Heating Control Systems

[SteamyTea]

After a slight hijack on another thread, I thought it best to start one a bit more dedicated to heating control systems.

Some of us like playing with wires and stuff, other are happy to buy in a system.

This may be a good place to combine the pooled knowledge, and pinch a few ideas.

If I get time today (decorating bathroom) I shall try and post up some thermodynamic theory.  But no promise on that.

[IanR]

For me, a climate/environment control system for a low energy homes needs to be able to coordinate ventilation and shading along with heating and, if possible, predict timing and amount of solar gain to avoid unnecessary early morning active heating, if within in a couple of hours the sun would have done the same passively.

 

This is all based upon my, to date, 1 week of experience of requiring some low level heating in a low energy home.  

[ragg987]

1 hour ago, SteamyTea said:

Some of us like playing with wires and stuff, other are happy to buy in a system.

I think keeping it simple for the end user is essential. IMO a "wires and stuff" approach does not lend themselves to this.

 

Of course we are curious so monitoring comes to the fore, most householders would not give a monkeys. At least not beyond "it's cold in here let me boost the thermostat" and "i need a hot bath so going to boost DHW".

[SteamyTea]

So if it was as simple as just plugging in a 'thermostat' or two then that would suit most people.

The logging can go in in the background without them knowing.

I was more thinking of a very simple device, basically a temp sensor and a knob to set the temp, the clever bit would be behind the scene in the 'box of tricks'

Not really any different to what most people are used to.

[Nickfromwales]

I think @JSHarris's thermostat with its selectable 0.1 or 0.2oC increments is the way forward for the majority of instances TBH. 

If any solar gain complimented a space being heated by the central heating system in that house, and it was controlled zone by zone, by said thermostats, then the sharp increase in room temp would be responded to far, far quicker and negate any complex multi-functional / global controls. 

Operating ventilation would need to opposite approach so it only activated upon sustained overheating.   

[ragg987]

23 minutes ago, SteamyTea said:

I was more thinking of a very simple device, basically a temp sensor and a knob to set the temp, the clever bit would be behind the scene in the 'box of tricks'

Sounds good. Your thermodynamics lecture is probably an excellent place to start - the controls are only one part of the whole system, the house, orientation, slab, windows, insulation etc all contribute as well. I suspect understanding these factors first drives your controller solution.

[IanR]

28 minutes ago, Nickfromwales said:

If any solar gain complimented a space being heated by the cebtral heating system in that house, and it was controlled zone by zone, by said thermostats, then the sharp increase in room temp would be responded to far, far quicker and negate any complex multi-functional / global controls.

 

wrt solar gain, I'm thinking of a predictive system rather than reactive.

 

With outside temps dropping to -2 overnight last night it looks like our house dropped its average temp by about 0.8 degrees (down to 20.6) between 22:00 and 07:00 (I block the ASHP from working between these times, not actually sure when the last time heat was put into the buffer, may have been much earlier)

 

At 07:00 it immediately wanted to heat the buffer, but I'd noticed the forecast said that it would be a bright morning, so blocked it from heating. By 10:00 the average house temp was sailing passed 22 degrees and is now (15:00) 23.5. I'm choosing not to dump any heat out as it's forecast a cold night tonight. Had I not manually blocked the heating this morning I'd have wasted a buffer full of warm water.

 

Now, had it actually been over-cast this morning I'd have needed that buffer heating up at 07:00.

 

What would be nice is a control system that was informed by the forecast so that I can stop checking the BBC Weather.

 

In reality I'm using Loxone to control my heating/shading/cooling/ventilation (not yet fully automated). And they do offer a weather service that requires a subscription to be paid. I haven't yet checked if this would provide my controls the info it would need , but I'd prefer to avoid a subscription if I can.

 

A dedicated climate control system would hopefully be more focused on just the climate side, rather that full home automation as Loxone is, and be simpler to set up.

Edited November 17, 2017 by IanR

[Jeremy Harris]

Last night was interesting, as @IanR has noted, as the temperature here dropped to just below zero overnight, our heating didn't need to come on this morning (or yesterday), but the house is now sitting at 21.5 deg C, just from solar gain - it's been clear and sunny most of today so far.  That's about 0.7 deg C warmer than it was this time yesterday.

 

So, a predictive system would need to look at a reliable forecast for local temperature, sunshine and wind speed (as @Stones has discovered, accelerated convective heat loss from the outside surface of a house increases a lot with increasing wind speed).  From the months I spent trying to get weather compensation to work with my home brew, microcontroller, system, which measured temperature to a bit better than 0.1 deg C, I suspect any predictive system is going to be tough to fine tune.  I've just had a look through my software archive for that project, and there were over 60 iterations of the code, and it still didn't work as well as a simple thermostat mounted on the hall wall!

 

The idea of being able to control solar gain is a good one, for those who have houses that are significantly affected by it.  The other factor that makes a significant difference to our house is the number of occupants.  An hour or so with two extra people in the house warms it up to a noticeable degree.  Not enough to be troublesome, just enough to notice.  When I had a group of six people visit the temperature increase was enough to trigger the cooling system to kick in, though.

 

 

[SteamyTea]

The problem with solar gain is that it is very location dependant.

On my cheap 'weather station', basically a clock, a thermometer and a pressure gauge, it is amazingly accurate in predicting sunny days.  The only outside measurement it can make is air pressure.  I wish I know how it worked as these things are sold world wide.

 

It may be possible to point one of those IR thermometers up into the sky.  The colder it is, the clearer it is.

The most reliable indicator that I know of is wind direction.  Not 100%, but pretty good in the UK.

 

Edited November 17, 2017 by SteamyTea

[MikeSharp01]

2 hours ago, Nickfromwales said:

I think @JSHarris's thermostat with its selectable 0.1 or 0.2oC increments is the way forward for the majority of instances TBH.

In the end a sensor is just a sensor the processing that goes on with its output is what counts along with the response times of the various processes across the system. I still wonder, although do not dispute that it works, if 0.1deg is fine enough because across a 10 deg range it is a 1% change which feels large. It is the old resolution / accuracy problem along with keeping the sensor in a stable environment, or adopting other technologies, that allows it to measure tiny changes without being affected by other things, such a drafts (not in a passive house). If you were processing this directly you would take a million high resolution samples every second and process them to remove all the unwanted 'noise' you can then programme the hysteresis at whatever level you want, subject to the resolution, and even adjust it for conditions. In the end the trend has to be picked out from the noise and it maybe that the noise is 0.05 deg C so 0.1 is about as good as you might expect.

 

1 hour ago, JSHarris said:

So, a predictive system would need to look at a reliable forecast for local temperature, sunshine and wind speed (as @Stones has discovered, accelerated convective heat loss from the outside surface of a house increases a lot with increasing wind speed).  From the months I spent trying to get weather compensation to work with my home brew, microcontroller, system, which measured temperature to a bit better than 0.1 deg C, I suspect any predictive system is going to be tough to fine tune.  I've just had a look through my software archive for that project, and there were over 60 iterations of the code, and it still didn't work as well as a simple thermostat mounted on the hall wall!

It is somewhat unclear to me how knowing it is going to be a cold night is going to help other than in preparing for heat input or provisioning, in the reverse case, for heat loss. I can see that knowing that it will drop does allow heat input to be made ahead of the curve perhaps based on most effective energy source. So you might lift the slab temperature using the ASHP driven by the PV array in preparation for a cold night and let the decrement delay take the cream off. This might be the place to look at a neural network technology just taking a feed from the met office and able to sense the environmental factors with the caveat that the system itself then has some predictive capacity based on what it senses in real time and it is able to adjust the weightings across the network to improve prediction based on experience.

 

2 hours ago, SteamyTea said:

It may be possible to point one of those IR thermometers up into the sky.  The colder it is, the clearer it is.

What emissive  'surface' would you point at? One of these might be a good option if a little pricey but the image processing would be great fun. Many years back I built a sun sensor in a pingpong ball it could measure the strength of the sun and took into account the various trignometric factors associated with where it was in relation to the sun and the time of year to the minute. I guess I could dig out the code, I will have it on a 8" floppy somewhere, and use that to help work out how much heat will be acquired from the sun and hence the amount needed to get ahead of the frosty night to come with free energy from the PV array (which itself is a measure of the suns strength but is very directional).

[Jeremy Harris]

The problem is simplified a fair bit by our perception of "cold".  For example, on a cold, clear night, standing inside in front of one of the reasonably decent double glazed windows in our old house felt decidedly chilly.  The reason was that the glazing was pretty poor at preventing long wavelength IR from escaping, and with the night sky on a clear night having a very low temperature (typically around zero deg C or perhaps a bit lower on a very clear night with not much water vapour around).  Do the same in our new house, with it's triple glazing and two internal pane faces low e coated and you can barely detect that you're standing in front of a window, as the additional radiative heat loss from your skin is very much lower.

 

The room temperature is the same in both cases, all that has changed is the way that your body has sensed it's own rate of heat loss, in this case radiative heat loss.

 

Similarly, our bodies are pretty sensitive at detecting temperature gradients from our peripheral regions.  This is one reason why we can stand in front of a roaring fire in a stone built pub, with stone flags on the floor, and still feel as if we're not that warm, despite the high heat level being radiated from the fire.  Lift our feet from the stone floor and we feel a lot more comfortable, as their local heat loss rate through the stone floor has decreased (I strongly suspect this is partly why people feel more comfortable with their feet up).

 

UFH feels subjectively more comfortable for the same reason, it reduces the rate of heat loss from our feet and so fools us into thinking we are warmer than we are.  Quite a few people have reported that they feel comfortable with a lower overall room air temperature when the heat source is coming from the floor.   Conversely, I remember a friends house, built in the 1970's, that had electric radiative heating in the ceiling.  That always felt cold, even when the room temperature was at a silly level (and his electricity bills were ludicrous).

 

In our specific case, much of the improved comfort has little to do with the low energy demand and low heat input, but a lot to do with pretty much every surface in the house being very close to being the same temperature, with no cold spots.  We perceive this as being comfortable, probably by a similar mechanism that tells us that being evenly wrapped in clothes, or in bed, makes us feel comfortable.

Edited November 17, 2017 by JSHarris

[MikeSharp01]

7 minutes ago, JSHarris said:

The problem is simplified a fair bit by our perception of "cold". 

You are right here. Last week we were running out of GAS (LPG switch over) so I dropped the stat down to 18 from 21 here in millstone manor. The other occupants went crazy but after a day or two they got used to it - we had a delivery today, the stat has gone up to 21 again and I feel sweltering!

[Lesgrandepotato]

I’m looking into whether I can override the displays of the heatmisers in code. So they say 24c at about 19.... that should do it :-) 

[Trw144]

4 hours ago, IanR said:

 

wrt solar gain, I'm thinking of a predictive system rather than reactive.

 

With outside temps dropping to -2 overnight last night it looks like our house dropped its average temp by about 0.8 degrees (down to 20.6) between 22:00 and 07:00 (I block the ASHP from working between these times, not actually sure when the last time heat was put into the buffer, may have been much earlier)

 

At 07:00 it immediately wanted to heat the buffer, but I'd noticed the forecast said that it would be a bright morning, so blocked it from heating. By 10:00 the average house temp was sailing passed 22 degrees and is now (15:00) 23.5. I'm choosing not to dump any heat out as it's forecast a cold night tonight. Had I not manually blocked the heating this morning I'd have wasted a buffer full of warm water.

 

Now, had it actually been over-cast this morning I'd have needed that buffer heating up at 07:00.

 

What would be nice is a control system that was informed by the forecast so that I can stop checking the BBC Weather.

 

In reality I'm using Loxone to control my heating/shading/cooling/ventilation (not yet fully automated). And they do offer a weather service that requires a subscription to be paid. I haven't yet checked if this would provide my controls the info it would need , but I'd prefer to avoid a subscription if I can.

 

A dedicated climate control system would hopefully be more focused on just the climate side, rather that full home automation as Loxone is, and be simpler to set up.

 

 

Yes Loxone do offer a weather service - it's an interesting idea that essentially we are nt so worried about forthcoming temperature, but actually potential solar gain. With the low winter sun eliminating the benefit of the overhang, I gained 4 degrees in my kitchen/dining room within just a few hours this morning.

[MikeSharp01]

34 minutes ago, Lesgrandepotato said:

I’m looking into whether I can override the displays of the heatmisers in code. So they say 24c at about 19

I tried that by moving the stat dial on the splines about 10 years ago but my son - the blighter, calibrated the stat with the kitchen thermometer and told me it was 4 deg out and that he had drawn an arrow on the dial where 21 was supposed to be.

[TerryE]

Nick, you might be interested in

though I am covering the actual details on the control system in the next part.  No Loxone or the like just a couple of RPis and ESP8266 modules at the moment.  I've used the ESP modules because I am more familiar with hanging OW bus, GPIOs and other sensors off them.  Also, if you get a polarity wrong during testing and fry an ESP  module, they costs less than £2 to replace -- though now that the I/O logic is stable, I could do all this on a Pi-hat to the second RPi, but why bother?

 

Given that in our case raising the overall house temperature by ½°C per day would double the daily heating requirement,  I feel that using something like an internal air temperature as the primary control sensor is a mistake.  For example, we've had carpet fitters and workmen working in the house over the past couple of days.  You just can't go around after these guys closing outside doors and nagging them when the are going in and out, especially when the house isn't really occupied and it's only going to add pence to the heating bill.  Having the front and back door open for half an hour is an effective way of dropping the hall air temperature by 5°C, but the last thing that I'd want to happen is for the heating system to go into overdrive to compensate.

1. Turning on the slab heating won't make any material difference for 3-4 hours
2. And by that time, (with the doors shut) the house will have largely recovered to its average temperature anyway, so the slab heating will only cause an evening overrun.

If I do want to use a ambient temperature estimate independent of slab  temperature then in my view it would be better to measure the temperature of an internal plasterboard wall and use its physical characteristics to filter out any higher frequency temperature variations that the heating system can't respond to anyway.     

[TerryE]

In our case we have a good passive-class internal frame and stone skin, so it takes about 2 days for a step change in external air temperature to effect the internal temperature.  There is no point in trying to forecast or even to measure external temperatures, IMO.  Quite a few weather services provide localised feed APIs which are zero-cost and good enough.    

[Lesgrandepotato]

@TerryE that’s exactly the use case I’m going to apply with the Fibaro door sensors - however I’m not going to try and infer a door is open. It’ll create an event in the system and if it remains open for more than 15mins then I’ll set the relevant stats to setback temp. 

[Stones]

22 hours ago, JSHarris said:

...as @Stones has discovered, accelerated convective heat loss from the outside surface of a house increases a lot with increasing wind speed...

 

Average wind speeds can be modeled and this is the approach I took when I was working out my heating requirement.  The difficulty, as I have found, is modeling the impact of the increased pressure differential driving heat loss that comes with very high / gale force wind speed, which we have more than our fair share of.

 

I'm 3 months off having a 12 month data set to be able to better quantify the impact such wind speed has, based on how the house has actually performed.

 

Modeling wind infiltration discussed here:

 

http://www.ebuild.co.uk/topic/16438-modelling-wind-infiltration/

 

 

[Gone West]

2 hours ago, Stones said:

 

Average wind speeds can be modeled and this is the approach I took when I was working out my heating requirement.  The difficulty, as I have found, is modeling the impact of the increased pressure differential driving heat loss that comes with very high / gale force wind speed, which we have more than our fair share of.

 

I'm 3 months off having a 12 month data set to be able to better quantify the impact such wind speed has, based on how the house has actually performed.

 

Modeling wind infiltration discussed here:

 

http://www.ebuild.co.uk/topic/16438-modelling-wind-infiltration/

 

 

Do you suffer from the effects of decrement delay or is it always windy enough so that heat build up on the surface of the building fabric is lost to the atmosphere.

[TerryE]

If you assume that the external house surface is at external ambient and the external skin is an essentially impermeable barrier, then wind will have three separate effects:

1. if the skin surface is wet and the air humidity is less than saturated, then you will get evaporative cooling which will reduce the skin surface temperature below ambient and there increase heat losses.
2. If the house is inherently leaky then small pressure differentials across the house will severely increase such air losses which bypass heat recovery.
3. Particularly if the MVHR inlet and outlet are on different walls then any pressure imbalance will degrade the effectiveness of the heat recovery.

I've put both our MVHR inlet and outlet on the same gable to  minimise the third, but where we are we rarely have strong winds.    In Jason's case, I would have though a good place to put the MVHR inlet and outlet would be in his sheltered "alleys" between the two wings of his house which would shelter them from the worst of air wind effects.

[Jeremy Harris]

The house external surface will be above ambient, though, as even with the best insulation there has to be some heat flow through the structure.  Also, there will usually be some external surface solar gain, even on a dull day, that will tend to make any external surface that is impacted by even diffuse sunlight a bit warmer than  ambient.

 

In the standard U value calculation for any external element there are two assumed surface heat transfer factors included (or there should be, if the calcs are being done properly), one for the internal surfaces and one for the external surfaces.  These factors are usually simplified to assume constants for surface emissivity (which determines the radiative heat loss) and still air surface convection loss rate, which depends on surface roughness (which effects both surface area and still air convective flow) and the angle of the surface (vertical surfaces have a lower convective heat loss factor than horizontal or angled surfaces).

 

What changes with wind is that the assumed convective surface heat loss rates in the U value calcs change, quite substantially in the case of a rough surface.  Add in a bit of moisture (driving rain or mist) with some wind and you also get a fair bit of evaporative heat loss, as well as increased loss because of the higher heat capacity of water vapour.

 

If the airtightness is good, and there is MVHR, then I doubt that there will be that much change in ventilation heat loss.  The resistance to flow of the air leakage paths will be high, whereas the resistance to flow of the MVHR, ducts and terminals will be a lot lower, so most wind-induced increased ventilation will probably flow via the MVHR, pushing the ventilation rate up, but still recovering some heat, albeit at a lower efficiency.

Edited November 19, 2017 by JSHarris

[SteamyTea]

Starting to seem to me that heating control is more about prediction that the actual switching on and off.

 

There are two statistical methods used to predict weather, both based on observation.

Frequentist:  This relies purely on the distribution of past events.  The output is a probability rather than binary.

Baysian: This is still based of past events, but allows a known event to be discarded.  The output eventually becomes binary, which is useful.

 

I am not sure how reliable getting a local weather feed from the WWW will be.  It is rather at the whim of the weather station owner and the feed could be pulled at any time.

 

[dpmiller]

As long as there's no massive step-change in daily weather, the normal adaptive turn-on based on historic rate-of-rise should be plenty to look after the seasonal changes for the average system.

Likewise approaching setpoint ther's plenty of fuzzylogic/ PID controllers around. In fact, I wonder if anyone has ever attempted to use a standard DIN panelmount controller for room or whole-house control; they generally have a pretty smart Autotune PID system that will sort the parameters during the first heatup and overshoot cycle...

Edited November 19, 2017 by dpmiller

[Jeremy Harris]

I still maintain that if you have a house with a low heat loss rate, even under worst case conditions, then a dead simple, but low hysteresis, thermostat should do the job. 

 

Say the house takes 10 hours to lose 0.5 deg C in winter (ours is a bit longer than that, more like 24 hours or so).  As soon as the temperature in the room drops by 0.1 deg C below the set point, the heating comes on.  The heating system then has several hours to pump heat gently in to the house before the temperature drops much below the set point.  Likewise, as soon as it has gone 0.1 deg C above the set point the heating will turn off, but the house tends to carry on warming up a bit over the next couple of hours, from the residual heat in the UFH pipes and the time taken for heat to travel from the warmer core of the slab to the surface.  It's this latter issue that makes the house comfort level so dependent on keeping the UFH flow temperature as low as possible - the more heat there is sitting in the core of the slab the greater the temperature overshoot when the heating turns off.

 

Overall this system seems to be able to control the house to around -0.2 deg C, +0.7 deg C normally, a fair bit better than the hysteresis on some of the pretty crappy thermostats that have been around for decades.  We do occasionally see an overshoot to around 1 deg C over the set point, but that's usually because of a bit of solar gain.  We never experience temperatures dropping more than about 0.2 deg C below the set point, no matter what. 

 

Personally, I can live with that.  That range of temperature variation around the set point seems perfectly acceptable to me, so I really don't see why there needs to be any more complication.  Best of all, it uses off the shelf stuff, so can be fixed quickly if anything fails.

Edited November 19, 2017 by JSHarris
typos