Controlling underfloor heating in passive house

[Dudda]

Would it also depend on how many loops you have in the manifold? Due to the floor area I'll have 9 loops all 90-100 meters long from the manifold. Allowing 0.12 litres per meter I'll have about 100 litres of water (~900 meters x 0.12)  in the pipes in my underfloor heating slab. Looking at @TerryE's picture in the other thread he has three loops so max 36 litres of water.

 

Should I have enough water in the slab and system then to avoid the need for a buffer tank?

[Jeremy Harris]

3 minutes ago, Dudda said:

Would it also depend on how many loops you have in the manifold? Due to the floor area I'll have 9 loops all 90-100 meters long from the manifold. Allowing 0.12 litres per meter I'll have about 100 litres of water (~900 meters x 0.12)  in the pipes in my underfloor heating slab. Looking at @TerryE's picture in the other thread he has three loops so max 36 litres of water.

 

Should I have enough water in the slab and system then to avoid the need for a buffer tank?

 

My guess is that with that sort of volume of water you may well be OK without a buffer, as our buffer is only 70 litres, and our UFH is about the same as Terry's, 3 loops of 100m each.

[Dudda]

7 minutes ago, JSHarris said:

 

My guess is that with that sort of volume of water you may well be OK without a buffer, as our buffer is only 70 litres, and our UFH is about the same as Terry's, 3 loops of 100m each.

Thanks. I should have also stated I'll be running it as all one zone too so will always be around 100 litres. If I was splitting it up into separate zones I would consider a buffer as you have.

[PeterW]

On 15/03/2017 at 16:23, Nickfromwales said:

 

I'd 100% be using the ashp for DHW uplift. Why have such a good resource and not use its full ( good CoP ) potential?

A buffer with DHW uplift coil ( so actually a TS ) is a no-brainer afaic,

 

 

Or use a standard tank with a normal coil and a solar coil to do the DHW uplift. 

 

Or use a standard 90 litre indirect tank and use the heat coil as DHW uplift and the bulk of the water as buffer with the ASHP connected directly. This would mean the ASHP could potentially connect directly to the UFH in the event the buffer was cold.  

[Barney12]

Another great thread. One very related question:

 

"Controllers" ............. are the ones supplied with the ASHP generally fine for the scenario's we're discussing here (UFH, Buffers, DHW etc)? Are they all unique to the ASAP (I note Samsung ASHP state you must purchase with their controller)? Or are there "better" generic controllers?

[TerryE]

38 minutes ago, JSHarris said:

Anything lower than about 30 deg C causes the heat pump to cycle too much and it will probably go into anti-short cycle mode, staying off for 15 to 20 mins even when there's a call for heat. 

 

39 minutes ago, JSHarris said:

Our 6 kW to 7 kW ASHP will modulate down to about 1.5 kW output, minimum, and that is usually far more than the UFH needs.

 

Jeremy, I feel that these two statements are mutually contradictory.  As I think that I explained on the Boffin's thread, you will get this sort of behaviour if you set your blender target temperature too low.  Our slab will take 3 kW at 21°C with the input at around 27°C and return at around 24°C, with the 3°C representing the delta temperature of heating our UFH circuits at a 1m/s flow rate which is what the Gunfos pump will do at setting I with our 3×100m loop configuration.    Have a play with the analytic steady state solution to the radial 1D heat equation.  You need the UFH circulation to be at least 3-4°C hotter than the baseline slab temperature to pump 10W/m into the slab, and you've got the 3°C ramp over the length of the loop.  If you limit your rercirc temperature to 25°C then you will be hard pressed to put a 1Kw into the slab.  Try to push any more at this blender setting, and your heater is going to have to cycle.  

 

 @jack feed his ASHP directly into the slab; no buffer tank.   IRC, he just runs his ASHP on a timer.  

 

Anyway, I'll be doing heating trials in a month and I will be able to characterise and compare our real slab and compare this to model predictions.

[TerryE]

1 hour ago, Barney12 said:

Another great thread. One very related question:

 

"Controllers" ............. are the ones supplied with the ASHP generally fine for the scenario's we're discussing here (UFH, Buffers, DHW etc)? Are they all unique to the ASAP (I note Samsung ASHP state you must purchase with their controller)? Or are there "better" generic controllers?

 

I may still decide to install an ASHP, especially if we need to cool the slab during the peak summer months.  I would still stick with my configuration of using the slab itself as my buffer store, but do as @jack does and set the ASHP to a target temperature of 27°C say and my blender valve set at 30°C so that the ASHP will not be slab limited, and run it in chunk mode as I describe in the Boffin's thread, rather than use a room thermostat.  The lags and time constants are such that you can't do this stably without  a buffer tank.  So all my Home Automation system will be doing is to turn the ASHP on for N hours overnight with a possible top up during the day. 

 

[Jeremy Harris]

20 minutes ago, TerryE said:

 

 

Jeremy, I feel that these two statements are mutually contradictory.  As I think that I explained on the Boffin's thread, you will get this sort of behaviour if you set your blender target temperature too low.  Our slab will take 3 kW at 21°C with the input at around 27°C and return at around 24°C, with the 3°C representing the delta temperature of heating our UFH circuits at a 1m/s flow rate which is what the Gunfos pump will do at setting I with our 3×100m loop configuration.    Have a play with the analytic steady state solution to the radial 1D heat equation.  You need the UFH circulation to be at least 3-4°C hotter than the baseline slab temperature to pump 10W/m into the slab, and you've got the 3°C ramp over the length of the loop.  If you limit your rercirc temperature to 25°C then you will be hard pressed to put a 1Kw into the slab.  Try to push any more at this blender setting, and your heater is going to have to cycle.  

 

 @jack feed his ASHP directly into the slab; no buffer tank.   IRC, he just runs his ASHP on a timer.  

 

Anyway, I'll be doing heating trials in a month and I will be able to characterise and compare our real slab and compare this to model predictions.

 

All I can say is that they are based on measurements on our system.  I tried running the UFH without the buffer tank valve open, and when doing that I turned the ASHP flow temperature down to 30 deg C.  When I did that the ASHP went into anti-short cycle mode, and kept shutting down for about 20 minutes, even when there was a call for heat (pretty much the same as a boiler with the same feature will do when asked to modulate down below it's lower limit).  The problem is that the return temperature rises to be very close to the flow temperature and the heat pump does not like that one bit.  There has to be a few degrees temperature differential between flow and return for the thing to stay turned on.

 

We have three 100m loops, too, and what happens is that the thermostatic valve pretty much closes, and the bypass valve opens because of the increased resistance, even with the ASHP internal pump on its lowest setting.  The bypass "short circuits" the ASHP and so causes the return temperature to rise to close to the flow temperature, whereupon the ASHP shuts down and enters the anti-cycle delay.

Edited March 15, 2017 by JSHarris
typo, missing ")"

[Nickfromwales]

31 minutes ago, PeterW said:

 

Or use a standard tank with a normal coil and a solar coil to do the DHW uplift. 

 

Or use a standard 90 litre indirect tank and use the heat coil as DHW uplift and the bulk of the water as buffer with the ASHP connected directly. This would mean the ASHP could potentially connect directly to the UFH in the event the buffer was cold.  

Then you have to introduce antifreeze to the whole system . 

 

[TerryE]

28 minutes ago, JSHarris said:

I tried running the UFH without the buffer tank valve open, and when doing that I turned the ASHP flow temperature down to 30 deg C.  When I did that the ASHP went into anti-short cycle mode, and kept shutting down for about 20 minutes, even when there was a call for heat

 

But did you open your blender to 30°C for this test and have you pump at the minimum setting?   If the ASHP isn't being throttled by the blender then the delta should be at least 3°C and the higher the flow rate the lower the temperature delta.

[Jeremy Harris]

20 minutes ago, TerryE said:

 

But did you open your blender to 30°C for this test and have you pump at the minimum setting?   If the ASHP isn't being throttled by the blender then the delta should be at least 3°C and the higher the flow rate the lower the temperature delta.

 

Not sure what you mean by "blender", all I have is the standard Wunda remote sensor thermostatic valve on the input to the manifold, controlled by a capillary pipe in the upper (supply) manifold.  There's no way I can run the supply to the UFH at 30 deg C, as the room temperature overshoot from doing that is pretty massive.  If I let the supply temperature go no higher than 25 deg C then things stay reasonably well-controlled, and 24 deg C supply temperature is better (but the valve does have a problem trying to hold 24 deg C, so it tends to fluctuate a bit).

 

What happens is that the thermostatic valve pretty much closes after a few minutes, because the supply manifold reaches the set point temperature, and the flow resistance increases on the ASHP circuit.  Before I fitted the bypass valve, the ASHP would then turn off with an over-pressure fault, because it sensed that the flow resistance was too high.  I added the bypass valve to overcome the over-pressure trip problem, but then that results in the return temperature increasing to virtually the same as the flow temperature, which then causes the ASHP to go into anti-short cycle mode.

[dogman]

Ok now i am getting really confused. I think i will have to spend Friday( when MBC are drinking copious amounts of guiness)   reading all the referenced posts and try and make some sense of the different systems.

Biggest change that appeals is the use of sunamps as i will have 4K plus of pv if the DNO allows an extension to my system ( i wont get any extra FIT though)

 

[TerryE]

3 hours ago, JSHarris said:

There's no way I can run the supply to the UFH at 30 deg C, as the room temperature overshoot from doing that is pretty massive.  If I let the supply temperature go no higher than 25 deg C then things stay reasonably well-controlled, and 24 deg C supply temperature is better (but the valve does have a problem trying to hold 24 deg C, so it tends to fluctuate a bit).

 

This is why your ASHP is cycling.  In essence there are two  broad strategies for heating the slab.  If you want to put a chunk of ΔQ joules of heat into the slab in a controlled fashion you can either:

• limit the power going into the slab, and you do this by limiting the ΔT of the heating water and hence you can allow the heat to be added over an extended period.  If you do this and run your ASHP at a higher power output, then this heat has to go somewhere and that's the buffer tank.  You heat up the buffer tank at a quicker rate and then trickle flow this heat into the slab.
• limit the energy going into the slab, and you do this by allowing the dynamics of the heat transfer dictate the ΔT as I explained in my modelling topic.  You don't actively control the  ΔT at all, except that you might want to set the maximum blend temperature as a safety stop.  If you aren't controlling the ΔT then instead you need to control the Δt for which this power is applied.  You compute this and apply the heat as single "chunk" (or possibly two chunks) per day.

In the case of my electric heater, putting in a known  ΔQ is easy since the power output of the heating element is known and fixed.  In the case of an ASHP, even if I don't know the exact power response of the ASHP, it doesn't really matter since 

              

where ΔT is the temperature difference across the manifolds, c_pthe specific heat of water and s the flow rate through the UFH.  In other words since s and c_p are fixed, the only thing that you need to control on is the integral of ΔT.  For example measure the ΔT once per minute and add these up.  Turn off the ASHP once this total exceeds some preset N Ks (degree seconds).

This is easy if you are willing to use one of these   instead of one of these  .

But even if you don't want to use an embedded server to carry out control, then simple alternative asynchronous control strategies could be used for example:

• Simply run the ASHP for a fixed but settable time period.  This will heat the slab by some overall temperature increment. 
• Keep the pump running but then have a dead window to allow the slab temperature to even out.  Say 4 hours.
• Trigger the next heating cycle when the water return temperature from the slab drops below some fixed but settable temperature threshold.  Say 21.5°C
• By a process of trimming set the ASHP on-period (as per the first bullet) so that the heat pump only cycles one or twice a day.

I realise that my whole approach has been documented as a set of evolving discussions on various topics, some of which includes maths which will lose some readers.   But my core approach is that it seems sensible avoiding having to buy and install a 100 ltr buffer tank when I already have a 10 tonne slab that I can use for this purpose.

 

Let me do a summary write up as a blog post.

[Jeremy Harris]

Don't forget that I spent over a year trying to fine tune the slab control system.  I know, beyond doubt, what is possible and what is not possible, not in theory, but in practice.  The key issue is that this is not anywhere near as simple a challenge to find a solution for as I originally thought it should be.  There is no way on this planet I'm going back to using a complex control system, been there, done that, and run around in circles in the process.  We now have a reliable, simple, off-the-shelf, control system that works - it maintains the house at a steady temperature with very little variation, and certainly not a big enough temperature variation to be noticeable.

 

I will admit to being a bit annoyed at just how well the simple control system works, having invested a great deal of time in trying to get the rather complex, multi-sensor  control system to keep the house at a steady temperature.  My switch to the simple room stat control came about by accident, in part, as I loaded a very simple on/off room temperature control code into the microcontroller system one Friday, simply because I wanted to stop the house getting too cold whilst we were away for a break and the system was still failing to reliably control the house temperature.  When we came back from holiday and I downloaded the logged data I was more than a little surprised to find that the simple room stat code had done a far better job of controlling the temperature than any of the complex multi-sensor strategies I'd been playing with over the previous year. 

 

That finding, together with a decision we made whilst away, which was to take out as much customised, home-built stuff as possible, and replace it with off-the-shelf alternatives that anyone could fix, led to the system we have now.  I've no intention of changing it, mainly because it just works very well.  I'm afraid that I have no appetite left for installing critical components in the house that are not fairly easily serviced or replaced when I'm not around to fix them.  Had I managed to get the original microcontroller system to work as theory suggested it should, then I would have been the critical failure point for the whole house heating system, if something happened to me their was every possibility that the house system could fail and there would be no way of fixing it.

 

It is very easy to fall into the trap of over-thinking things, just because you can, I found.  Sometimes the best bet is to just take the simplest solution that's available off-the-shelf and use it, even if you do think you can do better.

Edited March 15, 2017 by JSHarris
typo

[TerryE]

Jeremy, I understand that you have a system that works beautifully for you and given that you've already bought it, installed it and tuned it to meet your needs then there is absolutely no valid reason for you to consider changing it.

 

However my position is that using a buffer tank is not to only method of controlling the heating of an MBC-style slab and passive build.  There is an alternative approach and by adopting it, this considerably simplifies my potable water + UFH heating design and implementation.  At the moment my entire UFH system is under a shelf in an area some 0.8 × 0.3 × 1.4 m³ in volume.  The only mechanical component is my Gunfos pump.  I can add an ASHP simply within this envelope if I have to.

 

The only place I could put a buffer tank would be on the second floor in my storage room so if I went this route then I would need to house it and add pipework from the ground to the second floor.

 

We both have partners that lack our technical background, and so share the "what if I drop dead tomorrow" risk.  In my case my son-in-law is equally into RPis and ESP8266s and IoT hacking, so I have to implement my system in a way that he can understand and maintain it should this happen.  However I also suggest that your wife would also struggle to maintain or repair or find someone to do this for your system.  Yes it might comprise off-the-shelf components, but the system as a whole is configured in a way that is very different to typical UFH systems.

 

We are both pioneers / evangelists of this type of system.  We are adopting variant approaches, that's all.  Your design works.  I have designed and built mine largely informed by your blogs and experience, but with slightly different design goals.  And I am extremely grateful to you for this knowledge.   I will soon find out whether mine works as I've modelled it. 

 

As far as a "complex, multi-sensor  control system" goes, I think that we need to separate out the extensive logging and analysis (that I might need to do to validate and tune my concept) from its final stripped down implementation.  For example, I don't think replacing the analogue temperature gauges in the Wunda manifolds with a couple of DS18B20 temperature probes and linking them into Node Red, OpenHAB or Pimatic with a 3 line control rule to turn off the heating when the accumulated ΔT has reached some preset threshold is that complex an implementation. 

 

Others that follow are free to pick and choose and quite possibly either or both could end up as a standard template for future self-builders.

Edited November 20, 2017 by TerryE
Add Node Red

[ragg987]

I went with a "keep it simple" approach to zoning and control. Been meaning to do proper write-up but want to get through a winter and still experimenting a bit.

• approx 2,000m of UFH buried into our slab, split across 3 manifolds (and I think about 15 loops)
• manifolds do not have any pumps or blending valves. Each loop is generally at max flow to reduce resistance in the circuit - have throttled a couple of loops a bit to prevent overheat (e.g. in the basement)
• 7kW ASHP with onboard compensation controller linked to intelligent thermostats from the ASHP supplier
• i have created 2 zones as we wanted separate control in the annexe, to permit warmer room temps. Zone 1 covers 60m2, zone 2 covers 160m2.
• No UFH upstairs, we use direct electric towel rails in bathrooms and I have a 1kW heat coil in the upstairs MVHR as a backup..
• each zone has a room stat, pump and valve to operate it. Pump running at lowest speed.
• small 50l buffer for space heating - I do not consider this to be essential as we have 2,000m of pipe but put it in anyway - if zone 2 switches off then we have less than 500m in circuit
• the buffer is heated direct from the ASHP, and flow from buffer to UFH is direct as well - no heat exchanger or coils

One of my objectives is to run the flow at the lowest temperature possible as this has a big impact on COP - so there are no blending valves or heat exchangers in the system.

 

I think compensation control is a must for max COP - the flow temperature adjusts according to outside temperature (and hence heat loss from building). e.g. external temp of 10C gives flow of about 24C and external 0C gives flow of about 30C. The ASHP controller and intelligent thermostat work well together to calculate the optimum flow given the room setting, room and external temperatures. The ASHP modulates to meet the flow temp. I generally see a return temperature 2-3C lower than flow.

 

I keep heating on 24x7. The ASHP seems to run mostly at the lowest setting (I think approx 2 to 3kW heat output), when very cold outside it ramps up and in these warmer months it switches on and off as it cannot modulate so low - the 50l buffer must help in reducing the cycling, but I have no way of telling for sure.

 

Have set the controller to turn off the space heating completely at 16C external temperature - too early to tell if this is OK.

 

Blending valves are a compromise, in my view, as you would set them for a static temperature (e.g. 35C based on a very cold day) and hence rely on cycling on/off to maintain the correct room temp. Asking the ASHP to heat to 35C is inefficient if your flow needs to be at 25C. Same with heat exchangers - you heat the water about 3-4C warmer to maintain the same UFH flow temperature.

[ragg987]

2 hours ago, TerryE said:

• Simply run the ASHP for a fixed but settable time period.  This will heat the slab by some overall temperature increment. 
• Keep the pump running but then have a dead window to allow the slab temperature to even out.  Say 4 hours.
• Trigger the next heating cycle when the water return temperature from the slab drops below some fixed but settable temperature threshold.  Say 21.5°C
• By a process of trimming set the ASHP on-period (as per the first bullet) so that the heat pump only cycles one or twice a day.

My limited experience suggests this is difficult to put into practice. One problem is that heat demand is constantly fluctuating over the day, so how do you calculate the requirement for any given day. e.g. a sunny or windy day will decrease or increase the heat needed. Another is that slab response to heated water is delayed by about 5-6 hrs, so waiting for the water return to drop is probably too late and the room starts to cool before the next slug of heat takes effect.

 

I prefer to keep the water flowing all the time and adjust the flow temperature using compensation control. Slab temperature is even. I cannot see how cycling the ASHP a few times a day in the shoulder months is such a bad thing. And a small delta T of the slab vs room should minimise overheating.

 

The logic is already built into the ASHP controller, just add a room thermostat that can complement the controller (by which I mean one that does more than simple demand on / off based on room temperature).

[TerryE]

@ragg987, I understand your point about varying demand as a general issue, but in my case I have an MBC class passive twinwall frame with blown cellulosic filler and an outer 125mm stone skin. So the decrement delay from the external environment is a couple of days.   

 

Another point is that we aren't in love with acres of glass.  Our windows on the front elevation were limited by planning restrictions.  We have no side windows for the same reason.  To the rear we overlook a 1970s estate.  Our windows aligned on an ESE and WNW facing, so we catch the sun morning through midday on one side and late afternoon on the other, so my average solar gain varies from ½kWh/day in the winter to 2½Wh/day midsummer

 

To keep our house warm in the coldest months, I only need about 1kW averaged over the day which only requires the slab at a couple of degrees warmer than room temp (on average) to achieve, and solar gain isn't a large percentage.  Topping up the heat in a single 6 hour burst is only going to give a ~1°C ripple on room temp and I am quite happy for our house to swing from 21-22°C for example.  The rate of change (of the order of 0.1°C per hour) is really too slow to be noticeable.

 

At the moment we are planning to use off-peak electricity to do our bulk heating.  Detailed modelling supports that this will work fine in our house, but until I've collected a year's had data, I won't know for sure and whether my modelling has missed something.  At the moment I can't make the payback vs investment case for installing an ASHP, but if the data indicates otherwise or summer cooling becomes an issue, then I will add an ASHP.  But still no buffer tank.

 

But all-in-all your template seems well optimised for your use and house design.  Ours is a more compact cottage-style (70m² × 2 floors + 35m² warm loft rooms).

[Barney12]

7 hours ago, ragg987 said:

No UFH upstairs, we use direct electric towel rails in bathrooms and I have a 1kW heat coil in the upstairs MVHR as a backup..

 

Interested in your heat coil for upstairs. Do you any more details? Is it simply an on/off switched unit or connected to a thermostat? Have you used it in anger? 

[ragg987]

We

1 hour ago, Barney12 said:

 

Interested in your heat coil for upstairs. Do you any more details? Is it simply an on/off switched unit or connected to a thermostat? Have you used it in anger? 

We split the supply manifolds of the MVHR into "upstairs" and "rest of house". "Upstairs" is 4 bedrooms and 2 bathrooms - i.e. 4 supply and 2 extract ducts. Then added an in-line electric heater to the upstairs supply side. Detail design and supply is done by Gary at BPC.

 

For control, I use a simple on/off thermostat which is located on the landing. A Neostat-e, which is designed for direct electrical control of UFH, though we did away with the slab sensor it comes with.

 

We have used it in anger and it works pretty well. We find upstairs stabilises about 3C below downstairs when cold outside. So in those very cold weeks the thermostat would trip (setpoint 20C) and it would heat the whole of the upstairs gently. Timer set for early evenings when kids are bathing etc.

[Barney12]

38 minutes ago, ragg987 said:

We

We split the supply manifolds of the MVHR into "upstairs" and "rest of house". "Upstairs" is 4 bedrooms and 2 bathrooms - i.e. 4 supply and 2 extract ducts. Then added an in-line electric heater to the upstairs supply side. Detail design and supply is done by Gary at BPC.

 

For control, I use a simple on/off thermostat which is located on the landing. A Neostat-e, which is designed for direct electrical control of UFH, though we did away with the slab sensor it comes with.

 

We have used it in anger and it works pretty well. We find upstairs stabilises about 3C below downstairs when cold outside. So in those very cold weeks the thermostat would trip (setpoint 20C) and it would heat the whole of the upstairs gently. Timer set for early evenings when kids are bathing etc.

 

Thanks that sounds perfect for us and would mean I could do away with the planned radial circuit in case we found we needed some electric radiators for heat upstairs. 

[ragg987]

10 hours ago, TerryE said:

At the moment I can't make the payback vs investment case for installing an ASHP, but if the data indicates otherwise or summer cooling becomes an issue, then I will add an ASHP.

I finally clicked - you are using direct electric heating and not ASHP, so your solution revolves around that. Makes sense now!

 

I did evaluate direct elecric heating option using E7 and avoiding capital cost of ASHP, but given our larger footprint and hence larger heat requirement, I calculated we would need a massive thermal store of water or would end up with storage-heater syndrome if we used the slab as a buffer - too hot in the morning and cold by the evening. It was a leap of faith too far for me.

 

Are you using this already or still building? Would be interesting to see how it works - do report back.

[TerryE]

1 hour ago, ragg987 said:

Are you using this already or still building?

 

Here is the latest post with a picture.  Not a lot to it for the heating system to heat a reasonably large four bedroom house.  I've since completed the PRelV line plumbing and pressure tested the entire potable + UFH system.  I need to put in the electrics so I can commission this, and I plan to do this in the next few weeks.  At this stage, I will be able to do some commissioning runs to validate the modelling.

[dogman]

A picture paints a thousand words- or something like that

@TerryE

Will the pump draw well enough through the mixer/blending valve if only partially open?

 

If it does it will no doubt do the same for a buffer tank if used instead of a Plate Heat Exchanger

 

[Nickfromwales]

1 hour ago, dogman said:

Will the pump draw well enough through the mixer/blending valve if only partially open?

That's the beauty of it, it doesn't need to 'pull' ( draw ). The TMV is just a gate which opens partially / fully to allow heated water to be introduce into the already recirculating, free flowing body of water ( between pump, manifold the Ufh loops ). 

That circuit allows the manifold pump to freely circulate regardless of the state of the TMV, e.g. open or closed the pump sees no real difference as the loops typically will have no motorised actuators per loop arresting flow. If the TMV opens partially then water will be forced down the return and subsequently back into the flow, and that will be gentle rather than a surge so basically linear with the required amount of heat input. It's a beautifully simplistic, self managing arrangement .