Indeed, see my comment elsewhere (<=link). I particularly like the 'modbus hat for a raspberry pi', it's absolutely brilliant. I will endeavour to explain the jargon in this post, but have until now made a few assumption eg people know what LWT/RWT and COP mean. Perhaps I need to do a jargon translation post for absolute beginners.
Also relevant to this thread: I posted some charts earlier (<=link). The data for thse charts was extracted manually, a very tedious and boring process. The idea behind the monitoring system I have in mind is that the data for these charts and the charts themselves would all be collected and processed automatically, thus at a stroke putting an end to tedium and boredom.
Midea 14kW (for now...) ASHP heating both building and DHW
Indeed, see my comment elsewhere (<=link). I particularly like the 'modbus hat for a raspberry pi', it's absolutely brilliant. I will endeavour to explain the jargon in this post, but have until now made a few assumption eg people know what LWT/RWT and COP mean. Perhaps I need to do a jargon translation post for absolute beginners.
Also relevant to this thread: I posted some charts earlier (<=link). The data for thse charts was extracted manually, a very tedious and boring process. The idea behind the monitoring system I have in mind is that the data for these charts and the charts themselves would all be collected and processed automatically, thus at a stroke putting an end to tedium and boredom.
Not simply LWT/RWT and COP. The entire concept of PI, dashboards & modbus etc. Personally it's all beyond my comprehension. I doubt I'm alone in not being techy gifted.
Retrofitted 11.2kw Mitsubishi Ecodan to new radiators commissioned November 2021.
@morgan - one reason I called this post 'A Beginner's Guide' is to remind me to keep it as accessible as possible. I'll try to put together a glossary/jargon busting post. It might just help me to understand things a bit more as well!
As I said elsewhere, the Mrs Trellises of North Wales are never going to get it, or even want to get it (which is fine, horses for courses), but for those do want to get to grips with things, it should be made as pain free as possible. Something like:
LWT (leaving Water Temperature): the temperature of the water leaving your heat pump to heat your house/hot water. The hotter it is, the more heat energy it contains, all other things being equal, but it's greatest importance is in how it affects efficiency and COP (qv). Because of the way heat pumps work, the greater the LWT, the lower the heat pump's efficiency/COP. This leads to an obsession with having the LWT as low as possible that at the same time still provides enough heat.
COP (Coefficient of Performance): the standard measure of efficiency of a heat pump or indeed any heat producing device: how many units of energy do get out for each unit put in? It is an absolutely core concept. A standard electric mains heater has a coefficient of performance of 1, you put one unit in and you get 1 unit out. Heat pumps do better than this, sometimes a lot better, because they use the input units to extract heat energy from the surrounding air, so they can put out more units than they consume. A typical mid range COP might be 3, your heat pumps uses (consumes) one unit of energy, and puts out three units. The formula is simple: units out divided by units in. The units are usually kWh (on an electricity bill a unit is a kWh) so the formula is kWh out/kWh in. Three things to note: a COP can be determined over any time period, hour, day, week, month, year, forever, as long as you have the kWh in/kWh out for the period. Because COPs vary with ambient temperature, because heat pumps become less efficient at lower temperatures, it is conventional to use something called the SCOP (seasonal COP), in effect the average COP over a heating season, to compare heat pump efficiencies. The other thing to note is some manufacturers present the calculation in a slightly different way. It's the same calculation, but instead of using the better, simpler kWh out/kWh in, they use yield plus input divided by input, where yield is the extra energy yielded over and above what was put in. For example, instead of 3/1 = COP of 3, they use 2 (the yield) + 1 (the input) divided by 1 (the input) = COP of 3. This is obfuscation pure and simple, and should be abandoned. Thirdly, COP is sometimes expressed as percentage efficiency (300% is the same as a COP of 3), and is almost invariably expressed as a percentage when it is less that one, eg for fossil fuel boilers. 90% efficient sounds better than a COP of 0.9, but it means the same thing. This is how the heat pump industry gets to sell itself as amazingly efficient eg my heat pump is 300-400% efficient where as your top of the range stone age gas boiler is only 90% efficient.
Modbus: a protocol, in effect a language, that allows two parts of a system to talk to/communicate with each other. There are many variations (dialects), meaning translation between one language and another is often needed. The communication can be either way. It can be used to read (get) something, say a flow rate or temperature, or set something, eg set a desired temperature.
Raspberry pi: a mini computer about the size of a pack of cards (and probably just about as chancy). Often used by DIYers to provide the hardware on which to run small often dedicated (to their purpose) operating systems and programs that do the monitoring and controlling of heat pump systems.
Dashboard: a computerised equivalent of a car dashboard. It tells you how fast your heat pump is going, how many miles it has covered (kWh out), how much fuel you have used (kWh in), how hot your engine is (LWT) and numerous other useful things.
Others more expert than me, please feel free to correct any or all of the above, but remember the golden rule, this is a beginner's guide, so nothing too technical (or if it is, find a way to describe it in plain English).
@morgan - is the above accessible to you? Or am I still being too technical?
Midea 14kW (for now...) ASHP heating both building and DHW
One possibility I have considered is using an RC-4 type temperature data logger (around £25 on ebay), which is what I currently use to log hourly room temps in the kitchen, with an external probe attached to the primary flow in at my plate heat exchanger, and then compare that to the LWT reported on the Midea controller. If there is a consistent relationship, eg the temp at the PHE is always 95% of that showing on the Midea controller, then I can probably assume the Midea reported LWT is good enough. If there isn't a consistent relationship, then I might just shoot myself.
I would (and do) use a DS18B20 temperature probe (£2) taped to the pipe. With a bit of pipe insulation over the top the result is within a degree of the actual WT. (and as derek-m says, even if its a few degrees low, if its consistent then its fine)
As many probes as you like connect into an ESP32 board - outlay less than £5 from aliexpress as bare board and some basic DIY soldering (I learned to solder to do this, it'll win no prizes but it works) or £20-25 if you want uk suppliers with a built board. I have dozens of these probes - you can attach many to a single ESP board. I monitor temperature of boiler flow, boiler return, DHW cylinder top, middle, bottom, DHW mixer hot inlet, DHW mixer cold inlet , DHW mixer outlet, CH pump flow, CH Zone 1 flow, CH zone 2 flow.....
I think I am approaching the point where I am ready to start getting some hardware, and starting some monitoring, something simple, like room and outside temperature plus primary flow and return connections to the plate heat exchanger.
I have decided I am going to use Home Assistant (HA) as the main software behind the monitoring, on the grounds it seems the most supported/developed system (which isn't necessarily saying very much). HA can be run in various ways, but by most accounts it is best to run it as an operating system (OS, like Windows, MacOS, Android, Linux etc, but dedicated and adapted to run HA) in and of itself on a dedicated bit of hardware. The idea is you 'flash' a so called image of the HA OS (which is itself apparently based on Linux) onto the hardware, and Bob's your uncle, or rather HA is your uncle. I want to use wired connections wherever possible, and so I think my set up will end up something like this:
There is already some jargon in there, modbus RTU, RS-485, ESP32 etc, these are connection boards/protocols that I will come back to later. The first big question is what to use for the dedicated hardware, the main box in the middle of the diagram. It needs to be some sort of mini-computer, because it has to run the HA OS, and it will need ports (USB, ethernet etc) for the connections shown, storage for all the data HA collects and a power supply. The first problem is the the range of possible choices is almost over-whelming, it is where we enter the realm of Strawberry Flies Forever.
Possibilities include but almost certainly are not limited to:
a raspberry pi - these are ultra mini computers, and are currently in short supply world wide. They are also not cheap, and lack connections and storage => binned (Strawberry Flies Forever)
a true mini computer - there is a bewildering array of these, some are quite affordable especially second hand, and ports and storage are usually present out of the box => on shortlist
an old laptop, bought on ebay - affordable, will have ports and storage, and is the devil I know => on shortlist.
At the moment, of the two on the shortlist, I am leaning towards the old laptop. Not only is it the devil I know, it also will have its own screen and keyboard, which will make interacting with it much easier, compared to the mini computer, which will most of the time be a black box. I also know how to burn an image of an OS (downloaded form the HA website) onto a hard disk (I will go through the details when the time comes) so that step is taken care of.
The remaining two questions/concerns about using an old laptop are (a) power consumption: I need to find one that isn't power hungry, as it will be always on (turning off the screen for 99.9% of the time will help) and (b) what sort of storage? HDD or SSD? Given their proneness to sudden failure I am inclined to lean towards a good old fashioned HDD. Or maybe SSDs have improved now to the point where they do match HDDs for reliability.
Any thoughts or comments very welcome as ever.
Midea 14kW (for now...) ASHP heating both building and DHW
I used a couple of 0.1C accurate temperature sensors for flow and return - they’re pushed into thin copper tubes which are thoroughly soldered to the flow and return 22mm copper pipes, helping the sensors better measure the water temperature. The whole thing is insulated over the top - piccy shows before I insulated it all. There’s a few tubes, in case I wanted to add an extra sensor:
I used a couple of 0.1C accurate temperature sensors for flow and return - they’re pushed into thin copper tubes which are thoroughly soldered to the flow and return 22mm copper pipes, helping the sensors better measure the water temperature. The whole thing is insulated over the top - piccy shows before I insulated it all. There’s a few tubes, in case I wanted to add an extra sensor:
What type of sensors did you use and how did you read them?
The sensors are tsic506f, and they connect up to an mbed(old micro type I’m used to) that I programmed with a free Arm account. There’s a cheap flow sensor (2%), and an in line electrical meter (1%), that has an isolated output, so all these go straight into the mbed, and it can calculate the COP to 5% absolute accuracy. It works well, I don’t understand why COP measurements aren’t front and centre on every heatpump.
@cathoderay Excellent thread and great start, thank you.
Given the many different circumstances I would suggest anyone heading down this journey to ask themselves the following questions.
what outcome[s] they want to achieve
what type of journey they want to take
what their $$ budget is
what their time budget is
what precision and accuracy they are aiming for (part of the outcome but I called it out specifically). Accuracy is more important than precision, achieving accurate and precise data will likely be pretty expensive mostly because of the challenge of measuring flow rate, without which you really have no idea of how much heat has been "moved" into and out of the water.
After that analysis they should create a problem statement, strategy and guiding principles (similar to what you have done)
Accuracy vs Precision
So you can now chose from "turn key" e.g. homely, to gold plated part diy part "turn key" where all the hard stuff is done by really smart guys such as emonHP to relatively easy but quite time consuming diy such as home assistant with a gazillion ways of doing it, to quite niche and a different way of thinking with cloud and micro controllers only and quite a lot of diy such as toit.
My personal choice is home assistant + raspberry pi + 1 wire devices + blue tooth devices + modbus + esp32's + a significant amount of my own time.
Thank you all for your comments. A significant amount of my time is very pertinent, I have already spent many hours on this, which I am happy to do (it is my time, I choose what I want to do with it! But I have also set a time limit, see original post, to ensure I don't go down a rabbit hole and never come out again). One of the reasons for starting this post is to provide a guide for others, so they don't have to spend time reinventing the wheel. Those who have already travelled the long and winding road that leads to who knows where, and have posted details of what they have done here are making invaluable contributions. At the same time, please remember this is a beginner's guide. Jargon, if it must be used, needs to be explained. I for one can't do very much with 'I used a sidewinder Henley variant flashed with caprom latency probe to measure molecular dispersion in real time on the a-b axis adjusted for non-linear scaling connected to the H1 port on a Strawberry Fly with an ethernet top hat'.
I think precision and accuracy may be the same concepts as what I would call validity and reliability, same concepts but arose in a different field. Validity is accuracy, how well the measurement measures what it claims to measure, and reliability is very close to precision, being the closeness of repeated measurements. Taking my weight (say 60kg) and my bathroom scales as an example, the measurements from the scales might always be 65kg, which makes them reliable but not valid, or they may be clustered around, but not exactly 60kg, which makes them valid but not so reliable. If they are always bang on 60kg, they are both valid and reliable. The valid but not reliable clustered around 60kg measurements may be amenable to improvement by taking an average, as long as the error is random. If it is not random, ie there is a systematic displacement (error) in one direction, we call that bias. For those who haven't guessed, the field I am talking about is medicine. Bias, known and unknown, is the bane of all medical research.
Applying this to heat pump monitoring, I think we want both validity and reliability, or in the other language, both accuracy and precision. We need validity to be able to compare one system with another (or against a benchmark, eg COP > 4 is good), and we need reliability to compare the same system with itself, ie how it performs over time, in response to changes etc. Formally, these might be called inter (between) and intra (within) system comparisons. Inevitably better validity and reliability comes at a cost in time and money, so there has to be a trade off. In these situations, I often use the good enough principle: the thing doesn't need to be perfect, just good enough. This idea comes from psychology, the concept of the not perfect but nonetheless good enough mother. For heat pump monitoring, I am going to suggest that to be good enough, we need to aim for validity/reliability percentages in the low to mid single figures. Once we get to 10%, we're probably largely wasting our time. Two systems, or the same system at another point in time, might in reality be the same, but with a 10% error either way could appear to be 20% different. Ho hum.
For most of the measurements we need for COPs, this should be achievable. Both energy in, in kWh from a dedicated heat pump meter, or as current from a clamp sensor (a clamp round the supply to heat pump cable that measures current) x voltage x time, and temperatures can be easily and cheaply, and just as importantly, reliably and validly measured. The problem measurement as I see it is flow rate, it appears to need expensive invasive equipment. @robl, you mention 'a cheap flow sensor (2%)' - would you be able to tell us a bit more about what you are using? Midea units do report flow rates on the wired controller operational parameter pages (and so should be accessible via modbus, see block diagram above), others may do so as well, but we don't know how valid/reliable these reported figures are. An independent way of measuring flow rate is necessary if we are to be confident in our measurements.
Midea 14kW (for now...) ASHP heating both building and DHW
