Not a hugely glowing user experience. Interesting blog.

That's been very similar to my experience, I came to the same conclusion about the rounding up / down within the SenseComfort app making the calculations meaningless. Like you when I did the sums I was seeing 2-2.5, when I check the Working Figure on the ASHP control interface it's about 3.6. 

I would be interested to know what sort of MMSP you've got running? I'm keen to understand and optimise my system, but it's impossible without proper data.

Posted by: @editor

Not a hugely glowing user experience. Interesting blog.

I think this confirms that unless you have a proper heat meter to measure energy delivered, working out COP is at best an estimate and at worst a finger in the air. It also reassures me that while Mitsubishi/Melcloud isn't perfect, it's not that bad either.

@kev-m

I'm afraid that you are all making the same mistake, you are measuring the same parameter more than once with different systems. If you only measure the parameter with one measuring system, then it must be completely accurate. 😎 

I don't think that's the case Derek. The problem, as Cycleneil described, is the way the Vaillant app rounds numbers up/down gives wildly inaccurate results - bad data in = bad results out. Using different methods such as a proper MMSP (which is presumably more accurate) gives what appear to be more believable results.

@squonk

I was being facetious as usual.

Being an Instrumentation and Control Systems Engineer for over 50 years, producing accurate indications has been my 'bread and butter' for much of that time.

I fully agree that the rounding of data within an App, and thereby providing incorrect data is far from ideal. Providing incorrect data is probably worse than providing no data at all.

I am used to working with systems where accuracy is a prime requirement, but even then there are limitations as to how accurate the system can perform, and at what cost.

As I stated previously, accuracy costs money. An MMSP system, if correctly installed, should give reasonably accurate results, but don't expect it to be completely accurate.

The point that I was making facetiously, is that if you only have one indication that appears reasonable, then one has to assume that it is reasonably accurate. The problem occurs when you have more than one system measuring the same parameter, and the results differ, then it is difficult to decide which if any are indeed accurate. This is a particularly problem where data is totaled over a period of time, when slight errors can become much more apparent.

Posted by: @squonk

I don't think that's the case Derek. The problem, as Cycleneil described, is the way the Vaillant app rounds numbers up/down gives wildly inaccurate results - bad data in = bad results out. Using different methods such as a proper MMSP (which is presumably more accurate) gives what appear to be more believable results.

Not my blog - I just thought it was interesting and relevant to the discussions on here.

Posted by: @derek-m

@squonk

I was being facetious as usual.

Being an Instrumentation and Control Systems Engineer for over 50 years, producing accurate indications has been my 'bread and butter' for much of that time.

I fully agree that the rounding of data within an App, and thereby providing incorrect data is far from ideal. Providing incorrect data is probably worse than providing no data at all.

I am used to working with systems where accuracy is a prime requirement, but even then there are limitations as to how accurate the system can perform, and at what cost.

As I stated previously, accuracy costs money. An MMSP system, if correctly installed, should give reasonably accurate results, but don't expect it to be completely accurate.

The point that I was making facetiously, is that if you only have one indication that appears reasonable, then one has to assume that it is reasonably accurate. The problem occurs when you have more than one system measuring the same parameter, and the results differ, then it is difficult to decide which if any are indeed accurate. This is a particularly problem where data is totaled over a period of time, when slight errors can become much more apparent.

Ah, ok - I take your point! Sorry, been a long, mostly sleep-deprived week. Agree entirely