An answer appears to be yes. I say 'an answer', rather than 'the answer', because this is one night in one building, with the weather conditions that prevailed at the time. Other buildings and/or other prevailing weather conditions may - very probably will - give different results. But here is a documented record showing that I saved around 5-6 kWhs last night by applying a 26 degree to 12 degree setback between the hours of 2100 and 0300. Result (chart shows the last 48 hours, the first 24 hour no setback period for comparison with the second 24 hour setback period; I have also done the long overdue conversion of the energy in/out plots to bars rather than lines, with each bar showing the preceding hour's energy in/out value; note the bars are overlaid not stacked, ie each bar starts at zero):
There is zero energy use for heating during the setback. As the current normal energy use is roughly 1 kWh per hour, that means I saved 6 kWh of energy. Now comes the 64 million dollar question: how much extra energy did I use during the recovery (which happened fairly promptly, the room temp was back to the desired room temp of 19 degrees by 0700, having fallen to a nadir of 18 degrees during the setback, not a lot I agree, a consequence of a large thermal mass and mild but by no means unusual temperatures)? I emphasise extra energy because in continuous running (no setback), the heat pump would, in the prevailing conditions, use about 1 kWh per hour, what I want to know is how much extra energy I used during the recovery period.
The answer is, in the prevailing conditions, very little. The recovery boost is managed by my python auto-adaption script (see below for details), which increases the weather compensation curve endpoints (and so the LWT) when the room temp is below where it should be, as it was at the end of the setback. If you look at the first hour after the setback, the energy out (red bar) is visibly higher (as is the LWT in the upper chart) but the increase in energy in is minimal, 1.24 kWh for that hour, compared to 0.84 kWh in the hour before the setback, and 1.1 kWh in the second hour after the setback. That gives around an extra 0.24 kWh used, over and above the around 1 kWh I would have used anyway. I therefore saved 6 - 0.24 = 5.76 kWh last night by using a setback, and comfort was not compromised, the room temp was where it should be when I got up at 0700 this morning (and yes it was a bit cooler in the middle of the night, I know because I have to do what most middle aged and over men have to do during the night).
A result, I hope you will agree. Some things to bear in mind.
1. As noted above, and emphasised again here, this is one building over one night in the weather prevailing at the time. I will add more data once I have it, and it has something to show.
2. I call it a setback, because it is a setback of the room stat from 26 to 12 degrees, but its effect is to act as an on/off timer, because in normal circs the house will never fall to 12 degrees, meaning the heat pump will never come on during the setback. A heat pump on/off timer could used to do the same thing.
3. The parameter monitoring is pretty tight, data collected over modbus at one minute intervals, meaning we can really see what is going on. All the data bar the room temperature comes from the Midea controller, and so far as I have been able to verify its accuracy, it is not far out. The room temp (IAT) comes from a third party sensor in centre of the living area of the house.
4. I am currently running my DIY auto-adaption. This uses a python script that checks the difference between the actual room temp and desired room temp every hour and adjusts the end points of the weather compensation curve according to the difference. The current logic moves the ends of the curve one degree up or down from the baseline endpoints for every degree the actual room temp is below of above the desired room temp respectively, eg if the room temp is one degree below the desired room temp, the ends of the weather compensation curve go up by one degree. The maximum increase in the endpoints is limited to a three degree rise, to stop the heat pump melting itself down. One effect of this auto-adaption is that it can boost the output when the room temp is below where it should be, eg at the end of a setback period, and provide a so called setback recovery boost, meaning the heat pump doesn't take forever to get the room temp back to where it should be. The auto-adaption can also reduce output eg if there is solar gain, increase it if there is significant wind chill, for example from draughts from leaky window frames.
This topic was modified 6 months ago by cathodeRay
Midea 14kW (for now...) ASHP heating both building and DHW
The last 24 hours data, showing another setback and recovery, with minimal extra energy in during the recovery boost. This time the actual room temp fell just under two degrees below the desired room temp during the setback, and the recovery boost took a bit longer to get back to the desired room temp, currently 18.9 degrees (vs 19 desired), but this slight delay is probably because the outside air temp fell a bit around 0700 this morning. Overall energy saving much the same as the night before last, around 5.8 kWh:
Midea 14kW (for now...) ASHP heating both building and DHW
Useful data! Its interesting that it (apparently) takes 3 hours for the RWT/LWT to fall to room temp. Do you think this is real and if so does the pump keep going and is it UFH or rads? Also did you tell us what the OAT was during the setback?
Its interesting that it (apparently) takes 3 hours for the RWT/LWT to fall to room temp. Do you think this is real and if so does the pump keep going and is it UFH or rads?
Thank you! I think the fall takes as long as it does because as the LWT/RWT fall, and they approach room temp, the delta t between them and the room temp declines, and so the heat loss from the circulating fluid declines, and the rate of LWT/RWT fall declines, if that makes sense. Bear in mind the LWT/RWT are for the primary circuit, not the secondary radiator circuit (all rads, no UFH). I don't know if the primary circuit pump was still running, but I guess it was, that is why the LWT/RWT still has a delta t. The secondary circuit pump should be off, as the heating is off. I will try to remember to have a look late this coming evening to see what the pumps are doing. I'm guessing the LWT/RWT end up below the IAT because the primary circuit is mostly in a non-heated space (attic) and has a short length outside from wall to the heat pump.
Interesting to see, thanks. I had started running my own experiments (using the limited data from my HP's (unimpressive) app), but have discovered a weird bug which has rather undermined my dataset!
@josephiah - thanks. Likewise, I found Midea's app to be hopeless, with dropped data, and no easy way to get the raw data for further analysis. I also rather suspected "O, what a tangled web we weave, When first we practise to deceive!" might apply ie Midea obfuscate things to the extent we don't really have a clue what is going on. That is why I went over to using modbus over a wired connection to get the raw data, which gets recorded every minute to a csv file. It is still Midea data, but it is - I hope - the raw data from the sensors eg I get amps and volts and use them to calculate the energy used, which I can then verify against my external heat pump kWh meter. So far, it has been very reliable, and because the raw data is in csv format, it is very easy to plot it, either in a standard spreadsheet, or using more sophisticated such as plotly in python, which is what I used for the above plots. Although the screen grabs posted here are static, the actual plots in my browser are interactive, eg I can zoom in and out, a very useful feature.
I also use the wired modbus connection to write to the Midea controller, specifically to set the ends of the weather compensation curve (WCC), to implement my so-called DIY auto-adaption. This compares the actual room temp to the desired room temp, and adjusts the WCC accordingly, ie up if the actual temp is too low, down if it is too high. meaning my heat pump output is now set not just by the outside air temp, but also the inside air temp discrepancy, when there is one. My hope is - it is still too early to say with any certainty - that this DIY auto-adaption will help the heat pump manage things like recovery from a setback (without the auto-adaption boost it takes forever - 12 plus hours - to recover), and over-heating from solar gain.
I was very nervous about doing the writes - I thought the wired controller might object to being told what to do by a couple of thin wires stuck somewhere where the sun don't shine - but managed to persuade myself that in reality I was merely doing over modbus what I could already do manually on the controller, set the ends of the WCC curve. I have deliberately steered clear of directly setting the LWT, instead, I set the WCC end points, and let the wired controller set the actual LWT. So far so good - apart from what appears to be a weird bug on my system. Over the last few days, the DHW timer settings have failed to stick, and as a result, I have had to trigger a cylinder reheat manually to get hot water.
Until the bug appeared, the DHW tank settings were: target temp 50 degrees, reheat if 10 degrees or more below target temp, and only run this between 1300 and 1400 daily. For months this has run flawlessly, but over the last few days it has been in a coma: come 1300 hours, with a tank temp below 40, which should trigger a tank reheat, nothing happens. I have checked, and rechecked all the relevant settings, and they are all definitely as they should be.
Initially, I thought the bug appeared when I started using auto-adaption, on 30th October, but looking at the logs it is not entirely clear when it first appeared, it may have been over the preceding weekend, when the clocks changed. Whatever the cause, it has left me, like you, wondering why seemingly clear setting fail to produce the desired outcome. Very frustrating!
I will post the last 24 hours data shortly, may be interesting because I am pretty sure the OAT dropped to defrost levels overnight, it will be interesting to see how defrost cycles affect things, should they occur.
Midea 14kW (for now...) ASHP heating both building and DHW
Here is the last 24 hours' data. The spike on the left hand side is a manually triggered DHW cycle, and on the right hand side we have the setback and recovery period:
The OAT has gone into defrost cycle territory over the last few hours, at 4 degrees, but no defrost cycle occurred (they can be identified by the fact the LWT drops below the RWT); instead, we have less cycling, with longer periods of steady state running. We can also see, from the popup box, that the actual IAT has recovered to within 0.2 degrees of the desired IAT by 0700, from a nadir of 17.2 at the end of the setback period. The heat pump does use more energy during the recovery period, but it would have done so anyway, because the OAT was lower. Before the setback, it was using just over 1 kWh per hour, after the setback it is around 1.7 kWh per hour - let's say around 0.6 kWh more per hour. Let's say half of that is because of the drop in OAT, ie it would have been used even without a setback. That leaves an extra 0.3 kWh extra energy used to speed up the recovery over a four hour recovery period, a total of 1.2 kWh extra energy used. But I saved at least 6kWh, probably nearer 7 kWh, during the setback, meaning I still have a useful overall energy saving of around 5 to 6 kWhs overall. So far so good...but it is still early days.
Midea 14kW (for now...) ASHP heating both building and DHW
Thank you @cathoderay, this is a very interesting read, not just because my system is not a hundred miles away from your setup, but also in the way that I use it. I have only radiators but all K3 or K2 and well sized and I set back over-night and rely upon the clever Mitsubishi auto adapt to manage the house temperature according to the scenarios that I have set. I am delighted with the result in terms of comfort and cost but can only measure it anecdotally or by the finger in the air while you are giving me real world data. Great job!
@abernyte - thanks. Here is the last 24 hours' data, also showing the previous 24 hours for comparison:
For some reason, today's recovery is slower, and still hasn't quite got back to the desired room temp, despite a warmer OAT this morning compared to yesterday. It may have something to do with the weather, wet and windy. Nonetheless, the extra energy in during the recovery is far less than that saved during the setback, meaning overall a net saving.
Midea 14kW (for now...) ASHP heating both building and DHW
@abernyte - thanks. Here is the last 24 hours' data, also showing the previous 24 hours for comparison:
For some reason, today's recovery is slower, and still hasn't quite got back to the desired room temp, despite a warmer OAT this morning compared to yesterday. It may have something to do with the weather, wet and windy. Nonetheless, the extra energy in during the recovery is far less than that saved during the setback, meaning overall a net saving.
I suspect the recovery is taking longer today because the IAT was lower, and because you are limiting the permitted LWT increase to 3C, this in turn is limiting the rate at which the lost energy can be replaced. If you raise the LWT limit the system will recover quicker, but of course will use more energy in the process.
This is where the homeowner will have to decide how low they are willing to allow the IAT to fall before it is no longer acceptable.
@derek-m - I agree, the IAT nadir was lower, but only by 0.5 degrees, and I think that is only part of the story, the other part being it is very wet and windy this morning, and that means wind chill, as much from air leaks - still a problem in this old leaky building - as from direct chilling. It will have the effect of increasing he number of air changes per hour, perhaps by quite a lot, though I can't put a number on it, as no practical way to measure it.
I set the initial 3C limit for both ends of the WCC because the left hand end of the WCC is already close to its limit (56 @ -4, with the max available being 60 degrees) and I didn't want the heat pump to be pushed to hard. But that is the left hand end, and at the moment we are somewhere near the middle of the curve. One option would be to increase the right hand end more that the left hand end, which will increase intermediate values without pushing up the extreme left hand end.
17.0 degrees is about where I start to feel a bit chilly. Going below that overnight is fine, as long as I am back to around 18 degrees or more by breakfast time. Looking at the energy use values, I still have a lot of 'credit' left over from the setback, meaning I can afford to burn up a little more during the recovery, and still remain in credit. I will tweak the code in the python auto-adapt script and see what happens.
Midea 14kW (for now...) ASHP heating both building and DHW
@derek-m - I agree, the IAT nadir was lower, but only by 0.5 degrees, and I think that is only part of the story, the other part being it is very wet and windy this morning, and that means wind chill, as much from air leaks - still a problem in this old leaky building - as from direct chilling. It will have the effect of increasing he number of air changes per hour, perhaps by quite a lot, though I can't put a number on it, as no practical way to measure it.
I set the initial 3C limit for both ends of the WCC because the left hand end of the WCC is already close to its limit (56 @ -4, with the max available being 60 degrees) and I didn't want the heat pump to be pushed to hard. But that is the left hand end, and at the moment we are somewhere near the middle of the curve. One option would be to increase the right hand end more that the left hand end, which will increase intermediate values without pushing up the extreme left hand end.
17.0 degrees is about where I start to feel a bit chilly. Going below that overnight is fine, as long as I am back to around 18 degrees or more by breakfast time. Looking at the energy use values, I still have a lot of 'credit' left over from the setback, meaning I can afford to burn up a little more during the recovery, and still remain in credit. I will tweak the code in the python auto-adapt script and see what happens.
What you suggest is fine whilst the weather is reasonably mild, although inclement. My concern would be when the temperatures get colder, the IAT would likely fall even lower, but the heat pump would not have the reserve to recover, probably until much later in the day.
At the moment your setback isn't actually a 'setback', but more a 'switch off', which takes no account of the prevailing weather conditions and IAT. It may be preferable to lower your room thermostat to a more appropriate value rather than 12C, thereby limiting how low the IAT can fall.
An alternative would be to add additional Python code, to monitor both OAT and IAT and decide to reduce the setback period and increase the recovery period if the weather conditions are more severe.
If you would like to provide a copy of some of your csv data, I can model the possible outcome under different weather conditions.
