Do low energy houses really need heating controls?

[pdf27]

I've been musing a bit about how best to control the air temperature in a low energy house over the past couple of weeks as a bit of an escape from reality, and started to wonder about something.

 

As per @JSHarris's formula, underfloor heating will provide a power of = 8.92 * (floor surface temperature - room temperature)^1.1 W/m² - assuming a low energy house will never need more than 20 W/m2 and very rarely more than 10W/m², that means the surface temperature will never need to be more than 1-2°C above ambient. Perhaps more interestingly we can use this to calculate the average house temperature from the flow and return temperatures.

Liquid screeds seem to be in the region of 2-3 W/mK, with the pipes being between 25 and 150mm from the surface - so at 1m thickness you're looking at 3-5°C temperature drop per metre at 10W, realistically for 1/8m maximum conduction distance through the screed you're at <1°C. Importantly, this means that the temperature difference between the flowing water and the rooms will be small (2-3°C) and a pretty much linear function of power since at low levels of heat the ^1.1 term can be pretty much neglected. Very roughly, you're going to be looking at something like:

 

          power  ≈ 5 x (average water flow temperature - room temperature)

 

For a wet underfloor system, this means that everything needed to calculate room temperature is already known, provided the insulation constant of the floor can be measured:

          power = (flow temperature - return temperature) x specific heat capacity of water x flow rate

                      ≈ 5 x (average water flow temperature - room temperature)

 

And measuring the insulation constant of the floor should be straightforward - you need a room temperature thermometer to make the initial measurements, but doing so should be simple to automate and when completed should give very consistent temperature control. The downside is that you need to be able to measure flow rate and flow/return temperatures very accurately - this would potentially all be within a heat pump using existing parts (the P/Q/Power curve for the pump should be well known for instance), but would presumably be rather hard to DIY.

 

So what am I missing?

[MikeSharp01]

I guess if everything is perfect then your logic feels sound, other than it is rather an open loop - suppose you have a party and fifty people turn up? So it also look to me like you have answered your own question with a yes, its just where the control is happening that is the crucial point I suspect.

[pdf27]

If 50 people suddenly turn up, then the heat transfer should shift from floor -> rooms to the other way around, and the return temperature should exceed the flow temperature. Response times may be an issue though - you'll get an initial rapid response as the floor slab will be close to the target temperature and the air temperature will be a long way off leading to a lot of heat transfer from the slab, but I'm not sure how quickly the water circuit would pick this up - or even if a slow response would matter.

[Jeremy Harris]

Read back on my efforts, they may prove interesting.

 

I started from the premise that with a low heating demand and UFH in the slab, controlling the slab temperature should be quite good enough to control the room temperature.  Over the very small range of floor surface temperatures required, to give zero to around 15 to 20 W/m² heat output with a nominal room temperature of around 21 deg C, then the floor surface temperature variation can be considered to be near enough linear, making the maths and controls a bit simpler.

 

The theory goes like this.  If the slab is kept at the correct surface temperature for the prevailing heat loss (determined roughly by the inside/outside temperature differential) then the system should self-regulate.  If the room temperature drops, say because an outside door is opened, then the temperature differential between the slab surface and the room air increases, so the heat output increases.  As the room warms up the temperature differential decreases and the heat output decreases, until equilibrium is reached once more, where the heat input exactly matches the heat loss.

 

Sadly this doesn't happen in the real world.  I tried many variations of the control system to try and get floor slab temperature control to work, with dozens of firmware variations and three different sets of control hardware.  None worked well, and in the end I tried a simple, but very low hysteresis (0.1 deg C), room stat and was surprised to find that, as long as the flow temperature to the UFH is kept below about 25 to 26 deg C, this works very well indeed.

 

I have looked back at why the theory wasn't supported by the real world experience and reached a view as to what I believe to be the main problem, and that is that the incidental heat gains to the house often swamp the heating system heat output.  Unless the control system knows what these incidental gains are, it cannot maintain an even house temperature.  In our case we had significant unwanted solar gain, almost entirely in late autumn and early spring, when the low sun angle got under the external shading and penetrated deeply into the house.  I worked out that we could easily get over 1 kW of solar gain on a bright, cold, clear wintry day, and that was more heat than the whole house needed.

 

There were other really big incidental heat gains too.  Run a shower for ten minutes and you dump around 2 kWh of heat, of which a fair proportion heats up the air and gets recovered via the MVHR.  One very significant factor when I was working inside the house was the vacuum cleaner.  I have an ancient Vax that puts out around 1 kW of heat when it's running.  Half an hour of cleaning up around the place on a cold day would easily increase the room temperature a couple of deg C or more.

 

 

Edited December 16, 2017 by JSHarris

[SteamyTea]

1 hour ago, pdf27 said:

formula, underfloor heating will provide a power of = 8.92 * (floor surface temperature - room temperature)^1.1 W/m²

Was this a generic formula, or one created from measurements at @JSHarris house?

 

[pdf27]

I have been reading up on what you did, and this is to a large extent inspired by it. The idea is that if you have the flow & return temperatures plus the flow rate this will give you both the average slab temperature plus the net heat loss from it - providing the missing bit of information you needed.

The difference is that instead of making assumptions about losses based on the inside and outside temperatures (as in a standard weather compensation circuit), you measure heat demand directly.

Edited December 16, 2017 by pdf27

[Guest Alphonsox]

1 hour ago, SteamyTea said:

Was this a generic formula, or one created from measurements at @JSHarris house?

 

 

This is a generic formula for UFH - set out in

BS EN 1264-4:2009 Water based surface embedded heating and cooling systems. Installation

 

 

[ragg987]

Your logic of temperature differentials reflects my own case. The slab being maintained at a stable temperature is key, the temperature would need to be adjusted with heat demand.

 

The complexity comes with how do you measure heat demand, given your low energy build? A major factor will be external and internal temperatures, however solar gain and internal activity have a strong bearing on the demand. Apart from the "50 people" scenario, cooking, appliances etc add heat and drying clothes, fabric losses, MVHR or air leaks removes it. Most of these aspects vary minute-by-minute in a busy household.

 

I decided to use the logic built into the ASHP compensation controller to simplify this. So we have a room thermostat and external probe as 3 key inputs - i.e. internal temperature, temperature set point and external temperature. These inputs then drive all the logic to calculate optimal flow temperature and the system, essentially, self-balances. I currently leave it on 24x7 and let it do it's stuff. I experimented with using a timer on for 12 hours, but I found the COP of the ASHP took a massive hit when load was high compared to a smaller load over 24 hrs.

 

The only aspect that can catch it out is sudden swings in conditions, e.g. massive solar gain or lots of cooking, I don't think this is just a controller issue, the slab temperature simply cannot be adjusted quickly given the energy it holds. You would need to devise some form of predictive logic to pre-empt your optimal slab temperature. Not worth it IMO, you are chasing diminishing returns given your starting point is a low energy build.

[Jeremy Harris]

1 hour ago, SteamyTea said:

Was this a generic formula, or one created from measurements at @JSHarris house?

 

 

Broken down the formula just gives an approximation of the total heat emitted from a heated surface that approximates to a floor with an average emissivity and assumes a typical convection heat transfer factor.  It's simplified a lot by the inclusion of fixed factors for emissivity and convection heat transfer, but is good enough for most practical purposes.  Don't rely on it for value of Δt outside the range of a UFH system, though, or for emitting surfaces that are not horizontal or of a similar emissivity to a typical floor covering.

 

28 minutes ago, pdf27 said:

I have been reading up on what you did, and this is to a large extent inspired by it. The idea is that if you have the flow & return temperatures plus the flow rate this will give you both the average slab temperature plus the net heat loss from it - providing the missing bit of information you needed.

The difference is that instead of making assumptions about losses based on the inside and outside temperatures (as in a standard weather compensation circuit), you measure heat demand directly.

 

I did try this, but the time lag is very high, around 10 to 15 hours, whereas solar and incidental gain can give local heating effects within tens of minutes.  One of the greatest heat emitters in our new build are two grey stone internal window cills.  They can easily reach 35 deg C in winter sun, and are large enough to contribute a significant amount of heat once the sun has warmed them up.  The same used to be the case for the cream travertine in the hall, but fitting long wavelength IR external reflective film on the hall glazing has reduced that a very great deal.

Edited December 16, 2017 by JSHarris

[pdf27]

14 minutes ago, ragg987 said:

The only aspect that can catch it out is sudden swings in conditions, e.g. massive solar gain or lots of cooking, I don't think this is just a controller issue, the slab temperature simply cannot be adjusted quickly given the energy it holds. You would need to devise some form of predictive logic to pre-empt your optimal slab temperature. Not worth it IMO, you are chasing diminishing returns given your starting point is a low energy build.

This is where I'm not so sure - for say a 75m² slab then you only need the room to be 1.4°C warmer than the slab for it to be absorbing 1kW of heat. Realistically that means you can absorb several kW of heat before any temperature rise becomes significant. The question then is whether the slab surface temperature really is remaining constant and the issue is a lack of heat transfer between slab and air, or whether (as I suspect the case may be with Jeremy's system) the pipes are buried deep in the slab and this means that the limiting factor is the large amount of concrete which must be warmed up before any additional heat transfer to a heating/cooling system can take place.

 

14 minutes ago, JSHarris said:

I did try this, but the time lag is very high, around 10 to 15 hours, whereas solar and incidental gain can give local heating effects within tens of minutes.  One of the greatest heat emitters in our new build are two grey stone internal window cills.  They can easily reach 35 deg C in winter sun, and are large enough to contribute a significant amount of heat once the sun has warmed them up.  The same used to be the case for the cream travertine in the hall, but fitting long wavelength IR external reflective film on the hall glazing has reduced that a very great deal.

Your heating pipes are buried deep in your slab, correct? I've got to wonder if that has an impact - the deeper your control element is in the slab the longer the time constant involved, while the thermal buffering ("thermal mass" ) of the slab will be a function of the depth and thus heat capacity of the slab rather than the position of the heating pipes in it.

When you're running the circulation pump to distribute heat around the house with the heating off, do you get the same sort of time lag - i.e. if the pump had been off for a while and you turned it back on would you expect it to take a day or so for the heat to settle down around the house?

[ragg987]

4 minutes ago, pdf27 said:

This is where I'm not so sure - for say a 75m² slab then you only need the room to be 1.4°C warmer than the slab for it to be absorbing 1kW of heat. Realistically that means you can absorb several kW of heat before any temperature rise becomes significant. The question then is whether the slab surface temperature really is remaining constant and the issue is a lack of heat transfer between slab and air, or whether (as I suspect the case may be with Jeremy's system) the pipes are buried deep in the slab and this means that the limiting factor is the large amount of concrete which must be warmed up before any additional heat transfer to a heating/cooling system can take place.

Our UFH is in approx 50mm of screed with insulation below and engineered wooden flooring stuck on with adhesive. The slab is not going to be a constant temperature all over, there are going to be heat gradients all over the place. So, in order to maintain room temperature of 21C for a particular heat demand, we could have the following scenario:

Engineered floor at 22-23C

Screed below flooring at 24-25C

Screed at 30mm depth at 25-26C

UFH flow at 26-27C

 

The above are very simplified and do not take into account the impact of pipe spacing in the UFH. In reality there will be a compound temp gradient in 3 axes.

 

So for your 75m2 of engineered floor to absorb said 1kWh, the room temperature would need to be 23.4-24.4C. Too late and much too warm already. Plus the ability of the slab to absorb heat  would be limited as the temperature of the screed below the surface is already higher than the room temperature, nowhere for the heat to flow (apologies if my terminology is a bit casual).

 

In any case, forgetting the theoretical aspects, we do find that a couple of hours of winter solar gain will raise our open plan room temperature by about 0.5C and that when we cook and have people over the temp will rise by up to 2C. The advantage of having a room thermostat and closed-loop control is that as room temp rises, so flow temp will be reduced by the controller so (in time) it should balance out.

 

In our system, a flow of approx 29C is called for when external temperature is 0C according to the heat curve I have set. I have no means to measure the slab or floor temps.

[Jeremy Harris]

22 minutes ago, pdf27 said:

Your heating pipes are buried deep in your slab, correct? I've got to wonder if that has an impact - the deeper your control element is in the slab the longer the time constant involved, while the thermal buffering ("thermal mass" ) of the slab will be a function of the depth and thus heat capacity of the slab rather than the position of the heating pipes in it.

When you're running the circulation pump to distribute heat around the house with the heating off, do you get the same sort of time lag - i.e. if the pump had been off for a while and you turned it back on would you expect it to take a day or so for the heat to settle down around the house?

 

No, they are around 30mm from the surface, so not deep in the slab.  They are 16mm diameter UFH pipes, tied to the top of 6mm diameter reinforcing fabric that is sat on 50mm chairs.  The slab is 100mm thick.

 

Anyway, structural concrete has a typical thermal conductivity of around 1.5 W.m/K, whereas dry air has a typical thermal conductivity of around 0.025 W.m/K and water a thermal conductivity of around 0.6 W.m/K, so heat transfers out through a concrete slab pretty quickly and easily.

[MikeSharp01]

4 hours ago, pdf27 said:

If 50 people suddenly turn up, then the heat transfer should shift from floor -> rooms to the other way around, and the return temperature should exceed the flow temperature. Response times may be an issue though - you'll get an initial rapid response as the floor slab will be close to the target temperature and the air temperature will be a long way off leading to a lot of heat transfer from the slab, but I'm not sure how quickly the water circuit would pick this up - or even if a slow response would matter.

I still feel it is to, all intents and purposes, an open loop because the party will be over before the slab detects the party people but they will have been well cooked in their own body heat - I think Nyquist would have had something to say about this whole problem - well the aliasing and sampling bit anyway. This heating issue is one we all have, or will get it once the house is finished. Without an air temperature sensor, after all the actual thing you are trying to control, you won't be able to bring any third party system into operation to help with the rapid rise in air temperature - EG pull the MVHR into summer bypass & boost mode to deal with the at least the worst of the air temperature rise while the slab catches on. Even then you will need a watchdog on your system to ensure that the heat being removed by the MVHR, and unceremoniously dumped out of the vent, is not being made by a runaway UFH. Lovely and complicated!

[pdf27]

1 hour ago, ragg987 said:

Our UFH is in approx 50mm of screed with insulation below and engineered wooden flooring stuck on with adhesive. The slab is not going to be a constant temperature all over, there are going to be heat gradients all over the place. So, in order to maintain room temperature of 21C for a particular heat demand, we could have the following scenario:

Engineered floor at 22-23C

Screed below flooring at 24-25C

Screed at 30mm depth at 25-26C

UFH flow at 26-27C

 

The above are very simplified and do not take into account the impact of pipe spacing in the UFH. In reality there will be a compound temp gradient in 3 axes.

 

So for your 75m2 of engineered floor to absorb said 1kWh, the room temperature would need to be 23.4-24.4C. Too late and much too warm already. Plus the ability of the slab to absorb heat  would be limited as the temperature of the screed below the surface is already higher than the room temperature, nowhere for the heat to flow (apologies if my terminology is a bit casual).

Umm... are those temperatures measured ones? The temperature differences look rather too large to me.

• Room temperature 21°C at 10 W/m² - this means floor surface temperature is 22.1°C
• 18mm or so of engineered wood has a US R-value of ~0.75 (U=7.6 W/m²K in SI units) - at 10 W/m2 that's 23.4°C at the bottom of the engineered flooring or a 1.3°C delta T.
• Thermal conductivity for pumped screed seems to be about 2.2 W/mK, assume the pipes are on 200mm centres 30mm deep so the heat will have between 30mm and 105mm to travel. That gives U-values between 73 and 21 W/m²K and thus at 10 W/m2 temperature differences of between 0.13 and 0.48°C - trivially small. That means the pipe surface temperature will be 23.7°C.
• Assuming 200mm centres, there will be 5m of PEX-Al-PEX pipe per square metre of floor space - so a heat transfer of 2W/m of pipe. Thermal conductivity appears to be about 0.43 W/mK. I can't be bothered to do the integral so taking the pipe to be a 4mm thick plate which is 43mm wide (circumference at 14mm diameter) gives a temperature difference of 107 W/m²K and thus a temperature difference across the pipe of 0.4K. That means the mean water temperature will be 24.1°C.

So in a scenario where the slab is supplying 750W and the pump is providing 150 l/hr (0.04 kg/sec) - 175 W/°C temperature difference in the water. That means a flow temperature of 26°C and a return temperature of 22°C.

Now assume a sudden increase in provided heat of 2kW for a net surplus of 1250W. Initially the flow temperature won't change and if that continued then you'd be very warm indeed - 31°C. However, that's very unlikely to happen - an increase in the return temperature from 22 to 24°C for instance would indicate a halving of the demand for heat. That would imply a reduction in the mean water temperature target from 24°C to 22°C, and thus a flow temperature of 23°C - at which point the heat pump is already providing cooling to the slab.

 

There are going to be two time constants in operation here - that for the flow of water around the circuit and that for the slab to heat up.

• A 100m loop would apparently need a flow rate of 2.5 l/min for a conventional installation - that would have a fluid content of ~11 litres, so time constant for the water temperature to react would be less than 5 minutes.
• A 30mm thick by 1m² concrete slab will have a mass of ~70kg - that means 60 kJ of heat needs to be added to warm it up by 1°C. With a sudden increase in heating load of 25W/m2 that's going to take 2250 seconds (37 minutes) per °C temperature rise, and realistically rather longer than this since there will be a significant contribution from the volume of concrete below the slab. Treating the whole slab as a unit increases this time constant to 2 hours/°C. This does give you a significant impact on the rate of rise of air temperatures however - 

So in any such scenario the slab cooling simply isn't going to react - it'll continue providing the same background level of heat for at least a couple of hours. That would apply to pretty much any control scheme however, unless there is a very big source of heat or cooling available - a 75m² slab 100mm thick would weigh 17 tonnes and so have a heat capacity of ~4kWh/°C - providing rapid cooling would need something like a 100kW heat pump and you'd end up chasing your tail all day long trying to keep the room temperature constant.

 

Specifically with solar gain, however, I do wonder if the time constant wouldn't be somewhat different. In the majority of cases the sun will presumably be hitting the floor, which then warms up and so heats the rest of the house. There the sequence works for rather than against you - the room won't warm up significantly until after the slab has, and the slab warming up gives any heat pump the chance to switch over into active cooling mode.

 

This then raises a question - what sort of temperature swing are people comfortable with in the short term, e.g. when cooking? Slab cooling isn't going to cut it, nor are normal MVHR ventilation rates.

[Jeremy Harris]

There is a very definite impact when visitors arrive, as they heat the air in the house fairly quickly, at around 80 to 100 W per person.  Five people is roughly 500 W of pretty high heat transmission rate directly to the room air, and that does raise the air temperature in the house fairly quickly, certainly within 15 to 30 minutes the temperature will have increased by a degree or so. 

 

Circulating water in the slab has a longer-term effect of keeping the slab temperature even when the heating is off, and it is effective if there is an area of floor that receives solar gain, in that it will effectively cool that area (by moving the heat elsewhere in the slab) and has a disproportionately significant impact on lowering the air temperature.  The latter is mainly because the sun can quickly put enough energy into a small area of floor so as to raise the Δt  a fair bit - I once measured a floor surface temperature of nearly 30 deg C by our front door, before we fitted the external IR reflective film.  Taking this heat away and distributing it to areas where there is no solar gain reduces the overall floor to air Δt  and so decreases room air heating a fair bit, particularly when the time-dependent variables are accounted for (heat capacity, thermal resistance etc).

 

The real challenge is to come up with a way to tame a system where you only have control over a relatively small part of the heat energy input.  This is in contrast to a high heating requirement house, where generally incidental and solar gain in late autumn, winter and early spring is small, and the heating requirement is well and truly dominated by the heating system.

[ragg987]

1 hour ago, pdf27 said:

Umm... are those temperatures measured ones? The temperature differences look rather too large to me.

Not measured - I gave some numbers to illustrate a point. Your calculated values are probably right, but do they assume perfect conditions? What about the impact of boundary layer (transfer of heat from water to pipe) and non-contiguous materials (voids in screed, adhesive and flooring)?

 

Our system is currently flowing at 29C with a return of 26C at 1050l per hour (as reported by the HP). I calculate the power to be 3.7kW. House is 330m2 (though only about 220m2 has UFH), so calculated UFH output is 17W/m2. We have 150mm pipe centres. Interestingly, whole ASHP system input power is 750W so we seem to be achieving a COP of nearly 5!

 

Not sure I trust those numbers too much. The deltaT could be anywhere between 2 and 4 given the rounding inherent in displaying the flow and return (no decimals are displayed). This gives a range of 2.5 to 5kW heat output, and a COP of 3.3 to 6.7.

 

Coming back to your original question of do low energy house need heating controls, my view is they do need something that either regulates the flow temperature or turns heat on and off. And to achieve these requires data on room temperature at a minimum. This sounds like a control system.

[Jeremy Harris]

As another data point, the data logger in our house logs to a resolution on 0.0625 deg C, and generally we find that the ASHP runs at between 3.8 and 4.5 COP, significantly better than the spec.  Part of this comes from running it at a relatively low flow temp (40 deg C), part comes from running the UFH flow at around 25 to 26 deg C.  The UFH return is rarely more than about a deg or so below the flow, and the ASHP runs for around 1 to 2 hours every couple of days in this cold weather, as long as we are only using it for heating.

 

Room temperature is reasonably steady at between 20.5 and 21.5 deg C, and only seems to change very slowly under most conditions. 

[dpmiller]

given that heat tends to rise is it maybe risky to assume that added heat within the room (bodies, appliances whatever) is going to start moving into the slab in a reasonable timeframe?

[Jeremy Harris]

1 minute ago, dpmiller said:

given that heat tends to rise is it maybe risky to assume that added heat within the room (bodies, appliances whatever) is going to start moving into the slab in a reasonable timeframe?

 

Yes.  I've seen no evidence from data logging in our house that there is any appreciable heat transfer from the house to the slab.  If there is much heat transfer, then it's both very small and very delayed, such that it doesn't seem obvious from looking at the room temperature and slab temperature data.

 

The only exception to this is if we deliberately cool the slab to around 18 deg C.  That does have a noticeable effect on reducing the air temperature fairly quickly (within around an hour or so).

[ToughButterCup]

This is a meaty read... Perfect for a lazy Sunday morning. 

The answer is 'yes', then? 

[PeterW]

6 minutes ago, recoveringacademic said:

This is a meaty read... Perfect for a lazy Sunday morning. 

The answer is 'yes', then? 

 

Or the answer is “yes” but actually how they operate is the challenge..!!

 

I rewrote a Pid algorithm in python to control a sousvide unit and the biggest challenge was when the “shock” of the cold item went into the liquid. Monitoring it took a lot of effort and the problem seems to be when there is a large input - either way - that the pid algorithm has to compensate for. As @JSHarris points out, the characteristics of concrete, air and water are so different that changing one with another is difficult to both predict and control. 

 

I watch this with interest..!!

[Jeremy Harris]

35 minutes ago, PeterW said:

As @JSHarris points out, the characteristics of concrete, air and water are so different that changing one with another is difficult to both predict and control.

 

 

This is at the heart of the problem, as it all comes down to how quickly heat transfers between the deliberate and incidental heating sources, the room air, and the structure and furnishings.

 

Even if we just take the four materials that have the greatest influence, air, gypsum plasterboard, concrete and water, then looking at their heat capacities and thermal resistance characteristics shows that there is a significant challenge.

 

Listed in order of greatest mass heat capacity:

 

Water = 4181 J.kg^-1.K^-1

Plasterboard/plaster = 1090 J.kg^-1.K^-1

Air = 1012 J.kg^-1.K^-1

Concrete =  880 J.kg^-1.K^-1

 

Then listed in order of decreasing thermal conductivity:

 

Concrete =  1.5 W.m/K

Water = 0.6 W.m/K

Plasterboard/plaster = 0.19 W.m/K

Air = 0.025 W.m/K

 

The main problem seems to be that in many cases, the incidental heat transfer medium is air, heated by people, showers running, solar gain heating up surfaces that then heat the air, cooking etc, and air is both a lousy thermal conductor and has a pretty low mass and volumetric heat capacity. 

 

Things do look a little different if you look at the volumetric heat capacity of these materials, bearing in mind the fact that there will be a lot of plasterboard virtually in contact with the room air, so it has a potentially large impact on thermal buffering, limited mainly by its relatively poor thermal conductivity, which will tend to slow down the rate at which it absorbs or releases heat:

 

Water =4,179,600 J.m^-3.K^-1

Plasterboard/plaster = 3,037,830 J.m^-3.K^-1

Concrete =  2,112,000 J.m^-3.K^-1

Air = 1210 J.m^-3.K^-1

Edited December 17, 2017 by JSHarris

[pdf27]

1 hour ago, JSHarris said:

The main problem seems to be that in many cases, the incidental heat transfer medium is air, heated by people, showers running, solar gain heating up surfaces that then heat the air, cooking etc, and air is both a lousy thermal conductor and has a pretty low mass and volumetric heat capacity. 

 

Things do look a little different if you look at the volumetric heat capacity of these materials, bearing in mind the fact that there will be a lot of plasterboard virtually in contact with the room air, so it has a potentially large impact on thermal buffering, limited mainly by its relatively poor thermal conductivity, which will tend to slow down the rate at which it absorbs or releases heat:

 

Water =4,179,600 J.m^-3.K^-1

Plasterboard/plaster = 3,037,830 J.m^-3.K^-1

Concrete =  2,112,000 J.m^-3.K^-1

Air = 1210 J.m^-3.K^-1

In other words the only practicable way to deal with sudden temperature spikes in a low-energy house is to directly cool the air, either through a recirculating unit (air conditioner or similar) or through the ventilation system (MVHR bypass or opening a window), unless a replacement plasterboard-type material with much better thermal conductivity is available. Fermacell is 50% better than gypsum but still pretty rubbish.

I'm assuming sudden reductions in temperature aren't a significant problem - essentially that means opening a door or window, which is under the control of the inhabitants and so more likely to be accepted.

 

Using air as the main heating/cooling system of the house would work nicely with this, but tends not to be very popular. It's possible to come up with a partially recycled air heating system to break the link between heating and ventilation, or to go down the route suggested in the Passivhaus spec of using the ventilation air for heating and cooling. This is feasible (the 10W/m² spec comes from ensuring that duct temperatures at standard ventilation rates do not need to exceed 50°C) but doesn't seem to be particularly liked. Cooling at standard ventilation rates is much harder - to get the 10W/m² you need to be providing air at -10°C which certainly won't be comfortable.

 

My understanding (theory only - please correct if I'm missing something) is that there are three scenarios in which you would need to dump excess heat:

1. Lots of visitors come over at once. Realistically in this scenario you'd need to increase the ventilation rate anyway to keep the air quality high, so using the ventilation system for cooling is the obvious answer. When the house is in heating mode then the summer bypass should kick in and extract hot air, replacing it with cool air. In cooling mode that won't work, but if there is a water-to-air heat exchanger tied in to the heating system (which would be in cooling mode anyway) and between the outlet of the heat exchanger and the point where the summer bypass is teed in then this should work acceptably well. This may need a buffer tank to work well though, although you could I suppose use the thermal inertia in the slab to provide initial cooling until the flow temperature rises and the heat pump kicks in.
2. Cooking. This splits two ways - for ovens, the best answer is a well insulated oven which will also reduce energy consumption anyway. For cooking on a hob, local extract-only ventilation would appear to be the correct answer - a cooker hood would conventionally be positioned to pick up the hottest air, essentially isolating it from the rest of the house. This goes against normal practice however, and I'm not sure why - the MVHR system would have to be able to run with unbalanced supply and extract flows, and you'll need an airtight (servo-operated?) valve on the cooker hood extract pipe to ensure you don't suffer from backdrafting. This is a potential problem if you have a combustion appliance in the house, but I'm not sure this is a showstopper - the low heat demand makes wood burning awkward, gas boilers tend to be room sealed and induction seems to be slowly taking over from gas, so in the long run these are going anyway.
3. Solar gain. I think the answer to this is probably a combination of design (i.e. remove by design the risk of having to deal with 10kW of solar heating on a spring day), boosting the ventilation and always running the UFH circulation pump in daylight hours (as per @JSHarris) to increase the heat capacity of the area warmed by the sun.

[TerryE]

On 16/12/2017 at 12:47, SteamyTea said:

Was this a generic formula, or one created from measurements at @JSHarris house?

 

As @Alphonsox says it's a BS quoted power law fit. The way I read it , the non-unity power term really reflects the convective element starting to kick in for higher delta temperatures, but in the case of a low temperature passive salve, I'd just stick with a straight proportionality.  

 

Is it 7 or is 8 or what depends on such factors as the material type (roughness, and transmissivity) -- we have a slate floor and I am sure that a dark matt slate would have a slightly higher coefficient than a light polished tile. The coefficient comes out of a linearisation of the Stefan–Boltzmann law.  I used 7 rather than 8, but that's the sort of ballpark.

 

I've only had 3 days of living in our house (rather than working in it 7 days a week for years)) so I am still getting to grips with the dynamics and will do a blog post on this, but I think that the coupling between floors is less than I'd hoped.  Maybe its all of that acoustic insulation in the ceiling void, maybe its that the house hasn't reached a proper equilibrium vertically, but if our ground floor is at a comfortable 20°C ish, then the 1st floor is definitely a couple of degrees cooler.  I have my office on the first floor and it's enough to notice when I am sitting working at my PC.  At the moment, if the outside temp is around 0°C and the MVHR is 90% efficient, then the inflow air is at around 18°C.  I am considering putting an inline heater in the MVHR in-stream to lift the air temp to say 22-24°C to counter this effect.  Jeremy won't be experiencing this because of his Genvex.

 

At @jack mentioned in and earlier related topic, the temperature differences and thermal gradients are so low, that I don't see any material evidence of convective flow between floors.  The only material coupling is though the ceiling/ floor and MVHR circulation.

 

 

 

 

[Jeremy Harris]

18 hours ago, pdf27 said:

In other words the only practicable way to deal with sudden temperature spikes in a low-energy house is to directly cool the air, either through a recirculating unit (air conditioner or similar) or through the ventilation system (MVHR bypass or opening a window), unless a replacement plasterboard-type material with much better thermal conductivity is available. Fermacell is 50% better than gypsum but still pretty rubbish.

I'm assuming sudden reductions in temperature aren't a significant problem - essentially that means opening a door or window, which is under the control of the inhabitants and so more likely to be accepted.

 

Using air as the main heating/cooling system of the house would work nicely with this, but tends not to be very popular. It's possible to come up with a partially recycled air heating system to break the link between heating and ventilation, or to go down the route suggested in the Passivhaus spec of using the ventilation air for heating and cooling. This is feasible (the 10W/m² spec comes from ensuring that duct temperatures at standard ventilation rates do not need to exceed 50°C) but doesn't seem to be particularly liked. Cooling at standard ventilation rates is much harder - to get the 10W/m² you need to be providing air at -10°C which certainly won't be comfortable.

 

My understanding (theory only - please correct if I'm missing something) is that there are three scenarios in which you would need to dump excess heat:

1. Lots of visitors come over at once. Realistically in this scenario you'd need to increase the ventilation rate anyway to keep the air quality high, so using the ventilation system for cooling is the obvious answer. When the house is in heating mode then the summer bypass should kick in and extract hot air, replacing it with cool air. In cooling mode that won't work, but if there is a water-to-air heat exchanger tied in to the heating system (which would be in cooling mode anyway) and between the outlet of the heat exchanger and the point where the summer bypass is teed in then this should work acceptably well. This may need a buffer tank to work well though, although you could I suppose use the thermal inertia in the slab to provide initial cooling until the flow temperature rises and the heat pump kicks in.
2. Cooking. This splits two ways - for ovens, the best answer is a well insulated oven which will also reduce energy consumption anyway. For cooking on a hob, local extract-only ventilation would appear to be the correct answer - a cooker hood would conventionally be positioned to pick up the hottest air, essentially isolating it from the rest of the house. This goes against normal practice however, and I'm not sure why - the MVHR system would have to be able to run with unbalanced supply and extract flows, and you'll need an airtight (servo-operated?) valve on the cooker hood extract pipe to ensure you don't suffer from backdrafting. This is a potential problem if you have a combustion appliance in the house, but I'm not sure this is a showstopper - the low heat demand makes wood burning awkward, gas boilers tend to be room sealed and induction seems to be slowly taking over from gas, so in the long run these are going anyway.
3. Solar gain. I think the answer to this is probably a combination of design (i.e. remove by design the risk of having to deal with 10kW of solar heating on a spring day), boosting the ventilation and always running the UFH circulation pump in daylight hours (as per @JSHarris) to increase the heat capacity of the area warmed by the sun.

 

The problem with trying to move heat around with air is that you need to move very large volumes of it in order to have any appreciable effect, simply because air has such a low volumetric heat capacity.

 

Say you wanted to shift an unwanted 500 W of heat out of the house, and that the temperature differential you were trying to correct was 2 deg C (the house is too warm by 2 deg C).  The volumetric heat capacity of air at 25 deg C is about 1210 J.m^-3.K^-1, which is around 0.3361 Wh.m^-3.K^-1, so if you wanted to move 500 W of heat in one hour you would need to shift just under 3,000m² of air to lose 2 deg C.  That is about 10 times more air than our Genvex Premium 1L MVHR can shift on full boost.  Because the Genvex has an in-built air to air heat pump, we can deliver fresh air on a hot day at around 12 deg C or so, and that does make a slight difference.  It's nowhere near enough to compensate for significant overheating, but is OK for a tiny bit of comfort cooling if we can restrict the incidental/solar heat gain a lot.

 

 

Edited December 18, 2017 by JSHarris