@morgan  I am not qualified to comment regards the system needing a buffer/ low loss header etc, however I need to confirm have you tried checking the basic things such as are all the radiators balanced? Some installers tend to do a rough balance on radiators and let TRV's do the fine tunning, however I have found this is not good enough and balanced all radiators properly. It is also recommended that TRV's should be used as temp limiters and be set higher than the desired room temp so there is very rarely flow restriction in the system. @Toodles posted an excellent article recently for balancing rads.

The other item that caught me out was the calculation of the COP value and interpretation of the system data. COP = (Heat Generated (Kw) + Input Kw) / Input Kw. I was only getting values of 1.9, 2.1 etc until I discovered I was misinterpreting the data then my COP was over 3 once correctly established.

I do not have a Mitsubishi ASHP but is the photo showing "Delivered Energy" the total including the input electrical power or just what the ASHP has generated from the outside air do you know?

Regards

@technogeek 

My understanding is that COP is ascertained by > Delivered electrical energy (Kw) divided by the consumed electrical energy (Kw).

For example in my case above for February 2024 1327Kw by 527Kw = COP 2.5.

I'm afraid your calculation doesn't make sense to me.

Some of the manufacturers use the COP calculation method shown by @technogeek, but I don't think that is the case with Mitsubishi.

Could you please provide some graphs from MelCloud data?

How does the Ecodan electrical energy consumption compare with your actual electrical power meter readings?

@morgan For your information the calculation I gave is from here https://energyeducation.ca/encyclopedia/Coefficient_of_performance . The value of Qc in my case (Samsung) is the generated heat from the atmosphere plus the work put in (W in)(electrical energy) = QH divided by the work put in (W in)(electrical energy).

I was just trying to establish if your COP is calculated the same way or not, however if Mitsubishi details this information differently fair enough

Regards

I think @technogeek is looking at a Vaillant which does, I believe, present a figure for the 'free' heat generated by the pump, so the 'paid-for' input energy has to be added back in to calculate the total energy generated. I can't see how you can obtain a figure for heat from the environment in the first place except by taking the figure for the heat generated and subtracting the energy used to generate the heat!

 @morgan I don't follow the argument that your second opinion guy has put forward. If you have too low a volume of water available to defrost then you are going to have issues around slow and more frequent defrosts which will certainly detract from the energy efficiency of the system, but these should be issues in themselves that you can diagnose from looking at the performance record in MelCloud. However your COP for April is still low even though we (at least where I am) have not had temps dropping low enough to cause defrost cycle for weeks. Compulsory buffers seems old-school and needs to be justified for the way the system works rather than thrown in as a catch-all fallback.

I think that overall my system is somewhat comparable to yours, Ecodan, rads only, 70s house – it took around 10000kWh heat last year and so far this year has required 4400kWh. For a cost of 3100kWh last year (COP 3.2) and 1425 so far this (COP 3.4). But, as evidenced in various threads, the Ecodan reported energy used is typically way off, 3700 last year and 1690 so far this – it is only by having a meter on the supply that I know the true figure of kWh consumed. So hopefully not actually quite as bad as you fear.

What is the balance between HW and heating energy used? And how often does the tank reheat get triggered? The COP for that is always on the low side so may be dragging the overall number down if you are running a HW cycle more often than you need to. This may be affected by bad positioning of thermistors in the tank. An easy experiment is to switch the HW cycle off after it has run and then wait until the hot water is running too cold for you and check what the thermistor reading is on the controller display – you might be (should be?) surprised by how low it can get before the hot water actually runs out. Of course you might need a fuller tank for operational reasons at certain times, but a 24hr reheat is probably the least efficient setting.

I have an 8.5kW unit and I know that is larger than it needs to be for my house, so I think your 11.2kW unit is definitely too large for yours – you are keeping the house at 20° for about the same output. So I presume it only runs continuously at low temperatures. Unfortunately an issue not as simple as changing a setting or small bit of plumbing! I find the unit always likes to work to a LWE in the mid to high thirties, then just varying the cycle off periods – as much as I'd like the LWE to be lower and run with a lower COP I suspect that only increasing the rad sizes to allow for more heat output at lower temps will have an effect. So I would also suggest removing the TRV limits and allowing the upstairs rads to be part of the heating circuit at all times, otherwise you have less radiant capacity and the water temperature may be pushed up by the controller in order to deliver the heat it is expecting to.

@harriup There is a Heat Geek presentation on Youtube in which Adam Chapman fitted a Vaillant ASHP in a house with poor insulation, ill fitting windows and less than ideal pipework - even so the ASHP was producing a COP in excess of 3! Just saying… Toodles.

@toodles All those factors might result in a lot more heat being required, it can still be delivered in a very efficient way. Lack of efficiency seems to be about a mismatch between the way the pump produces heat and the way the building can extract it. I quite fancied having a Vaillant when ours was installed but ended up with an Ecodan because of supply issues at the end of the RHI grant period. And, like Morgan, I am constantly slightly disappointed by the apparent lack of efficiency which I increasingly put down to the software running it.

Now people have said all the sensible long things, I have a question - can you tell how often it is coming on and off each hour, and is it possible to see what sort of temperatures it is running the water at through the radiators?

Posted by: @guthrie

↑

Now people have said all the sensible long things, I have a question - can you tell how often it is coming on and off each hour, and is it possible to see what sort of temperatures it is running the water at through the radiators?

If you search the forum, you will find countless examples of people doing exactly this. The vast majority of heat pump controllers will give you this data live (you may need to work out what parameters to watch), many will give you historical data, and, with a few mods, relatively sophisticated monitoring is possible in some cases eg where a modbus connection is available. Here is minute by minute data from my heat pump for the last 48 hours:

There is one caveat: this data (apart from indoor air temp) comes from the Midea wired controller, ie it is Midea marking its own homework, and is not verified by third party independent sensors. That said, it does not appear to be wildly inaccurate. 

Posted by: @derek-m

↑

Some of the manufacturers use the COP calculation method shown by @technogeek, but I don't think that is the case with Mitsubishi.

Could you please provide some graphs from MelCloud data?

How does the Ecodan electrical energy consumption compare with your actual electrical power meter readings?

 

@derek-m 

Not in a position to provide those yet.  Will do so when I'm back home.

@harriup Does this mean your Ecodan reports electricity consumption 15% higher than actual measurements? 

If we assume something similar happens with @morgan - which we of course don't know but is a possibility - the COP would be around 3 instead of 2.5 for February 2024. Not great still, but definitely better. But let's await the data.