Tuesday, February 4, 2014

Smells, Moisture, CO2 and Indoor Air Quality in Tight Houses

In past blog posts, I've noted various areas in which we are in a technological revolution - including lighting, renewable energy, energy storage, not to mention high performance buildings and windows..

But smells!  The science of smelling/sniffing is definitely in its infancy, but the potential for advancement to yield results is huge.  I looked into this just today because I've been working on our ventilation system, and we happened to install the PAUL Novus 300 HRV (probably the most efficient machine on the market).  But effective and efficient ventilation goes well beyond the heat recovery efficiency.  The energy demands of the ventilation strategy have a lot to do with sensing and control.  (And this was the main reason we chose the Novus. It allows up to 4 channels of control.  )

The ventilator should slow down when users are not occupying the house, or windows are open, should ramp up when users are having vigorous activity in the house, or there are a lot of people in the house, and should operate with a big imbalance if there is a vented dryer in operation or a vented kitchen range hood.  What if 6 people took a crap nearly at the same time?  The indoor air quality could be very poor for a while.
So, to minimize energy demands and maximize effective provision of fresh air to house, I would consider sensing the following, in order of importance with first items most important;

  1. CO2
  2. moisture
  3. Pressures inside and out
  4. windows and doors
  5. smells (!)
(this listing is not including the CO sensor - which could be important if you have gas appliances).

Using CO2 sensing, one could minimize the delivery of fresh air to the house and not over-ventilate.  It would also take into account the presence of more or less people and plants, and probably even fires.  CO2 sensors on the market for HVAC applications (there are also lots for greenhouse applications) are about $300 with NDIR (recommended) sensing systems).  Companies like Honeywell, GE, Senseair all make them).  You can get desktop CO2 monitors on Amazon for about $150.

Moisture sensing is important to protect the building contents, but more importantly to protect the structure, and prevent mould formation, etc.  But don't forget to protect houses from dry conditions also - moulds wont' grow, but anything made of wood will shrink and crack - very low humidity is just as bad as high humidity.

Pressure sensing would compensate for any other exhaust fans as well as open and closed windows.

Smells! - what if someone were to be using a harsh paint, or a cleaner, or indeed, the washrooms.  The science of gas analysis (all smells are due to gases in the local atmosphere) applied to smells has incredible implications.  For example, this company http://www.enose.nl/ says they can make electronic sniffers that can detect pathogens in the air - this means one could potentially detect a virus in the air in one's house, and record that event in a log.  By smell detection, the house could 'know' when the washroom was being used, or when cooking was being done, or when there was a fire, or when laundry is being done, or when mould was growing in the walls (!), when animals have nested inside the attic, perhaps, or when the outdoor air is more polluted than the indoor air.
There is much to read online regarding electronic noses.  The process of technological smelling involves 3 steps:  acquiring a standardized sample, detecting the chemicals in the sample, and analyzing what was detected.  This last part involves a fair bit of software, including databases - because various odor events are logged as a certain configuration of results form the gas chromatography or whatever detection method was used. - the device needs to learn and grow its knowledge to be effective.

All the above is another way for me to feel good about all the low voltage wiring we've been putting into our house - we'll have the built-in ability to employ sensors all over the house because of this, and things are telling me the future of buildings will have a lot to do with sensing.

Thursday, January 30, 2014

How Best to Plumb a Drain Water Heat Recovery System

If you are not sure what a drain water heat exchanger is, google it or read the following:


Here's a couple pics of our installation:

Below I summarize salient points on this concept/product, and then give my thoughts.
  1. Cost is $500 to $1200 for the item.  Installation is extra.
  2. Go with the largest diameter and longest exchanger you can fit into your plumbing, assuming a vertical installation.  There are horizontal ones available from http://www.ecodrain.ca/en/how-does-it-work , in which case, I'm not really sure - but I think they have just the one size.  Leave at least a 12" (0.3m) of straight drain pipe above the exchanger to smooth out the flow.  You can see we've done this, and we've used 4" pipe
  3. It is still worthwhile to pump shower drains in the basement back up so they can drain into the heat exchanger.  Cost of these Gulper pumps with a kind of control that senses the water level in the drain is something like $300.
  4. Gather all your drains to one place, if possible.  In our case, we've done this with all the shower drains in the house, which all happen to be clustered together in the building.
  5. Savings of some 50% on water heating energy are possible depending on how things are plumbed in the house and cost of energy.
  6. Hot water recovery times can be dramatically improved - this, to me, is a very sure sign of energy being saved.
  7. The design requires a double walled heat exchanger - in other words, there must be an air space between the copper pipe carrying the drain water, and the copper pipe carrying the potable water - this severely limits the efficiency of the unit, and increases the costs - more on this later.
  8. Gravity film - surface tension effects cause the drain water to stick to the drain walls in a thin film - this is why heat exchange is arranged at the surface of the drain pipe.
  9. Simple payback periods range from 2 years to 10 years.
My Thoughts:

Leaving water in the Bathtub  One comment I've read is this:  Recover your 'waste' hot water heat by just leaving the shower water in the bathtub and letting it cool to room temperature before allowing it to drain away.  This is an excellent measure, and no cost to install, but it has some issues:   - first, it adds moisture to the house - good thing in winter, probably,  not so good in summer.  2nd, only possible with bathtubs, showers can't hold much water.  3rd, ring around the tub, 4th, it gets difficult to have multiple showers in a short period of time - such as on busy mornings with a family of 4, and 5th, don't assume it is 100% heat recovery.  The water in the tub cools only to room temperature, which is almost the halfway-point from the cold water inlet temperature to the 40deg C or so needed for a shower/bath.  So heat recovery is something like 60%, similar to a good drainwater heat recovery installation.

Efficiency:  The best units are limited to something like 60% efficiency - measured in terms of inlet and outlet water temperatures. Industrial heat exchangers are good for over 90%.  Even over 95% in some cases. A much more 'effective' design, would be to simply have a copper drain pipe inside of another larger, plastic pipe filled with the pressurized, cold water.  This would increase heat exchange efficiencies a lot (to probably above 80%, easily more), and reduce the cost of the units greatly as well, and also reduce the pressure drop incurred by the units we use today.  Given the huge potential for energy saving across a nation like Canada, one would think it is possibly worth the very small contamination risk - after all, if ever there were a leak, the pressurized water would go into the drain, not the other way around, and it would be fairly easy to detect - just check the water meter while all fixtures are off - perhaps an annual check would be worthwhile.  Imagine - if we could recover 95% of the energy used to heat hot water for bathing, small electric point-of-use hot water heaters would be so much more viable (read my post on POU Hot Water) - they could be built right in to shower fixtures, and this could lead to all sorts of interesting plumbing configurations (just plumb one line, for example - no need for both hot and cold?).  I can see Doc saying "What, are we in the Dark Ages?!".

Alternative Strategy: One way around this is to avoid the issue altogether.  Instead of directing reclaimed heat to the DHW system, transfer it to the space heating system instead - ie to a non-potable heat sink - such as a hydronic heating component, or direct to refrigerant in a heat-pump system.  This avoid the water contamination issue, and can recover much closer to 100% of the energy used for heating water.  Sadly, I know of nothing on the market that does this as yet, but it would be very easy to build something - a coaxial pipe heat exchanger is all you would need.
Another thing not being explored - use heat pipes to do the heat exchange work - this would probably increase the options for horizontal exchangers.
And yet another approach involves using an auxiliary tank and a pump - so we recover heat even from processes like clotheswashing and dishwashing, in which the hot drainwater is not expelled at the same time cold water is being drawn in.  Here is a link to one company to doing this:

Installation:  You can see there is a lot of copper in these things - the one pictured was a good 60 lb.  We came up with a simple way to mount the item effectively.  We used a 4" water closet flange.  We removed the small groove inside this flange so the 4" ABS pipe could be passed right through.  This is basically creating a bulkhead fitting on your pipe - but there is no break in the pipe.  Glue it on with solvent, and then mount to a couple of wood runners.  Makes a good support.  Notice also we've installed unions around the exchanger - this will let us more easily check its insides from time to time and clear out any build-up.

Water Pressure Losses from Drainwater Heat Exchangers:  Something not mentioned too often:  What about the pressure losses?  The one we chose to install uses a single 3/4" copper tube (about 60 feet of it) wrapped around the 4" drain.  In choosing a product, it is a matter of heat exchange performance and this is measured by the Canadian government testing apparatus, so after finding units that perform well, we looked for designs that retain water pressure the most, and most robust.  So how to plumb in order to retain water pressure?  If you have high water pressure, this may not be an issue, but keep in mind, water pressure is not free.  Someone, somewhere, must provide it, and I feel we must always think of our buildings as 'off-grid', so we want to minimize energy losses at every opportunity.  Therefore, buy a unit that performs just as well, but results in the least pressure loss.  then, I would plumb everything through it.  After all, we heat a lot of water to just room temperature via space heating - it sits there in the toilet, in the pipes, etc.  But to recover that heat using the drainwater heat exchanger, the cold side must flow - therefore, as per manufacturer recommendations, just plumb everything through it - I would still provide a small line to a drinking water fountain or icemaker, however.

Sunday, December 15, 2013

Low and High Voltage CAN go in the Same Box

All our lights are low-voltage.  I looked for boxes for the switches, but I was surprised to find the plastic LV boxes were expensive, and in my view, not that great.  I searched for a steel box and found one!  It is the Hubble product pictured below, the price is reasonable, (less than the LV mud-ring boxes) and the divider is removable (additional $1 for the divider), so the box can accommodate both regular household voltage and low voltage in the same double-gang box.  We used double-gang boxes for all our switches and plugs in the house - (almost all).  Most will have a CAT6 cable and an 18g. cable in the LV side, plus whatever goes on in the household current side.  From these, we can eliminate power adapters in the house by providing LV direct from the plug, and we also can use any 12V devices, and have DC lamps (so DC power direct to LED COB lamps).  DC-DC boost or buck pucks are available online now for about $5, so you can get almost any voltage you want by placing it in the box (we are supplying 28V), and it would remain within the electrical code rules as long the voltage remains below class 2 wiring limits.  In addition, every plug location becomes a data/voice/LV/automation/sensing/instrumentation node as well, without adding additional boxes all over the place.  So we can have things like motion sensing that controls things far away from the box, microphones, speakers, bluetooth, wifi, fire sensing, temperature, humidity, etc.
In many of the switch locations, there is no household current component, so the whole 2-gang box is for LV.

There are virtually no octagonal boxes in our project - Just one, actually in the utility room because I wanted to have a different source of lighting while the LV lighting system was being worked on.
You can see the LV wires come from above the plywood - they were incorporated into the acoustic floor assy above.  This could be done with 115V also, but you would likely need for it to be in conduit depending on the acoustic floor thickness.

Exterior Cladding is Finally Going Up

The exterior cladding is finally coming together, but its hard to photograph due to scaffolding, and the difficulty of getting a good camera angle.
We've made so many changes to the project it is no longer funny, so I've figured out now that one way to avoid changes is to build it fast!  Everyone seems very unwilling to change things once the money is spent and the construction complete.  Sadly, that is not our current situation!

Here's a rendering of what we are doing:

I offered to go up there and paint the roof white (because we both felt we chose the wrong colour shingles for it), but she's not keen on the white idea, so the brownish mistake will remain.  I find the house pleasing, though it is simple.

And here's a pic of the cladding going up:

Removable Basement Floors and Interior or Exterior Insulation Placement

A basic decision that one faces early in the design of a super-insulated building is the strategic choice of interior/exterior insulation placement and thermal mass.  This a strategic decision because it has far-reaching implications and ripple effects.  Think of the building as a shell on all sides, including the parts in the ground.  If we are designing an airtight envelope without thermal bridging, then we want to avoid having some of the insulation inside, and some on the outside - it can be done, but this frequently leads to thermal bridges and sealing problems.  For example, if we have insulation under the footings, (this being outside the structure of the shell), but then we want to have insulation inside the basement walls, how to connect the insulation under the footings to the insulation inside the basement?  The problem is there because in general, insulation materials are weak and soft, while structural materials are hard, but conduct heat.  To simplify the design and construction greatly, and improve the effectiveness of the insulation system, work to have all the insulation either outside the shell, or inside the structural shell.  Cross-overs are to be avoided.  In our case, we decided to place all the insulation inside the shell, and forego the thermal mass benefits - I believe thermal mass benefits are less well proven than insulation benefits, and that 'thermal' mass can be achieved without 'mass' (for ex. by the use of water - a very thermally massive material without much mass, that can be moved around).  
SO, here are more photos of our basement floors - they are above all of our interior insulation (about R55, or 15" of Roxul) above our basement concrete slab.  As posted earlier, they are removable, and they are a common material - regular construction lumber 2x12.  which means we can remove and replace pieces, but we can also remove and look underneath.  We're currently pretty happy with these floors, and the system feels very solid to walk on - as if the floors were resting directly on concrete.  It turns out the wood has shrunken a little in the 2 months since we installed it - but only the pieces that were wetter.  those nice planks in the 2nd photo have not shrunken at all.
Some astute observers have commented that the floors will allow moist interior air to go into the spaces below the slabs.  What will happen to this moist air when it reaches the cold concrete some 17" below?  Well, we have Tyvek under the floor boards in one area to prevent this bulk movement of air, but most of the floor is left without any kind of air barrier.  Since it is removable, we can make a correction if this turns out to be an issue, but I have a feeling the issue is fairly minor for a couple of reasons.  If we think of regular basements, many have no insulation under the concrete floors, and they are perhaps a bit damp on muggy, hot summer days, but often this problem is short lived in the Toronto climate.  In our case, there is a floor assembly blocking the bulk movement of air to some degree, and in addition, the space beneath our floors may be warm for much of the summer due to our under-floor (sub-slab) heat storage strategy.  This raises the temperature of the basement concrete slab right when the chances of hot moist air condensing on it may be highest, which should reduce that whole issue quite a bit.

However, as there could be a small concern, we did place some sensors at the bottom of the floor insulation, in three locations.  The photo below shows a small pump with tubing, a water level sensor, and a temp/humidity sensor in the background.  The sensors are inexpensive devices for Arduino, and cost about $5 each.  The pump was from Princess auto and was about $20.  We had some problems with our basement floor pour - there was not enough slope in some areas, and during the big Toronto flood in July 2013, we noticed a little water in three locations on the floor, and so marked these spots and placed these little pumps to transfer the water to the sump pit.
Later on, as the systems become live, we will be able to report the fluctuations in temperature and humidity at the bottom of our basement floor assemblies.
We will also probably place sub-slab soil temperature sensors as well, one day...

Sunday, October 27, 2013

Cellulose vs Urethane Foams - Again

I've noticed that in much older posts I reported on costs of various types of insulations.  More on this here. We have now finalized our cellulose insulation contract with Greensaver - a not-for-profit in the Toronto area.  Costs for dense-packed cellulose insulation for walls seems to break down like this, at least on our project:

Walls:  4085 cubic feet, at 3.5lb/cf density = 14,300 lb cellulose, at 33 lb/bag, we need 433 bags at about $10ea. - so materials for the walls are $4330.

Attic:  4183cf at 2.0 lb/cf = about 8400 lb in the attic, or about $2550 worth of material.

Labour in the walls is about twice the material cost, and labour in the attic, about 70% of material costs.

Much thanks to Climatizer insulation of Toronto for providing a fantastic price on the material for our project!  (They've had a tour of our house and took a step to support us as a contribution toward green initiatives).  Here are the bags we will be using:


In the past I've noted that spray-foams are about 10x the costs of cellulose and the other fibrous insulations. There is a bit more to note regarding this issue, which has some impacts on the 10x difference.  I discovered this video on youtube: http://www.youtube.com/watch?v=F26eIesDDQg&feature=player_detailpage showing the use of pour foam - liquid foams that you pour into things.  This is very similar or the same stuff that is used in the spray foam process, (it is basically the same, but additives may differ).  The video points to Aeromarine Products http://aeromarineproducts.com, where you can purchase the foams right from the website.  You'll notice you can purchase about 500 cubic feet of 2# foam for about $3900 - which works out to $7.80/cf, or about 10.8 cents/sf-R.  This is still a lot more than cellulose, at $0.303/cf, or $0.0072/sf-R.  Something like 15x the cost on the materials.  Here is a place you can purchase spray-foam kits: http://www.sprayfoamkit.com/products/spray-foam-kits, and they also give you the prices right online - I love it when they do this.  Here the price of spray foam on their largest kit works out to $14.3/cf - almost twice the price of the pour-foam.  This is some 30x the cost of cellulose, R for R (not accounting for the fact that cellulose takes about twice the space to achieve the same R levels - the value of space and the construction details required to build this space for cellulose are pretty variable - but then, we are also not accounting for the health and environmental footprints of the two materials, which are vastly different as well, with cellulose miles ahead on both accounts).  Note that labour costs are not included in the comparison, but given the labour portions noted above, we are still well ahead with cellulose.

Why is spray foam so much more expensive than the pour-foam?  Pressure vessels, and possibly additives - but mostly the pressure vessel/ hoses, gun, etc.  So the interesting point here is that if one must use urethane foam as a DIY, consider buying it in the liquid phase - that's what contractors do.  If you don't need to spray it on walls, but can pour it into a cavity, this is really the way to go.  In our case, we could have poured it into our walls - just like we will be 'pouring' the cellulose.

PS - We are purchasing larger amounts of mineral wool for our project as well, and Winroc has also given us excellent pricing on the material - again, to do their part to support 'green' projects.  Our cost for this material worked out to about $0.034/sf-R - about 4.7 x the cost of cellulose, not accounting for installation labour, which would reduce the cost advantage of cellulose, probably bringing it nearly even with the Roxul.

Tuesday, October 8, 2013

Above Floor Acoustic Assemblies, Basement Raised Floors, Porcelain Tile Exteriors

Damnation!  A power outage caused the loss of almost the entire article I was writing for this post.  Here it goes again.
Below are steel plates with screws holding the plywood floors in line with each other.  We had issues with this (plywood not flush with adjacent plywood) due to the exposed beam ceiling below - we used good-one-side fir plywood for the sub-floor, and it does not come in T&G.  With exposure to weather, the plywood warped and bowed.  The plates, with short screws, bring it back in line.  The steel was clear-coated.  Not shown are holes we drilled just after this step to accept low voltage wiring for lights.  In the background you will see the wood frame for a structural wall.  One must be careful when framing a structural wall over an exposed beam ceiling - or any ceiling.  Having the studs directly over joists is important to reduce warping of the plywood - but this is especially so in exposed beam ceilings.  We also beefed up the bottom plates of the walls to support studs which were between joists - see photo.  There is much to learn about open beam ceilings, in terms of construction finesse.  Some concepts are thus:
  1. Reduced height joists:  Joists in the exposed beam area that are not as tall as the regular joists are advantageous.  This is especially achievable if you have steel-beam joists, but heavy barn beams and other thick members would work as well.  The reduced height allows one to add some thickness of structure above the exposed beams - such as diagonal planking, acoustic floors, and sleepers and wiring chases.
  2. Use all galvanized/stainless hardware to prevent stains due to weather exposure.  Avoid making beams out of thinner, doubled members - It is difficult to double them properly due to the desire to avoid exposed fastenings.  Also, protect the assembly from wetness after the floor above is built.
  3. Beware the structural wall above:  Any wall built above exposed beams should have the studs aligned with the beams.  Point loads due to door openings are to be avoided to reduce uneven loading on the exposed beams, which would cause some beams to deflect more than others, or even twist.

On the right, we are laying in an above-floor acoustic treatment.  Above the 3/4" ply subfloor, it consists of 6mm cork, then thick foamed poly sill gaskets under 1x4 sleepers.  Sliced up roxul batts fill the spaces between sleepers.  This allowed us to run our low voltage lighting wiring above the sub-floor as well (covered by Roxul in the photo), so no wiring is visible in the exposed beam ceiling below.  Then we glued and screwed 3/4" T&G ply to the sleepers, again with short screws.  The sleepers hold the ply together, and the ply holds the sleepers in place. Much cheaper than two layers of ply.  Sound ratings unknown for this assembly, but the difference is very noticeable to our ears.  This we are doing all over the 2nd floor except in washrooms and other areas of hydronic radiant flooring.  Total thickness of this assembly not including the first sub-floor is just under 2". All the door openings were framed to accept this added height.

R-54 Insulation in the basement.  That is an R32 Batt under the 2x6 joists in the basement floor, and an R22 roxul batt will fit between joists.  Below these batts is a space made from PT lattice and landscape fabric, held up 2 to 3 inches using small PT blocks stapled to the batts.  This lattice assembly keeps the insulation off the concrete floor beneath, and provides an unimpeded 'surface' drainage path for any water in the basement to flow towards the sump system.  Any water which remains is able to dry through the insulation upwards.  Any water in the insulation can also flow downwards through the porous landscape fabric.  We were able to get the lattice material super cheap - it was culled material - an entire skidful for $50.  The landscape fabric was about $8 for 150sf.  We stapled it to the lattice.

In the next pic you see a 2x12 pine floor screwed to the joists in in the basement.  We buried the screws about 1/8" so we could later sand the floor and get a somewhat finished surface.  The boards will help to regulate moisture below the floor by passive moisture through them, but also by absorbing and releasing moisture.  The main reason for using this kind of floor - low cost (again, we were able to purchase the material for a great deal - 25% discount from regular contractor pricing) at about $1.25/SF.  This single layer of floor will become the finished floor (with sanding), so it means we can easily remove sections of flooring to inspect/service the spaces below in future.  We therefore install without overlaps and plan the layout to allow removal.  The floor feels extremely solid!  Air sealing is not required at this floor - this was determined from previous air-tightness testing on the building, so we know we are already down to Passive House levels of air-tightness.
The last photo shows large porcelain tiles we plan to use in the cladding.  These are 16x32" porcelain tiles.  The clips you see are stainless steel.  More on this later.

Monday, October 7, 2013

Geo-Exchange is a Largely Untapped Resource in Canada

A new study found ground temperatures in Canada have risen significantly in recent years.  Find this article here.

Jay Egg of Egg Geothermal in Florida writes in the Sept 2013 issue of Plumbing Engineer on some interesting approaches to ground coupling which were certainly new ideas to me.  With geo-exchange (or 'groundsource'), we normally think of a flowing liquid in a closed loop of pipe inside the ground.  This can be a regular pipe made of various materials, or it can be a completely sealed 'Heat Pipe' (look up heat pipes if you are not familiar with them http://en.wikipedia.org/wiki/Heat_pipe).  However, there are other approaches of strong merit as well.

One such approach is the open-loop concept, in which one pumps water from the ground, harvests its thermal potential energy, and re-injects this water into the ground.  This system has the significant advantage of reduced tendency for the local soil to drift in temperature, especially in dry periods.

Another approach is the standing column well (SCW).  This I need to read more about.
Perhaps Here:

A strong resource on ground temperatures in Canada:


Sunday, May 26, 2013

Super Energy Efficient Lighting for Low Energy Buildings

There is a lighting revolution going on right now - just as there is a solar energy revolution and renewable energy revolution, and energy storage revolution!  I don't remember anyone predicting that after the information age began, we would be going through a major shift in energy and infrastructure.  And this is not to mention Passive House.  I feel like saying people building standard houses today will face some significant obsolescence issues within a few short years - namely in energy, IT, and lighting.  At first these seem like parts of the house one can easily change - but the energy aspect is a big one, and certain things are really hard to justify changing - like missing out on good solar exposure, major thermal bridges, and insulation values that don't cut it.

About the lighting:  Cree announced in December they have an LED that offers 200 lumens per watt (!).
Here's the press release:

If you are in the big box hardware stores, take a look at the LED offerings.  They are all hovering around 60 lumens per watt right now, if you get a good one.  Most are at 50 lumens per watt, often less.

How to get highly efficient lighting without spending too much?  Low Voltage distribution.
Watch the Video on What Is Lumencache:  http://www.youtube.com/watch?v=7eULjpkf7oE

It seems the thing to do is to separate the COB (chip on board - which is the actual LED chip) from the power supply.  This has two advantages:  purchase a single, centralized, high-efficiency, high-efficacy power supply; and then reduce the cost of the actual luminaires.  All those backward-compatible LED bulbs you can buy to replace the bulbs in your existing fixtures have the significant issue in their design that they have to have an on-board power supply, which is cheaply made to reduce costs and therefore, not that efficient. This also means the cost of those low performance power supplies is included in every bulb.  Eliminate this circuitry and you can improve the lighting.

So, we wire the house as per the Lumencache strategy, but there is another issue to handle.  LED's are pretty sensitive to voltages and current.  Say we have a living room with 12 LED luminaires, each one with nothing more than the COB in the luminaire.  We need to arrange these luminaires into groups that match the voltage and current output of the central LED driver system, or at least keep the demand within the range of output levels available at the central power source.

Welded Stainless Steel Floor - Waterproof

There are a few options for waterproof exterior decks/balconies.
  1. Use roofing on the deck - for small areas, a single EPDM membrane - black or white - over a wood frame.  I would think something needs to go on top of this to protect the membrane and make it more suitable to deck furniture and foot traffic - don't know what.  Regular built-up roofing is not my preference - I've been on this kind of balcony before and found the odour quite strong - so I assume the tar and rubber are off-gassing - anyway, it makes the space unpleasant.  I also would not do this over a deck which has had rigid insulation applied under the roofing membrane.  There is a mushy feeling underfoot, and a sense of doubt as to whether one can place furniture on the surface.
  2. Extruded aluminum deck:  These are generally powder coated extruded aluminum with either a rough or a smooth finish, such as available from Wahoo decks.  These work well, I think, but they are pricey at about $19/SF.  In our case, I couldn't find anyone local who imported them.
  3. Rubber membrane installed in a slanted scallop shape under the deck - it drains to one side - not an elegant solution, though cheap and simple.  Perhaps more suited to decks over soil, to carry water away from the house.
  4. The option we chose:  welded metal.
Welded stainless flashings not yet installed.
Why?  Cost of stainless is low right now - about $1.75/lb.  (it was about $4.55/lb ten years ago).  If you calculate the volume we are buying (1/8" thick plate is 18 cubic inches per sf), and multiply by the density of 0.3lb/cubic inch, you get the weight per sf.  Multiplying by price, we get about $9.45/SF - not too costly for smaller areas and it's an easy purchase from local sources.  I figured on a small area, I wanted to try this, and the thermal expansion issues would be minor - I expect this could be an issue on large expanses outdoors - the stainless steel has been in place a month or more now, and it is obvious that despite the high reflectivity, it does get quite warm in the hot sun - like a roof, so thermal expansion is something to keep in mind, as they do on steel roofs.  Stainless steel will also accept tons of abuse without any maintenance, feels solid underfoot, doesn't smell bad, and modifications can be made by welding to it....

I always wondered why people don't employ this solution in their projects for roofs and balcony floors, but even for interior floors.  What I found was that it was a fair bit of work, but not ridiculous, and the result is (I'm hoping) going to prove very practical and durable.  The steel frame in grey primer you see in the photos is regular structural steel, 3" deep C-channels welded to 6" deep hollow beams.  It is far stronger than it needs to be, but we ended up doubling up the joists from what is in the photos - It is a slightly difficult calculation to figure out the floor deflection as a diaphragm, and my rough numbers turned out to be close enough, but I didn't trust them, figuring the floor was supported on all sides.  The deflection of the diaphragm proved noticeable enough and it was simple to add additional joists so they ended up at 9" on centre, originally about 18" OC.  The floor feels very rigid now.  Originally, I thought I would have it plug welded from above and ground smooth, but the mag drill wouldn't stick well enough to the steel through the 1/8" stainless, so we mig-welded from below.  We tig-welded the seams from above (with some mig tacking from below) as the photos show - this is the reason for the tarp.

One thing I really like about stainless - I always feel it is easy to add to it with welding.  You needn't grind off the paint as you would have to with steel, and then re-paint after welding.  I knew I'd want to add some fittings, possibly cleats, hold-downs, etc.  This is easily done on exposed stainless.  The flashings were all welded on.  Most are 22ga, bent on site with a heavy brake - this proved tricky to weld to the 1/8" plate.  The flashing under the door was 14ga, fitted into a grove in the PVC door frame - it is just a window placed on the floor.

Distortion:  There was some distortion.  We had a 1/4"/ft (1 in 50) slope on the floor.  In some areas there was enough distortion to upset this slope and cause pooling - mainly on the long seam you can see near the 6" deep beam, where small sections were welded on.  I'm not too concerned, but we'll be working to correct this and see where it goes.

We used 304L for all the stainless - I'd done a little stainless tig-welding before, and I was familiar with the significant movement of carbon and chromium within the material in the heat-affected zone.  This causes non-stainless areas to appear near the weld afterwards - this is why you need to passivate the stainless after welding (dip into an acid bath to remove the surface areas of non-stainless material).  In our case we would not be giving acid baths to the stainless.  The 'L' in the material designation stands for 'Low carbon', which reduced post-weld corrosion quite a lot.  As it turned out, we did have some surface corrosion appearing on our project, but it was not due to the welds, which seemed to work well.  It was actually due to the use of wire-brushing with steel brushes.  We switched to stainless brushing after noticing this and all was well.

Welding was with both 308 and 309 rod - yes, welding regular steel to stainless is entirely fine and routinely done.

The reflection off the stainless is very bright - it is blinding, actually.  I was hoping this would be a good thing to raise the light incident on the deeply inset glass balcony door - but we did plan to have real wood-slat tiles or boards to block this reflection anyways.  I also plan to do some solar experiments on this balcony, so the reflective surface may be an interesting feature for this.  I've not seen anything on how to integrate this effect into the Passive House thermal modelling.  There are solar reflectors you can purchase that actually track the sun, and reflect it into a given window on the house, such as these:
http://www.egis-rotor.de/helio_us.html  This would be a significant energetic effect on the house - one that might be an attractive solution in some cases.  This one is really cute:  http://wikoda.com/.
However, if this were part of a design, I would place the reflection-receiving windows high on the wall of the house reduce glare inside the building.

The experience with this has had me consider other metal flooring solutions:  A carbon steel sub-floor/finished floor in the basement (the beauty of this is that the subfloor is the finished floor, reducing labour and materials, and the floor plates - I'd use 4'x2' plates, screwed down - can be removed to gain access to the space beneath)......and also radiant stainless steel staircases (designed well, a radiant staircase can be good for both cooling and heating, since the stair is both a floor and a ceiling).

C3x5 (75mm deep, 7.5kg/m) C-channels welded to 3"x6" deep hollow steel tube beam - Beam wall was 3/16" or 1/4" thick.

Too bad we've not cleaned off the dark residues from the welding for the photo.  Need more photos of this part of the project.