[Greenbuilding] South-facing windows on living spaces lose heat on cloudy days...

[nick pine]

John Straube <jfstraube at gmail.com> (aka God? :-) writes:

> ... sunny days are NOT like coin flips

Sure, there are differences, but in my engineering experience, a 
statistically-unbiased model with equally-likely cloudy (0 Btu/ft^2 on a 
south wall) and sunny (2X average) days at the average monthly outdoor temp 
is useful for back-of-the-envelope sizing, before refinement with a 
simulation. (A 3-state Markov model can  match longer actual weather 
patterns more closely, but the next step for me is a simulation using 
measured TMY2 hourly local weather data.) I use a metric from PE Norman 
Saunders to find the worst-case month for solar house heating: divide the 
average amount of sun on a south wall by the indoor-outdoor temperature 
difference to find the month with the lowest "Btu/HDD" figure of merit.

The coin-flip model can also be useful for softening the brains of people 
who imagine or delude themselves into thinking that direct gain houses can 
have high solar heating fractions (eg 90%) or low fuel bills or firewood 
consumption (eg 1/2 cord per year) in cold climates, despite simple 
physics.. ("How does the house keep itself warm on cloudy days? How often do 
cloudy days happen?" How much do you use that wee woodstove?") This works 
best with people who are familiar with simple physics. Norman Saunders 
called what he did "just physics" -)

>... you cant reach high solar savings fractions using average monthly or 
>even daily data.

True, but that's just the first step. Norman started by assuring a house can 
keep itself warm on an average day in the worst-case month with passive 
solar air heating, THEN added enough heat storage for 5 cloudy days in a 
row. A design based on average monthly weather data works well if a house 
stores heat for a few cloudy days. Norman said 10 days can further increase 
the solar heating fraction, with diminishing returns... 5 days is more 
economical. This ballpark design can be tweaked with simulations using TMY 
or 30-year measured hourly local weather data.

> Monthly average data can get major savings like 50%., but to get into the 
> 60's and higher is really hard in all cold climates.

High solar heating fractions are  extremely difficult  with direct gain, 
because windows on liviing spaces lose heat on cloudy days! With infinite 
thermal mass and insulation behind it, a good south window on a living space 
can barely keep  itself  warm on an average December day and night in 
Minneapolis...

It's far easier with air heaters or low mass sunspaces with heat storage for 
a few cloudy days and DHW preheating. A 40'x60' house with an R40 ceiling 
and R30 walls and 96 ft^2 of U0.25 windows and 30 cfm of air leakage and a 
30+96x0.25+1504/30+2400/40 = 144 Btu/h-F conductance and an average 65 F 
indoor temperature needs 24h(65-17.9)144 = 163K Btu on an average December 
day in Minneapolis. If an R30 floorslab with perimeter insulation loses 
24h(65-44.9)2400/30 = 39K Btu to 44.9 F deep ground, totaling 202K and 300 
kWh/mo of indoor electrical use supplies 34K and 1 ft^2 of R2 100 F air 
heater with 80% solar transmission gains 0.8x820-6h(100-22)1ft^2/R2 = 422 
Btu/day, we can heat this house with (202K-34K)/422 = 400 ft^2 of solar 
siding, eg  8'x50' of the south wall. It needs 840K Btu for 5 cloudy days in 
a row, eg 840K/(140-80)/62.33 = 225 ft^3 of 140 F water cooling to 80 in a 
4' tall x 8.5' diameter indoor sheet metal tank. If the 219 ft^2 tank 
surface has R30 insulatiion, it can supply 100x24h(140-65)219ft^2/R30/168K = 
8% of the house heat on an average day with a small solar water heater to 
keep it 140 F.

>The highest solar savings fraction that has been documented is now a sola 
>seasonal storage system in Alberta and it is drifting toward 90%

A 90% solar heating or "solar savings" fraction? Indoor tanks wouldn't waste 
80% of the solar heat...

> I dont know if a tank in a single house is a good solution.

Heat leakage from an indoor tank can help warm a house. And should everyone 
suffer if one homeowner wants to leave windows open in wintertime?

> I have seen houses in Europe with exterior 10 000 gallon tanks (that 
> should do it for an efficienct house!) and I have seen 2000 gallon tanks 
> inside houses in Fairbanks Alaska... Somewhere in that range is likely the 
> answer for a modest good house aiming for high solar heating fractions.

Eli Talking wants to turn whole basements into tanks. An old Swedish rule of 
thumb said make the collector the size of the house footprint and make the 
heat store volume the same as the house volume...1000 gallons work well in 
my simulations. The 1700 gallon Minneapolis tank above would be smaller with 
more attic or slab insulation, but the house might cost more.

>I just looked at the Drake Landing site...
> It is -20C and has been below -15 or so for a while
> Looks like the 52 houses are drawing 310 or so kW.  This is about 6 kW per 
> house or 20.5 kBtu/hr for a -4F temperature and 1500-1800 sf above-grade 
> house square footage.  That is not bad, but it could likely be dialed down 
> by 25-50% with a reasonable upgrade in insulation  windows, airtightness.
> They have two tanks, each with almost 30 000 gallons. So that works out to 
> 60000/52= 1150 gallons per home.  If you check the annual report, 80% of 
> the solar energy put into the ground is lost, but over 95% of the energy 
> put into the tank is able to be used.  So maybe 2000 gallons might be 
> enough for one home, and should be enough for a super insulated one. Note 
> that they have 8 solar panels of about 4'x8' each to collect the heat 
> (that aint cheap either).

If most of the house heat comes from air heaters and inherent house mass on 
average days, we need fewer collectors to trickle-charge a hot 
well-insulated tank. And houses with air heaters that lose no heat on cloudy 
days can have smaller tanks.than houses with lots of direct gain windows 
that lose heat on cloudy days, right?

> See www.dlsc.ca and look at the most recent newsletter, or just browse.
> All other documented buildings, including highly insulated ones with solar 
> air heating, are much less.

John Christopher's rock bed storage CSI building in New Hampshire has been 
heated with 98% solar power and 2% fan power for many years..

> Norm Saunders probably got some of his homes to 90%+ (100% if you were 
> willing to freeze) but this was never documented.

After lots of conversations and visits with Norman over the last 30 years 
(he died 4 years ago, at age 91), I believe his houses were very close to 
100% solar heated, starting with monthly weather data averages and air 
heating walls and low mass sunspaces and heat storage for 5 cloudy days in a 
row. Some of his houses had long track records, more than 10 years. Norman 
was a chemical engineer by training, not an academic, but he was designing 
transistor circuits around 1950, while most EEs were still using vacuum 
tubes :-) His Z80 controllers recorded performance data on cassette tapes 
which customers mailed back to him every month, but I doubt this will ever 
be documented in an academic sense. Chemist Harold Hay also designed houses 
that were very close to 100% solar-heated. Harold Hay and Steve Baer often 
denounced academic and government scientists whose careers depend on never 
inventing houses that heat themselves :-)

Lawrence Lile <LLile at projsolco.com> writes:

>> > Greetings from Missouri, where we are treated to a month of clouds in a 
>> > typical November.
>>
>> NREL says 720 Btu/ft^2 of sun falls on the ground and 1020 falls on a 
>> south wall on an average 46.2 F November day with a 54.7 high in St. 
>> Louis.

December is the worst-case month for solar house heating, when 940 Btu/ft^2 
falls on a south wall on an average 33.9 F day, with 26 Btu/HDD, morer than 
820/(65-17.9) = 17.4 in Minneapolis.

> NREL can say whatever they want, but they haven't bothered to look outside 
> at the coating of lead that substitutes for skies around here.

You might find that depressing, but NREL has been measuring real hourly 
weather data with instruments in St. Louis since 1961.

>Weather averages aren't very good at representing extremes that happen 
>rather regularly, but not always on the same date.

Then again, NREL's 30-year hourly weather files are accurate histories, as 
are the TMY files compiled from those files.

>"Average" January around here is supposedly 30F because it might literally 
>be 0F or it might be 60F on any given date.   EVERY January the ponds 
>freeze over thick enough to walk on, requiring a couple of weeks below 20F 
>to achieve that.  Maybe they freeze over in the first two weeks, maybe the 
>last two weeks, but they always freeze over. If the temperature was really 
>30F all the time they would never freeze over. EVERY January we get at 
>least one day dipping below 0F.   EVERY January we also get a few freak 60F 
>days and maybe a week of 45F to raise the average to an unrealistic number.

NREL's 30 year record min and max are -18 and 76 F.

>Anyone who dresses for 30F every day is going to freeze their ass off, on 
>average. Anyone who designs a house for 30F is also going to freeze their 
>ass off.

NOT if it stores enough heat for a few cold cloudy days.

> I am pretty tired of hearing about average weather being anywhere near 
> representative.  It isn't.

Averages are averages. Do you have a problem with that? :-)

>On average, year-round, the temperature here is 52F.  Should we design 
>houses for 52F outdoor temperatures?

We might, if we live in caves with long time constants.

>... Real heating systems must cope with peak loads, not averages.

Agreed. Whatever distributes the heat from the tank (eg a radiant floor or 
fan coil units) or rarely-used wee woodstoves should have the capacity to 
keep the house warm at ASHRAE's 99% 2 F or 97.5% 6 F St. Louis winter design 
temps. The ASHRAE Handbook of Fundamentals says "Two frequency levels are 
offered for each station representing temperatures that have been equaled or 
exceeded by 99% or 97.5% of the total hours in the months of December, 
January, and February (a total of 2160 hours)... In a normal winter there 
would be approximately 22 h at or below the 99% value and 54 h at or below 
the 97.5% value... Snelliing (1985) and Crow (1985) found that the duration 
in a single episode of extremely low temperatures below the 99% and 97.5% 
levels can continue for 3 to 5 days."

Amen.

Nick