So I had a look at a Hitachi Yutaki ASHP manual and indeed there is an option of room temperature compensation, which I quote:

"Room Temperature Compensation

If room compensation is enabled, the calculated OTC supply set-point is adjusted based on the difference between room temperature and room set-point in order to reduce the room error. The amount of room influence can be adjusted by the room temperature compensation factor setting (can be adjusted, default 2).

Example

Compensation factor = 2,

SR1 (room set-point) = 20,

TR1 (room temperature) = 23

=> S1 (supply set-point) reduced by 2 * (23 – 20) = 6C "

Even though this is a rather crude approach, I think it does show the future, which is very well predicted by your reported actions here.

I presumed when Gemini was writing the automation code for me, that it was data farming from your forum posts about your set up

Spooky! I think you already know I don't use AI much, but if I did, and found myself asking it a question and finding the answer seemed curiously familiar, I might end up thinking I need to start taking the tablets again!

It does work well- thanks for all of your help

A pleasure, and thank you for documenting what you have done here. I am sure the other who have contributed have also been glad to do so. 

My auto-adapt does pretty much the same thing, infrequent but useful adjustments.

I presumed when Gemini was writing the automation code for me, that it was data farming from your forum posts about your set up 😆 

It does work well- thanks for all of your help, and everyone else who has advised along the way.

Ok some more info on the new setup which may be of interest- specifically the comfort offset automation which adjusts the flow temperature if the IAT is outside of the comfort parameters, and an interesting observation over the last couple of days now that it is a bit warmer.

The above shows the comfort offset history for the month, so it will adjust the LWT by 2 degrees C either way, depending if the IAT is undershooting or overshooting. It hasn't actually run that much, which is what I was hoping/expecting. On 13th March it must have been windy and needed a boost. On the other days where it has offset -2, this will be down to solar gain.

What the adjustment effectively does when warmer, is that it turns the ashp off- so if the set LWT is then below 29 degrees C which is the lowest temp that it will run. This actually works perfectly. As per this graph (red arrow where the comfort offset has pushed the set LWT below 29)

I think now I can confidently say this is our ideal setup in terms of control strategy, and I can probably leave as it is.

It is worth summarising/highlighting the following perhaps:

-our ashp was up and running early to mid Oct 24 so homely had been set up for a few weeks by the time we reach the earliest comparison month below (Nov)

-we also generally have our IAT higher now than we did in the initial stages post install (probably averaging closer to 21.5 rather than 20.5).

-homely was removed around August 25. It was quite cold in our house for that first winter we had it as the max LWT had been set at 48 (the setting on OUR ashp which you cannot change unless you pester homely by email 😆 ) which wasn't sufficient to keep our house warm. 

-I was spending a fair bit of Nov and Dec 25, having removed it and then using native controls, fiddling with the WCC settings to try and get them just right

-mid February was when I changed the control strategy from using WCC mode in the midea controller to using HA automations instead to set the outlet flow temp. This uses a similar method compared with our old WCC automations to slightly adjust LWT if IAT deviates from our comfort range. This is definitely the way to go if you have a wired controller like mine that will turn everything off when accessing serviceman settings- much better.

-whilst I thought pure WCC worked well initially, it is definitely a challenge on cold windy days so realistically I'm not sure this is a viable strategy unless you want to be monitoring and manually adjusting, or at least not for our house. 

November 2024 (average OAT 9°C): 675kWh. COP 3.77

November 2025 (average OAT 9°C): 684kWh. COP 3.78

December 2024 (average OAT 9°C): 824kWh. COP 3.87

December 2025 (average OAT 8°C): 928kWh. COP 3.74

January 2025 (average OAT 5°C): 1383kWh. COP 3.15

January 2026 (average OAT 6°C): 1230kWh. COP 3.37

February 2025 (average OAT 7°C): 1089kWh. COP 3.31

February 2026 (average OAT 8°C): 733kWh. COP 3.94

I appreciate that it isn't an exact science particularly with the average OATs as it depends if there has been a prolonged cold snap, more windy etc. Gives an idea though.

For example, have a look at this diagram

That is a very abnormal period, almost all defrosts (LWT below RWT) and two steady state runs (both afternoons, second one also has a DHW reheat). It is not normal Midea slow cycling, that chart tells you nothing about normal Midea cycling. It does tell you a bit about steady state running, when the actual LWT does approximate to the set LWT. 

In the second chart, the OAT is mostly around 10°C, and that is in the normal slow cycling range.

Now, why does this happen ?

That is indeed the question those of us who are interested in this problem are asking! But I am not sure I understand your possible explanation yet...

To figure out the exact details, 1 minute plots of a couple of cycles would be very helpful.

My temperature charts are one minute plots, I can do a very zoomed on one (a couple of normal Midea cycles) for you if that would help, let me know.

First thing I noticed, is that cycling is exaggerated when set LWT is less than 40 C. For example, have a look at this diagram:

As you can see, as a result of OAT being below 10 C, set LWT is above 40 C and leaving LWT is around there.

However, in this diagram:

set LWT seems to be below 40 C, causing the system to cycle.

Now, why does this happen ? Well, I think because of "water temperature hysteresis of activating heat pump" logic:

coupled with lower LWT limit in heating mode (dT1SH is 5 C and lower LWT, 35 C for my setting).

To figure out the exact details, 1 minute plots of a couple of cycles would be very helpful.