JamesPa

It needs access only occasionally so temporary balustrading will suffice. I haven't yet worked this out in detail but my plan is to mount the unit on unistruct spanned so it bears directly above the walls, so the weight is supported directly on the walls. The same unistruct base could support removable balustrading that can attach to the unistruct when in use, probably 'dropping in' and pinned.

That's really helpful thanks. I plan to mount mine on beams bearing on the walls, not directly on the deck, which should reduce the noise transfer enen further. Your post gives me the confidence to proceed. Now I just need to find an installer who doesn't want to charge me an arm and a leg and put in loads of stuff I don't need! I would DiY it, but my LPA is being stupid over noise levels so it needs to go in under PD, which means 'installed to MCS planning standards'.

Thanks. Any problems with vibration or noise transmission to the inside? What's below?

Scissor lift most likely I don't particularly like the solution, but literally the only viable alternatives are a long-run split, which I don't like either, or under a window at the front of the house, which i quite like (qith the right 'meaningful' design) but neither my wife nor the planning department do like.

I confess to not really understanding what the valid, practical, use case is for this type of water heater in a typical domestic situation. Fundamentally you are taking heat energy from somewhere, so unless you don't need that heat energy and it renews itself from somewhere which you are happy gets cooler, you are inevitably robbing energy from yourself. In summer, if you use it to cool the house and warm hot water then yes, it might make sense (bus so does PV and a solar diverter), but in winter its more difficult to see, unless you have a source of genuinely waste heat from some other process which is otherwise not contributing to warming your building/stopping it cooling. Most of us don't! Hopefully someone can put me right and convince me that these are not just exploitative marketing.

Pretty much the only location for my ASHP is on the (easily accessible) flat roof of the garage. The roof span is 2.7m and Im thinking about whether it would be sensible to support the ASHP on separate beams bearing on a part of the deck which is supported immediately underneath by the roof beams and is directly above the walls (so that effectively the support bears directly on the walls). Thus there would be absolutely no chance that the roof deck becomes an amplifier for vibration and no doubt that the load can be supported. The unit (Mitsubishi 11.2kW) weighs 120kg. The load tables say that Unistruct P1001 will take the load easily, so that sitting on the usual type of feet for aircon is one option although that means it will be 175mm or so above the deck. Am I over-thinking it, or is this a sensibly prudent approach? Does anyone have any experience of a moderately heavy unit on a flat roof. In a commercial environment aircon on a flat roof seems pretty much the norm, but for some reason domestic installers don't seem to like the idea. All commentary/inspiration welcome.

Pretty much. Phew, we got there in the end but... Unfortunately I don't understand the logic to your 4P Buffer. What you describe happening can be explained by the buffer being fully mixed, so it could just be a volumiser in your flow or return line (return would be better). The their part controller does add to the evidence that your package is a one-size fits all package that de-risks the installation for the installer. Not really a problem as long as its able to keep the property warm. it does seem like correct buffer 'control' (by the heat pump) might matter. Otherwise the buffer tank de-stratifies because either the CH flow or the HP flow dominates. Many of the HP control units for which I have read the instructions can have buffer temp probes. Are these important I wonder.

Interesting. Lets see how @IanRresponds to my latest attempt (above) to rationalise what is going on. If I have it about right then it might be important that the HP both knows about the presence of, and also knows the temperature of, a 4 port buffer (but not so a 2 port buffer), because its that knowledge that effectively compensates for differential pump flows and ensures that, over time, the thermocline doesn't drift to the top or the bottom of the tank.

Nothing wrong with a2a, I have it at work in a couple of buildings. It's cheap to install and you get warm air quickly. It delivers a good CoP. Multisplit, for some reason, seems to be more expensive than multiple single units. However the fans do make a noise and some might not like the breeze. Also, without a reserve of hot water, the defrost cycle can be a bit cool. Domestic buildings tend to have more smaller rooms than commercial which isn't such a great match to A2A, but a big open-plan house might be. Its really about what you prefer and layout. Given that wet systems are well established in the domestic market I would expect A2A to remain a minority, but that doesn't mean it is not worth considering.

Yes that's exactly what I mean. For this to happen water must actually move upwards in the tank, but I agree that the integrity of the boundary could be maintained, at least until it pushes so far up the tank that its nearing the top Whist the analogy with DHW is useful to understand turbulence within tanks, it can only be pushed so far. We run DHW for minutes and in any given run extract only a portion of the tank volume. We run CH continuously and the volume passing through them is a fairly short period of time well exceeds the tank volume. I think its time to 'do the math', it may well be that we agree on the outcome, if not actually the mechanism. Lets consider an 8kW heating system operating at approximately full load, with a 200l tank and (since this is a UFH thread), a deltaT across the emitter of 4C. Roughly 0.5l/s of water must be delivered to the load. If the HP pump and the CH pump are operating at exactly the same speed, the tank water will remain static except at top and bottom where 0.5l/s will pass laterally between input flow and output flow/input return and output return. A static thermal profile will establish. Lets call this the 'baseline' If the HP pump is stopped, then it will take 400s, 6.6 mins, for the contents of the tank to cycle round the CH system, the original thermocline will have moved fully through the tank (and possibly replaced by a new one at a lower temp) and the water exiting the tank will be (roughly) at the initial return temperature. If, alternatively, we have both the HP pump and the CH pump on and the HP pump is operating at half the speed of the CH pump, it will take 800s, 13 mins, for the water initially at the bottom to rise to the top. But rise to the top it must because otherwise there is more water leaving at the top than entering at the top. The thermocline will be pushed up the tank relative to the baseline. Eventually half of the water exiting the tank at the top will have come from the bottom, heating up as it is pushed up, but not quite as high as the original temperature. If the system remains like this it will eventually reach a new steady state where there is a flow temp difference, however small, across the buffer tank. If, we have both the HP pump and the CH pump on and the HP pump is operating at twice the speed of the CH pump, it will take 800s, 13 mins, for the water initially at the top to rise to the bottom, cooling as it falls. Eventually half of the water exiting the tank at the bottom will have come from the top, cooling down as it is pushed down, but not quite as low as the original temperature. If the system remains like this it will eventually reach a new steady state where there will be a return temp difference, however small, across the buffer tank. I imagine what happens in a properly designed system is actually none of these. I expect that the HP can deliver more than the maximum load required by the CH. If the HP pump switches off for a time whilst the CH pump continues running (because the demand from the buffer tank is satisfied but not the demand from the load), the thermocline moves up with the excess water entering at the bottom. However the HP pump switches on again before the cool water reaches the top. The HP pump then, for a while, delivers water at a faster rate than the CH pump, and the thermocline moves down again as hot water is pushed down the tank. This cycle can continue indefinitely, the control system of the HP effectively balancing the input and output to the tank even though the pumps themselves are not balanced. The water in the centre of the tank gently oscillates up and down the tank, but plausibly never leaves it. Thus, over time, the appearance of a quasi steady state is maintained (it would be interesting to build a glass tank, introduce some dye into the water, and watch it move!) Is this more or less correct in your view?

Im not now sure if we are agreeing or disagreeing! Assuming that both HP flow and CH flow are at the top and HP return and CH return at the bottom, lets start with an extreme example. Suppose we switch off the HP pump but leave the CH pump on. Nothing will flow through the HP circuit. Over time cool water will come in at the bottom (from the CH pump) and slowly both the actual water and the thermal gradient will move up to the top. Eventually the water at the top will be more or less the same temperature as the water coming in at the bottom. It will take a while to get to this point because the circulating water in the CH system will continue to give up energy to the house, and thus return to the tank cooler, until it has cooled down completely. That's similar behaviour as a DHW tank when you switch off the element. If we switch the HP back on but it is running at a lower flow rate than the CH pump, then there will be a slower movement of the, initially cool, water molecules from the bottom to the top, at a rate equal to the difference in pump flow rates. If this movement were not taking place, then more water would be entering at the bottom (at the flow rate of the CH pump) than is leaving at the bottom (at the lower flow rate of the HP pump), and more water would be exiting at the top than is entering at the top,. This can only be resolved by the movement of water from bottom to top. The water may well move 'slab-like', but it must move. As the cool water molecules are pushed up they will, if the heat pump is on, exchange energy with the warm water molecules, so the thermal gradient will be maintained. If everything is left for a while, it will eventually reach a steady state. The movement doesn't, if its sufficiently slow and is not too turbulent, destroy stratification, but it does involve energy exchange. In principle the energy exchange could be purely by conduction, but more likely it also involves water mixing in the top stratum as the rising cooler water encounters the turbulence likely caused by the hot water flowing in and out of the ports. If the movement through the tank is sufficiently slow and not turbulent (other than, possibly, at the top) I would expect that the penalty would be small, quite possibly negligible. This could be verified by measuring the temperature drop from flow inlet to flow outlet across the buffer tank. Basically Im saying that I agree that a 4 port buffer tank with little turbulance operated with pumps that have reasonably similar flow rates will not incur a penalty. However If the flow rates aren't reasonably similar (particularly if the CH flow is faster than the HP flow) or there is significant turbulence, then this may not be the case.

Sorry but so far as I can see there must be if the flow from the pump on the hp side is not equal to the flow from the pump on the ch side. The only way to balance the flows into and out of the tank if the two pumps are operating at different flow rates, is for there to be a top to bottom (or bottom to top) flow as well as a lateral flow. Of course as the water flowing from top to bottom moves through the tank it will heat up or cool down by robbing heat from (or donating heat to) the body of water in the tank. But one way or another it must move through the tank and must exchange heat with the water it mixes with. I'm not suggesting that this means that stratification is necessarily upset, as you say that will depend on flow rates (and times - eg if one pump is off for long enough the stratification will change), but still exchange of energy takes place.

OK Im being slow today. The reason the caleffi 3 port buffer diagram has 2 pumps is that, without 2, water from the HP would just circulate through the buffer. Also, but turning on the CH pump and turning off the HP pump, the Ch can be fed from the buffer. Likewise by turning on the HP pump but not the CH pump, the HP can dump heat into the buffer. That's makes for a lot of flexibility that a bypass valve alone can't achieve. Incidentally caleffi have something to say about stratification "Buffer tanks connected to heat pumps tend to have minimal temperature stratification. This happens because most heat pumps have recommended flow rates of 3 gpm per ton (12,000 Btu/hr) of capacity. A typical 4-ton air to water heat pump operating at these conditions would “turn over” an 80-gallon buffer in less than 7 minutes. Those flow rates, especially if introduced vertically into the tank, create lots of internal mixing.'. As per above this may or may not be true (btw 1 ton = 3.5kW in sane units)

If the flow rate of the pump on the heat pump side of a 4 port buffer tank is H, and the flow rate on the central heating side of the tank is C, the H-C must flow from top to bottom (or bottom to top) to balance flows. If H-C is small in comparison with C, the buffer is well designed and the flow rates are sufficiently low that there is no turbulence, then it is indeed plausible that this difference in flow rates of the two pumps is the only source of mixing. A bottom to top flow will cool the flow to the CH slightly, so the flow temp from the HP will need to be increased a little. This does impact efficiency, but if the delta T across the emitters is small and the top to bottom flow is also small, then its again plausible that this is a small effect. A top to bottom flow will warm the return slightly, and increase its volume. There must be a penalty for this, if only the excess pump energy, otherwise we would have a perpetual motion machine, but again its plausible that this is a small effect. So with reasonably well balanced pumps (and in particular, perhaps, a pump on the HP side that is a little faster than that on the CH side), a well designed buffer tank and a flow rate sufficiently low that there is not material turbulence in the buffer tank, its plausible that the degradation is small and outweighed by the gains. Some measurements of the temp drop across a 4 port buffer tank (flow to flow), which sadly nobody seems to be in a position to post, would help understand this. Of course if the flow rates are close it still begs the question, why is a 2 port buffer tank (and one fewer pump) not sufficient, but of course there are so many variants of CH design and requirement that there are almost bound to be cases where it isn't! My enthusiasm for Occam's razor still suggests to me, however, that a 4 port tank and 2 pumps should not be deployed unless there is a clearly identifiable reason to do so. Whilst on buffer tanks I do think the Caleffi 3 port tank is worth a bit more discussion, For some reason, which I don't currently understand, most of their schematics, including this one, show 2 pumps. But a variant of this with a single pump and a bypass valve arranged so that the flow goes either direct to the CH system (when there is demand) or through the buffer tank (when there is not) might have some interesting properties. If anyone does want to explore this perhaps it should be a separate thread!

Relays are certainly one option. I think you need 2 to make a NOR. Bypass valves rely on differential pressure (I think), so (crudely speaking) a bypass valve opens when the alternative route is blocked. Put a tee in the flow, one leg goes to the heating circuit, one to the buffer. The one to the buffer has a bypass valve in it. The bypass will open and water flow via the buffer tank when the heating circuit shuts down (I think). It might be worth checking out the three port configuration advocated by caleffi https://idronics.caleffi.com/magazine/27-air-water-heat-pump-systems I still don't quite get these but the logic seems prima facie believable.

Is the latter really the case. Surely system volume (which determines hp run time) is system volume whether the buffer is 2, 3 or 4 ports or indeed if the volume is in the emitters. If the HP switches off but the circulator pump keeps running, heat is still delivered to the emitters however many ports the buffer has. Again I admit I may be missing something, what is it that I'm missing?

I don't believe that's the case, they are designed for stratification, ie. no vertical mixing. Sorry but I'm missing why you would have 4 ports if you don't need mixing? When I say that they are 'designed for mixing' what I'm really saying is 'thats part of their function'. Of course it's also called 'hydronic balancing' but doesn't that inherently imply mixing? If the flow from the HP is switched off and the flow to the ch is on, then the water circulates vertically through the buffer defeating the stratification.

It might be that one pump, a 2 port buffer in the return and a bypass valve is the optimum in this (quite common) circumstance, or the 3-port combined buffer/bypass (which I don't yet fully understand) described here: https://idronics.caleffi.com/magazine/27-air-water-heat-pump-systems The former, at least, would guarantee no mixing and eliminate the need for the second pump. Of course there may be other reasons for the second pump in your system and we don't yet know whether any mixing actually occurs, so this is pure speculation. But it is what I would now default to with the benefit of the wisdom on this forum, whereas 2 years ago, although I was uncomfortable with the 'always fit a 4 port buffer tank and 2 pumps' mantra, I couldn't objectively defend my discomfort).

my initial response was "Could well be, and I agree that this would to some extent mitigate the efficiency loss of the higher flow temp." Thinking about this again, I'm now not so sure. Whilst the flow temp may never reach 55C, it quite likely that the refrigerant is at whatever temperature/pressure corresponds to a flow temp of 55C. It depends on how the HP operates. I imagine (but don't know) that there is a target refrigerant compression ratio for any given target flow temp and that, as soon as you ask for a higher flow temp, the compression ratio is hiked up. If this is the case then the HP will be operating at the CoP corresponding to a flow temp of 55 even though the flow temp itself has not yet reached 55.

Nowhere so far as I know. But 4 port buffers are designed for mixing (otherwise why have 4 ports?). They may well have baffles to encourage some stratification but, if there is mixing, water must be forced past the baffles. Mixing inevitably reduces heat pump efficiency but, if you need it you need it, and thus are obliged to suffer the degradation. If you don't need it then its easily avoided by using a 2 port buffer if system volume is the issue, or no buffer if it isn't. Basically don't use 4 ports unless you need 4 ports. Am I missing something or is it as simple as that? (for simplicity I've left out discussion of the 'idronics' 3port combined buffer/bypass mentioned in another thread. That deserves a bit more attention as it may have some interesting potential in systems where volume is marginal but mixing is not required).

Is the take away from all of this, if you don't require mixing for some identifiable reason, use only one pump and either a 2port buffer on no buffer?

I can't imagine why a 4 port buffer would be designed to mix. The internals of mine are designed very much to not mix. My controller and buffer allows for three buffer temps to be used, top, centre and bottom, but I'm not using that option on my setup. The fact that there are 3 pockets within the buffer for temperature measurement suggests the manufacture believes there will be stratification within their buffer. I agree with all of this which is why it would be good if we had some actual measurements of temp drop across a buffer tank so we could see how well design translates into practice. Others have suggested that there is no or almost no stratification and some facts would help. Of course I'm not suggesting you should make the measurements, only that it would be good if someone did (for the avoidance of doubt I don't have a buffer tank so cant make a measurement).

Interested to know if anyone has ever attempted AND achieved this? Isn't that what WC aims to do. I get the impression that several here have WC set up well and other controls are temp limiters, to kick in only eg when solar gain takes over, which comes jolly close.

Could well be, and I agree that this would to some extent mitigate the efficiency loss of the higher flow temp. Its still better to adjust the flow temp so that a constant flow matches the demand though, thats the lowest possible flow temp = highest efficiency, and also the most comfortable building because there is no hysterisis.

Just a thought, if your system is as illustrated by @IanR and you have only one pump you might want to run a pipe from B to C (with stop taps at B, C and in the middle of the pipe). This would enable you to bypass the buffer on the flow and, if there is complete mixing in the buffer as you suppose, reduce your WC curve by ~4C gaining perhaps 8-15% system efficiency. If it doesn't work open the taps at B and C and close the mid pipe tap. A summer job i would suggest.