@tasos and @cathoderay thanks. I have some history graphs set up with all of those parameters. The slight issue is I mistyped some code with the wrong entity name in my recorder config (essentially the part of HA that decides which entities record historic data, and for how long). Thus there is some key data missing in these graphs (the set outlet flow temp) but I've just fixed this going forward.

A small tweak to my curve automation settings has now stabilised the IAT very well. Having cross referenced energy usage by looking at consumption data for a few other similar weather days (in terms of average OAT) I actually can't identify much if any difference in efficiency.

Here’s a current graph showing a bit more info. The set outlet temp line you can see was ‘stuck’ at 40C until this evening when I changed the recorded history settings. It’s currently running at set flow temp of 35C.

Nice steady IAT. So far not really any different to using the midea custom curve 9, but none of the downsides. Incidentally the comfort offset adjustment hasn’t run. I’d expect this to only run if very cold and windy, or lots of solar gain.

@benson Thanks for new graph. Initial reaction is that fluctuation of the various quantities is quite inexplicable and needs some more parameters to be plotted on top of them, at the risk of being too messy. Here are my thoughts:

1. Set LWT remains the same, despite the rise of OAT by 5 C. I assume this is because of the steady IAT. This means that either your house does not respond to OAT fluctuations or the temp sensor is in a position that is not affected by OAT a lot.

2. Actual LWT fluctuates a lot and does not seem to follow set LWT. I can only assume DHW demand affects this. So can you tell us the DHW rule ? 

3. Actual RWT also fluctuates along with LWT. I hypothesize that the system is trying to maintain Δtemp = 5 by changing pump speed. Can you plot this parameter along with the rest?

Of course, all of the above are just hints.

@tasos ignore the set LWT value on the graph before around 1700hrs yesterday. It was just stuck at 40°C on the HA statistics and not representative of the actual value. It varied between 35 and 39 °C ish from memory for that preceding period, influenced by OAT.

Unusually in the last 12 hours or so we've had a very steady OAT of 9/10 °C so it has been set at 35 °C.

Point 2- not sure what you mean. I'm seeing what I'd expect to see and the DHW runs are pretty obvious, no? I have quite long reheat schedules in the evening so it will reheat a couple of times normally when the family start using the shower.

Point 3- absolutely. The midea uses flow rate to maintain ΔT from what I can see. Here's an updated one this morning.

Interestingly comparing yesterday (after change to manual WCC via HA automation) with Monday (pre change using custom curve 9):

16/02. Average OAT 6.47. COP 3.9. kWh consumption 37

20/02. Average OAT 5.47. COP 3.9. kWh consumption 31

HA automation default settings: 48°C @ -4 OAT. 29°C @15

Curve 9 default settings: 49°C @ -4 OAT. 29°C @15

Posted by: @benson

↑

Incidentally the comfort offset adjustment hasn’t run. I’d expect this to only run if very cold and windy, or lots of solar gain.

My auto-adapt script only makes infrequent changes. If the weather compensation, however achieved is where it should be, then most of the time the IAT will be in range. @tasos, it is important to note that the IAT element of the control is a secondary fine tweaking control, most of the time the control comes directly from the OAT. This intentional, because all other things being equal, the OAT is the determinant of the heat loss, and so the energy needed to keep the IAT in steady state.

More generally, the 20th Feb chart looks much as I would expect, apart from the Set LWT which you have explained. On the left there is some steady state running with occasional defrosts consistent with the low OAT, on the right there is typical Midea slow cycling at more moderate OATs, along with some DHW spikes.

Posted by: @tasos

↑

I hypothesize that the system is trying to maintain Δtemp = 5 by changing pump speed. Can you plot this parameter along with the rest?

Posted by: @benson

↑

Point 3- absolutely. The midea uses flow rate to maintain ΔT from what I can see. Here's an updated one this morning.

I'm still not so sure about this. If you look at the flow rate in the second chart, in effect it only has two values, ~0.8 and ~1.6 (and occasional brief spells at ~1.25), and if you look closely I thin you will find the set changes are in step with the ramping up and down of the actual LWT, with the current (amps in) following suit. The unit is modulating by turning itself on and off, ie cycling. Because of this, the LWT/RWT delta t changes constantly. The right hand part of the second chart:

shows what I normally see, the LWT oscillating about the set LWT with a mean actual LWT that looks very like it matches the set LWT. I think this is just Midea's way of doing things. As you collect more data, the patterns will become clearer.

Posted by: @cathoderay

↑

I'm still not so sure about this. If you look at the flow rate in the second chart, in effect it only has two values, ~0.8 and ~1.6 (and occasional brief spells at ~1.25), and if you look closely I thin you will find the set changes are in step with the ramping up and down of the actual LWT, with the current (amps in) following suit. The unit is modulating by turning itself on and off, ie cycling. Because of this, the LWT/RWT delta t changes constantly. The right hand part of the second chart

Ah yes. A fair point. I wonder what triggers the change in flow rate then? Does your unit, and mideas in general, stick to these same flow rates, more or less?

Posted by: @benson

↑

I wonder what triggers the change in flow rate then? Does your unit, and mideas in general, stick to these same flow rates, more or less?

It seems the R32 units tend to show this step pattern (mine certainly does) with a limited number of parking slots, the later R290 ones less so, but still not perfectly smooth continuously variable flow rates. 

Thanks for recent data. Here are some more questions.

Posted by: @benson

↑

I have quite long reheat schedules in the evening so it will reheat a couple of times normally when the family start using the shower.

I apologize, I am not using DHW and I am not familiar with schedules. What is your exact schedule ?

Posted by: @benson

↑

Why is set LWT dropping at around 17.00 for no apparent reason ? Both OAT and IAT are quite steady.

Posted by: @benson

↑

The unit is modulating by turning itself on and off, ie cycling.

I wonder why it is cycling ? I thought this should be avoided. Is some thermostat intervening ? Is the unit oversized ?

Posted by: @tasos

↑

Is some thermostat intervening ? Is the unit oversized ?

No, and not really. Standard midea behaviour at these sorts of OATs...

Posted by: @tasos

↑

I am not using DHW and I am not familiar with schedules

DHW are the big peaks going up to 57 ish LWT.

Posted by: @tasos

↑

Why is set LWT dropping at around 17.00 for no apparent reason

As said ignore that bit. It wasn't running at set LWT point of 40. Home assistant just wasn't storing the correct data  before that time. Basically in home assistant there's various ways of configuring what entities have recorded historical data, and for how long i.e. 1 month, 1 year etc. Set LWT didn't have any recorded historical data before around 1700 yesterday evening.

Posted by: @tasos

↑

I wonder why it is cycling ? I thought this should be avoided. Is some thermostat intervening ? Is the unit oversized ?

We have pretty much established by direct observation that Midea R32 units (and perhaps R290 units, less data available) modulate at most OATs (anything above ~8°C OAT which is a lot of the time) by cycling. It is what they do, and as it is slow cycling, typically around once an hour, we don't believe it causes much if any harm. Midea units are not known for pegging out early from too much cycling. They cycle even when the unit's rating is reasonably well matched to the building (as mine is). My setup has no intervening external thermostat, the room stat is set high (26°C) so it is always on (always calling for heat) in the heating season.  

@cathoderay If I can see correctly, there are periods on the graph where LWT equals RWT and during those periods the unit shuts. If this is the case, it means there is no heat drawn by the house. How can this be ?

I have drawn some time lines to show this point.