- two years ago when our ASHP was new I recorded daily data. With hindsight this was just confusing and I ran down a rew dead ends as I optimised system efficiency. Now I just record weekly data and only ever adjust anything weekly. (For non-HP reasons I do keep daily max and min temperature but compile weekly averages.) 

- Newton's law of cooling says that the rate of heat loss by a warm body to its environment is proportional to the temperature difference between them. To keep your home at a steady temperature the energy pumped into the home must match this loss. At average OAT 10C I my ASHP consumes around 65kWh per week; 125kWh when it is 5C and 219kWh in the only 0C week I've observed. This is looks pretty consistent to me.

- Wind speed is a significant factor in heat loss. But I don't record it because we are in a sheltered location and it would be even more of a faff than all the other metrics. But consumption seems higher on windy days... subjectively. 

Lastly, a few thoughts:

- I count myself as very lucky to have an energy consumption meter just for the ASHP. Anyone reading this before having a new installation- take note!

- it is possible to get good results by playing with your weather compensation curve. Up to end October I've consumed exactly the same energy in 2024 as 2023 (2309kWh vs 2310) but at CoP 3.5 vs 3.1.

- A low-loss header can be set up to work just fine- I have one and can't see any need to change it. (Having pumps with a digital read-out of their flow-rate/power consumption helps. The installers set them both to max but even then the CoP was still around 2.8.)

Given my experience I'm still puzzled by claims that ASHPs work well even in sub-zero temperatures. How is this so in locations such as continental Europe and Scandinavia? Do they have much lower humidity than in the UK?

Unfortunately UK houses are notoriously poorly insulated because we cheapskate on this (as we do on so many things).  This obviously means that any heating system, however it is 'fired' has to work harder than it would at the same temperature where the houses are better insulated.  Also our electricity prices are high, partly because we load 'policy costs' onto electricity not gas.  However it is also true that many mainland European countries have a lower humidity than the UK.  All of these do contribute to higher heating costs.

Sub zero is actually less of a problem that an or just above zero, because sub zero the amount of water in the air drops dramatically (which is why -4 feels, to many people, better than +4).    

But in persistent cold conditions it can be stuck in frosting/defrosting for much of the day.

Whilst defrost cycles are expected when temperatures are zero or just above, what you describe here really shouldn't be happening and does suggest something could be improved with your system.  

Can you tell us a bit more eg

• how many radiators?
• size (capacity) of heat pump, model number?
• flow temperature at design temperature?
• have you got a low loss header, 4 port buffer or volumiser (probably the most important question)
• Weather compensation parameters
• floor area and rough construction of house
• measured SCOP
• annual heat energy to house (as measured eg by the heat pump)
• how long does a defrost cycle take

No worry of you don't know all of these, tell us what you do know, it might collectively give us some ideas

Sorry I was being loose with my language  - I was referring to the fact that our total kWh per year for the last 3 years has been pretty constant.

It's OK, that's what I thought you meant, same total kWh per year, and what I was picking up on was the fact that the winters haven't been the same, some colder than others, and for that reason you would expect the annual use to be different, noticeably so, all other things being the same. That said, it might depend on when the year runs from to, 1st Jan to 31 Dec is actually half each from two separate heating seasons, and that might by accident somehow even things out. It might be interesting to look at 1st Aug to 31st July years, and see what that shows. I'm zooming in on this as part of my general interest in understanding the relationship between OAT and heat pump costs (kWH and separately £££) and efficiency (both recent COP and SCOP). It's clear from the charts that many of us have produced that OAT is, unsurprisingly, a key determinant of these outcomes.  

And perhaps comforting to think that when we experience extremes, then they are likely to be short lived and thus, as we must live with them, they are liveable with!

Very true, but we have to bear in mind that chart is for London NW3, where the constant use of high capacity air fryers no doubt keeps the OAT a degree or two above what we country cousins experience.  

Mike

The annual figures since 2022 have been very consistent.

This again has me wondering - OATs in winters 2022/23 and 2023/24 were quite different for most of the UK:

which suggests the annual figures should not be the same...

I quite agree about avoiding spurious accuracy bit I'm inclined to think reading to the nearest integer value shouldn't explain the minimum OAT distribution (volatility). However, an inclination to think X isn't evidence for X so I thought I would check a plot for similar data. For some reason most of my normal sources of weather data seem to be FUBAR (surveillance capitalism at work perhaps?) but I did find daily minimums for London NW3 for 2023 and got this histogram (after rounding the numbers to the nearest whole number): 

Make of this what you will. To me, it is perhaps more normal-ish than your distribution, but at the same time is sufficiently non-normal to mean that volatility may be normal, so to speak. 

This is a fairly large timber frame, leaky construction but last January when we had 15 days with sub zero temperatures it coped no problem although it chugged through 60kWh/day in doing so.  £15 per day on the coldest days I can live with, yes I am fortunate, perhaps other Grant users can comment on your units performance better.

The ASHP is an older Grant model. I've looked all over for a data sheet with a SCOP value but can't find one. The COP of 2.66 is from the sticker on the back of the unit. Looking at more recent Grant models they appear to have improved dramatically since then. The latest R32 model with a similar output has a claimed COP of 5.28 and a SCOP of 5.2! If true it would be enough to offset the price penalty of electricity compared with gas. In my dreams.

I understand that the KWh will be greater when the OAT is below Zero but at what cost? In that data series the one day when the OAT hit a Minimum of -6C, was 21st January 2023, when we used 93 kWh. That compares with our average daily usage across the entire year of 29 kWh per day. (About 8 to 10 of that is for all other non-ASHP uses). Despite this the DHW still did not reach normal temperature and we struggled to get the room temperature to its usual 20C.

I have daily data from the start of 2021. But there was definitely a problem with the set up then (since rectified) as our annual kWh is now only 2/3rds of the level at that time. The annual figures since 2022 have been very consistent.

I'm less concerned these days with further efforts to tweak my system (I can accept a monthly energy bill, now,  of circa £200) but I'm still interested to see how my experience compares with others. It's one of the few benchmarks that enables you to judge whether the results are good, bad or typical.

Given my experience I'm still puzzled by claims that ASHPs work well even in sub-zero temperatures. How is this so in locations such as continental Europe and Scandinavia? Do they have much lower humidity than in the UK?

My ASHP goes into a defrost cycle at the drop of a hat as temperatures approach Zero. It's already happened once this season on 10th October when the Min OAT was 2C (humidity 97%) But in persistent cold conditions it can be stuck in frosting/defrosting for much of the day.

Mike

However - one thing I do notice about you most recent chart, with the histogram for days with each minimum OAT, is that they do not look normally distributed beyond a vague overall bell shape. It seems most odd that the min OAT with the most days (10 degrees 45 days) has two of the smaller counts on either side 9 degrees 9 days, 11 degrees 11 days. Of course that might happen by chance, and it may be that minimum OATs are not normally distributed, but the variance would certainly make me double check my sensor and data (and also the normality assumption).