Thanks for the analysis of the capabilities of the various smart TRVs.  Its interesting that Wundasmart and Wiser work by PID, which suggests that, with a bit of tweaking of the PID parameters, it might be possible to turn them into the device we really need for heat pumps, ie one which basically settles at the right level of 'opening', a kind of automated LSV.  However it seems clear from what you have said that nobody, other than Adia, has really got their mind round the ASHP case yet which is why I (and many others) advise against using them at the current time and relying on more simple control approaches or even just pure WC.

Posted by: @madbilly

↑

At some point I will do a £/degree days cost comparison of smart TRVs vs the traditional approach, but it's unlikely to be a large difference based on what I've seen, so the decision is probably more about convenience and personal preference.

Good luck with this one (but I look forward to reading it)

The problem, as you say, is  that its unlikely to be a large difference.  So far as we can tell so far micro-zoning (in time or space), unless extreme, typically makes a few percent difference to fuel consumption.  In some cases it increases consumption, in others it decreases consumption.  Unfortunately the experiments are nigh on impossible to carry out because controlling the external conditions (unless you have a house in a lab) is impossible, and whilst its possible in principle to model it theoretically (I, too, have tried) the simplifications that are necessary to make the model practical are such as to cast doubt on the results.   Several robust discussions on this forum (for example this recent one, which is a continuation of over two years of analysis/discussion) and elsewhere have failed to reach a definitive conclusion that is applicable other than to the actual system on which experiments have been carried out. 

The solution, I fear, is either properly controlled experiments, which requires a house in a warehouse - these do exist but we dont have access) or 'big data';.  The likes of Octopus have (or shortly will) have the 'big data', whether they will release any meaningful insight we don't yet know.  Those of us with a keen interest (which includes me) are likely to continue our amateur attempts at analysing the situation both experimentally and theoretically, but I am increasingly pessimistic that much further progress will be made.  What is really needed is a couple of PhD studies!

If you did get any useful results out of your heatpunk house model it would be great if you shared them!

Posted by: @madbilly

↑

I am generally happy with the result, however the bedrooms are always warmer than the living room which I suppose is the point about insulation and the living room is basically heating the rooms above it.

The simple, low tech, solution to this (assuming its a problem) is to turn down the LSVs in the bedrooms.  It does work, honestly!  Disappointing if you like tweaking things, but effective.

I think that most of the others work by PID as well, I just didn't confirm it with tech support. I think a DIY solution is possible using TRVs that expose more control, e.g. generic Z-Wave or Zigbee smart TRVs, but that would require also programming it all - fun if you've got the time!

I'm not sure any analysis of mine will be useful, but I'll share it if I think it tells a story.

The LSVs were already turned down to 1/8th turn open and I realised that there was actually no flow! So I had to turn open to 1/4, and I think I'll find I have to turn down the TRVs (not smart currently) to avoid the rooms being too hot.

Posted by: @madbilly

↑

The LSVs were already turned down to 1/8th turn open and I realised that there was actually no flow! So I had to turn open to 1/4,

Many lsvs go from full off to full on in a fraction of a turn, they are basically isolation valves.  There is a whole section on this (comparing different lsvs) on the heat geek site.  The property is called, at least by heat geek, 'valve authority'

Yes, I'd read that and already knew that LSVs exhibit a kind of logarithmic behaviour, but knowing it unfortunately doesn't change the actual valves I've got.

@madbilly - I think you may risk over-complicating this. Re-reading your original post, there is a mis-match between your requirements, which amount to a wide range of ad hoc room temperatures and zones (which is fine, each to his own!), and the way a heat pump works. Indeed, when I was writing the ABC Guide to Heat Pumps, I very nearly included a section on when not to install a heat pump. One of the scenarios was ad hoc zonal heating... The problem is that heat pumps are by design low and slow devices, tortoises rather than hares, and they are not well suited to bursts of localised activity interspersed with inactivity.  

But you are where you are. As you have a Midea heat pump, as I do, you have direct access to a wide variety of parameters through the modbus interface. Some of the parameters are read only, but others can be both read and set, which gives you a considerable degree of control. I have my Midea wired controller hooked up to a headless mini PC running Linux over a wired connection, and I both read data and set values using python scripts. For historical reasons, I have Home Assistant installed, but I don't actively use it much now, apart from using the main dashboard as quick overview of what my system is doing. HA's main purpose now is to provide the underlying OS and networking that my python scripts rely on.

As an example of a simple but very useful extra layer of control, I have an auto-adapt script that adjusts the LWT based on any discrepancy between my desired room temperature and the actual room temperature. If the house is a little cooler than it should be, then boost the LWT a bit, if the house is a lot cooler than it should be, then boost the LWT quite a lot. This is done by setting the end points of the weather compensation curve, eg if the house is a lot cooler than it should be, then add three degrees to each end of the weather compensation curve. I haven't done it, because I don't see weather forecasts as being that accurate, but there is nothing to stop you pulling in online forecasts and making similar adjustments eg if forecast hours of sunshine > 6 over the next 8 hours then lower the LWT a bit, if forecast wind = NE and speed > 30 knots then up the LWT a bit etc etc. Again I haven't done it, but you could also set up pretty much any on-off schedule for your heat pump, and even add scheduled boosts eg if hour > 5 and hour < 8 then boost the LWT by 2 degrees. 

But back to over-complicating things (the python scripts don't count, they are not complicated once you know the basics!) - I spent ages balancing my radiators. Without realising it, I ended up throttling my system, a good many of the lock shield valves were too far closed (part of an attempt to get a pair of reluctant radiators to warm up), and my system struggled in cold weather (basically not enough water moving round the system). Based on some wise advice from this forum, I opened up all the lock shield valves (TRVs were already fully open) and whoomf everything warmed up very nicely. The cold rads were still cold, but as I have open indoor doors the heat evens out very easily, and as I run the system low and slow once it is in steady state it pretty much stays there. Even better, I was able to reduce the baseline weather compensation by no less than four whole degrees, and my COP has increased.

There is a lot of debate about whether setbacks (and subsequent boosts, which is what your ad hoc requirements amount to) save money or not. The theoreticians are inclined to suspect they do not save money, and may even waste it, while the empiricists are inclined to think setbacks do save some money, but find it very hard to prove the point, the main problem being that no two houses are the same, and even worse, the same house exists in a constantly changing environment. If I did use less energy last week compared to this week, is that because (a) I had a setback last week but not this week (b) the mean OAT has fallen (c) I had guests staying last week (more human and cooking warmth) (d) it was gloriously sunny last week (e) it's blowing force nine from the north east this week of (f) a combination of some or all of these things?

If I was in your position I would give consideration to opening the system up (all valves lock shield and TRVs fully open) and then putting my efforts into tweaking just the weather compensation curve over modbus, along the lines described above. That way, I would have the simplicity of just one thing to tweak, rather than trying juggle a myriad of spinning plates all at once.