer)
Does this suggest that around half of performance is down to system temperatures and around half other factors that contribute to % carnot?
The correlation between % carnot and system temperatures is very week which suggests that Im not leaking a strong temperature related signal into the % carnot figures… these should be non-correlated…
The majority of systems that achieve a SPF of 4 or above have a mean flow temperature when the heat pump is running on the coldest day <40C and more typically ~35C.
A mean flow temperature when the heat pump is running on the coldest day <40C is not a guarantee of a SPF of 4 or above as we can see by the systems in area 2. There are also other factors at play.
I think it’s worth highlighting and taking on board that the majority of systems that do get a SPF of 4 and above on HeatpumpMonitor.org are running very low mean flow temperatures on the coldest days. 35-40C is significantly below what is more commonly designed for, 45C is usually seen as low and 50C is becoming more common. Design target temperatures are however often over-estimated given over-inflated heat loss calculations and so the in-practice flow temperatures can be lower.
There is a risk of improving the accuracy of heat loss calculations but not updating our thinking on target flow temperatures at the same time.
Max 2 hour mean flow temperature on the coldest day vs mean flow temperature when running for the whole day.
Many systems on HeatpumpMonitor run flow temperatures that are hotter for a part of the day on the coldest day, e.g higher outputs required after night time set back. While there is a wide spread of values here, the maximum 2 hour mean temperature is typically about 2.5C higher than the mean flow temperature when running for the whole day:
It’s important to bear in mind that the relationship between the design heat loss/demand & temperature and the measured values does depend to a degree on the time period that we are averaging over.
Do we have a new, more recent coldest day that could be used for this type of analysis? I’m thinking around Nov 21st, where the average from to 0° at my location.
