Posted by: @cathoderay

↑

Posted by: @jamespa

↑

I find this quite offensive.  All the analysis I have done has been published and none of it is particularly complicated.  The most recent is a very simple regression analysis of your data, and two graphs illustrating the output.  The underlying spreadsheet has already been shared.

No one likes being told they come across as being patronising and arrogant, all the more so if they really are that way. I wrote that earlier remark (knowing full well how it might be received) because I think there have been occasions in this thread where things have got close to that state of affairs, and it is not a pretty sight, not to mention it being a useless way to get the dumb to see the light.

If you read my post more slowly, you will notice that I moved from named individuals to more general comments. You do generally publish your methods as well as your results, which is very much to your credit. My comments were general comments about the toxic combination of black box science and arrogant 'scientists'.

And come off it, of course I 100% accept the laws of physics, and invoke no special magic in such things - such foolish delusions are merely in the minds of the readers. In medicine I am not so sure, maybe sometimes magic does happen there, but that will no doubt cause a terminal fuse to blow in @derek-m's mind, so I won't explore that one any further.    

To avoid any further confusion, since I appear to be the target of numerous personal attacks.

I will categorically state that "I have no problem with CathodeRay operating his heating system in whatever manner he thinks appropriate".

@jamespa - I have no doubt at all in my mind that you are a fair and good-natured person, based on what I have seen on this forum. You have also genuinely helped me and I am sure many others understand what is going on. Nor do I mind robust debate. My comments as such are about the more general tone, perhaps at times as perceived by others.

Sadly my own profession is more than capable of Olympic grade arrogance, and no one bar none likes being told they are wrong than a doctor, who all too often is wrong. It is more than possible that in the great scheme of things we have the gold medal for arrogant ignorance; even the blow back from a well-meant but misguided attempt to do good can be painful to behold. Unfortunately, it is also the case that an appreciation of these things tends to come later in life, after decades of activity unaware of such things. Perhaps they need to teach more humility at medical school!  

@cathoderay thank you for the comments above.  We all need to be mindful of how others perceive (or may perceive) us.

To get back to the topic, I genuinely think we may be onto something with a plot of adjusted energy vs degree-minutes.  In one graph one can see

• how good or bad the correlation is
• what savings or losses setback is likely to give
• under what conditions savings are likely (or losses)
• whether the conditions where savings occur happen frequently or rarely (the frequency of the dots, whatever the colour)

Furthermore no specific experimental technique is required other than collecting the data and recording when setback was operational and when it was not (which could of course be deduced from the data given a bit of effort with spreadsheet formulae).

It needs to be replotted as energy in not energy out to give cost savings, but that's an easy task which I will do for the data I have.  I want to take the time though to tidy up the spreadsheet, so that the same technique can easily be applied again and to data (if any) from others, so it will be a few days before I post it.

So I would encourage you (and others) to collect more data, both with and without setback, so we can get some results which pass the test of statistical significance.

Here is a reminder of where we are currently at with the experimental data, with the health warning (for anyone who comes to this thread and does not read all that precedes) that the difference between the two curves, although apparently obvious, does not yet pass a test for statistical significance so this is NOT a conclusion, just an intermediate result.

Posted by: @jamespa

↑

I genuinely think we may be onto something with a plot of adjusted energy vs degree-minutes

I agree, and I hope you will agree that this is what beneficial collaboration can achieve, which I think it is fair to say is still happening here, despite the background noise (said to remind all that there is good as well as bad in this thread).

I am still collecting data as before, and will switch to setback mode from time to time. Now that your spreadsheet has matured (or will soon do so) to the point where it can be used more generally, on data from other heat pumps, it might be helpful to say what the data needs to include, ie the minimum data set. I think it needs to have:

Minute by minute data (possibly the hardest condition to meet for many?) 

OAT (outside air temperature, whether from heat pump or another sensor) and IAT (indoor air (room) temperature)

Assuming conversion to energy in rather than energy out, kWh in, either from an external kWh meter, or from the heat pumps internal monitoring

I think those are the only things that are necessary at the minute level. 

There also needs to be some way of characterising the thermal mass of the building - perhaps a brief textual description.

I may well have missed something out, if so, I am sure @jamespa will make the necessary additions (or subtractions). 

Posted by: @cathoderay

↑

Posted by: @jamespa

↑

I genuinely think we may be onto something with a plot of adjusted energy vs degree-minutes

I agree, and I hope you will agree that this is what beneficial collaboration can achieve, which I think it is fair to say is still happening here, despite the background noise (said to remind all that there is good as well as bad in this thread).

I am still collecting data as before, and will switch to setback mode from time to time. Now that your spreadsheet has matured (or will soon do so) to the point where it can be used more generally, on data from other heat pumps, it might be helpful to say what the data needs to include, ie the minimum data set. I think it needs to have:

Minute by minute data (possibly the hardest condition to meet for many?) 

OAT (outside air temperature, whether from heat pump or another sensor) and IAT (indoor air (room) temperature)

Assuming conversion to energy in rather than energy out, kWh in, either from an external kWh meter, or from the heat pumps internal monitoring

I think those are the only things that are necessary at the minute level. 

There also needs to be some way of characterising the thermal mass of the building - perhaps a brief textual description.

I may well have missed something out, if so, I am sure @jamespa will make the necessary additions (or subtractions). 

 

Funnily enough Id just posted a list on a new topic started.  You more less have it, but there are a couple of other necessary factors

I need the following at intervals of no less frequently than 1 hour, preferably more frequently, but no more frequently than every minute:

• Flow Temp, Return Temp, Flow rate (or some other means of calculating the energy supplied to the emitters)
• IAT, OAT (with a resolution of a quarter of a degree or better)
• Date and Time
• A measure of energy supplied to the heat pump, either from the heat pumps own data (volts and amps , W or kWh) or a separate kWh meter
• A flag yes/no, 1/0 or whatever) to indicate whether the heating in any given time period is DHW or Space Heating

together with a very brief outline of the system including how it is operated (24x7, with setback etc). 

I don't actually need a way of characterising the thermal mass, but it might prove useful, eg a brief textural description of the building.

I need the time sequence data ideally for at least a couple of months where there were a variety of OATs

Posted by: @jamespa

↑

-- Attachment is not available --

There's really no meaningful difference between those two lines at all. People can set their setbacks if they want and not worry about it, or people can feel warm and cosy all day and night and not worry about it. Everyone wins. Yay!

EDIT: I was trying to quote @JamesPa's graph that showed blue and orange dots that more or less fell on top of each other, and two lines that were more or less the same line but someone nudged the ruler a bit, but it didn't show up. Hopefully my completely unbiased description is enough for you guys to know what graph I'm talking about!

@derek-m, @cathoderay

OK, this is where it gets complicated!  

From the data provided by @cathoderay I can compare the energy delivered to the house with an without setback, and correct these for the energy that is stored from or released to the fabric.  So far the experimental evidence suggests that there may be an advantage (order 5-max 10% at the rare extremes) in implementing setback when its cold, but when its mild there may be a disadvantage of the same order, 'advantage' being defined as energy delivered to the emitters. 

The theoretical calculations are not dissimilar.

But of course what we really want to know is, is there an advantage in terms of energy consumed from the grid (ie delivered to the heat pump)?   I started to extract this from the data, but realised that this is where it gets complex.  Basically, the question is - what correction should I apply for the inevitable fact that, at least for some of the recovery period, the energy consumed is artificially low because the house is cold and that is not a sustainable position?

My initial thought was to apply a proportional correction on a daily basis, ie, to correct the energy consumed by the same factor as the energy delivered is corrected.  However this is subject to the criticism that, if I do apply setback, its likely (in the case of a 12 hr setback, less so in the case of a 3-6hr setback) that the recovery period  will occur during the day, when the COP is higher.  So a daily proportionate correction wont quite represent reality.  But what is reality, and more to the point what is a realistic approximation to (model of) reality?

Is the answer this: at the current time the experimentally verified difference in energy delivered with and without setback is sufficiently small that trying to hypothesise what might have happened (in terms of energy supplied) if setback was not enabled is a fruitless exercise, even more so because the 'pure' modelling results also suggest a relatively modest saving if any. 

Incidentally the experimental (tentative) result (based on one system and not yet statistically significant) that the advantage of setback is marginal (in terms of energy delivered to the house) applies, without further correction, to fossil fuel systems also.  It makes you think - why we bother with all these complex controls.  I guess the answer is - because capitalism (and consumerism) drive saleable innovation irrespective of its actual utility, the key being saleable not utility!

Surely, with a bit more experimentation and modelling, there is a paper in this!

Comments please

@jamespa

I'm afraid that I have no comment to make.

@jamespa Would you kindly but briefly explain the term "degree minutes per day" and how they relate to heat pump running costs. Also why no mention in your calcs of the potential savings by not running the circulating pump(s) during the set back periods?

Posted by: @jamespa

↑

My initial thought was to apply a proportional correction on a daily basis, ie, to correct the energy consumed by the same factor as the energy delivered is corrected.  However this is subject to the criticism that, if I do apply setback, its likely (in the case of a 12 hr setback, less so in the case of a 3-6hr setback) that the recovery period  will occur during the day, when the COP is higher.  So a daily proportionate correction wont quite represent reality.

James, can you remind me what is your definition of setback, in terms of temperature? What are you defining as target daytime room temperature and what is the number of degrees below this temperature is your setback? Or are you not targeting room temperature at all?

if you are using weather compensation how are you operating a setback at night? I don’t control Weather compensation since I am using Auto Adaptation which reacts to set room temperature. 

@jamespa I think it is better to measure energy consumed of course, but then you also need to take into account things like Time of Use tariffs. For example, I actually increase the flow temperature during the 6 hour cheap rate period, using the house as a heat battery. I don't think this saves me much money quite frankly, as the higher indoor temperature leads to greater heat loss at a time when the OAT is low.

But in any event, if you have a ToU tariff, having an overnight setback will save you less money than if you don't.

I agree, however, with your general conclusion, that there is no difference between setting or not setting a setback, and that the above should be left as an exercise to the reader.

Posted by: @iaack

↑

@jamespa Would you kindly but briefly explain the term "degree minutes per day" and how they relate to heat pump running costs. Also why no mention in your calcs of the potential savings by not running the circulating pump(s) during the set back periods?

Yes of course.  I will answer the second question first, because its a shorter answer, then I will move onto the first.

There is no mention of savings due to not running circulating pumps because

a) their energy consumption doesn't feature in any of the experimental data or the model which are being discussed and 

b) in almost, if not all, all situations of interest, the energy consumed/power needed by a circulating pump is negligible in comparison with the energy required to heat the house. 

Circulating pumps are typically a few tens of watts whereas heating a house is typically of the order of a kilowatt or two.  In any situation where the energy use of a circulating pump is significant in comparison with the heating energy, the latter is so low that we really don't need to bother too much with the discussion about setbacks, as the house will be largely heated by the waste heat from electronic devices and the people living in it (ie its a passive house ++)  

Degree-minutes - Simple explanation

Degree minutes are just a finer grain variant of degree days (see https://www.degreedays.net).   Basically its temperature difference between indoor and outdoor * time.  The energy lost from the house is, with the exception of some relatively small perturbations which I discuss below, directly proportional to this  Thus the amount of energy that you need to supply to the house to maintain its temperature is also (with some small perturbations) directly proportional to this.  The graphs presented show that the theory is supported, actually remarkably well, by the experimental data.  Exactly the same concept is used when installers calculate an expected annual energy consumption.

Degree-minutes/degree-days are a handy way to enable energy demand on different days, with different temperature profiles, to be compared on a like for like basis.  Without this concept one requires two identical days to make comparisons, which only happens in a lab. The data that @cathoderay has donated is a minute intervals, so in the calculations I get to degree minutes per day by simply summing the values of (IAT-OAT) for a days worth of data.

For fossil fuel boilers its relatively straightforward to get from degree-days/degree minutes per day to cost of running, because the energy consumed by the boiler is directly related to the energy supplied to the house.  For a heat pump its more complicated, because the energy consumed is (energy supplied to the house)/COP.  So the concept has to be used with greater care when applied to heat pumps, which we are being careful to do.  Its nevertheless still valuable as the correlation curves demonstrate.  

Degree-minutes - More complex discussion

In principle the heat loss from the house is proportional to (IAT-OAT)*time.  There are stated some perturbations to this such as wind (which will increase the energy loss), possibly rain, and certainly solar gain.  These will cause small apparently random departures from the linear relationship for which its not easily possible to compensate (and for which I have not attempted to compensate because I don't have the data.  Heat supplied to the house by appliances other than the heating system, and by humans, will also cause small perturbations for which we cant readily correct (but are likely to be fairly constant, so not of particular concern) However they wont materially change the nature of the relationship namely linear.

Also affecting the energy demand is the house fabric itself.  If the IAT changes because of setback or because the control system isn't 'perfect', then the fabric either gives up energy (if the IAT falls) or requires energy to heat it up (if the IAT rises).  This is in addition to the energy lost from the house to the outside world.  The house behaves as (in fact is) an energy store.  This is sometimes called 'thermal mass' and is a quite significant effect, a house without thermal mass would respond instantly to the heating being turned on or off, which we all know is not the case. 

Unless the house remains at the same temperature (or any calculations compare two points in time when the house is at the same temperature), this is a further perturbation to the relationship between degree minutes and energy demanded by the house.   However in this case we know that in the long term the perturbations must sum to zero, because eventually the house will be returned to its design indoor temperature (that's what the heating is for!).  So if we are trying to work out whether setback is saving money (or indeed work out anything else which requires us to know and compare energy demand/consumption - which is most things of interest!), we either need to look at (and between) two points in time where the house is at the same temperature, or correct for the energy stored in/released from the fabric.  Doing the latter potentially gives us more experimental opportunities, so its worth considering.  Not doing it leads to false conclusions, as demonstrated earlier in this thread.

We can work out the correction factor by looking at the correlation between degree-minutes and energy supplied to the house and doing a best fit.  We can also work it out by measuring the rate at which the house cools when the heating is switched off.  If the two ways of working out the same variable come to a broadly similar result, we can have more confidence that its roughly right.

The indoor temperature is the most important factor affecting the energy stored in the fabric (as opposed to lost from the fabric) of a tolerably well insulated house.  However the outdoor temperature also has an effect.  There is of course a temperature gradient between inside and outside and changes in the OAT affect this gradient.  Mostly the gradient is in the insulation, which also has a low thermal mass, which is why OAT is less important when calculating the energy stored.  However the factor can be calculated and applied as a further adjustment.  Again we know that, over time, the adjustments sum to zero because OAT eventually returns to whatever value it started at.

end of more complex discussion

Hopefully that helps explain the concepts.  Scroll upwards for a summary of where we are at in answering the question posed in the thread, both by experiment and by modelling.  Degree-minutes are mostly relevant to ensuring a sound interpretation of experimental results, but may help the modelling eventually also.