Posted by: @cathoderay

↑

Lastly, how does it compare to the various open source home automation projects like Home Assistant and domoticz? Clearly it is proprietary, and it only manages an ASHP, but does it do that better or worse than the open source projects. The latter, so far as I can tell are still at the 'enthusiasts' stage, fiendishly complicated to set up, and more often than not definitely do not work 'out of the box'. 

I'm pretty sure that Home Assistant definitely does work "out of the box", but only a minority buy a Home Assistant box. I'm not even sure anyone sells domoticz boxes yet. Of course, you probably want it to do something in a certain way, and that's where the difficulty starts. If someone chose the parts to be compatible, using things that are known to work well with HA or MQTT or whatever, then it wouldn't be too difficult, but of course, few people think of that before they hit problems with the manufacturers' junk and consider building their own controllers, so most of us are trying to cobble together what we already had, which is never as easy. At least it's often possible!

How does Homely compare to open home automation? Well, it'll cost more and it'll either work almost immediately, or it won't work properly and there probably will be nothing you can do to fix it. And one day, they'll stop selling you the service, it won't work properly any more and its users will be back to square one, hoping enough people with similar hardware supported open home automation for it to rescue them.

We also interviewed Karolis from Homely and I was skeptical about the longevity of this project in our “off the record” conversation. I think they’re battling to get back end access to manufacturers’ heat pumps so that their kit can control those pumps which is why it’s only rolled out on two brands. There’ll be warranty concerns too. It’s messy in the long-term. 

I still don’t get why this level of ‘intelligence’ isn’t included from the likes of Daikin, Samsung, Mitsubishi, etc. out of the box. Maybe Homely’s play is to get acquired by one of the big guns.

Posted by: @editor

↑

I still don’t get why this level of ‘intelligence’ isn’t included from the likes of Daikin, Samsung, Mitsubishi, etc. out of the box. Maybe Homely’s play is to get acquired by one of the big guns.

Maybe it's cheaper not to, they sell plenty of high priced units without it and they don't think it'll help them sell more than it costs to develop the feature? 

@editor Thanks, that's very interesting to know. 

I have a Samsung 8KW system but find the controls infuriating to use. We had the heat pump installed as part of an extensive renovation project. With two small kids and a whole load of decisions to make, when the plumber recommended the Samsung we were happy to go for it after doing some internet research.

But it's different to live with a product than to read reviews. We are happy that the unit can comfortably provide all the heating and hot water we need, but it is very difficult to determine if could be optimised further to reduce running costs.

I'm not an engineer and don't have loads of free time to spend regularly tweaking parameters, so the idea of a system like Homely appeals to me. I totally agree that the controls out of the box should be much better, but for the people out there that have systems installed, retrofit systems might be the only option?

Posted by: @mjr

↑

Posted by: @ecodannewcastle

↑

There's so much of interest in this thread. I know there is much controversy about compensation curves but to me it's a strange concept that the outdoor temp would be given any primacy over the indoor supply.

Well, the future indoor temperature is mainly a function of the current outdoor temperature determining how much heat your home loses now, isn't it? To me, it's even stranger that anyone would ignore such vital information and wait for it to manifest in unexpected changes in the indoor temperature. I can see that you might get away with it if you have heating that can respond quickly like an air-to-air system in a few rooms, but you'll spend most of the time at a "wrong" temperature if you try it with air-to-water supplying a whole house of low-temperature radiators or underfloor.

 

When I look at these systems I often put my industrial control systems hat on.

So the indoor temperature would be the Process Variable (the bit we are trying to control), but if it responds very slowly to change, then the controller needs a good dose of help.

In the industrial sense, the outside air temperature measurement would be termed a 'Feed Forward' input, the objective being to help the controller predict the future.

Of course having all this helpful information is of little use if the controller does not have a clue how to use it correctly. It would appear that most manufacturers require much better software writers, that actually understand control engineering. Maybe they are just reasonably good at producing shiny apps, or is it 'smoke and mirrors'?

Good pushback to my question. I agree that the outside temp is best predictor of future inside temp. But, there is a leap then to assuming the air to water systems cant heat quickly, I think.

For example, my well-insulated mid-terrace, 4 floors:

- overnight, with flow temp 30, outdoor temp say 2-5 degrees: 1 unit per hr approx x 8 hrs = 8 units

If I decide to not heat overnight, and have a relatively not warm morning, I can run at flow temp 40 between 3 and 5pm when outdoor temp is higher for approx 4 units) then come right down to 26-30 degrees for the rest of the evening and be toasty warm.

What I suppose I'm saying is - give users control over flow temps by hour (programmable), and choices over room temps (programmable) and you rely less on the best guess solution at present.

Interesting conversation and very happy to be proven wrong!

Posted by: @redzer_irl

↑

I have a Samsung 8KW system but find the controls infuriating to use. We had the heat pump installed as part of an extensive renovation project. With two small kids and a whole load of decisions to make, when the plumber recommended the Samsung we were happy to go for it after doing some internet research.

But it's different to live with a product than to read reviews. We are happy that the unit can comfortably provide all the heating and hot water we need, but it is very difficult to determine if could be optimised further to reduce running costs.

I'm not an engineer and don't have loads of free time to spend regularly tweaking parameters, so the idea of a system like Homely appeals to me. I totally agree that the controls out of the box should be much better, but for the people out there that have systems installed, retrofit systems might be the only option?

I think your statement above sums up the bulk of homeowners’ experiences with heat pumps. When you buy expensive kit, you just want it to work. I like to tinker and learn how things work, but not everyone is that way inclined and many people don’t have the time with full time jobs and kids to look after. 

Reviews are helpful, but when it comes to heat pumps, installation and design is probably more important than the unit itself.

So I can see the appeal of Homely, but given that much of the underlying tech is already built into the units, I’m not sure why they’re not delivering what Homely offers.

I’d also be very keen to see data and case studies of Homely installed on a system that’s horribly inefficient due to design shortcomings and whether it can make things better, or whether it’s best to fit on a system that’s already running well.

Posted by: @editor

↑

I’d also be very keen to see data and case studies of Homely installed on a system that’s horribly inefficient due to design shortcomings and whether it can make things better, or whether it’s best to fit on a system that’s already running well.

This is really where this thread started. On the face of it, Homely seems as if it might be the answer, but when it came to finding real world examples of it working in practice, precious little came up. On the other hand, some other rather less comforting things did come up, questions about transparency, control, ownership of data and longevity. So far, the only real world feedback to appear here has been negative...

The other question I can't avoid is do I want an ex-economist/investment banker in charge of my heating system? These are the folks that engineer financial crashes, using a science that makes astrology look robust... 

I'm still trying to get my head round the relative importance of indoor vs outdoor temps. I tend to use a slightly different terminology to @derek-m that comes from health care, structure (how many beds you have), process (how many hip replacements you do) and outcome (how many patients end up pain free). If this terminology is applied to heat pumps, there is only one outcome measure/variable, indoor temperature, which puts it in pole position. Outdoor temp then becomes just another process variable, like (and no more or less important than) the thermal characteristics of the building. You could even say the indoor temp provides feedback on a summary of all the other variables (eg a leaky building might still be cool on a warmer day or a tight building might be cool on a cold day), in other words, however you got there, the indoor temp tells you what the real world outcome is, and so what you need to do with the flow temp. If the indoor temp is at the design temp, then the flow temp is right, if it is below design temp, flow temp needs to increase, if it is above design temp, the flow temp needs to decrease. Is there any reason why it isn't that simple?

On a by the bye which is sort of relevant to this thread, I have found a way of downloading the Midea app data from my mobile phone to my PC, by using Total Commander with the ADB (Android Device Bridge) plugin. You sort of get root access to the phone without rooting it. You can also do something similar by doing a whole phone backup to PC using the command line version of ADB and then hex edit the backup file header to make it into a decompressible tag.gz archive. What I get is a log folder with 15 files, one for today and one for each day in the last fortnight (I also get most of the apps own files, mostly javascript). Each of these 15 files is about 20-40kB in size, and has the extension xlog. I am convinced these files have a lot of not all of the daily data in them and had high hope I was just a text editor away from finding out (xlog files can be xml log files) but when I opened the xlog files they are mostly garbage, most likely binary files, but with occasional readable lines eg a file path that tantalisingly makes sense. But the bulk of the file remains unreadable. Has anyone any idea how we might read them? I would post a sample, but as they may well contain sensitive data (passwords etc) I am reluctant to do so. In case it provides a clue, the file header (in hex) is 06 00 02, but that doesn't appear in any of the standard lists of file headers/types.         

Posted by: @cathoderay

↑

If the indoor temp is at the design temp, then the flow temp is right, if it is below design temp, flow temp needs to increase, if it is above design temp, the flow temp needs to decrease. Is there any reason why it isn't that simple?

Only that you need an initial value for flow temp when starting up or the external temperature changes rapidly. That's where your weather compensation curve can help.

@cathoderay 

If the indoor temp is at the design temp, then the flow temp is right, if it is below design temp, flow temp needs to increase, if it is above design temp, the flow temp needs to decrease. Is there any reason why it isn't that simple?

Whilst it may sound simple in theory, it is a little more complex in practice. Unfortunately I don't know precisely how the algorithms inside a heat pump controller perform, so I must presume that they operate in a similar manner to those inside an industrial controller.

If we analyse what happens in the real World, it may become a little clearer.

Say the indoor temperature is stable at 20C, with an outdoor temperature of 10C, the heat energy supply and the heat energy demand is balanced, with a LWT of 35C. The homeowner feels a little cold, so turn the setpoint up to 21C. The heat pump controller measures the temperature difference between the desired indoor temperature and the actual indoor temperature, so increases the LWT to 36C. After 30 minutes the heat pump controller finds that the indoor temperature has only increased to 20.2C, so increases the LWT by a further 1C to 37C. After a further 30 minutes the indoor temperature has only managed to reach 20.5C, so the heat pump controller increases the LWT to 38C. A further 30 minutes later the indoor temperature has increased to 20.8C, but the heat pump controller insists on adding a further 1C to bring the LWT to 39C. A full 2 hours after the initial setpoint adjustment the indoor temperature eventually achieves the 21C setpoint and the heat pump controller is happy. Simples. 😎 

But there is a slight problem.

At a LWT of 39C, the heat energy supply exceeds the heat energy demand, so the indoor temperature continues to increase, and reaches 21.3C at which point the heat pump controller wakes up and reduces the LWT to 38C. After a further 30 minutes the indoor temperature is now at 20.6C, so the heat pump controller panics and reduces the LWT to 37C. You've guessed it, 30 minutes later the indoor temperature has reached 20.8C, and the heat pump controller is getting quite distraught, so reduces the LWT temperature again, this time to 36C. A further 30 minutes later the heat pump controller discovers that the indoor temperature is no longer increasing, so breaths a sigh of relief, but to be on the safe side reduces the LWT to 35C. Not so Simples. 🙄 

I feel certain that I don't need to continue the sorry tale of the heat pump controller, for you to realise that just monitoring and responding to changes to the indoor temperature will cause the LWT to oscillate, which in turn will cause the indoor temperature to swing above and below the desired temperature, and that is just with a 1C change in setpoint. With luck the system response would create a 'damped' oscillation, and the indoor temperature would eventually settle at the desired 21C, but if the system response is particularly slow then it is possible to create a 'forced' oscillation, when the temperature swing of both the LWT and indoor temperature increases rather than reducing, and both the heat pump controller and the homeowner become very unhappy. 😥 

Going back to the original steady state condition, 20C indoors, 10C outdoors and a LWT of 35C.

If the outdoor temperature starts to fall from 10C to 5C, what do you think will happen to the indoor temperature? Initially nothing. Even if it takes several hours for the outdoor temperature to fall from 10C to 5C, Dependent upon the insulation level and thermal mass of the property, it could be many more hours before the indoor temperature starts to respond. Since we now have a 5C change in the difference between indoor temperature and outdoor temperature, I confidently predict that the heat pump controller will experience absolute hysterics by the time the full effect of the outside temperature change is felt inside. 😜 

This is why outdoor temperature measurement can be even more important than indoor temperature measurement, with an ideal control system utilising indoor temperature, outdoor temperature and LWT in the control process.

@derek-m the same oscillation would occur if the weather comp didn’t match the property, and the system was running on an on/off stat.

Please don’t think I’m banging the homely drum again like some sort of fanboy, but it’s precisely this ‘figuring out the weather compensation curve’ that homely aims to do, so that installer aren’t tempted to 1. Guess, and 2. Not enable it at all

There’s a perfectly good reason (in my mind at least) why the heavy lifting (the flow temperature requirements for a given property, efficient recovery from a setback period etc) are done on AWS, and that’s to remove any requirement for field firmware upgrades when evergreen want to roll out new features or fix bugs. 

Posted by: @cathoderay

↑

Posted by: @derek-m

↑

I forgot to mention that this is not just theory but actually works in practice. I have an industrial type controller which controls the indoor temperature by simply measuring the temperature in the hallway and the temperature of the radiator in my 'Man Cave' and achieves a temperature of 20.5C +/- 0.1C for most of the time with a deviation of +/- 0.2C on the odd occasion when the outdoor air temperature changes quite rapidly. It achieves this without the need for any wi-fi, cloud or even outdoor temperature measurement.

Derek - thanks for such a detailed explanation which makes sense. Your last paragraph (quoted above) is particularly interesting, especially the last sentence. Thinking about it, weather compensation using outdoor temp is perhaps a red herring, a convenient selling point for marketing, but ultimately it is never going to do very well, too many other variables. Earlier I mentioned using the current measured indoor temp for load compensation, but thinking about it, why not do as you do, have it (and some way of monitoring the LWT or rad temp) as the only inputs. The basic point, which I didn't spell out clearly enough, is that the only output that matters at the end of the day (hour/minute) is room temp - that is the be all and end all of a heating system. Everything else, in a language I am familiar with, structure-process-outcome, is a process variable. Why don't manufacturers do this, ie regulate (vary, rather than on/off) the heat pump output on the basis of the deviation of the actual room temp from the desired room temp? Variations in thermal mass, current weather, whatever, could be taken care of either dynamically (system measures itself and then tweaks itself accordingly), or by manual presets in a controller, eg a high thermal mass building would be set to have a bigger boost. If the system knows the date, it could adjust the boost to accommodate seasonal changes. Gain, latency etc, are all simple and should be easy to adjust. Even the basic autopilot on my boat has these things, to accommodate different types of vessel, heavy and slow to react vs light and skittish etc. 

That your system can achieve "a temperature of 20.5C +/- 0.1C for most of the time with a deviation of +/- 0.2C on the odd occasion when the outdoor air temperature changes quite rapidly" says it all really. That is very tight control indeed, and in practice shows you don't need outdoor temp, weather forecasts, and all the paraphernalia of wifi, clouds, companies logging your home data and not letting you see it, and all the other IoT baggage that comes with 'you will be a QR code, and you will be happy' (a reference to the WEF, 'smart' cities and techno-fascism, for those who haven't joined up the dots in the code just yet).

It'as precisely this control strategy that is (I think) employed in Homely - indoor temperature sensor to know and learn the thermal response of the property, outdoor sensor from the modbus data on the heat pump, flow temperature from the same, target temperature from the UI,

The control algorithms are run in a high level language as a web service, on AWS. These algorithms can easily be updated or improved without needing any on-site firmware updates.

@grahamh swears by it on his Samsung install.