Part Thirty Five – Hot Water And LEDs

This thirty fifth entry was published originally by JSHarris on the 13th December 2014 and received 2,255 views on the closed forum


I thought a bit more detail on the hot water system and some of the pitfalls of using LED lighting might be useful. I’ve added an instant water heater to the thermal store mixer valve output to allow for occasions when the thermal store temperature drops below that needed to deliver useful hot water. I’ve also found a few failings with using LED lighting that are worth sweeping up into one entry, in the hope that it may help others not make the same mistakes.

The hot water is provided by a combination of three energy sources now. The thermal store is pre-heated to around 35 to 40 deg C by the air source heat pump, whenever that is running in heating mode. It is then boosted up to around 70 deg C by the immersion heater, which is only driven by excess electricity from the photovoltaic panels. Whenever the panels are generating more power than the rest of the house is drawing, the excess is fed to the immersion heater. On bright days this will easily boost the thermal store up to temperature, but on cloudy winter days I’ve found that it often does little or nothing to increase the thermal store temperature at all. I only discovered this recently, with the onset of winter and the rapid drop in PV output.

I had planned a way to boost the thermal store temperature on dull days, by using a time switch connected in series with a thermostat on the thermal store. This was set so that in the early hours of the morning, before we needed hot water for showers, the time switch would turn on. If the thermal store wasn’t up to 65 deg C half way up the tank then the thermostat would also turn on and bypass the excess PV energy diverter, feeding mains power directly to the immersion heater. As soon as the thermal store was up to temperature the immersion would turn off.

This worked OK, but I hadn’t anticipated just how little excess PV there would be on some days, which meant that the immersion was drawing a lot more power from the grid than I’d anticipated. This power would be drawn irrespective of whether we actually needed that much hot water the following day or not. It also meant that the thermal store was always very hot and so the losses from it were higher than they needed to be (around 1.5kWh per day), which wasn’t a real problem in efficiency terms, as it’s inside the house, but was a problem for the bedroom next to the services room, which got a bit too warm from the heat leakage.

After some digging around I discovered a couple of German companies that made instant water heaters that would accept warm, or even hot, water at the input. Externally these look like electric shower units, but inside they are a lot more sophisticated. Unlike an electric shower, these heaters have an electronic power control system, that regulates the amount of electrical power delivered to the heating elements in order to control the set output water temperature. This means that they only ever draw as much electricity as is needed in order to increase the incoming water temperature to the set point. If the incoming water is already at the right temperature, the heater doesn’t turn on, it just lets the hot water through. If the incoming water is 5 deg lower than the set temperature the heater will supply just enough power to increase it to the set level, and no more. The instant water heater also only uses power when you need hot water, a bit like a combi boiler,and the thermal store can sit at a lower temperature, preheated by the ASHP, with lower standing losses.

I settled on the Stiebel Eltron DHC-E 8/10 model, mainly because Stiebel Eltron in the UK were very helpful on the ‘phone, and understood what I was trying to do. They provided clear technical data that I needed and this showed that their heaters would work fine with warm or hot water at the input. I wish more UK companies would be as helpful, as getting technical information from some of them can be near-impossible. The model of heater I have chosen can deliver a maximum of 9.6kW, which means it can maintain the ~10 litres/minute needed to run a shower at 40 deg C right down to an incoming water temperature of around 29 deg C if it needs to. The ASHP can fairly quickly fully re-charge the buffer and thermal store to 35 to 40 deg C, so in practice the input to the instant water heater is never likely to drop as low as this, meaning we’ll have plenty of hot water no matter what the weather does. They do a larger 12kW model, but I worked out that we should never need more power than the 9.6kW model can deliver. The 12kW model would have needed an extra consumer unit to have been fitted, too, as the maximum rating of RCBO that I could fit to one of the spare points in our 10 way unit was 50A, OK for the 9.6kW unit (the 12kW one needs a 60A breaker).

I’ve redrawn the underfloor heating and hot water system in two parts, the heating part is the most complex and is all on the ground floor:

This system has sort of evolved to deal with some of the things I’ve learned about the way the house responds and to allow for the ASHP to provide some floor cooling in summer. One thing I’ve discovered is that it is very useful to keep the UFH circulating pump running during daylight hours, even with no heating requirement at all. Just circulating water around the slab evens out the temperature a great deal, taking heat from the areas of floor that get a bit of sunshine and moving it to the areas of floor that never see any sunshine at all. The result is that the slab runs at a much more even temperature, even with the heating or cooling off, and makes for a greater apparent comfort level in every room. The only time control I have is a time switch that turns both the heating controller and the circulating pump on in the morning and off in the evening.

The thermostatic controller is home made and has a three way switch, off, heating mode and cooling mode. In heating mode the thermostat does two things. It controls the slab to the set temperature by running the ASHP, which heats both the buffer tank and the UFH supply to the ASHP set temperature (currently 40 deg C). The flow to the UFH is set to about 28 deg using the mixer valve on the manifold, so although the ASHP may be putting out water at 40 deg C into the buffer tank the UFH is only drawing water at a lower temperature, When the slab is at the set temperature it turns off the heat to the floor with an electrically operated thermal valve on the manifold. The ASHP will continue running in hot water mode if the buffer tank isn’t up to the set temperature. Currently I have this set to 33 to 35 deg C at the mid-point of the buffer tank, which means around 38 to 40 deg C at the thermosyphon outlet at the top of the tank. Once the buffer tank is at 35 deg C then the thermostat turns the ASHP off.

In the summer, the thermostat controller can be set to cooling mode. In this mode the flow to the buffer tank is turned off with an electrically operated ball valve (as there’s no need to cool down a tank full of water under the airing cupboard). The ASHP is set to cooling mode (just by opening a dry contact), the UFH thermally operated valve is left open and the operation of the thermostat is reversed, in that it now turns the ASHP on if the floor is above the set point and off when it gets below it. In practice there’s no need to cool the floor below about 18.5 deg, as that is more than enough to ensure that the warmer spots where the sun hits the floor are cooled down.

The hot water system is a fair bit simpler:

Pre-heated water from the 70 litre buffer tank heats the 260 litres of water in the thermal store fairly quickly. With the ASHP running flat out it can take the buffer tank and thermal store from 10 deg C to 35 deg C in around 75 to 90 minutes, although it should never need to do this in practice, as our hot water demand is far less than this each day. In the summer, the excess PV heats the whole thermal store to 70 deg C, via the immersion heater right at the bottom. If there isn’t enough excess PV to heat the thermal store and it only gets to around 35 to 40 deg C, then the instant water heater provides enough boost to provide all the hot water at 42 deg C that we need, especially given the fairly rapid recharge from the ASHP (which can deliver around 6 to 8 kW). In effect, with the ASHP running and the instant water heater running the system can deliver a constant 16 to 17 kW of “instant” heat to the hot water, and a lot more than this for the period when stored heat from the thermal store is being used as well. This is the power available once the thermal store excess heat has been used, i.e. when the thermal store has cooled down to below 29 deg C.

I did some measurements to see what our hot water demand was, using a mix of our usage at our existing house and the flow rates from the taps and shower at the new build. Our current shower (run from a thermostatic mixer valve and a Vaillant gas combi boiler) delivers water at 38 deg C and is far and away our greatest daily hot water demand, using around 150 litres of water at this temperature per day, which with our incoming cold main temperature of around 8 deg C is a daily energy use of around 5.25 kWh.

The shower at the new build delivers water at 9.2 litres/minute when fully on, so a bit less than the shower at our current house and the incoming cold water temperature at the new build is about 9 deg C. This means that a shower running at 38 deg C and 9.4 litres/minute needs a hot water system that’s capable of delivering around 19 kW to the cold water feed for the maximum duration of a shower (which is around 10 minutes, the average is around 7 minutes). 10 minutes at 19 kW is 3.19 kWh. With the buffer tank and thermal store at only 35 deg C it can deliver around 2.3 kWh before the temperature of the store drops to below 29 deg C (assuming no recharge), so around 2/3rds of the energy needed for the shower will come from the ASHP. The instant water heater is only required to provide less than 1 kWh of additional energy to boost the water temperature by a few degrees for a 10 minute shower, which seems a reasonable price to pay for a fairly simple system. The price is the other advantage of this system of boosting the temperature from an ASHP, as the instant water heater only cost £240, plus around another £30 or so in installation materials (50A RCBO plus some cable and pipe fittings, although I had some spare 10mm² cable and pipe fittings, so only had to buy an extra RCBO).

Here’s a photo of the installed water heater, next to the thermal store and above the hot water manifold:

If we are lucky enough to get some useful PV input to the thermal store in winter, then that just directly reduces the amount of power (and so energy) that the instant heater needs to provide. Overall I think this seems a pretty good solution, although had I not been doing this as a DIY job then I’d have opted for an unvented hot water tank, with an indirect coil, rather than a thermal store. There would have been a few advantages in doing this. I’d have had a smaller tank, which would be useful in terms of taking up space. I would have had more usable hot water, as there would be no losses across the heat exchanger coil in the thermal store and the overall rate of heat loss would have been lower (as the combination thermal store we had to use has a higher heat loss rate because of the integrated header tank).

There are some downsides to using an instant water heater of this type, and the main one links directly to the use of LEDs. This instant water heater controls the power by pulsing the power on and off to the heater. The heater draws around 40A or so when it’s pulsed on, and these pulses produce a noticeable voltage drop on the supply (our loop resistance is 0.26 ohms, so every pulse from the water heater drops the supply voltage by around 10V or so). Most appliances won’t even notice this sort of drop, but any LED driver that is dimmable on the mains side will pass that drop directly on to the LED, making it flicker. Unfortunately, all of the LED power supplies I’d bought were dimmable, so all the LEDs flickered like mad, almost like strobe lights, when the water heater first kicked in and got a slug of cool water from the pipe work. The flickering noticeably reduced after a few seconds, but it was clear that it was going to be a real nuisance in the bathrooms.

The fix was easy, just replace the dimmable LED power supplies with non-dimmable ones. This worked a treat, except for another problem. Some of the non-dimmable LED power supplies I tried produced massive amounts of radio frequency interference (RFI). In fact, I suspect the vast majority of the Chinese stuff sold on eBay and the like will be like this, as I’ve now seen several examples of similar, very poor, supplies from Chinese suppliers (often pretending to be based in the UK). All of the dodgy supplies I’ve seen have been correctly marked, with a proper CE mark and all the relevant approvals, so are obviously fakes. I’ve taken a couple apart and they were both bloody dangerous, because they had no effective isolation between the input and output, and were unlawful to use, as they had no form of radio interference suppression, or shielding at all.

The 12V LED supplies that I know are OK are those sold by TLC, and they are slim enough to go into a normal downlighter hole. There are probably others that are OK, but I’d advise buying a sample, wiring it up and seeing if it causes radio interference (the ones I tried completely killed the radio anywhere in the house, so the problem is easy to detect). Having looked around on the web, this seems to be a very common problem, with the market flooded with fraudulently marked Chinese made junk. You really have no way of knowing what you’re buying unless you’ve bought and tested a sample, as even some UK suppliers are selling this junk, without realising that the approval marks are fake.

The biggest headache from interference has been the new panel lights I’ve bought. I bought a panel light for the hallway and that was fine, no interference at all, so the power supply unit was probably OK in terms of the approval marking being genuine. I then bought some very similar round panel lights to replace the MR16 downlighters in the kitchen, mainly because we wanted a slightly softer look. I spent half a day removing the MR16 downlighters and rewiring the lighting wires so that there was switched mains with an earth at each location (previously they’d all been running at 12V, from a single supply unit). Once I’d fitted all the new panel lights I was very impressed with the light quality and brightness (these are 3W LEDs) but found that the radio interference was diabolical. These are constant current driven fittings, with each having a mains-driven 300mA constant current power supply. I’ve taken one of the power supplies apart and it is absolute junk. No screening or suppression components at all, inadequate creep distances between the high voltage AC side and the DC side, no form of filtering on the DC side at all, yet it carries all the right approval marks on the case to show compliance with the EMC Directive, Low Voltage Directive etc.

So, now I’m stuck with 12 very nice panel lights that are unusable. Not only does the interference blank a radio anywhere in the house, but it also blanks the radio in my car parked on the drive, so I suspect they may well be causing problems with our neighbours, too. I’ve been back to the supplier, who was allegedly in London. Sadly this proved to be untrue, the supplier is really in China and was just drop-shipping from a warehouse in the UK. I’ve had a broken English exchange with the supplier, but it’s clear that he doesn’t care and cannot fix the problem. All he has suggested is that I ship the units back to him for a refund, but shipping them back means sending them to China! This isn’t viable, because of the high shipping cost, and I can’t be bothered to fight the case for selling illegal stuff, as I am certain this is just one supplier out of hundreds that’s doing the same thing.

The alternative is to look at ways of running these lights another way. There’s nothing wrong with the light units themselves, they are quite nicely made and have a good appearance, far more discreet than the downlighters I had fitted before. There are a few ways to provide the 300mA DC that these lights need. I looked at buying some mains powered constant current drivers, but it looked as if I could easily have to buy several samples and test them to find out if they were OK, as many looked suspiciously like the ones that came with the lights. I also looked at DC to DC constant current drivers, but most looked a bit expensive and again there was no guarantee that they would be interference-free. I decided to do some measurement on a light to see what its characteristics were. The voltage across the light is fairly constant with temperature, from around 9.5V to 10V when they are drawing a constant 300mA. The lamps don’t get particularly warm when run like this so seem to be well within their rating, and the light output seems the same as when they are run on the dodgy supplies they come with. Looking at the LEDs they use, I think they could be safely run at up to 350mA, or down at 250mA with little loss of light output, which means they don’t really needs an accurate constant current supply, just one that will limit the current to no more than a safe maximum.

One way to drive LEDs from a DC supply is to just put a resistor in series with the LED. The major disadvantage usually cited against doing this for lighting is that it reduces the efficiency. This is true, as the resistor just wastes power as heat, but there is a balance between having something simple and cheap and having something super-efficient. A good switched mode DC to DC constant current power supply might be around 85% efficient. Using a 6.8 ohm resistor in series with the supply to each LED and running them all from a 12V DC supply gives an efficiency of about 83%, which is good enough for me. Each resistor will only waste about 0.6W, which is negligible overall, and less than the power wasted by the old MR16 LEDs (which also used resistors to control the LED current, but more of them). To do this modification I need to make up some leads with a 2.1mm connector and an in-line 6.8 ohm 1 W resistor for each light, and rewiring the kitchen back to run from 12V (easier, as the main change I had to do was reinstate all the cut off earth wires when I converted it back to 240V – the cables were OK but all the earths had been snipped off as they were only feeding 12V). I can go back to using a single 5A 12V DC supply to feed the 12 panel lights plus the LED strip lights under the wall cabinets, a total current demand of about 4.3A. I’ve ordered a load of resistors, 2.1mm inline connectors and some heatshrink sleeving (enough to do more lights than we have) for a total cost of around £20, so I should be able to get a safe, low interference lighting installation installed after a bit of faffing about.

I’ve edited this to add some more information on the worst of the LED drivers I have.

This is the outside view of the driver, with the mains input on the left and the constant current DC output on the right:

You can see that it appears to have approval and certification and carries the CE mark to show that it complies with the EMC directive and the LV Directive.

This is what’s inside the box:

As you can see, there’s not much there, just a switched mode driver chip, fed by half wave rectified mains that’s barely smoothed. There are no suppression components at all, so this unit will transmit big current spikes back into the mains wiring at the switching frequency it runs at, generating a wide spectrum of radiated and conducted harmonics in the mains wiring of the house.

Here’s a close up of the mains input side, with the smoothing capacitor lifted up:

And here’s a close up of the “DC” output side:

If you look at this you can see that the “DC” output isn’t DC at all, it’s just a rectified pulse train from the small transformer with a tiny little capacitor to take the worst of the edges off. What’s fed to the LEDs is a high frequency pulse train that will radiate radio frequency interference of a wide hunk of the radio spectrum (up to and including FM radio). At the very least there should be some smoothing on this, plus some filtering to keep high frequency noise from going down the leads.

The other noticeable point is that there is no screening whatsoever, This unit is like a small wide band radio transmitter, and with no screening it will radiate noise freely.


ProDave 14 Dec 2014 12:01 PM:

As always lots of useful information. I have a feeling my hot water system will be near identical to your system, though I am still favouring an unvented hot water tank rather than a thermal store.

The LED light thing, It is a shame nobody polices these false directive claims. I guess you should be reporting them to trading standards as they are the ones that are supposed to enforce these things. Now if they were on sale in a shop in the UK I am sure they would take an interest, but an ebay seller shipping directly form China, not a hope I am afraid. So it sounds like your idea of re modelling the power supplies is a better idea.

An alternative thought on that, since they want a constant current of 300mA have you considered wiring a bank of lights in series and feeding them from a higher voltage constant current source? Then just one power supply per switched circuit rather than one per lamp?

I think you are convincing me that CCFL lamps are still the better bet. the extra energy efficiency of LED’s seems to come with too much fiddling around to make it work properly and solve all the issues.


 jsharris 14 Dec 2014 12:26 PM:

I’m with you on preferring an unvented hot water tank. It would probably give more usable hot water, would have lower standing losses and would take up a lot less space. My only reason for not fitting one was the rules and regs, had it not been for the need to get a plumber in to do the installation and sign off I’d have fitted an unvented tank without a doubt. The added cost and faff of having to get a plumber in would have pushed the cost up too much, plus I’d have probably had trouble explaining the rather non-standard way I have things set up.

 There is a distinct possibility that the thermal store may be swapped out for an unvented tank post-completion, though, as I reckon that I could probably sell the combination thermal store for a fair price to offset the additional cost of buying a replacement unvented tank.

The world of LED lighting is a nightmare, and it’s not just ebay where there are massive problems with unsafe and falsely marked products. I’ve bought samples from other online retailers in the UK and had Chinese crap that was clearly unsafe and unlawful delivered. In each case I’ve sent the stuff back for a refund, with a note that the certification was fake and that they should check things with their suppliers. I even bought a fake light in B&Q! It was CE marked, but looked identical to a known to be fake Chinese one I already had. Sure enough, when I looked closely at the B&Q one it was a fake too. They were pretty good about it and have taken them off their shelves, but my guess is that there is so much of this stuff out there that not even the big retailers can keep it out of their supply chain,

The problem with daisy-chaining a lot of constant current lights together is that you quickly reach the ELV limit of 60V DC, so cannot then use a DIY solution. In the case of these lights I could wire 6 together as an absolute maximum, perhaps 5 to be safely inside the limit, as each has a forward voltage of around 10V. I have 12 of them in the kitchen, so I’d need to pull through a fair few extra cables above the ceiling, not an easy job as the void is filled with insulation and getting wires through the joists would be a bit of a game, even though they are open posijoists. The only access I have is through the 72mm holes in the ceiling.


jsharris 14 Dec 2014 02:59 PM:

I’ve just edited the above entry to include some photos of a pretty poor LED driver.


ProDave 14 Dec 2014 03:23 PM:

I guess it’s the old story, you get what you pay for?

Re the unvented tank and needing it to be signed off, on my last build neither NHBC or Building control had any problem with me doing it myself (though they did argue between themselves exactly where the vent pipe should terminate)

If I find on the next house that they won’t accept it, then I will disconnect the mains water in, disconnect the final connection of the vent pipe, then get a plumber with the right ticket to come and “connect it and commission it” knowing they only have to (re) install 2 short bits of pipe.


 jsharris 14 Dec 2014 04:27 PM:

Interesting point about the unvented tank sign off. The daft thing in my case is that right next to my thermal store I have a 100 litre pressure vessel working at 5 bar that is my clean cold water buffer supply, and that’s fed from a 300 litre pressure vessel down in the water shed that’s also sitting at 5 bar. Neither require any sign off for building regs. The unvented cylinder would be fed from the 2.8 bar pressure regulating valve on the cold feed, so would work at a bit over half the pressure of the pressure vessels, yet that needs a G3 sign off by someone registered with a competent person scheme (or building control if they have someone who is suitably qualified – many don’t now).


 notnickclegg 15 Dec 2014 11:05 AM :

Great post Jeremy. I’ve been hemming and hawing about committing to an unvented cylinder, and I think this post is the final push I needed.

Edited to add: I just saw one of your replies to your part 34 posting, which answered all of my questions except the last one about the position of the immersion element.

Regarding your ASHP pre-heating, do you set that to 35-40 because of a limitation of the ASHP, because you don’t want to overheat the buffer tank, or because you assumed that the PV would supply more power year-round than is actually the case (or for some other reason)?

I ask because the AHSP we’re looking at goes up to 55 degrees. Even at the lower COP you get at that temp, it’s still roughly twice as good as an immersion heater.

Also, do you think the performance of your system would be better if you had an immersion element higher up in the tank?




 jsharris 15 Dec 2014 06:00 PM:

Thanks for the kind words.

The immersion element is right at the bottom, so that maximum use can be had from diverted excess power from the PV panels.

The ASHP is currently set to 40 deg C, as, although the theoretical COP is still reasonable at higher temperatures in practice the thing goes into frequent defrost cycles and takes forever and a day to heat up the tank in cold weather if it’s set to above 45 deg C. What they don’t tell you in the spec is how long the defrost cycles last or how much heat the ASHP takes from the buffer tank and floor during defrost. The flow temp to the UFH drops to about 12 deg C during defrost, and the cycle takes around 20 minutes to run when the outside temperature is around 0 deg C, with the flow temperature set to 45 deg C. This means the ASHP has to work hard to re-heat the buffer tank during the next cycle, before it then decides it needs to defrost again within maybe an hour or so.

If the flow is set to 40 deg C it hardly ever seems to need to defrost, and when it does the cycle only lasts for around half the time. I’m going to hazard a guess that many of the problems people have with ASHPs is when they try and run them at a high temperature, thinking they are going to still get a reasonable COP (which they will when it’s running) but not appreciating that when the outside temperature is around freezing point the true COP will be very poor, due the constant reversing of the ASHP to defrost (in defrost mode it just reverses and runs in cooling mode, extracting heat from the heating and hot water system in order to melt the ice on the external unit). This is very humidity dependent, so is a real issue around zero deg C, but less of a problem when it’s below about – 5 deg C or above about +5 deg C.

The performance would be worse with the immersion higher up, as the thermal store will stratify and the hottest water will always rise to the top. This means it makes no sense to put the element part way up, as that would then only end up heating a part of the thermal store. The same goes for a properly designed unvented tank, that should also be designed so that the cold water coming in at the bottom to replace the hot drawn from the top is fed in via a diffuser, to prevent destratification. It’s worth looking around, as some unvented tanks are better at maintaining stratification than others.


ProDave 15 Dec 2014 07:13 PM:

A question. Would a tank with two immersion heaters make sense?

Low down one for solar PV dump. Higher one for occasional economy 7 top up and for legionnaires precautions (would heating say 3/4 of the tank to a high temperature be enoggh to kill legionnaires or do you really have to heat the whole tank)


 NSS 15 Dec 2014 07:50 PM:

Out of interest, would these repetitive defrost cycles happen irrespective of whether it’s a monobloc or split system ASHP?


jsharris 15 Dec 2014 08:17 PM:

I think that you could you could use a mid-point located E7 top up heater, to avoid paying too much for direct electricity, as long as you only had an immediate need for the top part of the tank’s worth of hot water.

The defrost is an inherent issue with any ASHP, and is just a function of the external evaporator being very cold and getting iced up with the humidity and evaporator temperature are such as to allow ice to form. It’s only really a problem when the evaporator gets well below freezing point and when the air has enough moisture in it to condense and freeze on it. If the evaporator temperature can be kept just above freezing point, or if the natural on-off cycling of the ASHP is infrequent enough to allow ice to melt between operational cycles, then it wont need to defrost, or if it does the defrost periods will be less frequent, and probably last for a shorter time.



15 Dec 2014 08:46 PM:

Okay, thanks Jeremy


 notnickclegg 16 Dec 2014 08:56 AM :

Hi Jeremy

I should have been clearer – I meant an additional immersion heater rather than moving the sole heater to the middle.

My thinking was that stratification will allow more useful heat to be extracted when there isn’t enough energy to heat the whole tank. So starting with the tank at, say, 30 degrees, putting your limited PV into the top of the tank might give you 50 degrees at the top of the tank, whereas heating the whole tank will leave you with a tank full of 40 degree water. The former might be a more useful situation. The lower temp in the bottom of the tank would also allow a better COP if the ASHP is in water heating mode, too.

Trade-offs include the extra cost of an immersion heater boss and element, and extra complexity (probably want a thermostat that switches to the lower element once the top half of the tank is up to temp).


 jsharris 16 Dec 2014 03:45 PM:

There’s only really one reason for fitting a mid-point immersion, and that’s if, for some reason, you want to limit the amount of available hot water. The only application where this is normally useful is in the case of an E7 hot water system, where the lower immersion is run from the E7 supply and a boost immersion is fitted towards the top so that when on daytime boost only a part of the tank is heated, to save money.

A single immersion at the bottom will always heat the top of the tank first anyway, because of convection within the tank. One part way up would do exactly the same, but would only heat a part of the tank before its thermostat cut in.

An immersion at the bottom can be easily limited to only heat half a tankful if you have a need for this (leaving the bottom cool) by fitting a tank thermostat part way up the tank.

The tank should stratify better with the immersion at the bottom, although if you wanted to be sure of this there are some devices around that will reduce convection within the tank. Some are internal to the tank (diffusers, plates or the central “chimney” type system) some are external to the tank (like the external immersion heaters fitted into a branch plumbed in top and bottom.

Overall I can’t logically see why an additional immersion in the middle would give any significant advantage in terms of stratification, if this is a concern then maybe just fit an external immersion like the Willis unit, as this will ensure that there is always a small volume of very hot water available pretty quickly after the immersion is turned on,


 notnickclegg 16 Dec 2014 05:03 PM:

I was under the impression that there’s a reasonable amount of mixing in a tank without stratification devices, which is why internal flumes/chimneys and external devices such as the Willis immersion existed in the first place.


 jsharris 16 Dec 2014 06:17 PM:

There doesn’t seem to be any mixing at all in our thermal store, as far as I can tell. With the bottom mounted immersion heating the tank the water gets hot at the top and stays cold at the bottom for ages. I have a tank thermostat about half way up and that consistently reads a lot lower than the top of the store whilst it’s heating, so my guess is that there isn’t much mixing at all going on.

Putting an immersion part way up is still going to cause convection currents so could still cause mixing, I think, but just like with the immersion at the bottom this is fairly insignificant in practice (or seems to be).

I don’t think this is a problem unless you have something like a pump feeding water under pressure into the bottom of the tank, with no sort of diffuser or separator. Any sort of diffuser that will limit the potential for incoming cold water to stir things up should do the job, and as long as the immersion is just above this it should be fine. I suppose there would be little extra cost in adding another boss and immersion, together with some sort of changeover switch between the two, but my guess is that you’ll probably end up running on the bottom one the whole time.


 ProDave 17 Dec 2014 03:01 PM:

A bit more about unvented hot water tanks and sign off.

Today I bumped into our local Gas Safe plumber on a job, not seen him for a while. So I asked him if he had the ticket to sign them off.

His answer was no, he has never been asked to prove anything, building control never ask for it, and he doesn’t know anyone who has done the course. He just fits them and connects them as per manufacturers instructions, job done.


 jsharris 17 Dec 2014 05:04 PM:

I’ve had a look around and it’s a one day course plus registration of one of the Competent Person Scheme bodies on the government register. According to the regs it is a legal requirement that any G3 installation be signed off by a competent person that is registered as such (unlike the ventilation rules where membership of a competent person scheme isn’t specifically required) so my guess is that it’s just another one of those rules that gets ignored a lot.

I’ve been doing a few calculations, using 42 deg C as the DHW delivery temperature, and it seems that my 260 litre thermal store would only deliver around 23 litres of hot water at 42 deg C if charged to 45 deg C, yet a 150 litre unvented cylinder at the same temperature would probably deliver most of its 150 litre capacity at the desired temperature, making it a much better bet. The usable capacity of a thermal store is pretty poor when its only run at a low temperature. This is a plot of our 260 litre thermal store DHW performance when fed with cold water at 8 deg C and delivering 10 litres per minute of hot water:

If charged to 70 deg C it can deliver masses of hot water, but if only charged to 45 deg C (the red dotted lines) then it can only deliver enough hot water with the boost heater running, and that won’t maintain the desired 42 deg C DHW temp if the incoming water temp drops below about 29 deg C (I’ve allowed for a 1 deg C loss in the pipework and cut off the plot at 30 deg C).

A smaller unvented tank with very good insulation would seem to be a better bet, as it could still be heated to a high temperature when there’s excess PV available, and provide all the hot water needed that way, but would also provide more useful warm water when only heated to, say, 40 deg C by the ASHP in winter, with less power needed from the boost heater because of the greater useful volume of hot water available.


 stones 18 Dec 2014 10:21 AM :

“ProDave, on 15 December 2014 – 07:13 PM:, said:

A question. Would a tank with two immersion heaters make sense?

Low down one for solar PV dump. Higher one for occasional economy 7 top up and for legionnaires precautions (would heating say 3/4 of the tank to a high temperature be enoggh to kill legionnaires or do you really have to heat the whole tank)”

I had been planning exactly this when I was looking into replacing my existing system with an ASHP. My thought had been two immersions at the bottom of the cylinder, one in the control of the ASHP for the weekly / fortnightly legionnaire cycle and occasional top up, and a second for Solar PV controlled through whatever solar diversion device was fitted. Getting a cylinder made is no issue, there are several companies who will custom make a cylinder at very little extra cost. Having initially said there wasn’t any issue doing this, one of the installers belatedly advised that I couldn’t locate the solar PV immersion below the heating coil, as it would cause a (high pressure) fault in the ASHP.

I wasn’t able to get a definitive answer to this (one installer said that DHW demand was measured from the return temp of the heating coil in the DHW cylinder, another said is was by cylinder thermistor, Daikin technical said both!) I’ve put my plans on hold meantime so have not done any further research. I would suggest that if considering this, it would be worth asking how the ASHP determines if there is DHW requirement.


 ProDave 18 Dec 2014 10:43 AM :

I can see a potential issue with an immersion directly below the heat input coil.

But surely you wouldn’t be using the heat pump while that immersion was on would you?

And in any case if it’s anything like Jeremy’s system and what I am proposing, the heat input coil to the SHW tank is thermo syphoning from the buffer tank, not directly from the heat pump.

My point of two immersion heaters is we have a requirement for a morning shower and SWMBO likes it hot!!!. So I am trying to find the most efficient way to get enough hot water in the morning, and then using the heat pump / solar pv at other times.

So a heater say half way up would heat enough for a shower.

My final solution will probably be one that gives a lot of flexibility, so 2 immersion heaters at different heights to experiment with, and an instant in line modulating water heater.


 stones 18 Dec 2014 12:26 PM:

In your proposed set up its not going to be a problem as the ASHP will only be heating a buffer. The issue seems ton be for cylinders with input coils fed directly by an ASHP. – lets say you wanted to program the ASHP to heat your DHW cylinder to 45C overnight to make sure you had enough for morning showers. If the PV had charged the cylinder up the day before, there is potential for the temp within the cylinder to be above the maximum 55C return temp the ASHP will accept before it registers a fault. The question is how the ASHP determines whether there is any DHW requirement, and in the case of Daikin at any rate, it appears that it in some way must sample the temperature of the return from the input coil. The answer of course is to have the PV dump immersion above the input coil, but this then means that you will not be able to heat the bottom 1/4 – 1/3 of the cylinder (with PV).


Leave a Reply

Your email address will not be published. Required fields are marked *