"Lean" Design

[pdf27]

In my day job I'm a mechanical engineer, and one of the big themes is how you work low cost (be that build or continued operation) into the design from the very first concepts. Some of it is well known (the Toyota Production System, Kaizen, Just In Time, etc.), but other aspects not so much - for instance each bolt is assumed to cost £1 in lost time, etc. so you should design out fasteners whenever possible. Some of you may be familiar with this graph:

It doesn't quite apply to buildings, but the fundamental concept that cost is committed to extremely early in the design cycle, far earlier than 90% of people normally realise does apply.

 

Since it directly applies to what I'm hoping to do in a year or two, I thought it would be interesting to apply these principles to the Passivhaus standard, and see where the logic takes me. Please feel free to jump in and rip this to shreds - I'm trying to ensure that I have a good grasp of the fundamentals driving cost when talking to my architect in the near future, since cost is one of the major hurdles for us.

 

Fundamentally the Passivhaus standard has two requirements - one for heating (15 kWh/m²/year OR 10W/m² at the design condition) and what is effectively a limit on imported energy. Historically the limit on imported energy has been very hard to meet, leading people to follow the 15 kWh/m²/year criterion and this has led to very well insulated houses where high performance glazing is used to provide a lot of the required supplemental heating.

However, with heat pump performance having drastically increased in the past few years and PV becoming very cheap I'm not at all convinced that concentrating on this requirement is still sensible: PV and a heat pump can essentially be used as a controllable diode, shifting solar heat into the building when needed, while the improved heat pump COP means that the penalty from no longer using "free" solar heat from the windows is smaller.

Interestingly, the 10W/m² is much closer to a comfort criterion - achieving this means everything has to be well insulated with no drafts or cold surfaces anywhere. In cost terms the two approaches are quite different however - the energy criteria encourages the use of (expensive) high performance windows with increased area, while the power criteria encourages the use of relatively cheap, thicker insulation with less glazing. Heating system capital costs are also dominated by the peak heating power, while at anywhere close to 15 kWh/m²/year unless you're burning banknotes you won't be spending a lot to keep warm.

 

From this I think a number of conclusions follow:

• The use of the alternative 10W/m² criteria should be the starting point unless other design criteria (e.g. wanting large south facing patio doors onto a rear garden) mean that a lot of solar gain has to be inherent to the design.
• Wall, roof, and underfloor insulation need to be as thick as possible without affecting the cost very much. When the current aircraft carriers were being built the RN followed the mantra "steel is cheap and air is free". Cellulose insulation costs about £10/m² of wall area to increase the thickness by 100mm and depending on the timber frame system used the associated timber costs should be quite modest. If I'm interpreting the PHPP modelling done when we were hoping just to extend and refurbish our house correctly, the energy impact of going from 300mm to 400mm is about the same as shifting from Part L minimum glazing to quite nice triple glazing.
• Use glazing to provide light and make a room a nice place to be, not to provide heating. This is likely to reduce total glazing area (and hence cost), and possibly help slightly with overheating.
• If going this route, the requirement for heating will be increased. To avoid this turning into a wall-of-text, I'll address this in a subsequent post if there is any interest.

[ProDave]

I took your approach. I designed the house to have window sizes appropriate to the room sizes and the local vernacular. And included as much insulation as reasonably practical.

 

However I did add on the "sun room" (not yet finished so I can't yet comment if it will work as I expect)

 

The sun room has large windows on 3 sides but a proper well insulated roof (so not a conservatory)  It is connected to the main family room via proper quality exterior double glazed doors (even though they are really internal doors)

 

The theory is, if we need some additional solar gain, open the internal doors to let some heat in from the sun room.  If we don't need the solar gain, keep the internal doors shut, and also open windows in the sun room to cross ventilate it and cool it down.

[newhome]

7 hours ago, ProDave said:

The theory is, if we need some additional solar gain, open the internal doors to let some heat in from the sun room.  If we don't need the solar gain, keep the internal doors shut, and also open windows in the sun room to cross ventilate it and cool it down.

 

I have a similar room connected to the kitchen / family room by double doors. It does get hotter in there than most other rooms due to the additional glazing / solar gain and I open the double doors if I want that heat in the main living space and keep the doors closed if not. So as you describe. 

[Triassic]

Applying your approach would suggest the house be factory made using ‘standard” components, with just enough insulation and windows of an optimum size. From what you’ve said, this is where modelling comes in. You could model various designs at a pre-production stage to identify the house that meets your predetermined passive house design criteria. 

[pdf27]

19 minutes ago, Triassic said:

Applying your approach would suggest the house be factory made using ‘standard” components, with just enough insulation and windows of an optimum size. From what you’ve said, this is where modelling comes in. You could model various designs at a pre-production stage to identify the house that meets your predetermined passive house design criteria. 

Mostly, yes - although the components should be standard to the factory, not the house since that is where you get the economies of scale. Doing this is likely to mean you over-design on the insulation (since it is pretty cheap), enabling you to use standard sizes of windows rather than trying to tune them for the house. For example, if you're doing a timber frame structure using JJI joists, you would do everything in the building at the 400mm or maybe 450mm depth. That adds a bit of cost to the timber compared to an optimised structure, but you take engineering time out of the calculations and the part count goes way down. It also means standardising the windows to a small number of sizes (basically picking from a menu to ensure that there is enough daylight inside) - it needs to be remembered that in a lean design you aren't trying to optimise the details, but optimise the build cost. That is likely to mean that you aren't trying to design for "just enough" or "optimum size": you're overdesigning on cheap components to allow you to underdesign on other parts and still meet the targets.

 

Doing as much as possible using standard techniques in a factory is absolutely critical though - the impact of specialisation is something which goes back to at least Adam Smith, and it also means using standard detailing, standard components, etc. to ensure that as many as possible of the raw materials and subassemblies used are factory made in volume too.

This, incidentally, is why I think timber frame or possibly SIPs (depending on the tooling requirements, which I'm not familiar with) is the best way to go for a "lean" design - it maximises the amount which can be done in controlled conditions and allows for maximum standardisation of parts.

[Tennentslager]

I designed my hut based on standard sheet sizes so the everything was a multiple of 1.2.

I.e. room dimensions 4.8m wall height 2.4 etc.

Meant less cutting and no waste although not always as doors need to go where suits and that isn't always 1.2 from a corner!

[Jeremy Harris]

Interestingly, the frame company that built our house seem to work on multiples of 400mm as far as they possibly can, as that then fits their standard way of making the prefabricated panels.  My initial drawing dimensions were all subtly changed so they fitted multiples of 400mm, which in my case only made a very tiny difference, as I'd already designed the house so that rooms like the kitchen, utility room, WC and bathrooms would all end up able to accept multiples of 300mm, 400mm, 500mm and 600mm units (we have built-in furniture in all those rooms.  Doing things like this saves labour cost at second fix, as there are no infill pieces needed - standard cabinet sizes just fit the available spaces near-perfectly.

 

Another advantage was the saving in plasterboard and the labour needed to fit it.  Whole sheets of metric board fitted with little need for cutting - the ceilings were all made to be 2400 high and the walls and ceilings being multiples of 400mm wide, plus them being dead square (in part because all the panels were factory made in jigs) made it easy to board out with minimal wastage.

 

On the topic of fasteners, our frame isn't bolted to the slab, it's fixed with expanding hollow "nails", which are a lot quicker to install.  Most of the frame fastening is already done in the factory with nail plates and a hydraulic press, and having seen them being assembled I can say it;s a very quick process - just a few second to make a frame sub-section that if hand nailed would take several minutes.

 

The key problem is getting those that design houses to understand, and work to, the sort of production techniques that allow a high degree of standardisation in factory-manufactured sub-assemblies.  We have a long way to go to get those who design houses to start thinking about designing cost out at the initial design stage, IMHO.  I get the distinct feeling that there has been a big divide between architects, who design spaces and features primarily on the basis of aesthetics, and who then metaphorically chuck their design over a wall to engineers who have to struggle to find ways to actually realise the design and solve the structural and other problems.

[Ed Davies]

I very much like your approach @pdf27 .

 

The equivalent to your £1/bolt metric I use is £30/watt - a rough estimate of the cost of provision of energy off-grid in winter. E.g., for calculation of how high grade windows and doors are worthwhile. There should be an equivalent metric for on-grid. Not just the p/kWh but for the marginal cost of generation capacity: how much does an extra watt of wind turbine cost or an extra watt of gas power (including the well in Qatar, the LNG plants, the ships, etc).

 

I got rather mocked and dismissed by some Passivhaus worthies on Twitter for mentioning this approach. While Passivhaus has a lot to recommend it (it's probably the best standard currently available for, say, a design/build contract) I don't like the way it allows solar gain via windows but not by PV [¹]. Similarly some on GBF have argued for a “fabric first” approach which encourages bigger windows but doesn't allow for “eco-bling” (PV panels) which might actually bring in more net watts per pound.

 

[¹] Yes, I know that more recent iterations of the standard do take on-site generation into account but I think it's still rather indirect. Much better, IMHO, to include expected generation (and variation of generation) directly in the modelling of the house's performance.

[pdf27]

36 minutes ago, JSHarris said:

The key problem is getting those that design houses to understand, and work to, the sort of production techniques that allow a high degree of standardisation in factory-manufactured sub-assemblies.  We have a long way to go to get those who design houses to start thinking about designing cost out at the initial design stage, IMHO.  I get the distinct feeling that there has been a big divide between architects, who design spaces and features primarily on the basis of aesthetics, and who then metaphorically chuck their design over a wall to engineers who have to struggle to find ways to actually realise the design and solve the structural and other problems.

It's one of those things that should be a big surprise and isn't: outside of the automotive industry, engineers actually tend to be pretty bad at designing for low cost manufacture (it's called the Toyota Production System for a reason) and nobody else really does it at all. Fundamentally there are a large number of competing things that you're trying to optimised, and optimising for low cost inevitably means crippling something else.

 

It might well be best to follow a three-stage approach:

1. Architect comes up with the general concept - size, shape, etc.
2. Timber frame company is selected and adjusts the design to fit with their standard detailing and build procedures. This stage might be a little tricky as most design to a U-value of 0.15 in order to minimise frame costs - however I suspect the optimum is to improve the frame performance in order to allow for poorer-performing (cheaper) windows.
3. Timber Frame design is run past the Architect again to check for any aesthetic or functional problems, and plans submitted for approval.

The key behaviour is to ensure that the design considers how to build it at a very early stage, and is optimised for the chosen build system. The two skills are very different, however, so it is unlikely that you're going to come across someone who is good at both.

 

2 hours ago, Ed Davies said:

I very much like your approach @pdf27 .

 

The equivalent to your £1/bolt metric I use is £30/watt - a rough estimate of the cost of provision of energy off-grid in winter. E.g., for calculation of how high grade windows and doors are worthwhile. There should be an equivalent metric for on-grid. Not just the p/kWh but for the marginal cost of generation capacity: how much does an extra watt of wind turbine cost or an extra watt of gas power (including the well in Qatar, the LNG plants, the ships, etc).

 

 

That's a very painful calculation because measuring it in Watts requires accurate knowledge of exactly when the power will be required. If you have an insulation system with a long time constant (i.e. high decrement delay) then the heat capacity of the house will be high relative to the losses and it's really easy to shift the consumption of heat around in time to match availability. When you do this, there is essentially zero marginal generation capacity required, but you do consume energy (already allowed for in the PH standard).

If you have a short time constant, either because of the insulation system chosen or simply because it isn't very good, then how storable the power you chose to use is becomes important. With gas there is a huge amount of capacity in the system to deal with exactly this problem, and you move back to an energy consumption problem (the cost of a gas well is pretty much all capital, amortised over the total amount of gas extracted, so power isn't a terribly helpful metric). With electricity the time of use does matter and this is somewhere that the PH standard (while much improved) really doesn't fully account for the problem.

Having said that, the current charging system for electricity in the UK essentially doesn't penalise use at the wrong time of day for low energy houses, since they tend to have low electricity consumption and so can't benefit from E7/E10 very much due to the higher standing charges.

 

2 hours ago, Ed Davies said:

I got rather mocked and dismissed by some Passivhaus worthies on Twitter for mentioning this approach. While Passivhaus has a lot to recommend it (it's probably the best standard currently available for, say, a design/build contract) I don't like the way it allows solar gain via windows but not by PV [¹]. Similarly some on GBF have argued for a “fabric first” approach which encourages bigger windows but doesn't allow for “eco-bling” (PV panels) which might actually bring in more net watts per pound.

 

[¹] Yes, I know that more recent iterations of the standard do take on-site generation into account but I think it's still rather indirect. Much better, IMHO, to include expected generation (and variation of generation) directly in the modelling of the house's performance.

I fully agree with the concept of "fabric first", but I think it needs to be reinterpreted slightly. "Fabric" should really be "things that are difficult to change" rather than "things which form part of the structure". Changing the size of a window is quite hard, but replacing 2G with 3G isn't at all - as all the double glazing companies out there prove. In fact it's probably easier than fitting PV (especially roof-integrated PV), since the access is very much easier and safer.

Heating systems are something else that should really be treated as part of the fabric, but generally aren't (note: "heating systems" rather than "boilers"). I posted this graph in another thread last week, but it's very relevant to this:

This is COP data extracted from data published for a 5kW Samsung ASHP, and extrapolated for 25°C flow temperature (the unit can do 25°C, but they don't provide performance data). The key point is that shifting from water at 50°C to 25°C roughly doubles the COP and halves the consumption of electricity for heating over the lifetime of the house. You'll often find comments suggesting that it's just fine to use radiators and a heat pump in a Passivhaus because the heat demand is so low. Perfectly true in so far as the low COP won't hit your energy bills very badly - but if the design objectives are comfort and low energy consumption then it becomes a critical factor.

The physics are very simple here - if you want a small dT between water and air, you either need a very high heat transfer coefficient (blowing a lot of turbulent air over a warm surface, essentially) or you need a very large surface area. One is very noisy and probably costs quite a lot of energy in the long run, the other is dead simple and is regularly used - wet underfloor heating.

Achieving this shouldn't be too hard - very crudely, at 10W/m² the floor needs to be 1°C warmer than the room air to pass this much heat on by natural convection. Carpet is ~0.25 m²K/W, so for a 20°C room air the slab needs to be at 23.5°C: challenging for underfloor heating, but realistic if the pipe centres are close enough together - things get a lot easier at 19°C, for instance. It should also be noted that the 10W/m² value is at the design cold weather condition - it would be straightforward to apply weather compensation to increase flow temperatures in cold weather. Essentially in this case you'd end up with a straighter operating line as at colder temperatures you'd lose out on the performance recovery slightly due to increased flow temperatures.

Anyway, apologies for the slight digression there, but the point is that densely-fitted underfloor heating pipes (possibly upstairs too as this will roughly halve the power density required and so reduce the flow temperature target) should be treated as part of the fabric - they're a pain in the neck to retrofit, and are a key part of the building performance. By comparison the windows probably shouldn't be since they are easy to change.

[Jeremy Harris]

As pointed out before, be very wary of only looking at COP, it can be extremely misleading in terms of real-world performance, especially in the sort of cool and damp weather common in the UK heating season.

 

Far better to use SPF, which takes account of the true energy used versus sensible heat out through the operating season, taking account of all defrost measures.  These defrost measures can range from actively switching the heat pump into reverse to more subtle measures such as turning the heat pump on and off, or modulating it up and down, to allow the evaporator to naturally warm and reduce the icing risk.  The COP also varies a great deal with cycling and the control methodology, as the EST showed in the adjunct to their series of heat pump trials, with COP dropping to as low as 1.5 when an ASHP was caused to short cycle due to a light demand.

 

All measures intended to reduce icing, or that induce short cycling, impact on the SPF, but not necessarily on the published COP, as COP is usually measured without taking proper account of any defrost system or short cycling.  There's the added problem that ASHPs may tend to have defrost controls that may be optimised to give a good COP under the two standard air/water test conditions, rather like the infamous Volkswagen emissions cheat. 

 

Edited to add:

 

Just found this quote, from here: http://www.hodkinsonconsultancy.com/ashps-and-the-code-in-the-british-climate/ that sums up some of the problems with just using COP in the UK climate (my highlight of key points about COP and icing):

 

Quote

Whilst ASHPs can reduce absolute emissions of carbon dioxide when compared with conventional gas heating, there are a number of issues to keep in mind:

• The British climate: The ASHP product selected must be proven to maintain performance at acceptable levels in British winters (low temperatures and high Relative Humidity (RH)). In these conditions, the external unit freezes.
• Rating of systems: The heat pump must be correctly sized with respect to the heat losses of the dwelling so as to minimise frosting of the external unit and consequent loss of performance.
• Heat distribution: Underfloor heating is ~50% better than enlarged radiators in terms of Coefficient of Performance (CoP).
• CoP: This factor must be treated with caution as it is an instantaneous measurement and does not take account of varying external conditions throughout the year.

 

The article is worth a read, as it highlights many of the points that I've found out by experiment with our heat pump and the UK climate.

Edited September 2, 2018 by JSHarris
Added info

[MikeSharp01]

6 hours ago, Triassic said:

Applying your approach would suggest the house be factory made using ‘standard” components, with just enough insulation and windows of an optimum size. From what you’ve said, this is where modelling comes in. You could model various designs at a pre-production stage to identify the house that meets your predetermined passive house design criteria. 

I think you can do this with a site built design its just about iterating towards a series of local optima and then choosing which one you are going for. In @pdf27's initial post I sense that one dimension is missing, which is part of most mass production design work, and that is the nature of value to the customer in an infinitely customisable product - I agree this is very much taken into account in terms of what the end user wants and would be prepared to pay for. However in self building I feel that we are working towards and infinitely customisable product, as in approaching any number of local optima, the challenge comes in taking the 'cost' as a driver not so much for the design but mostly for the construction optimisation - as @JSHarris points out in respect off kitchen sizes. In a curious way this closes the loop on the production / manufacturing engineering cycle it EG 'now rework the design so we can make it' which is what @pdf27 points out in his modified 3 stage process. - inevitably embodies some aspects of compromise though. If one was looking towards solving the housing crisis as your main objective you would take the manufacturing engineering approach directly and every time. Why can a house be more like a car - which we discussed on the old forum IIRC.

 

 

Edited September 2, 2018 by MikeSharp01

[pdf27]

1 hour ago, JSHarris said:

As pointed out before, be very wary of only looking at COP, it can be extremely misleading in terms of real-world performance, especially in the sort of cool and damp weather common in the UK heating season.

 

Far better to use SPF, which takes account of the true energy used versus sensible heat out through the operating season, taking account of all defrost measures.  These defrost measures can range from actively switching the heat pump into reverse to more subtle measures such as turning the heat pump on and off, or modulating it up and down, to allow the evaporator to naturally warm and reduce the icing risk.  The COP also varies a great deal with cycling and the control methodology, as the EST showed in the adjunct to their series of heat pump trials, with COP dropping to as low as 1.5 when an ASHP was caused to short cycle due to a light demand.

 

All measures intended to reduce icing, or that induce short cycling, impact on the SPF, but not necessarily on the published COP, as COP is usually measured without taking proper account of any defrost system or short cycling.  There's the added problem that ASHPs may tend to have defrost controls that may be optimised to give a good COP under the two standard air/water test conditions, rather like the infamous Volkswagen emissions cheat. 

That's actually exactly why I was quoting a full set of curves for one unit, rather than trying to predict the consumption of a particular unit: because it's a set of test data under standard conditions, it gives reproducible curves which show what happens when particular variables are changed - in this case flow and outside air temperatures. In this particular case it gives a qualitative understanding of the impact of flow temperature on the thermodynamic efficiency of the cycle, which is essential to understanding how the system as a whole should be optimised - something SPF doesn't give you because I haven't been able to find equivalent data for SPF at different flow temperatures on the same unit.

As it happens, I'm not altogether convinced that SPF gives all that accurate a value in any case - I really need to do a long post covering that however, which will follow on after this. I was writing it, but it got eaten while I was out at the park with the kids. Essentially the problem is that it is written around a situation with no PV and with a short time constant/low decrement delay structure - true for legacy structures but not necessarily a new build.

 

43 minutes ago, MikeSharp01 said:

I think you can do this with a site built design its just about iterating towards a series of local optima and then choosing which one you are going for. In @pdf27's initial post I sense that one dimension is missing, which is part of most mass production design work, and that is the nature of value to the customer in an infinitely customisable product - I agree this is very much taken into account in terms of what the end user wants and would be prepared to pay for. However in self building I feel that we are working towards and infinitely customisable product, as in approaching any number of local optima, the challenge comes in taking the 'cost' as a driver not so much for the design but mostly for the construction optimisation - as @JSHarris points out in respect off kitchen sizes. In a curious way this closes the loop on the production / manufacturing engineering cycle it EG 'now rework the design so we can make it' which is what @pdf27 points out in his modified 3 stage process. - inevitably embodies some aspects of compromise though. If one was looking towards solving the housing crisis as your main objective you would take the manufacturing engineering approach directly and every time. Why can a house be more like a car - which we discussed on the old forum IIRC.

The thing is, Lego is almost infinitely customisable and yet all the models are made from a relatively small number of parts. The key I suspect isn't asking someone to "rework the design so we can make it" but asking them "please make me something a bit like this as cheaply as possible", while being aware of the key cost drivers (e.g. thicker wall insulation versus more glass) and designing for low cost with these at the earliest possible stage. One thing to be aware of is just how hard this is to do in practice - manufacturing companies have legions of people doing this and still make an utter pig's breakfast of it frequently.

[Jeremy Harris]

I'm absolutely convinced, from personal experience and measurements, that SPF is a heck of a lot more accurate, as a measure of efficiency, than COP.  The bottom line is that you can't fiddle SPF; it's the actual measured input energy over a year of an installed heat pump, operating in the UK climate, relative to the actual measure sensible heat that that unit has delivered over that period.

 

COP is a short duration, standardised measurement at two different outside air and delivered water temperatures (for an ASHP) and takes no account of relative humidity, demand variation, defrost cycling etc, so will always be hopelessly optimistic.

 

 

[MikeSharp01]

7 hours ago, JSHarris said:

expanding hollow "nails"

Have you got a link for these?, sound interesting, found this in Oz but not much coming up here.

 

Ah may have found them is it this http://www.timco.co.uk/media/documentation/tds-en.pdf product, if so are there also hold down straps at intervals on the frame as well?

Edited September 2, 2018 by MikeSharp01

[Stones]

27 minutes ago, JSHarris said:

I'm absolutely convinced, from personal experience and measurements, that SPF is a heck of a lot more accurate, as a measure of efficiency, than COP.  The bottom line is that you can't fiddle SPF; it's the actual measured input energy over a year of an installed heat pump, operating in the UK climate, relative to the actual measure sensible heat that that unit has delivered over that period.

 

COP is a short duration, standardised measurement at two different outside air and delivered water temperatures (for an ASHP) and takes no account of relative humidity, demand variation, defrost cycling etc, so will always be hopelessly optimistic.

 

 

 

My own experience mirrors your view.  COP, whilst perhaps useful to determine how an ASHP will perform at various pre-determined points, does vary quite substantially over the year as it is affected by a number of variables.  

 

The performance of our ASHP, (COP and SPF) is described here:

 

 

 

[Jeremy Harris]

12 minutes ago, MikeSharp01 said:

Have you got a link for these?, sound interesting, found this in Oz but not much coming up here.

 

They didn't look like those Aussie ones.  They were Irish, and our building inspector hadn't seen them before, so asked for one as a sample.  I think they are these: http://www.timco.co.uk/fasteners-fixings/masonry-anchors/express-nails

[MikeSharp01]

2 minutes ago, JSHarris said:

http://www.timco.co.uk/fasteners-fixings/masonry-anchors/express-nails

Yep - thanks, found them, looks good but did you also have hold down straps at intervals around the perimeter?

[Jeremy Harris]

16 minutes ago, MikeSharp01 said:

Yep - thanks, found them, looks good but did you also have hold down straps at intervals around the perimeter?

 

No, the sole plate was just nailed down to the concrete ring beam with those Timco fasteners.  We can't put anything around the outside, as the "foundation" at frame level on the outside is just 200mm thick EPS, with the non-load bearing half of the Larsen-type truss bearing on it.  Only the inner, load bearing half of the truss frame is secured to the 200mm x 200mm reinforced concrete ring beam that runs around the edge of the passive slab.  With several tonnes of cellulose inside the frame I doubt it's going anywhere.

 

Just checked, and each one of those fasteners is good for a 1.5 kN working tensile load, and at a guess there are over 100 of them holding our frame down (there's one in every 400mm wide bay, I think), so that's about 150 kN of hold-down force, or around 15 tonnes.

[pdf27]

OK, so having dealt with the first of the two criteria (15kWh/m²/year or 10W/m² at design cold condition - essentially the comfort limit) it's time to take a look at the second, Primary Energy use:

The key thing to take away from this is IMHO that storage of electricity is heavily penalised - as is the use of gas in any way. This is where the behaviour of the heat pump / hot water control system gets really interesting, particularly the smart grid functionality.

A house which is well insulated and has quite a bit of heat capacity (i.e. has a high decrement delay) can actually store rather a lot of heat energy relative to demand. A 150 mm concrete slab, for instance, has a specific heat capacity of 880 J/kg.K - if you're willing to accept a 2°C swing in the internal temperatures, then given a density of 2400 kg/m³ that's 633,600 J/m² of energy stored. At a design heat loss of 10 W/m² (i.e. 10 J/sec.m²) that means it will take 17.6 hours for the energy in the concrete slab alone to soak away - in reality you're looking at 2-3 days in even very cold weather before an unacceptable temperature drop should be seen thanks to all the other material storing energy present, provided you're willing to run the house a little bit warmer when the energy is available. Going for thicker walls and smaller windows will help with this.

The fundamental point here is that the COP/SPF/TLA of choice performance of the heat pump isn't actually that important - the critical thing is IMHO whether it is easy to automatically turn up the thermostat when clean electricity is available. Doing that shifts from "short-term" or "seasonal" storage (probably mostly the latter, I suspect) to "direct use", with a huge gain in system efficiency - far greater than you will get by messing about with different defrost strategies. Incidentally, this also means that storing hot water for extended periods of time (either in a big cylinder or something like a Sunamp) has a similarly big benefit. Increasing the effective size of the hot water storage, either directly or through the use of shower heat exchangers, etc. when coupled with some sort of smart grid programmer will potentially mean that even resistance heat will have quite a high effective SCOP compared to a gas boiler.

 

This is also where including PV in the building gets very interesting. Sheffield Solar has downloadable data showing the national installed amount of PV and generation in half hour increments since about 2015, which combined with degree-day data to give some idea of the ratio of electrical consumption to output can be used to make a crude spreadsheet model of how much of it can be used for heating and hot water (if you're using a resistance heater like a Sunamp the calculation is vastly easier). The one I've been playing with uses the entire south facing roof for PV and essentially just trips a relay to turn the thermostat up and bring the heat pump on when there is enough power being generated, but even there given how most heat goes to hot water something like 70% of energy consumption for heating and hot water over a year comes from PV.

 

From this I think the conclusions are:

1. The energy stored in the structure both as heat and hot water makes a serious difference to performance. A high decrement delay structure is very valuable - we keep coming back to a cellulose insulated timber frame - and everything you can do to increase effective hot water storage helps too. A shower heat exchanger looks like a very good idea indeed since this both reduces heat consumption and increases the effective size of any given hot water storage device.
2. PV is heavily encouraged by the new Plus/Premium standards, but PV on the building is also the only way at present you can really guarantee that you're directly using primary energy since you know when you're exporting energy to the grid. That means fitting PV should effectively make a big improvement to the PER factor.
3. Smart grid controls also mean a lot, again by pushing up the PER factor. SG Ready controls on a heat pump would be my preferred route, but even immersion heater controls such as those found on a Sunamp or Immersun-type device will help a surprising amount.

None of these are particularly expensive, but all help get away from the idea that the windows are your primary heating source. One thing I don't know, however, and haven't been able to find out is whether the latter two are actually catered for in the standard - I think they should be, but that doesn't mean that they necessarily are.

[Jeremy Harris]

Spot on about the long thermal time constant of a house with a well-insulated concrete slab and high decrement delay insulation.  Both @TerryE and I have found that the room temperature changes really slowly with changes in the outside air temperature.  Left to it's own devices, with no heat input, the house settles at a temperature a few degrees above the mean diurnal temperature range over a few days, due to solar gain.  We don't heat our slab at night (during the heating season) at all.  In winter the slab gets heated once every couple of days for an hour or two normally, unless it's really cold, when we may have to heat the slab for and hour or so every day.

[TerryE]

In terms of the total specific heat of the internal environment, the slab is the biggest element, but don't discount the SH of the plasterboard and effectively ½ of the cellulosic filler.  The other big element in our case is the ringbeam and two transverse structural beams at 300×300mm cross-section adds another big chunk.  I haven't even got around to buying my ASHP yet and we survived happily last winter on my 3kW Willis heater heating the slab overnight on E7 low tariff.   Our daily top is pretty much a simple calc based on average outside temp, and we can stay within the cheap rate 21kWh overnight down to about +3°C below that and we need a daytime boost.  (Of the top of my head, our internal slab area is 70m² and we have 3 floors so the total floor space is around 200m²).   The slab responds almost exactly as my simple 3D heatflow model predicted.

 

Our DHW is Sunamp again heated overnight by E7.

 

My house performs within 10% of my initial JSH-style spreadsheet that I did in the early planning stages.  But as Jeremy has said, the MBC warm slab and cellulosic filled Larson strut construction pretty much eliminates all thermal bridging.  Our one design cockup was that our slab has an outer ringbeam bearing our Cotswold-type stone outer skin that is mechanically coupled to the main slab with 200 odd 20mm rebar sections which is a fine mechanical solution but one hell of a thermal bridge.  Luckily I picked this up during the slab construction and was able to mitigate perhaps 80-90% of this flaw (see my blog post on this for more details);  I suspect that the unmitigated residue is why our actuals are maybe 5-10% worse than our initial estimates.

 

I've never bothered working it out in terms of annualised kWh/m^²/yr, but based on our actuals over last winter through spring and summer (which was a cold one) our total annual heating requirement is around 4,500 kWh/m²/yr .  We were required to conform to a traditional cottage style (= small windows) on an SW principle access so we have little solar gain and weren't allowed PV.  But this is around 3W/m² averaged over the year so I am not sure where the discrepancy is with your 10-20 figure.  DHW is on top of course as is normal energy use for electrics, lighting, cooking, but these combined less than double this, so our total energy use is well under 6W/m² averaged year-round.

 

As well as the overall heat balance, you also have to consider internal heat flows and heat gradients.  At our design point we only have the groundfloor UFH.  We have a slate floor throughout and usually bare footed indoors -- cheap and low maintenance; the floor also feels nice and cool underfoot in the summer and warm in the winter.  We don't have any upper floor heating, so no radiators or the like and associated pipework.  Tell a lie: we do have an oil-filled electric towel rail in our master bedroom ensuite, but the only time that we have turned it on so far was during commissioning tests.

 

Even so our upper floors are maybe 1½°C cooler than the ground floor in winter.  This suits us for the bedrooms, but I do find that my office is a little too cool for my liking, so I have a small Dyson fan that I use as an occasional  top-up when I am working in the office and it is very cold outside.   I think if you were to double or quadruple the overall systematic losses, then the main design consequence would be that you'd need some form of upstairs central heating with all of the complexities that that brings.  Stick to 400mm cellulosic filler or better, IMO.

[pdf27]

4 hours ago, TerryE said:

I've never bothered working it out in terms of annualised kWh/m^²/yr, but based on our actuals over last winter through spring and summer (which was a cold one) our total annual heating requirement is around 4,500 kWh/m²/yr .  We were required to conform to a traditional cottage style (= small windows) on an SW principle access so we have little solar gain and weren't allowed PV.  But this is around 3W/m² averaged over the year so I am not sure where the discrepancy is with your 10-20 figure.  DHW is on top of course as is normal energy use for electrics, lighting, cooking, but these combined less than double this, so our total energy use is well under 6W/m² averaged year-round.

10 W/m² is an alternative Passivhaus certification criteria, setting the maximum permitted heat loss per square metre of internal area at the design low temperature condition - averaged year round will be far lower than this, and is catered for by the alternative 15 kWh/m²/year condition. 15 kWh averaged out over a year is 1.7 W equivalent - note that this is looking for inherently low-cost ways to meet the requirements for Passivhaus certification, not low cost ways to build a comfortable home that is very cheap to run.

 

The 20 W/m² is what the underfloor heating would have to deliver at the design cold condition in a two storey house if it is only fitted into the slab - 10 W/m² for the ground floor and 10 W/m² for the first floor. It's essentially a plant sizing condition which sets either the pipe spacing at 25°C or the flow temperature at a given pipe spacing to deliver this amount of heat - important if you're using an ASHP.

 

4 hours ago, TerryE said:

As well as the overall heat balance, you also have to consider internal heat flows and heat gradients.  At our design point we only have the groundfloor UFH.  We have a slate floor throughout and usually bare footed indoors -- cheap and low maintenance; the floor also feels nice and cool underfoot in the summer and warm in the winter.  We don't have any upper floor heating, so no radiators or the like and associated pipework.  Tell a lie: we do have an oil-filled electric towel rail in our master bedroom ensuite, but the only time that we have turned it on so far was during commissioning tests.

 

Even so our upper floors are maybe 1½°C cooler than the ground floor in winter.  This suits us for the bedrooms, but I do find that my office is a little too cool for my liking, so I have a small Dyson fan that I use as an occasional  top-up when I am working in the office and it is very cold outside.   I think if you were to double or quadruple the overall systematic losses, then the main design consequence would be that you'd need some form of upstairs central heating with all of the complexities that that brings.  Stick to 400mm cellulosic filler or better, IMO.

One of the potential problems I personally have is that my wife is extremely sensitive to variations in temperature, and as a result we will need cooling in summer. Extending the UFH circuit upstairs will solve this, and make the heat pump's job a little easier too - I'm still trying to work out if there are cheaper ways of doing the same thing though. We'll certainly run the UFH water through a wet duct heater which will give a watt or two of extra heating/cooling, and some sort of electric towel rails in the bathroom too I suspect - I really don't know if that will be enough in our particular case though, and it's one of those that you can't retrofit later. Upstairs UFH is a lot of money for what you get though...

 

10 hours ago, JSHarris said:

Spot on about the long thermal time constant of a house with a well-insulated concrete slab and high decrement delay insulation.  Both @TerryE and I have found that the room temperature changes really slowly with changes in the outside air temperature.  Left to it's own devices, with no heat input, the house settles at a temperature a few degrees above the mean diurnal temperature range over a few days, due to solar gain.  We don't heat our slab at night (during the heating season) at all.  In winter the slab gets heated once every couple of days for an hour or two normally, unless it's really cold, when we may have to heat the slab for and hour or so every day.

I'd be very worried if I was coming to different conclusions to everyone else on here!

If you look at the Sheffield Solar data, it's pretty rare that you'll have several days on the trot with little or no generation above the plug load requirement over the course of a winter - limited to December and January, pretty much. To me that means there is the potential to significantly exploit the thermal inertia of the house and thus reduce the fraction of energy taken from long term (seasonal) storage.

[ProDave]

This 15W / m² thing

 

Our house is about 147 square metres and using Jeremy's heat loss spread sheet, I worked out the peak heating demand when it's +20 inside and -10 outside i about 2200W  That woks out at 14.96W /m²

 

We set out to build a well insulated low energy house but with no particular certification to any standard, but does this mean we have (almost?) achieved passive house standard then?

[Ed Davies]

1 hour ago, ProDave said:

This 15W / m² thing

There are two things in Passivehaus like that but neither are that:

 

1) 10 W/m² as the peak heating load in the design-worst case conditions. So you're 49.6% over that if -10 is your design worst case.

 

2) 15 kWh/m²/year as the total energy use for space heating.

 

Both are, I'm fairly sure but not certain, final energy use. You get to choose, you can meet either of these criteria, you don't have to meet both.

 

4 hours ago, pdf27 said:

our total annual heating requirement is around 4,500 kWh/m²/yr

 

I really hope there's a spurious “/m²” in there!

[pdf27]

Pretty sure there is - the power values quoted match up for that being a total.