@jamespa - I'm planning to alternate setback plus recovery boost, ie the auto-adapt script, with periods of standard 24 hour running, building up roughly equal durations for both. We will get there!
Midea 14kW (for now...) ASHP heating both building and DHW
For those forum members who have a fetish for spreadsheets, I have been running some raw data (kindly provided by CathodeRay) through a modeling toll, with the attached results.
@derek-m I'm not certain what these are telling us. Am I comparing calculated vs actual iat, in which case there seems to be a small divergence which initially looks as if it might be progressive through the day, over several days is just "noisy" (which is unsurpring given that we know the real world to be noisy), so overall a pretty good match
Is this correct. If not perhaps you could summarise what you believe this shows
4kW peak of solar PV since 2011; EV and a 1930s house which has been partially renovated to improve its efficiency. 7kW Vaillant heat pump.
Not sure information this helps your modelling but here goes nothing:
our house is empty and has been slowly cooling down since Saturday
the initial drop is around 3.5 degrees overnight then half to 1 degree for the next 4 nights. In fact the last 24 hour show there was no noticeable drop at all. But the oat had risen to 10c.
It continues
This continues with smaller and smaller drop s each day until today.
the 4 day chart below shows it better
As background, the house is 180sqm, Low thermal mass to all external walls. Insulated envelope. Brick internal downstairs walls and plaster walls upstairs. 2 cold areas - 1 on old fireplace and one expanse of bifold glazing. Mostly underfloor insulated except one living room floor. Room stat is in coldest room
does this help esp ref James’ comments on holiday setbacks etc? And general ideas on heat loss vs heat retention.
Not sure information this helps your modelling but here goes nothing:
our house is empty and has been slowly cooling down since Saturday
the initial drop is around 3.5 degrees overnight then half to 1 degree for the next 4 nights. In fact the last 24 hour show there was no noticeable drop at all. But the oat had risen to 10c.
It continues
This continues with smaller and smaller drop s each day until today.
the 4 day chart below shows it better
As background, the house is 180sqm, Low thermal mass to all external walls. Insulated envelope. Brick internal downstairs walls and plaster walls upstairs. 2 cold areas - 1 on old fireplace and one expanse of bifold glazing. Mostly underfloor insulated except one living room floor. Room stat is in coldest room
does this help esp ref James’ comments on holiday setbacks etc? And general ideas on heat loss vs heat retention.
here is yesterday’s temp drop chart. I forgot to include
@derek-m I'm not certain what these are telling us. Am I comparing calculated vs actual iat, in which case there seems to be a small divergence which initially looks as if it might be progressive through the day, over several days is just "noisy" (which is unsurpring given that we know the real world to be noisy), so overall a pretty good match
Is this correct. If not perhaps you could summarise what you believe this shows
Hi James,
I need to go 'hunting and gathering' at the supermarket, but I will reply later. I deliberately did not add any comments to the previous post because I did not wish to influence anyone's interpretation.
One thing upon which I would appreciate your advice. The modeling tool shows that the WC curve is actually a curve. If the WC settings are adjusted such that IAT remains constant at 0C and also at 15C, applying values different from those two values shows a divergence of IAT from the requested value. As would be expected, the divergence is most pronounced in the 7C to 8C region.
The modeling tool uses linear WC, so the calculated IAT diverges with changes in OAT. I don't know if any of the manufacturers use linear or non-linear WC, so don't know whether to correct the curve within the calculations. Any thoughts?
@derek-m I'm not certain what these are telling us. Am I comparing calculated vs actual iat, in which case there seems to be a small divergence which initially looks as if it might be progressive through the day, over several days is just "noisy" (which is unsurpring given that we know the real world to be noisy), so overall a pretty good match
Is this correct. If not perhaps you could summarise what you believe this shows
Hi James,
I need to go 'hunting and gathering' at the supermarket, but I will reply later. I deliberately did not add any comments to the previous post because I did not wish to influence anyone's interpretation.
One thing upon which I would appreciate your advice. The modeling tool shows that the WC curve is actually a curve. If the WC settings are adjusted such that IAT remains constant at 0C and also at 15C, applying values different from those two values shows a divergence of IAT from the requested value. As would be expected, the divergence is most pronounced in the 7C to 8C region.
The modeling tool uses linear WC, so the calculated IAT diverges with changes in OAT. I don't know if any of the manufacturers use linear or non-linear WC, so don't know whether to correct the curve within the calculations. Any thoughts?
The wc is a curve because of the emitter power vs ft curve, (which I have been told is linear for ufh, but have not verified).
So far as I can tell there are various approaches to wc used by manufacturers. I can't honestly remember which manufacturer uses what, I just mentally noted the different approaches as I read the manuals during my discovery phase.
Many use linear only. Its an approximation but will deviate from perfection by a degree or a bit more at some temperatures depending how you set up as you know.
Vaillant have a series of pre programmed curves to choose between. I haven't worked out exactly what they are meant to approximate, but they are curves not straight lines.
Some use multi point linear, so you can program more than two fixed points and then it presumably does linear interpolation between them.
Then of course there is the hybrid approach like the mitsubishi adaptive mode, where the wc is used as a base but is fine tuned by measuring the room temp using an undisclosed algorithm.
I have yet to find one where you can program the exponent as quoted on radiator specs, although it's possible that's what vaillant is in effect.
So if you want to precisely simulate multiple manufacturers you need to allow a variety of wc curves. Whether this is worth the bother depends on what the objective is. I expressly simulated a variety of wc curves in my wc simulation posted earlier, but this was a single purpose simulation designed specifically to explore how important wc is and nothing else, so obviously it mattered to the outcome. In the early days of my discovery I thought that the precise wc algorithm might matter in terms of efficiency, but my wc simulation convinced me that linear is good enough so I stopped trying to remember which models do what!
For most purposes the precise curve won't matter too much, thus for a more general simulation I'd be tempted to use either 'linear' or 'perfect', or alternatively have a wc lookup table which can be adjusted to suit (as in my wc simulation). To an extent manufacturer specific results are not very interesting, what's interesting is general principles, so unless it's likely to matter I'd perhaps not spend too much effort modelling it.
This post was modified 1 year ago 4 times by JamesPa
4kW peak of solar PV since 2011; EV and a 1930s house which has been partially renovated to improve its efficiency. 7kW Vaillant heat pump.
Thanks for the reply James, I think for the time being I will follow your advice and stick with the linear model.
I started a reply to your previous question earlier, but then had a re-think when I was looking closely at the raw data. Initially I just took the LWT value for each hourly interval, on the hour, but I then realised that the value may not adequately reflect the LWT value for the full 1 hour period. I then averaged the LWT value for each 1 hour period, which was the values used within the modeling tool results that I posted. In formulating a more detailed description I have since realised that even the averaged 1 hour value may not be totally correct, since the heat pump was not running continuously during each 1 hour period, but was in fact cycling.
Although it is more time consuming to do so, I am going to average the LWT values for each minute that the heat pump was actually operating, which may produce more accurate results.
Here is the actual Set LWT vs OAT (ambient temperature) for my heat pump for all available dates (late March 2023 to date, when running, always in WC mode):
At least some of the spread, but not all, will be due to having various WCC endpoints set during the period (the WCC left hand set point has been variously between 58 @ -4 and 55 @ -4). The gradient between 5 and 15 degrees ambient looks fairly linear to me, despite the spread. Above 15 degrees ambient (WCC right hand set point), it flat lines, as expected. What I did not expect to see is what looks like flat lining below 5 degrees ambient. The Midea reported ambient temp is also somewhat suspect: it never drops below zero, when in reality it almost certainly went below zero, on at least three occasions in April 2023. I really need an independent OAT sensor but as previously noted, there does not appear to be a weatherproof OAT that has modbus output and will run off USB voltages (5V).
Midea 14kW (for now...) ASHP heating both building and DHW
After trying various options I decided that the hourly average of all the LWT values was probably the correct method since the water pump keeps running throughout the whole period. It is therefore probable that the heat pump will be heating the home when the compressor is running, but may also be slightly cooling the home when the compressor is stopped.
From the 1 week's worth of raw data supplied, I was surprised to find that the heat pump would appear to be cycling continuously, at a rate of 2 or 3 times each hour. Although cycling was predicted by the modeling tool at OAT's of approximately 8C and above, the raw data showed that cycling was still present at OAT's as low as 4C. This of course makes accurate comparison between the real World data and the theoretical predictions more difficult to accomplish.
A further complication is that most of the raw data values are integers, which of course can influence the accuracy of results within small operating ranges. Averaging was used to provide some degree of precision, and should hopefully improved accuracy of data values. Of particular note is the fact that the OAT value was seen to increase by 1C or even 2C when the compressor and fan were stopped, but then reverted to the previous value shortly after the heat pump restarted operation.
I have decided to re-check the data using more accurately averaged OAT values, since OAT is one of the more important parameters used within the modeling tool. It may take some time, but as the guy in the movie said "I'll be back".
After trying various options I decided that the hourly average of all the LWT values was probably the correct method since the water pump keeps running throughout the whole period. It is therefore probable that the heat pump will be heating the home when the compressor is running, but may also be slightly cooling the home when the compressor is stopped.
From the 1 week's worth of raw data supplied, I was surprised to find that the heat pump would appear to be cycling continuously, at a rate of 2 or 3 times each hour. Although cycling was predicted by the modeling tool at OAT's of approximately 8C and above, the raw data showed that cycling was still present at OAT's as low as 4C. This of course makes accurate comparison between the real World data and the theoretical predictions more difficult to accomplish.
A further complication is that most of the raw data values are integers, which of course can influence the accuracy of results within small operating ranges. Averaging was used to provide some degree of precision, and should hopefully improved accuracy of data values. Of particular note is the fact that the OAT value was seen to increase by 1C or even 2C when the compressor and fan were stopped, but then reverted to the previous value shortly after the heat pump restarted operation.
I have decided to re-check the data using more accurately averaged OAT values, since OAT is one of the more important parameters used within the modeling tool. It may take some time, but as the guy in the movie said "I'll be back".
Thanks for the update but I confess Im now a bit lost. 'The raw data' is presumably actual data from a real heat pump, was it @cathoderay 's or someone elses.
And you say it was cycling at 4 whereas the model predicted cycling at 8, what load was the model based on (is it just the case that the actual load is less than the modelled load - I guess another way to put this is at what power did it start cycling and is this consistent with manufacturer data/model?
Of particular note is the fact that the OAT value was seen to increase by 1C or even 2C when the compressor and fan were stopped, but then reverted to the previous value shortly after the heat pump restarted operation.
Do you mean the OAT, because if so its a measurement error (most likely because the OAT sensor is inside the casing)
And you say it was cycling at 4 whereas the model predicted cycling at 8, what load was the model based on (is it just the case that the actual load is less than the modelled load - I guess another way to put this is at what power did it start cycling and is this consistent with manufacturer data/model?
The real world cycling varies a bit, but is typically around twice an hour. The last 24 hours data:
Of particular note is the fact that the OAT value was seen to increase by 1C or even 2C when the compressor and fan were stopped, but then reverted to the previous value shortly after the heat pump restarted operation.
Do you mean the OAT, because if so its a measurement error (most likely because the OAT sensor is inside the casing)
The 1-2 degree increases are I believe because the heat pump has stopped blowing out cold air while the compressor is off, see the pattern in the above chart. What you are actually seeing is a decrease in the OAT when the compressor is running (try standing by a heat pump when it is running, and you can feel the cooling effect).
The OAT sensor is inside the heat pump enclosure, meaning it is affected but whatever the heat pump is up to, as above, but overall it correlates well enough with the OAT from a nearby Met Office WOW website OAT record most of the time. Of more concern is that it seems to censor OAT values below 0 degrees (see earlier plot of Set LWT vs OAT), which means the data for periods when the OAT is below zero aren't correct.
Midea 14kW (for now...) ASHP heating both building and DHW
Thinking about installing a heat pump but unsure where to start? Already have one but it’s not performing as expected? Or are you locked in a frustrating dispute with an installer or manufacturer? We’re here to help.
