Yes, but not in the old way. I’m now doing it by using a delta each minute from the energy report rather than seeing it in the ListDevices call.
Sounds about right.
Sounds plausible.
Design flow temperature: 45.
EPC says:
Annual space heating: 20,468
Annual DHW: 2,975
My quote from my supplier says this:
Peak Heat Load: 11.2kw
Annual Heat Loss: 19,200kwh
Seasonal Performance: 3.4 (the system is 340% efficient)
Expected EPC deemed heat: 20,000kwh
Renewable Heat Factor: 0.70
Total Renewable heat: 14,117 kwh
Renewable Heat Incentive tariff: 10.71p per kwh
Year 1 RHI: £1,512.00
Total RHI: £11,240.64
Heat Pump Running cost: £1,027.76 per annum
Expected fuel saving v’s Oil: £598.00 per annum (based on 45p per L)
7 Year Total Benefit: £15,689.87
It’s actually using something like 800GBP of electric a year.
As well as the “Greta” benefits from not buying 2,500L of oil a year, that also means we’re not paying 1,250GBP to heat the house / water.
The RHI certainly only impacts the purchase price, the running costs are still paid by us.
I mentioned it because it was insane. I’ve never had another day like that. It shows that you need to be sensible about how you use it because if you leave it going wild it will cost more than you’d like.
When it’s very cold I expect I will use that much. When it was -16 I used 10GBP a day in oil.
@muhrix here’s my graph of daily costs for heating and hot water versus the average temperature for the day. So where the x-axis says 0 °C that inevitably means it was colder than 0 °C for a chunk of the day.
I think my house is about the same as yours. We have the bathrooms with radiators set to 24 °C all the time and I think the lowest setting is 21 °C except during the night when the radiators drop down to 17 °C but that doesn’t really matter because we turn off the heating overnight to help us sleep.
On your heat loss calculations, the peak heat load is 11.2kW.
Do you know why your installer recommended a 14kW instead of 11.2kW?
Is it because that ASHP model has a slight dip in output at very low temperatures?
What a nuisance. Sounds like you have got a reasonable solution though. Looking forward to hearing about the next time they change the API
I know you don’t want to release your code but I’m wondering what your strategy is for controlling the pump? It’s not that cold yet (for Scotland, but maybe my outlook is biased) and yet Ecodan’s dodgy energy/heat numbers are saying I’m only getting 250% (I’m looking at the “CoP” key from the Report API). Can I do better?
If you could give me some pointers to the parameters you’re monitoring and the controls you’re touching I would be keen to replicate your results.
I sat down to re-organize the algorithm code to release it last Saturday, but completely ran out of steam after 30 minutes I’m afraid. I’m somewhat busy at work for a few weeks, but then I have some holiday where I’m hoping to get organized. If I’m lucky I can get the algorithm released.
Here’s a question for you. At the moment my code reads the input data from the JSON files I fetched from MELCloud. I have about 6 million of them, but only the last 100 are interesting (mostly for Legionella review, otherwise it would be a handful). However, it could use the data from EmonCMS. I didn’t do that because it would be another link in the chain that might break as I mess about with my EmonCMS install. What would you imagine as the data source to drive the algorithm?
BTW, if you are looking for the lessons from all my testing, I documented most of it in these posts:
Basically the algorithm is just a manifestation of those suggestions. Maybe that would help you decide what to do? Certainly in the last few weeks I’ve found turning the heat pump off the be the most important aspect as you can see in this thread.
I see. Based on your experience, would you say your ASHP is oversized? I’m asking because I wonder how conservative/realistic heat loss calculations actually are (one would prefer to err on the side of caution in this case).
I watched/read @TrystanLea’s video/post comparing the calculations with actual usage of his own house and found that very informative and super interesting!
@MyForest please correct me if I’m wrong, but based on your posts, I believe those would be:
Controllables:
Room target temperature and its schedule (or, simply, call-for-heat)
DHW target temperature and its schedule
Desired flow temperature
ASHP operation mode (to switch it off)
Observables (based on Mitsubishi’s EMP3 kit for MMSP):
Not just based on my experience, but also on lots of other people’s opinions too. The fact that I’m turning it off when it’s 0 °C outside is a give away. It does scale the inverter, but it doesn’t go low enough.
Yes, but we have no room stat, we have TRVs on each radiator and no call-for-heat. We’re using the return temp sensor as a proxy for the temperature of the water in the radiators.
+1
I get the others, but not flow rate from the EMP3.
I’m using a few other things outside the EMP3.
If the solar power is high I assume the house is bathed in sunshine so I don’t heat it (on a sliding scale taking temperature into account too).
If it’s windy / humid that has a noticeable effect on the effective temperature and that’s what I use in all my calcs (notably for the weather compensation curve).
I have sensors on some household devices and if they are in use I keep the heating on (e.g. my office monitor). EmonCMS makes that really easy.
There’s a few other things, like deferring Legionella runs until it’s warm and also giving the Solar iBoost a chance to do the job so the heat pump doesn’t run at all for times like July 20th to July 24th. You’ll no doubt get that with your solar thermal.
To be honest all those extras are nice-to-haves which make it more comfortable or just easier to get right. They improve the efficiency a bit but not enough to worry about I suspect.
Bopping this here for a bit of light relief from all the woe.
You can see the pump was turned on (manually, by me) and the algorithm spotted this and brought the desired flow temp right down to 25 °C. That allowed it run wonderfully efficiently even though it is about 4 °C outside (about 375%).
Then the algorithm watched the flow getting warmer and gently walked the heat pump up to warmer and warmer desired flow temp.
A small pump would have been able to keep runnig gently at say 37 °C but my big pump doesn’t scale down enough. There are many radiators closed off at 01:30. So it just blundered along getting warmer until eventually the algorithm stepped in because it had been run at the max-desired-temp for a while. It’s not really a problem running like that, but I’ve found the people are comfortable even if you turn the heat pump and water pumps off at that point. The heat just slowly emanates from the radiators.
We can also have a laugh at 02:00 where the bug in the “delta” calculation means it missed a minute and then doubled up on the next one.
So there we go. Maybe there should be a Sunday evening program “One man and his heat pump” where you get to watch people try to get their heat pumps to behave whilst commentated on by Trystan and John.
Considering that Melcloud stopped returning per-minute data, I started thinking of an alternative way to tap into the M-BUS to fetch the data without affecting the MMSP setup of the EMP3 kit.
Yep, it does feel that way, but if that’s the price needed to sort out your problems then maybe it’s OK. The rest of the kit is costing you a lot more than that @muhrix
Are those numbers accurate? > 600% efficiency and then < 100%?
On the first graph, are produced/consumed hot water and space heating combined? The are both red/green.
It appears that the short term solution is to find a way out of using Melcloud to fetch monitoring data. And then hopefully convince Mitsubishi that they should at least keep the data stream consistent.
Not really. It’s what the monitoring package is reporting but we’re mostly taking them with a pinch of salt. Often it’ll say it’s producing a few kW of output for just a few hundred watts of input. In the immediate time scale it is misleading, but over say 15 minutes it does seem to add up to the right numbers. My graphing tries to smoosh that about a bit without going overboard. Sometimes it looks ridiculous like this.
I’d be more likely to believe those 100% numbers because it was operating incredibly inefficiently because it kept switching back and forth.
The are distinct, but I chose to graph them with the same colour to make the graph less noisy. They are mutually exclusive so there should be one or the other. There generally is, except for crazy flip-flop cases like this when it’s just confusing.
Yep, that’s the plan. In fact I work for a company trying to help people get control of “their data” (for various definitions of “their data”). I think this is a good example where I consider that I have as much right to use the original data as anyone else. I will be making a bit more noise about this soon, I just need to sort some other things out first.
Hello Murilo, I was about to ask on the forum if anyone is monitoring a new R32 Ecodan, and I stumbled across your message. I have just fitted a 6kW unit, and the control algorithm seems not to be a good as my 10-year-old 5kW (same as Trystans). Basically, mine it seems to avoid steady-state minimum-power running, but cycles instead. I would be keen to know what others are finding. I wonder who else might be monitoring a new Ecodan. I have seen David Bowen’s graphs… too much for my slow brain!! I think his is an older model.
I’m monitoring a 2021 R32 8.5kW Ecodan since a few days ago, using pymelcloud and despite the report above, I seem to be getting readings updated every 5-ish minutes, not hourly.
Ours seems to settle comfortably into steady-state running, but it’s cold enough this week that I doubt this is minimum power!
I’m going to assume John’s graph was from 2022-01-04. I was struck by how it looked like my graph from the same time range. (Ignore my measuring hiccup at 20:00).
My 14kW ranged up to a max of 10kW output and was around the nominal CoP at the flow temp (the gold line on Efficiency). Of course a CoP of 2.74 looks pretty dreadful considering the outside temp was 2 °C. The only saving grace is the hour it was switched off meant it wasn’t using any power during that time - it averaged 1.64kW consumption over the 3hr time window and 4.49kW output).
@johncantor As we were discussing, I think that turning mine off at 18:55 meant I avoided the defrost you had at 19:17. However, I still had a defrost about the same time as your second one.
This evening it’s colder at -1 °C and I’m seeing longer runs with two defrosts (19:43 and 21:03). You can see the algorithm resets the desired flow temp after a long defrost which caused it to ramp back up gently.
Your graphs are mind blowing David. I need time to understand them properly. I offer you 2 graphs. 1st is my old 5kW unit on FTC3. It seems to modulate as I would expect wrt tracking the fixed flow setting. It has served me well for 10 or more years. Same unit as Trystan’s.
2nd is 6kW R32 unit on FTC6. It seems to be behaving oddly… Not worked out why yet.
(I’m looking at the “CoP” key from the 
