If you haven't already you might want to re-read my earlier post https://renewableheatinghub.co.uk/forums/renewable-heating-air-source-heap-pumps-ashps/should-i-or-shouldnt-i/paged/2#post-20960 which has a link to a worked example of spatial zoning and draws some analogies between this and on/off working.

The $64,000 question with on/off working is, where is the sweet spot (ie for how long does the 'off' period need to be to reduce the energy consumed?).  Conventional wisdom is that it is >12 hrs hence the 'keep it on 24/7' mantra.  But quite a lot of 'conventional wisdom' in respect of heat pumps is questionable.  I have never seen any quantitative figures for this and IMHO it really needs a model or some experiments under controlled conditions (more or less impossible in a house - needs a lab) if only to get a feel for it.

Posted by: @mike-h

↑

@derek-m I take your point that adding more variables will make the resulting model much more complicated. However, the purpose of my simpler model was to suggest that there may be indeed a very good scientific explanation to support running your heat pump 24/7. Up until now, I have been far from convinced. Now I am not so sure.

 

 

From the tests and observations I have performed, much depends on the climatic conditions, the thermal mass of your home, personal preferences along with lifestyle.

In the previous example, if the heat pump is switched off for 1 hour the potential energy loss from the home will be 6 kWh. The reduction in indoor temperature will be dependent upon the thermal mass and the rate of heat loss. If the property has a large thermal mass then the reduction in indoor temperature may be quite small, but if the property has a small thermal mass then the reduction of indoor temperature will be more noticeable.

To bring the indoor temperature back up to the original desired level, will require the 6 kW of thermal energy to be replaced, which will be true whether the thermal mass is large or small. A large thermal mass means the indoor temperature will not have fallen too much, but will still require 6 kW of thermal energy to bring the indoor temperature back up.

With a fossil fuel boiler it would just be necessary to run it at high output for a period of time to recover the lost heat, with marginal effect to its efficiency. You could of course do the same with a heat pump, but its efficiency would be seriously reduced.

If you are happy to have say a 1C reduction in indoor temperature overnight, then one way could possibly be to lower the temperature setting mid evening, say 8pm. The heat pump may switch off for a number of hours as the indoor temperature gradually falls. The heat pump then may need to switch back on in the early hours of the morning, to have sufficient time to increase the indoor temperature back to the required level. Unfortunately this would probably require the heat pump to operate during the cooler part of the day, when its efficiency will be reduced.

An alternative to consider, particularly for those who are not at home during the day, would be to have a small A2A ASHP, to warm the areas used during the morning prior to going out for the day. An A2A heat heats the air directly, so can quickly heat a room by several degrees in a short period of time. It would then allow a A2W heat pump to be run gradually during the warmer part of the day, so that the home is up to the desired temperature by the time the occupants return.

I suppose the same method could be used where occupants are present all day.

Posted by: @jamespa

↑

it really needs a model or some experiments under controlled conditions (more or less impossible in a house - needs a lab) if only to get a feel for it

This raises, at least for me, some interesting questions. I don't get the modern obsession with modelling. When I was growing up, models were something you played with, not something you worked with. Models never are the real thing, that's why they are called models, and, far worse, when the model is used to make predictions, it can get things very wrong indeed. We see this very regularly with weather forecasts, no model has yet been devised that can predict what the financial markets will do, and when quasi-epidemiologists start modelling outbreaks of infectious diseases things can go diabolically wrong. During the recent pandemic, certain doctors who should have known better, found themselves unable to produce any medical evidence that showed that masks work, and in desperation - because they believed masks should work - turned to ever more bizarre models that apparently were supposed to convince us that masks worked. It was all stuff and nonsense, all here's the answer, now where is the evidence, all bogus baloney bollox.

In medicine, it is well known that what is shown in the lab may have little bearing on what happens in the real world. The problem is that the former environment is controlled, where as in the real world things are usually anything but controlled. Other factors are also at play eg the Hawthorne effect (people 'perform' better when an interest is taken in them), and behind all this is the paramount importance of asking the right questions, for it is always better to get a good enough answer to the right question than a perfectly correct answer to the wrong question. 

My trains carrying heat loads and leaky baths aren't models, they are analogies (ideas) that I use to understand the processes that are going on. Because I can easily understand how a faster train can deliver more goods, I can transfer that idea to a higher flow rate delivering more heat energy, ditto if the train off loads more goods (bigger delta t). That said, I am delighted if people want to extend them, and even put some numbers on them, but I don't think that turns them into models, it is more a refinement of the analogy.

I'm inclined to think the way we will get the best answer to the timed or 24/7 heating question is by using real world data, ie trying both and see what actually happens. Sure, it won't be a randomised controlled trial, but it can be done as a good observational study, with important variables collected in real time. Even the data I already routinely collect is more than good enough to do a relative comparison (it can't be absolute, because I don't know how accurate the Midea variables are, but if I use the same data collection system, then I can get a relative comparison) between timed and 24/7 heating. I already know that if I just use a standard weather compensation curve, two things happen: (a) I do definitely use less energy (and so save money) but (b) the standard weather curve doesn't use the internal temp, only the outside temp, and so doesn't bump up the LWT during the recovery phase, and as a result the house takes forever to warm up again. I just need to find a way of getting a tolerably quick recovery phase.

The answer for me (with a Midea system that I can control though modbus) is to tweak the LWT according to the difference between the desired indoor temperature and the actual indoor temperature, for example, if desired temp minus actual temp is greater than two, then new LWT equals current LWT plus two. I have yet to work out the exact implementation details, but I am sure you get the general idea. The question to the timed of 24/7 question can then simply be seen as which is greater: the energy saved during the setback, of the extra energy needed to get the higher LWT during the recovery period?

Note that all the above is done without a single model in sight. It instead involves routine data collection, some basic sums, and a simple control routine.     

I believe Ecodan's may already have a tweakable LWT capability built in called something like auto adaption, and I'm pretty sure Homely does much the same thing, meaning the idea is available for a number of other heat pumps brands.

Like @derek-m, I also suspect other elements of the weather (sun causing warming, rain/wind causing chilling) are significant, and there is no reason why I can't record these as well using a near to me Met Office WOW weather station, and record them alongside outside ambient temperature.  

I fully accept that my results aren't transferable to other buildings, but they will be a good enough answer for me, and might well apply to similar buildings. More generally, if broadly similar buildings get broadly similar results, the the conclusions drawn will very likely be good enough, though not perfect. The problem with trying to get a universal answer that covers all possible settings and conditions is that it is the ultimate Sisyphean task. 

Posted by: @jamespa

↑

Posted by: @jamespa

↑

it really needs a model or some experiments under controlled conditions (more or less impossible in a house - needs a lab) if only to get a feel for it

 

 

This raises, at least for me, some interesting questions. I don't get the modern obsession with modelling. When I was growing up, models were something you played with, not something you worked with. Models never are the real thing, that's why they are called models, and, far worse, when the model is used to make predictions, it can get things very wrong indeed.

Personally I both play with models (railways) and work with them.

Experiment and theory complement each other. and modelling is not a modern obsession, the ancient Greeks were creating mathematical theorems which are, after all, nothing more than a model.  Newton modelled the solar system in mathematics, and tested it against the observations.  It was almost right, good enough for hundreds of years and still good enough for most practical purposes.  And weather forecasting, if you read the forecasts correctly and in particular look at the maps, is remarkably accurate given that we live in one of the less stable parts of the world from a weather perspective.

Theory helps us understand trends, key factors and yes, make predictions.  But without experimental data to show its correct its valueless.  Experiments confirm (or deny) theories, but there are only so many you can do and its not always practical to determine experimentally something one wants to know (like, in general terms, how is this likely to perform in my house).

The problem with experiments in this area is that we cannot generally control the conditions, unless there is a lab set up.  That limits how and when we can actually test some things.  On the other hand heat pumps are remarkably well documented so an element of theory can get us results that we would struggle to get with experimental information, because separating out the variables is almost impossible (again, unless you have a lab).

So I am firmly of the opinion that a bit of both are necessary and that neither alone will do. 

Obviously everyone is entitled to their own opinion, but I doubt we will make much further progress technologically without continuing to use models to complement experiments.

Posted by: @jamespa

↑

modelling is not a modern obsession, the ancient Greeks were creating mathematical theorems which are, after all, nothing more than a model.

I need to formulate (maybe even model?) what are currently rather vague thoughts about terminology, but what I am getting at is the idea that in days gone by we would, for example, fit a regression line, now most people would call that modelling, but I would still call it fitting a regression line (and would know that while interpolation is usually fine, extrapolation is usually not fine). I'm bothered that 'everything' is 'modelled' these days. Perhaps I worry that it is just sloppy language, using a umbrella term rather than a more precise one, in keeping with the general degradation of both written and spoken English.

As it happens, I am currently doing a study on just how accurate (or inaccurate) shipping forecasts are (more specifically, winds in the Inshore Waters Forecast). I'm probably modelling either the forecast, or the weather, or both, or maybe even modelling one against the other! The Met Office tried to do it about a decade ago, and basically gave up - too difficult to do - and haven't as far as I know published anything since (I don't count a Met Office spokesperson telling how wonderful their forecasts are). Don't hold your breath for the results, though, it is hard to do, as the Met Office found out.     

Posted by @derek-m

In the previous example, if the heat pump is switched off for 1 hour the potential energy loss from the home will be 6 kWh. The reduction in indoor temperature will be dependent upon the thermal mass and the rate of heat loss. If the property has a large thermal mass then the reduction in indoor temperature may be quite small, but if the property has a small thermal mass then the reduction of indoor temperature will be more noticeable.

To bring the indoor temperature back up to the original desired level, will require the 6 kW of thermal energy to be replaced, which will be true whether the thermal mass is large or small. A large thermal mass means the indoor temperature will not have fallen too much, but will still require 6 kW of thermal energy to bring the indoor temperature back up.

I don't think the statement I have highlighted in bold is correct. If the thermal mass is smaller, then the indoor temperature will fall to a much greater degree than if the thermal mass is larger. Therefore the heat loss will be smaller due to the lower temperature gradient between indoors and outdoors. My spreadsheet 'analogy' suggests that a smaller thermal mass is much quicker and more economical to reheat back to the original indoor temperature than a larger thermal mass - if the bath contains 50 L instead of 200L the volume of water required to restore the depth to 20cm after 5 hours with no inflow is considerably less. If the heat pump is only off for an hour, the larger versus smaller thermal mass effect will not be as noticeable, but becomes very significant for longer periods with the heat pump off.

Posted by: @mike-h

↑

 

 

 

In the previous example, if the heat pump is switched off for 1 hour the potential energy loss from the home will be 6 kWh. The reduction in indoor temperature will be dependent upon the thermal mass and the rate of heat loss. If the property has a large thermal mass then the reduction in indoor temperature may be quite small, but if the property has a small thermal mass then the reduction of indoor temperature will be more noticeable.

To bring the indoor temperature back up to the original desired level, will require the 6 kW of thermal energy to be replaced, which will be true whether the thermal mass is large or small. A large thermal mass means the indoor temperature will not have fallen too much, but will still require 6 kW of thermal energy to bring the indoor temperature back up.

I don't think the statement I have highlighted in bold is correct. If the thermal mass is smaller, then the indoor temperature will fall to a much greater degree than if the thermal mass is larger. Therefore the heat loss will be smaller due to the lower temperature gradient between indoors and outdoors. My spreadsheet 'analogy' suggests that a smaller thermal mass is much quicker and more economical to reheat back to the original indoor temperature than a larger thermal mass - if the bath contains 50 L instead of 200L the volume of water required to restore the depth to 20cm after 5 hours with no inflow is considerably less. If the heat pump is only off for an hour, the larger versus smaller thermal mass effect will not be as noticeable, but becomes very significant for longer periods with the heat pump off.

You are indeed correct, I should have said approximately the same.

As you correctly point out, as the indoor temperature reduces, so does the heat loss, therefore a home with a smaller thermal mass would lose less energy over a longer time period, and therefore would require less thermal energy input to bring the indoor temperature back up.

@cookie197 Welcome to the forum!

Reading about your decision dilemma I have been through a similar experience over the last 12 months. To echo the point of doing your research and getting familiar with the technology is of great help in the decision making. I spent practically 10 - 12 months learning and looking at different technology options that fitted my requirements and goals.

I live in rural Cambridgeshire and my journey started with an existing wood burner, high efficiency system oil boiler (6 years old) with DHW storage and up to date modern standard home insulation (which reducied my running costs by 35% when installed).

The one lesson I have learn't on my journey so far is that no one solution fits all. After careful thought I decided that reducing my carbon footprint as much as possible but without increasing my running costs was the key goals for me, with a longer term goal of eventually (if electricity prices allow) decommissioning the oil boiler, however your goals will be different.

Based on those goals I looked at the grant on offer here in England and decided against it due to a number of reasons, mainly being that it was too restrictive and did not allow me to have the system that was right for me and I would be happy with. The main requirement of the grant was my oil boiler had to be decommissioned and removed, which was not an option for me.

With the bulk purchasing process of oil and the volatility of energy markets at present, for my goals to be acheived I needed a flexible system that allowed me to choose the fuel that provided the cheapest running cost for the present weather condition.

To acheive this I have had a 12Kw Samsung High Temperature heat pump installed in parallel with the boiler back in June 2023, will be replacing my hot water cylinder with one that is compatible with HP's in October (Does not require such a high flow temperature to acheive the same cylinder temperature meaning better HP efficiency) and changed my electricity tariff to Octopus Go. To help reduce our overall household carbon footprint I persuaded my Wife to go electric for her next car (to help offset the oil boiler emissions).

My experience so far with the HP for creating DHW is it is cheaper to run (even with my old cylinder) compared to using the oil boiler. This is based on an average 1Kwh electricity consumption twice a day. Once in the early hours @9.5p and once in the afternoon @31p. Empirical testing with the oil boiler showed it used a litre a day (I have an oil flow meter) @63p / litre (when I filled up). If I was on a standard tariff I would break even on cost for this time of year (August / September). At the time of writing oil is now 75p / litre in our area so even on a standard tariff my HP would be cheaper to run. As the weather gets colder however this will change and I have the option to switch over to the oil boiler at that point to keep costs down.

An additional reason for replacing my DWH tank is my solar hot water system has come to the end of its life. For my particular situation (as mentioned one solution does not fit all cases) the solar system saved me approx £110 a year and was effective 6 months of the year. A home battery, that I can buy cheap overnight electricity 365 days a year and use during the day (and drive the HP in the summer for DHW) will potentially save me £560 a year, so that is the next item on my shopping list instead of replacing the solar system. For other people a solar system may work very well and they get the full benefit but for our household and lifestyle sadly it does not have as much impact on the finances as a battery (without solar panels).

When the heating season starts it will be interesting to see how the HP performs against the theoretical figures (slight adjustments may be needed to match the HP performance to the actual heat loss characteristics of the house) which indicate the HP should be cheaper to run until the outside ambient temp drops to 8C then it will be cheaper on oil. However if electricity comes down in price, that break even temperature of 8C will get lower.

To conclude I would definitely support the advice of getting familiar with the technology (to help in decision making and eliminating the cowboy installers) but also decide on your own goals to help shape the system that is right for you and works with your lifestyle. This will require a holistic approach as there are a number or variables that can effect your end result.

My examples above are what I have considered and designed for our household but yours will no doubt be very different.

If the type of system offered by the grant people fits to your goals and needs then happy days its a no brainer but if not, you will never be happy and will look at the technology with a negative bias. Knowledge and goals are key!

Good luck

Posted by: @cathoderay

↑

As it happens, I am currently doing a study on just how accurate (or inaccurate) shipping forecasts are (more specifically, winds in the Inshore Waters Forecast). I'm probably modelling either the forecast, or the weather, or both, or maybe even modelling one against the other! The Met Office tried to do it about a decade ago, and basically gave up - too difficult to do - and haven't as far as I know published anything since (I don't count a Met Office spokesperson telling how wonderful their forecasts are). Don't hold your breath for the results, though, it is hard to do, as the Met Office found out.     

I find that with weather forecasts you need to be careful how you 'read' them.  For example a '30% probability of rain' could be because you are edge of a front, and the uncertainty in precise trajectory means that it may or may not hit you.  So you may or may not get rain.  Equally it could be because there are scattered patches of rain passing through, in which case more than likely one or more will hit you.  So you will get rain, its the timing that's uncertain.  The weather map tells you this in a way that a narrative cannot.  I did once learn how to translate the spoken shipping forecast into a weather map, which serves the same purpose, although of course todays weather maps are much more detailed.

Posted by: @jamespa

↑

I did once learn how to translate the spoken shipping forecast into a weather map

That's going back quite some time! I used to do little sketches to achieve the same thing. 

Posted by: @jamespa

↑

For example a '30% probability of rain' could be...

This allows me to introduce another of my pet concepts, which I call the collapse of the probability function. I normally apply it to medical risk, but it applies to anything where the outcome is binary, ie you either have a heart attack or you don't, or it either rains where you are or it doesn't. Take a doctor's waiting room which just happens to have 10 patients in it who all just happen to have a 10% risk of having a fatal heart attack in the next year. You are now fairly sure one of them is going to peg out over the next 12 months, but you don't know which one, or when they will peg out. Medicine deals with this uncertainty by treating all of them, eg put them all on statins, even though nine of them don't need to be treated because they were not going to peg out anyway (I know its more complicate than this, I'm just keeping it simple to make the point), yet they may well get side effects from the treatment (ie all harm with no benefit). If that's not bad enough, then the whole shebang is a sort of Schrödinger's waiting room, where you have to wait 12 months before opening the waiting room door to find out which patient died.

30% probability of rain actually tells you very little about whether you will get rained on. It does mean you can say you are more likely to get rained on than someone placed somewhere the forecast is 5% probability of rain, and less likely to get rained on than someone placed where the forecast is 90% probability of rain, but you still don't know whether you will get rained on' let alone when. Indeed, a 30% probability is a particularly tiresome forecast: high enough chance not to ignore it, but on the balance of probabilities, you probably won't get rained on. To take the umbrella or not to take the umbrella...

The main problem with analysing the shipping forecast is that despite the many constraints, the actual wording is free form text. E-NE 3-4 is straight forward enough, but SW 4-5 becoming cyclonic for at time then veering NW 6 later in the west presents a whole load of possible variations for how to put that into something a computer can process. Getting actual weather data is relatively easy these days, the Met Office very much to their credit make the data available for thousands of weather reporting stations through there WOW website which can be downloaded. 

The Inshore Waters Forecast also allows another interesting comparison: yesterday's 0600 24-48 hour outlook with today's 0600 0-24 hour forecast. If the Met Office's super computer model can accurately forecast the weather up to two days ahead, then the two, the previous day's outlook and today's forecast, should usually be pretty much the same. But often they are not...

This forecasting uncertainty is why I am not in any hurry to incorporate forecast weather (which I believe Homely does) into my heat pump control logic. Too much pain for too little gain. Instead, I plan to use the difference between the indoor desired temp and actual temp to tweak the LWT, by changing the set points on the weather compensation curve. In simple terms, if the room is 2 degrees colder than it should be, then move each end of the weather compensation curve up one degree, or whatever. I might even go so far as to say that the desired vs actual room temp difference contains among other things a proxy net summary for the combined actual weather effects over the last few hours - eg a period of solar gain might be reflected in an actual room temp that is above the desired room temp - meaning there is no extra benefit to be gained from faffing about with predictive models. 

The key thing for me is always to remember that predictive modelling, however apparently sophisticated it is, is always at its heart nothing more that glorified whatiffery. Which brings me to I hope clearer thoughts on models. Yes, you could say Pythagoras' theorem is a 'model' that models the relationship in length between the sides of a triangle, but it is actually more a description of the relationship than a model, and therefore should not, I suggest, be called a model; instead, it is a description. Yes, Newton described the solar system in mathematics, and tested his theory against the observations, but that is (as you say) the scientific method, not modelling. My suggestion is that we reserve the term modelling for cases where a system is analysed and reduced to mathematical terms that are then used to make usually but not always routine predictions for management purposes, in distinction to using occasional observations to test whether a mathematical description of a system is correct. In a way, it's the direction of travel that is different: predictive modelling looks forward in time, to predict the future, scientific observations look backwards in time, to test whether a prior hypothesis/theory is correct. They may look similar, but, I suggest, they are not.         

@cathoderay Schrödinger's waiting room is a concept ive never thought of before 🤣 Counter-intuitively there is 35% none of the patients in the waiting room will have a fatal heart attack in the next year, but there are also the effects of non-fatal heart attacks to consider. Off piste I know but i cant resist a probability calculation!

@technogeek Hi, thank you for this - another very useful approach. I hadn't heard of anyone running oil and HP together, but I like the idea of having the flexibility to switch between them. Ours is a combi, so we don't currently have a water tank at all, and I don't know if it can be adapted to heat one, or somehow plumbed so that it doesn't need to. I'm diligently watching all the HeatGeek videos on YouTube etc, trying to learn as much as I can, but it's a steep learning curve for someone with little technical knowledge to start with! I suspect (but don't know for sure) that the Welsh Govt grant will also depend on us getting rid of the oil boiler, but in any case, I'm coming to the conclusion that the deal on offer is probably not the best for us. There are separate grants for solar panels, which was going to be my next thing, whether it's "deal or no deal", but your comments on the home battery are also definitely worth further investigation. We've been invited to see a complete HP system operating next week, which I hope will make things clearer in my head. In the end I think it will come down to cost, as some of the more complex systems I've seen are likely going to be way out of our price bracket, unless the Welsh govt. will support people going down the Design-It-Yourself route. There is a HeatGeek operator not far from us, and later on I'll ask them to come and advise us. The grants are available until March 2026, so I'm not going to rush into anything ... from what I read, the technology is developing all the time, along with opinions on making the best use of it. To be honest, my biggest concern is that our desire to go green, without sufficient understanding of the myriad variables, will lead us into exactly the situation you describe, and we'll end up with a system that doesn't work very well, costs a small fortune to run, and may even hamper the prospective sale of the property in a few years time when my daughter leaves Uni. Perhaps I'm just being too cautious, but I know the system we have can get us through the winter relatively cheaply, and it's a bit like "better the devil you know ..."

Thanks again