@jamespa - agreed. But never one to sit still, I have been thinking about another possible cause for the low OAT deficit, which does involve, indeed is caused by, cycling: the average LWT (and so average radiator temps) is below the set LWT. When we design a system we assume a steady state, but in reality my system is effectively never in a steady state. Thus a rad size based on a LWT of say 55 degrees only actually gets that LWT for part of a cycle. Much of the time the LWT is less, and that might now mean the rad is too small... Consider this, the last complete space heating hour (1300-1400 is DHW time):

For a lot of that hour, from about 1210 to 1245, the actual LWT was below the set LWT, and so the mean rad temp will have been less than it should have been. At around 1215, it looks as though the rads would have had a mean temp of only 35 degrees, about 15 degrees below where it should be. The original rad size calculations did have a margin in reserve, but is it enough to cover this deficit? This is of course a defrost cycle, but they are fully known about, and surely the design calcs should take their inevitable occurrence into account?

Working all this out is rather fiddly - nothing is steady state, actual temps are not design temps, and of course the original heat loss calcs on which the rad sizes were based are we now know over-estimates etc. Nonetheless, I am going to have a go at finding an hour when things were as close as possible to design conditions, and see what was going on then. 

Posted by: @cathoderay

↑

When we design a system we assume a steady state, but in reality my system is effectively never in a steady state. Thus a rad size based on a LWT of say 55 degrees only actually gets that LWT for part of a cycle. Much of the time the LWT is less, and that might now mean the rad is too small... Consider this, the last complete space heating hour (1300-1400 is DHW time):

One might expect (hope) that the heat pump takes this into account in the control circuit, by accumulating differences and overshooting to compensate.

I am not sure if mine does.  It certainly has a concept of 'energy integral', which is the integrated difference (in degree minutes) between actual FT and target FT, and you can see some evidence of overshoot followed by recovery in the defrost cycles below (target FT=38-39C).  I cant quite convince myself that the overshoot fully compensates for the deficit, but it certainly goes some way and the net deficit isn't large.

During this cold weather I've been thinking about the effect of system volume during defrosts.  The smaller it is the harder the heat pump has to work to recover temperature (although ultimately it has to put in the same energy, however it can afford to do it over a bit more time).  The rads in my bedrooms are on TRVs and so can be out of circuit during defrost.  I have just adjusted the lockshield on my biggest rad with a TRV, with the aim of getting it permanently into circuit whilst preserving the temperature differential.  I intend to  do a couple more and see if the dips in FT reduce in magnitude.  Its not simple however, because some of the LSVs leak unless they are fully on and the TRV4s on the other end, which in principle have a 'balancing' feature, are proving difficult to adjust with the crappy plastic key they provide, presumably because the balancing ring has seized to the body!

Posted by: @jamespa

↑

you can see some evidence of overshoot followed by recovery in the defrost cycles below

I agree, but is that by design or by accident? There is also some evidence of overshoot at the end, not beginning, of the cycles in the chart I posted above, but not in the chart I will post shortly which looks at what happens when running at design OAT (-2 degrees).

I have all my system in play, all my TRV valves are fully open, and I am reasonably confident the system volume is adequate, at least as adequate as anything can be against the perversity of a defrost. I know they are necessary, but to pump all that heat into the house, and then suck it out again... The massive K3 rad in the kitchen holds 15.8L, the living room K3 rad has 13.6L, and there are ten radiators in all. 

My current living room IAT is 17.3 degrees, and the kitchen is a bit colder, at 16.5 degrees.

Posted by: @cathoderay

↑

I agree, but is that by design or by accident?

It claims to use the energy integral as part of the cycling control loop, but is silent about how the defrost control loop works, so I dont know!

Posted by: @cathoderay

↑

My current living room IAT is 17.3 degrees, and the kitchen is a bit colder, at 16.5 degrees

Im sorry to hear that, thats a bit cool.  Im sure that will influence your attitude towards Freedom/your installer.  

Righty ho, here we go! The question is: does the lower mean LWT caused by cycling critically reduce output at low OATs?

Current best estimate of heat loss at -2C degrees OAT = 8.6kW. This is based on the 2023/24 heating season using hour data but excluding DHW hours, setback days and hours when the OAT was below 5 degrees (because odd things start happening then). The chart and regression equation look like this:

Next we need a period when the OAT was -2 degrees. Here is such a period from the night before last: 

Unsurprisingly, yet more surprises! There seems to be an alternating pattern for the cycles, slower rise followed by faster followed by slower rise etc, with matching compressor activity. I have no idea why that happens. Nonetheless, the set LWT is where it should be at 52 or mostly 53 degrees (the left hand end of my WCC is 56 at -4), and for at least some of the time the heat pump gets there, and even achieves periods of steady state running and apparent modulation of the flow rate. However much of the time the actual LWT is below the set LWT. We can nonetheless calculate the mean temp (mean of mean LWT and mean RWT) for the entire period, and it comes out at 45.4 degrees.

However this is the mean at the heat pump, the mean at the radiators will be lower because of pipework and PHE losses. I have no way of measuring them, and so will have to use a guesstimate, lets say 10%. That gives a mean rad temp of 40.86, let's say 41 degrees.

Now this is quite a lot less than the mean rad temp of 50 degrees used in the original radiator calculations. Here is what that calculation, which is based on Stelrad Correction Factors, looked like back in early 2022, when I knew rather less than I do now). Ignore the Utility Room, it isn't really part of the habitable rooms, and various other quirks:

As you can see, all the rads are adequate, with a mean rad temp of 50 degrees. This is even the case with the at the over-estimate of heat loss in use at the time.

Now what happens if we enter a more realistic mean rad temp, and compare the total rad output to what we currently believe to be the heat loss at -2 degrees OAT? 

The total rad output at this mean rad temp is 8939W or 8.9kW. This is just, but only just, above the current best estimate of heat loss at -2 degrees of 8.6kW. It seems that the over-sizing of the rads caused by over-estimating the heat loss may just have saved the day.

However, the margin is small, 300W, or three 100W incandescent bulbs in old money, and the losses in the system only needs to be a bit less for the heat supply to fall below the heat demand...

Posted by: @jamespa

↑

Im sorry to hear that, thats a bit cool.  Im sure that will influence your attitude towards Freedom/your installer.  

I shall endeavour, as I always do, to keep relations between Freedom and me as warm and as cordial as can be. 

Posted by: @cathoderay

↑

...

The total rad output at this mean rad temp is 8939W or 8.9kW. This is just, but only just, above the current best estimate of heat loss at -2 degrees of 8.6kW. It seems that the over-sizing of the rads caused by over-estimating the heat loss may just have saved the day.

However, the margin is small, 300W, or three 100W incandescent bulbs in old money, and the losses in the system only needs to be a bit less for the heat supply to fall below the heat demand...

I confess Im a bit mystified.  How was the WC curve determined.  If it was determined experimentally, then cycling and losses in PHE etc would be, at least in part, accounted for and the values would end up set above the theoretical ones to get the right house temperature.  Obviously if the WC curve was a product of whatiffery then not that's not the case, but it doesn't seem like you!

Posted by: @jamespa

↑

Obviously if the WC curve was a product of whatiffery then not that's not the case, but it doesn't seem like you!

Quite right, definitely not a product of whatiffery! Instead, empirical trial and error, which is ongoing. One thing I have picked up recently is that different seasons probably need slightly different curves, eg winter weather often has other severe elements above an beyond the OAT itself plus less solar gain so the curve needs to go up a bit in winter.

I am also mindful that the lower OAT end is already near the upper limit of sane LWTs. It is currently as I said 56 at -4 degrees, 56 degrees because my auto-adapt script (the one that measures desired vs actual IAT and makes appropriate WCC end point adjustments) has the potential to add another three degrees (but no more), taking it up to 59 degrees, at which point steam might well start coming out of my ears if not the heat pumps vents.  

Posted by: @jamespa

↑

If it was determined experimentally, then cycling and losses in PHE etc would be, at least in part, accounted for and the values would end up set above the theoretical ones to get the right house temperature. 

Good point. It is also in effect another way of expressing my idea that the IAT is the sum of all that has gone into it, including things both known and unknown.

I think the key thing here (the consequences of actual mean LWT being lower than set LWT) is that my system is borderline. There may be things I have got wrong, possibly as simple as typos, or factors I have not taken into account, and 300W is not a lot of headroom. 

Posted by: @cathoderay

↑

I think the key thing here (the consequences of actual mean LWT being lower than set LWT) is that my system is borderline. There may be things I have got wrong, possibly as simple as typos, or factors I have not taken into account, and 300W is not a lot of headroom. 

I think you are very likely right unfortunately.  This is a somewhat surprising conclusion given that the system was presumably designed by Freedom and given the industry tendency to over-estimate rather than under-estimate.

Its a pity you don't have a reason (or do you/can you concoct one?) to persuade Freedom to authorise stripping out the PHE and secondary pump.  Together with fixing whatever is causing the under capacity at low OATs, its very likely this would give you more than adequate margin and better efficiency.  It would also simplify problem diagnosis dramatically.  Unless the parts are particularly inaccessible it could probably be done reversibly.  It would be good to understand whether the pressure drop across the PHE is likely to be less or more than the pressure drop across your rads.  If its likely that the PHE is more, that would be another bonus.  The converse is unfortunately equally true!

Posted by: @jamespa

↑

This is a somewhat surprising conclusion given that the system was presumably designed by Freedom and given the industry tendency to over-estimate rather than under-estimate.

Short version: the Freedom Heat Pumps Tool Kit contains not one but two errors. If you leave both in place, you get a (meaningless) MCS pass, if you correct just the heat pump output, you get an MCS fail, but if you correct both you once again, and now meaningfully, get an MCS pass which, as it turns out, more by accident than design, is an over-estimate, but not by a lot. 

Long version...   

This is a whole can of worms with several different chambers in the can all full of their own mutant worms. There are however three main divisions. The first is design competence, the second is who is responsible, and the third is that at the end of the day, the first two probably don't matter!

Freedom's business model is a sort of pseudo franchise model. The offer training and installation kits to installers, who then do installations the Freedom way. The emphasis is on speed and ease, with their main heavily Freedom branded installation manual titled 'Quick Installation Manual' to set the tone. It has lots of white space and pictures and Freedom blue everywhere. All that is missing is Bob the Builder Quick Tips. In the interests of speed and ease, the instructions are direct, even blunt at times, eg the infamous every system must have a header buffer or PHE instruction. They also provide their now well known Freedom Heat Pumps Tool Kit spreadsheet, which really is quite something, because it does just about everything that needs to be done to design, make compliant (eg it produces a DNO form) and even certify (it even fills in most of the the MCS certificate) a system. In this sense, the Freedom Heat Pump Tool Kit is the designer, and the designer is the Freedom Heat Pump Tool Kit.

In my case, the Freedom Heat Pumps Tool Kit went through several iterations and even versions. This happened largely because of the vagaries of the grant system, including a long detour via a grant administrator 'preferred' installer whose main claim to fame was incompetence, and the constraints of being a listed building, which meant I had to get both Planning Permission and Listed Building Consent. The latter is why I went for a Midea unit, because, unlike for example a Samsung unit, it is relatively low profile, and this has the major benefit of reducing visual impact. Time pressure was also a major factor. I finally said enough was enough with the time wasting and incompetent grant administrator preferred installer in the days before Christmas 2021, which left only three months (the grant closed on the 31st of March 2022) to appoint another installer and get the installation completed and signed off.

To speed things up, I went back to an installer I had previously contacted, before the LAD (local authority delivery) grant was an option. Of all those I contacted, he was my preferred installer. As the quote details firmed up, we settled on a Midea 16kW unit with mostly K3 rads and a design flow temp of 50 degrees. The choice of K3s and a high flow temp was a necessary compromise to get radiator wall prints (amount of wall taken up) that were tolerable, a lower flow temp or less emitter dense rads would have turned my listed building into a radiator warehouse.

Fate still had one more card to play. In the week before the installation was due to start, it turned out Freedom had no Midea 16kW units in stock. It was the time when as I recall it a fire in a Far East company supplying silicon chips and a container ship going aground in the Suez canal meant a run on Midea units because Mitsubishi units were unavailable, or whatever. They did however have a 14kW unit (at the time we didn't know about the dip switches that can turn a 14Kw unit into a 16kW unit). We re-ran the spreadsheet (which did need a small modification to expose the 14kW model data), and the result came back borderline positive, provided we upped the flow rate to 55 degrees. It was MCS compliant, but only by 1%. What we didn't know at the time was that Freedom had over-estimated the Midea 14kW unit outputs in their toolkit. With hindsight, there was a red flag, hidden in cell AL90: "the midea figures are mostly made up gh had to average them" but at the time we took the view the data wouldn't be there unless Freedom had confidence it was good enough to be there. One might even say that if Freedom didn't have that confidence, then it was reckless to include it.

All of which is to say the heat pump sizing part of the design was done by the Freedom Heat Pumps Toolkit, albeit with data entry by my installer and the minor tweak by me to expose the Midea 14kW data. All the Midea models are in the data, including the 12, 14 and 16kW models, but of these three, only the 12 and 16kW models appeared in the relevant drop down, I just added the 14kW one to the drop down, to make it available in the spreadsheet, as that was the model Freedom had in stock. 

Another undoubted factor that didn't help was the extreme time pressure created by the grant deadline. Basically, it was go ahead with the 14kW unit or cancel the installation: there wasn't time to wait for a 16kW model to arrive (estimate was months). This is what the final immediately pre installation 'Your system' worksheet in the tool kit looked like: 

Note how the system is given the MCS green light: in cell B17 for 'Will the system qualify for MCS?' the answer is YES. This cell is filled in by the speadsheet, the spreadsheet's logic and formulas provide the answer. Likewise cell B50 shows the heat pump can supply 101% of the load, which is another way of saying it is an MCS pass (the heat pump must supply (at least) 100% of the load), so this also tells, us does the graph above, that this is an MCS pass, but only just (by 1%).

Faced with a clear statement from the spreadsheet that this was an MCS pass, albeit borderline, and given the deadline pressure and the fact was I exhausted by months of fighting red tape, incompetent installers and all the paraphernalia put in place to discourage heat pump installations, I agreed, and it went ahead. The alternative was going back to oil central heating. What I didn't know then, but I do know now, is that I should perhaps have paid more attention to "the midea figures are mostly made up gh had to average them". I say perhaps because if I had, the result would almost certainly have been a cancelled installation, and a return to oil central heating for me. Would that have been the right outcome?

Buried in that long account is an explanation of why the heat pump wasn't (at this stage in the proceedings, at least according to the Freedom Tool kit) substantially oversized relative to the heat loss estimate provided by the spreadsheet: it is because the first choice (and oversized) 16kW model wasn't available at installation time. It was 14kW, or nothing, and as 14kW was an MCS pass, albeit a borderline one, we went with it. What we didn't know at the time was that the "mostly made up" figures were over-optimistic. Everything is so easy with hindsight, what you have to consider is what we could see, and know, at the time.

This is what happens if you put the correct Midea 14kW unit output figures in (figures from the Midea manual): 

As you can see, it is an MCS fail.

Who is responsible for this error (giving the design an MCS pass when in fact it was an MCS fail)? In the event, at the time, both my installer and I relied in good faith on Freedom's Tool Kit. We did know there might be an error - the "mostly made up" figures, but we also knew they were averages of something, and who was to say they weren't at least good enough to use? Why were they there in the first place if they weren't fit to be used? I suggest that what both my installer and I did, which was to rely on suppliers data, was entirely reasonable, and that any fault must lie with the supplier for providing incorrect data.    

HOWEVER - there is ONE HUGE BUT - the MCS fail is based on also taking the Freedom Heat Loss calculation at face value. We now know, many many months down the road, that the Freedom Heat Loss calculator (it is not alone in this) almost certainly over-estimates the loss, whether by accident (whoops) or design (perhaps a covert way to ensure over-size heat pumps are fitted). My actual heat loss at design OAT is almost certainly a LOT less than the Freedom estimate of 12.3kW. Based on the empirical data presented earlier, it is likely to by around 8.6kW. If that is the case, and there is every reason to believe it is, then the system, or at least the capability of the heat pump part of the system, becomes an MCS pass again. All can breath a sigh of relief, at least on that point.

To summarise succinctly (the short version): the Freedom Heat Pumps Tool Kit contains not one but two errors. If you leave both in place, you get a (meaningless) MCS pass, if you correct just the heat pump output, you get an MCS fail, but if you correct both you once again, and now meaningfully, get an MCS pass! 

All of which is to say that the reason my system fails at low OATs in practice is almost certainly not due to the raw capability of the heat pump being insufficient. At relevant low OATs, the heat pump should deliver just over 11kW, more than enough to meet a demand of just under 9kW. Instead, for some other reason yet to be identified, the on paper capability of the heat pump does not in practice get delivered to the house.   

Thanks for that, I think there is much comfort to be drawn because of:

Posted by: @cathoderay

↑

All of which is to say that the reason my system fails at low OATs in practice is almost certainly not due to the raw capability of the heat pump being insufficient. At relevant low OATs, the heat pump should deliver just over 11kW, more than enough to meet a demand of just under 9kW. Instead, for some other reason yet to be identified, the on paper capability of the heat pump does not in practice get delivered to the house.

... with which, based on everything you have told me, I agree. 

My heat pump journey was (is still...) over two years long and frustrating, due in equal measures to a hopeless local planning authority and to the 'industry'.  Is sounds like your heat pump journey has been, if anything, longer and more frustrating than mine.

I guess we have to comfort ourselves that we have chosen to be early adopters, so can expect this sort of nonsense.  I would, frankly, be more comforted if I could see the aggressive actions to sort it out, informed not just be the industry, but by others (such as ourselves) who understand it but have no vested interest.  I hope there may be at least a couple of people in DENZ that have sufficient engineering background, guts and visibility to challenge the industry robustly.  We cant expect (most) politicians to do so, however well intentioned or determined.  They simply don't have the technical background and thus can all to easily have the wool pulled over their eyes.  Unfortunately, if @transparent is right (and I have no reason to disbelieve him) critical engineering skills and the ability to challenge are sadly lacking in DENZ.

I particularly like the Freudian slip, in that @jamespa  is abbreviating the Department of Energy Security and Net Zero to DENZ.

Can we assume that James thinks they're not actually delivering the Security aspect of their departmental title...?
In which case, I'm inclined to agree!