At some point I suspect @transparent may appear and say we are travelling off piste. If necessary, we could move the Midea heat pump performance posts to a thread I started not long ago, In the Bleak Midwinter...

Talking of which, @starkers and @el_presidente, have you had your heat pumps running in sustained (several hours or more) cold OATs? I'm not sure when either of you had your heat pumps installed (mine was Jan/Feb 2022).

Posted by: @cathoderay

↑

Talking of which, @starkers and @el_presidente, have you had your heat pumps running in sustained (several hours or more) cold OATs? I

Ours was installed November 22, but I've only had access to the dashboard for the past few months. (It is a Freedom supplied setup, as you guessed).  I've yet to encounter a period when the IAT had real problems keeping up, but we run our house at 17C target temp which is under the design temp by quite a bit, and we burn wood on days when we think it'll be very cold and/or agile is expensive, so I've not really confirmed the HP can handle extremes yet.

By the way, at the risk of derailing even further- I think our model is a 16kW one, which is identical to yours save for some DIP switches being changed (If I remember right the 12,14,16kW units are physically identical and the max power is set at the factory using these). Have you considered bumping it to 16kW and seeing if it keeps up with low OATs?

Meanwhile, Transparent has been busy installing cables below floor level...
... and topics going off-piste weren't featuring highly in the mind.

However, I should point out that the proposed Government ESA software for controlling devices in the home, would love to have access to our heat-pumps via Modbus.

This is the same issue which @cathoderay and I discussed over here.

If they were to map our location, tariff, Smart Meter consumption and ASHP firmware version into their snazzy new database, they could turn off our heat-pumps in a 'smart' way rather than simply cutting the power-input.

Modbus communication would also allow monitoring...
... which would be collated according to location,
and supplied to NESO so they can better balance supply & demand.

Having Modbus/RS485 on a heat-pump is a two-edged sword.

Yes, it can provide useful data to the homeowner,
but would you consider allowing a data-handling agent to access it on behalf of Government?

Posted by: @transparent

↑

Having Modbus/RS485 on a heat-pump is a two-edged sword.

Yes, it can provide useful data to the homeowner,
but would you consider allowing a data-handling agent to access it on behalf of Government?

To some extent agile tarrifs / the DFS is doing this work for them- homely is filling that need by handling these things automatically. I don't see why the govt would mandate it unless the situation was very dire.

@cathoderay, sorry, I forgot to attach the following for your previous request. This is one full day rather than two weeks as shown previously, with the flow/return temps included as well. Hovering over the plot I see that the average flow rates are consistently 27-28 l/min for the heating cycles but 29-30 l/min for that single DHW request. The previous 26l/min might be averaging when showing more datapoints.

Posted by: @starkers

↑

I've yet to encounter a period when the IAT had real problems keeping up, but we run our house at 17C target temp which is under the design temp by quite a bit, and we burn wood on days when we think it'll be very cold and/or agile is expensive, so I've not really confirmed the HP can handle extremes yet.

I think choosing to run at 17 degrees and using supplementary heating on very cold days may make all the difference. In cold weather mine manages to get to 17 degrees (or just under), so if you are only aiming for 17 degrees, you may well get there, all the more so with supplementary heat. My target in the living rooms is 19 degrees, once it gets to 17.5 and below I start to notice, and the old chilblains start to appear! It should be noted 17 degrees is actually 4 degrees below the MCS requirment of 21 degrees for living rooms, a significant shortfall.

Posted by: @starkers

↑

By the way, at the risk of derailing even further- I think our model is a 16kW one, which is identical to yours save for some DIP switches being changed (If I remember right the 12,14,16kW units are physically identical and the max power is set at the factory using these). Have you considered bumping it to 16kW and seeing if it keeps up with low OATs?

Luckily @transparent has been busy railing his cables at home rather than worrying about derailing here. Yes I do have those dip switches (apparently Freedom have now removed them folks, were buying cheaper 12kW models and using them as 16kW models) but I have yet to flip the dips, on the grounds that a 14kW model with a rated output of around 11 and a bit kW at zero OAT/55 degree LWT should be able to meet a 10kw demand out of the box. It is bit like buying a car that says it can do 100mph, and then finding it will only do 80mph (down hill and with a following wind). If that happened I'd expect the manufacturer to do something about it.

Posted by: @transparent

↑

Having Modbus/RS485 on a heat-pump is a two-edged sword.

Yes, it can provide useful data to the homeowner,
but would you consider allowing a data-handling agent to access it on behalf of Government?

After the said agency (very probably based in China - not at all far fetched - where are the Midea servers?) has stolen your data, it will soon move on to the next logical step, installing a modbus connection inside your head. You will own nothing, and be happy. 

@starkers - even more interesting! If we zoom in on the relevant bit 

where the red line is the flow rate (L/min), the magenta line is the LWT and the khaki line is the RWT, then you heat pump is basically working in on-off fossil fuel mode. In fact it spends more time off than on in the period shown, even outside the overnight period, when it looks like you may have a setback. The circulating pump is in effect either on at about something under 30L/min, or off.

Here for comparison is my heat pump's behaviour, from a steady state cold period last January: 

I agree with @cathoderay

@starkers you really do need to get rid of that on/off cycling.
That's not how a heat pump is meant to operate.

The Outside Air Temperature (OAT) sensor should be the input which allows the heat-pump to regulate the flow in order to achieve the target room temperature.

Every time the heat-pump starts it uses a lot of power until the compressor has achieved the required pressures for both sides of the heat-exchanger.
That's why manufacturers require the electrical supply to be fused at 32A, rather than the 10A or so which you'd expect for a COP of around 3.0

Nor is cycling good for the grid.

I can see heat-pump cycling when I look at current on the 3-phase feed from a local substation.

Note how the cycling on/off diminishes greatly between the two purple lines I've added.
The current being drawn at night is still quite high because this transformer supplies an area with a number of heat-pumps.
But it's the cycling which causes surges in the Neutral conductor, which then gets hot.

@cathoderay , @transparent I don't disagree with you that on/off cycling is bad and I could be making a big assumption here, but I thought this was homely doing what it was supposed to. By this I mean the flow temps and rate plot cropped like that miss the time of use tariff and setpoints.  Bear in mind throughout that plot homely is maintaining a temperature of 17C with a variation of about 0.5C, and it's trying to overheat the house in cheaper periods so it can glide through more expensive ones. Since the tarrif is half-hourly intervals I have seen it sometimes go ham for 30 minutes in a cheap slot surrounded my more expensive ones.  The outside temp is varying considerably more than the inside one, and it's not like the house is swinging up and down like a yoyo during those periods where it is calling for heat.

Here's an example 24h period recently when it was properly 'cold'- max 6C, min 2C, on the 28th Nov. It looks like the HP was running constantly over that period and maintaining flow temps steady. You can see the flow rate is a bit more choppy during the expensive period of electricity- this was homely trying not to run in that timeslot as much as it could while still hitting the 'green' setpoint region of 17C:

@cathoderay in particular in your case, it sounds like your HP is operating at the limit of what it can handle (note, not what it should handle- your maths looks right to me but it's clearly struggling to keep your setpoint) so I'm not surprised yours is pretty much always running. It could be mine is way oversized but I doubt it- 190sqm of 1930s house with an unfinished uninsulated basement and only 150mm of loft insulation. When I went over the numbers that Freedom used it checked out that the 16kW unit would be ever so slightly above the requirements and 14kW would be a bit below. That said, I'm running it at 17C IAT so it could be because I'm so far below design temps it doesn't have enough range to run slow and steady on warmer days, hence the short cycling on the plot above, where the OAT was mostly about 10C.

Posted by: @starkers

↑

it sounds like your HP is operating at the limit of what it can handle (note, not what it should handle- your maths looks right to me but it's clearly struggling to keep your setpoint) so I'm not surprised yours is pretty much always running.I

A heat pump should be always running.

The initial survey is to calculate the heat loss for the entire house.

That provides the power output required from the heat pump.
If the installer fits one which is larger than that, then it will cycle on/off to maintain the desired temperature.

In attempt to 'correct' the error of supplying too large a heat-pump, installers then fit buffer tanks or volumisers.

See this discussion on Low Loss Headers which explains the issues.

Posted by: @starkers

↑

I thought this was homely doing what it was supposed to

I think one perhaps important consideration here is precisely this - what exactly does Homely do, and as part of that, how does it monitor and report variables? Much the same applies to Home Assistant monitoring, though at least in theory we can see what it does because the code is open source, even if written by what one observer called python sadists. We have already seen for example how aggregating data can obscure important short term changes. Homely's charts aren't helped by the fact they are often overcrowded, and the tendency to use gee-whizz axes (cropped axes used to make changes appear more dramatic than they actually are - often seen in marketing materials eg to make turnip production is up look more impressive than it is, hence gee-whizz). When axes are cropped, it is best to make this explicit, but best of all, don't crop them.

I think the first take-away about what happens when you add Homely to a system is you lose standard weather compensation, all the more so if you use variable tariffs. It seems Homely will give a higher priority to tariff rates than to what the OAT is. If you look at the OAT and flow temps in your most recent chart at 1500 and 0300 the flow temps are very similar, but the OAT is not. Likewise the period after an hour of so after 0900 is not following weather compensation, the OAT is rising, but so too are the flow temps. It would also be interesting to know what the average LWT/RWT delta t was during the period shown in the chart.

The flow rate line continues to suggest some on/off operation, but I suspect a charting anomaly has changed what are step changes into slope changes. Rather curiously, the best steady state period is overnight, between 000 and 0600. The IAT also to my eyes looks quite volatile, but that may be a gee-whizz turnip production is up effect. Nonetheless, it looks like the IAT rose by over one degree between 1200 and 1500, and fell by one degree between 0600 and 0900. These are quite rapid changes by heat pump standards - sometimes it takes days to get my house to warm up by one degree. Thermal mass and other factors do of course influence these rates.

One of the things I have learnt about monitoring and trying to understand heating system is that they are highly complex dynamic systems where nothing stays the same. Add in something like Homely, and its get even harder to see what is actually going on. Is there any way you can actually get at the Homely data, ie the numbers behind the charts? Maybe Homely has a 'download your data' button somewhere (and it is your data). If not, it may be possible, if you are au fait with this sort of thing, to do some developer mode sniffing behind the scenes in your browser. In particular, look at the contents of the files listed under 'Network' - it is very likely one of them will have the raw data. 

The Freedom spreadsheet heat loss calculator almost certainly over-estimates heat loss, especially for older buildings like mine (the assumed wall U values are to high). Depending on which version of their toolkit you used, there is also something else to bear in mind: in the 2021/2022 versions they guessed the Midea heat pump outputs (there are a lot of hidden cells in those spreadsheets, you need to un-hide them to see the inner workings). For example, the 'top copy' version of the calculator sent to MCS for my installation (14kW badge) showed the output at design temp as 12.4kW, whereas the Midea Engineering Data manual shows it is closer to 11kW, not a trivial difference.

For your system, I think the fact you run the system aiming for 17 degrees IAT is significant, because it will significantly reduce your heat loss. By how much is a moot point, as 'professional' (five most terrifying words in the English langauge? 'Leave it to the professionals') estimates vary widely, between around 3% and 13% per degree (it is telling we don't even have an agreed figure for this most basic of numbers). Let's use a middle number, 8%, in which case a reduction from MCS's 21 degrees to your 17 degrees might reduce your heat loss and so demand by something between a quarter and a third.

If you can get hold of your raw data for hourly IAT, OAT and energy out (ie delivered to the house), you can determine your actual heat loss by finding a periods when the IAT is stable (when energy out ie heat loss must equal energy delivered to the house), and then plotting energy delivered against OAT. There will of course be some spread, but you should also see a linear relationship. Here's my plot for the 2023/24 heating season. Note that this is for all hours (including some when the IAT wasn't exactly where it should be) in all days when I didn't have an overnight setback, with DHW heating hours excluded: 

The R squared value isn't too bad at all, and by reading off the energy out when the OAT is at design OAT (-1.8) we can see what my empirical heat loss is: give or take, 8kW, which is considerably less than the Freedom spreadsheet value of 12.3kW.

We can also see the heat pump struggling and failing to keep up with demand at lower OATs. At higher OATs the correlation is good, but at lower OATs, as the OAT approaches zero, it is less good, as the data points start to fall below the regression line (which will in fact have the effect of pulling the left hand end of the regression line down a bit, hence my 'give of take' heat loss above. If I did a plot using only OATs above 5 degrees, I might get a design heat loss closer to 9kW - but life's too short etc). We can see this failure of the heat pump to keep up with the loss at lower OATS even more clearly by plotting the residuals (actual - predicted energy out against OAT), with the residuals becoming increasing negative as the OAT approaches zero: 

Those negative residuals at lower OATs are what I want to fix. To do that I need to understand how Midea heat pumps regulate output, and why mine is failing in that regulation at lower OATs.    

Posted by: @transparent

↑

Posted by: @starkers

↑

it sounds like your HP is operating at the limit of what it can handle (note, not what it should handle- your maths looks right to me but it's clearly struggling to keep your setpoint) so I'm not surprised yours is pretty much always running.I

A heat pump should be always running.

The initial survey is to calculate the heat loss for the entire house.

That provides the power output required from the heat pump.
If the installer fits one which is larger than that, then it will cycle on/off to maintain the desired temperature.

 

In attempt to 'correct' the error of supplying too large a heat-pump, installers then fit buffer tanks or volumisers.

See this discussion on Low Loss Headers which explains the issues.

This is true when the figure of merit being optimised for is COP (or, more specifically, when input kWh is being minimised). Homely isn't using that figure of merit though. To make an extreme example, it may be cheaper in terms of total cost of heating, to run the HP flat out at a higher flow temperature for 30 mins when the electricity cost is 5p/unit at a COP of 2.0 and then turn off for 30 mins than to run for one hour at a COP of 3.0 when the average electricity price is 15p/unit.

That said, I'm open to this being a genuine problem with how things are configured at my end. The issue is that homely is a black box from a controls standpoint- I don't think there's anything I can adjust at my end byond the setpoint and schedule. For those more familiar with the installer side of homely, do you know if anything can actually be tuned/tweaked in terms of the temperature curve?

@cathoderay I'll see if I can have a dig around and work out if I can access the data in a format I can play with- the plotting is done locally so the data must be being presented to plotly.js somehow.

Posted by: @starkers

↑

the plotting is done locally so the data must be being presented to plotly.js somehow.

That is good news. If you can find the script/code that does the plot, it will have a line in it that grabs the data, and that will tell you where it is. It'll probably use pandas with something like:

import pandas as pd

df = pd.read_csv(r'modbus\mideadata.csv') 

depending on what format your data is in.