I have Ebac ASHP since 2m - very decent unit, and looks good

One thing I don't quite understand ... running open-loop (no buffer tank) how does one know if I have sufficient water volume (system capacity)? Pure estimation? Drain the system and see how much water comes out?

Radiator manufacturers publish the volume of water in their radiators.  For any given height and type its proportional to length so they often just give a figure per m.   If you are talking standard rads it wont vary much from manufacturer to manufacturer so the figures from Stelrad or Myson for example are good.  You can estimate pipe lengths from your house size (you dont need to know the exact figure or routing and from that come up with a system volume.  Obviously draining the system is another approach!

Given your numbers I would say you want to accommodate at least 1kW of over capacity in the mid range, ideally 2kW.  Assuming a flow temperature hysteresis of 5C (which is a reasonable value), and say you want a minimum of 20mins run time, that means you need ~ (20*60*2000)/(4200*5) l of water in the system = 115l.  if you have the space you could just put in a 100l volumiser!

If you need extra system volume use a 2 port volumiser not a 4 port buffer tank.  4 port buffer tanks, low loss headers and plate heat exchangers, all of which separate the heat pump from the emitters, are bad news.  Lots of installers put them in sadly, dont allow it.  4 ports bad, 2 ports OK.

Here are the figures for the Vaillant 5kW.  Possibly very slightly worse than EBAC or Mitsi.  Its not always easy to find figures for minimum output sadly.

One thing I don't quite understand ... running open-loop (no buffer tank) how does one know if I have sufficient water volume (system capacity)? Pure estimation? Drain the system and see how much water comes out?

I have a feeling that a calculated heat loss (with the various estimated parts of it like air exchanges) will produce a higher number, but I think in no case should I accept a unit over 5 kW. I have some interest in Ebac heat pumps and got them to give me some info on how low it will scale. It was more info than I expected, and I have nothing to compare it to... but maybe you can give your thoughts. Page 6 of the attached file!

Re the Ebac heat pumps

The heading on the table is presumably meant to be kW not kWh!

- the 9kW one will be cycling almost continuously with a min output of 2.7kW at 30C  Personally I would avoid.

The 5kW (6kW if you operate at 35C, if the numbers take into account defrost) one could still cycle at 7C (o/p1.96kW at 30C), but definitely better and possibly you wont do much better than this - here is the Mitsubishi R290 6kW data (the 5kW is same at bottom end) - remarkably similar!

You probably need to make sure you have a reasonable system volume to ensure slow rather than rapid cycling, may be worth adding a volumizer depending on the system volume calculations.  Earlier today I reviewed a post with a 14kW heat pump in a 7kW house.  The striking thing is that on times are so short even close to the onset of defrost cycling that it never really gets going.  Im thinking that the system volume is too low in this one as well as the pump being oversized, probably something to be avoided.

I have a feeling that a calculated heat loss (with the various estimated parts of it like air exchanges) will produce a higher number, but I think in no case should I accept a unit over 5 kW. I have some interest in Ebac heat pumps and got them to give me some info on how low it will scale. It was more info than I expected, and I have nothing to compare it to... but maybe you can give your thoughts. Page 6 of the attached file!

If you can find a ~4 kW heat pump that is genuinely a 4kW heat pump, not a software downrated 5-6kW one, that could be a good choice otherwise you are probably looking at a 5kW heat pump and a bit of batch heating in the shoulder season.  The extra capacity will help with DHW but you need to be sure you have enough system capacity (ie water in the system) adding a 2 port volumiser (not a 3-4 port buffer) if necessary.  Do check that is not a downrated pump of higher capacity.  The 5kW Vaillant for example is 5kW but the 3kW Vaillant is the same hardware software limited.  I think that all the Daikin R32 models 8kW or less are the same, not sure about their R290s.  Dont know about other makes off hand.  Physical dimensions, compressor type (if specified) and refrigerant change are the give-aways.  If they are all the same for 'larger' and 'smaller' models then they are the same machine.

Is 3kW sensible given construction and floor area (if you provide some details Im sure various will comment).  If so then its probably as good a value as any.

Don't allow anyone who wants to fit a 4 port buffer, low loss header or plate heat exchanger between pump and emitters anywhere near.

I'm not clear though on how to express that, or extrapolate, this heat power usage into the typical worst-case energy needs that a heat loss survey would use. Can you or anyone help me? I think in my area they use about -2 degrees C as the low temp range.

Ok... I did another one, still not using the DegreeDays concept, but using another snapshot. From 11:55pm on March 3rd to 8:55am on March 4th I used 29.52 kWh of gas. At 90% efficiency that's 26.568 kWh of heat. Timespan is 9 hours so 2.952 kW each hour.

Average OAT was -1 C. IAT 19. So thats 147.6 watts per degree. If we measure to a standard of -2 degrees to +16 degrees (house equilibrium point) that's 18 degrees, so 18 x 147.6 = 2.657 kW. That's my house's heat loss.

Maybe right?

Quite possibly not, unfortunately.

Houses have a significant heat capacity (thermal mass) which means that measurements over short periods of time can be significantly distorted by the preceding conditions.  Also any particular day can be further distorted by solar gain, wind etc; 2.6kW is small enough to be affected quite significantly by these (if the loss were larger, perhaps less so). 

I would personally not feel comfortable rely on less than several weeks worth of measurements to determine house loss, and I would want to plot them vs degree days or average OAT so that I could get a feel for the scatter and thus likely error in the measurement.

Here is the data I used, measured daily over two years (I have a smart meter so this is relatively easy).  Each point is one period of 24 hours.  If you take the data for 9 degree days (OAT 7.5C - a pretty typical mid season OAT) there is a factor of 2 difference between the lowest consumption day and the highest consumption day.  You don't know which of these you have measured.

For the avoidance of doubt I remain convinced that this is a good method (possibly the best we have) for determining the loss, particularly of properties where the fabric construction isn't uniform (ie much of our housing stock which has been subjected to miscellaneous fabric upgrades of varying effectiveness), but you need enough data to deal with the scatter effectively.

Average OAT was -1 C. IAT 19. So thats 147.6 watts per degree. If we measure to a standard of -2 degrees to +16 degrees (house equilibrium point) that's 18 degrees, so 18 x 147.6 = 2.657 kW. That's my house's heat loss.

Maybe right?