Thermosyphoning freeze protection

[Nick Pine]

Andrew Swingler suggests thermosyphoning as one form of freeze protection
for evacuated tube collectors above a heat storage tank. This might be the
only form of freeze protection needed for freeze-tolerant heat-pipe tubes
below a pressurized header with some insulation on the header and its supply
and return pipes, and a pipe with minimal elbows and a circulation pump that
doesn't block flow when stopped.

Cool water weighs about 62.67 - 0.003T lb/ft^3, with T in degrees F. The
density difference caused by the temperature difference between up and down
pipes causes a pressure difference proportional to the height of the water
column, making water flow through the resistance of the pipe loop. With 16'
of height and a dT temperature difference, dP = 0.048dT lb/ft^2.

Bill Shurcliff says a pipe with radius r and length L in feet and pressure
diff dP has laminar flow Q = Pir^4dP/(8MuL) ft^3/s. Viscosity Mu is about
6x10^-7 lb-s/ft^2 for 32 F water. With, say, 32' of 1/2" pipe, 16' up and
down, Q = Pi(0.25/12)^4dP/(8x6x10^-7x32') = 0.004dP ft^3/s or 43dT lb/h,
which moves 43dT^2 Btu/h. If the heat pipe collector header is 6"x6"x8' long,
with 16 ft^2 of surface and R4 insulation and a 16/4 = 4 Btu/h-F conductance
from 60 F water temp to outdoor air at, say, -20 F, (60+20)4 = 320 Btu/h
makes dT = sqrt(320/43) = 2.7 F, if I did that right Water goes up at
60 F and returns at 57.3 F, well above 32 F. At -40, with about 400 Btu/h
of heat loss, the water might return at 60-sqrt(400/43) = 57.0 F.

Nick

 

[Fred B. McGalliard]

Nick. Isn't this a description of a convection driven water chiller?

 

[News]

Andrew Swingler suggests thermosyphoning as one form of freeze protection
for evacuated tube collectors above a heat storage tank.

In short it takes heat from the stored water in the tank below and emits
this to the night sky "every" night. Not very efficient at all.

If you have a roof mounted water solar panel one ant-freeze method is to
have two stage frost control. A pipe stat senses the water in the panel is
on 0C - the pumps runs, however it does not extract heat from the tank, so a
motorised valve will have to be in place. Moving water is less likely to
freeze. Another stat senses if the water is about to freeze, then a the
pumps cuts out and the panel water empties into an open vented tank in the
attic. Dependent on local climate of course.

 

[daestrom]

Andrew Swingler suggests thermosyphoning as one form of freeze protection
for evacuated tube collectors above a heat storage tank. This might be the
only form of freeze protection needed for freeze-tolerant heat-pipe tubes
below a pressurized header with some insulation on the header and its supply
and return pipes, and a pipe with minimal elbows and a circulation pump that
doesn't block flow when stopped.

Cool water weighs about 62.67 - 0.003T lb/ft^3, with T in degrees F. The
density difference caused by the temperature difference between up and down
pipes causes a pressure difference proportional to the height of the water
column, making water flow through the resistance of the pipe loop. With 16'
of height and a dT temperature difference, dP = 0.048dT lb/ft^2.

If freeze protection is the goal, keep in mind the density of water stops
increasing with dropping temperatures when you get to about 39 degF. From
there on down to 32, the density actually decreases (i.e. 34 degree water
'floats' above 39 degree water). If your water nears 39 degF, your
thermosyphon will stop, allowing the water in upper section to freeze.

daestrom

 

[Nick Pine]

If freeze protection is the goal, keep in mind the density of water stops
increasing with dropping temperatures when you get to about 39 degF. From
there on down to 32, the density actually decreases (i.e. 34 degree water
'floats' above 39 degree water). If your water nears 39 degF, your
thermosyphon will stop, allowing the water in upper section to freeze.

That might happen when 60-sqrt(Q/43) = 39, ie Q = 43x21^2 = 18,963 Btu/h
= (60-T)4, ie T = minus 4,686 F

Nick

 

[News]

News, Look up "Evacuated tube collectors".

Missed that.

 

[News]

Holy smokes!! This is beginning to look like a patchwork quilt not
the right direction here....

Pretty well standard way in commercial systems, which is coming over to
domestic. Look at the Honeywell web site.

You can do the calcs, but I would not trust them in cold climate. I would
have a mechanical/electrical backup to be sure. I have seen auto frost (wax
filled) valves that allow a panel to drain onto the roof, if it becomes
that cold. The best is an open vented system that drains back into a tank
when no useful heat is available. If useful heat at the collectors, the
pump takes the water out of the tank. Simple and the best in cold climates.

 

[Nick Pine]

...the evacuated tubes are almost always plumbed for ~3/4" tube...
This of course would aid the thermosyphon effect...

Sure, since the flow increases as r^4. We might also use the lower viscosity
at 68 vs 32 F, 3.4 vs 6x10^-7, and the more rapid change in density, 0.01 vs
0.003 lb/ft^3-F.

With 16' of height and a dT temp diff, dP becomes 0.16 vs 0.048dT lb/ft^2.

A 3/4"x32' pipe around 68 F might have Q = Pi(0.375/12)^4dP/(8x3.4x10^-7x32')
= 0.034dP ft^3/s or 1269dT lb/h, which moves 1269dT^2 Btu/h. If the heat pipe
collector header is 6"x6"x8' long, with 16 ft^2 of surface and R4 insulation
and a 16/4 = 4 Btu/h-F conductance from 68 F water temp to outdoor air at,
say, -20 F, (68+20)4 = 352 Btu/h makes dT = sqrt(352/1269) = 0.53 F. Water
goes up at 68 F and returns at 67.47 F. At -40, with 432 Btu/h of heat loss,
the water might return at 68-sqrt(432/1269) = 67.42 F.

...the idea here is ultimate simplicity and elegance leading to low cost,
added reliability and ultimate consumer acceptance.

Sounds great. Steve Baer says those freeze-valves are notoriously unreliable.

But I'd still like to see a differential thermostat to lower the nighttime
collector heat loss and the pump energy and make the pump last 20 vs 5 years.
Smart electronics can be very reliable and cheap in volume. To reduce sensor
wiring, we might run the pump for a moment once in a while after dawn to
sense the collector temp via the return pipe, or sense the tiny continuous
upgoing and downgoing thermosyphoning temp diff and run the pump when it's
close to zero.

"If freeze protection is the goal, keep in mind the density of water
stops increasing with dropping temperatures when you get to about 39 degF...

...This is why the system would be engineered to keep the temp above 45 deg
- and as nick points out - this may not be so difficult.

As long as there is a heat source below, eg an unfrozen house. It also needs
some bare surface to gather house heat. We might wrap some electric heater
tape around the upgoing pipe, with a thermostat, for very cold nights.

If the heat pipe collector header is 6"x6"x8' long, with 16 ft^2 of surface
and R4 insulation and a 16/4 = 4 Btu/h-F conductance...

R4? hmmmm. maybe conservative. maybe not. In anycase they could be
manufactured Rmore.... maybe R16 even?

Looks like R4 is plenty. A little foam or fiberglass inside the header
enclosure. No insulation might be OK, except in a very strong wind, but
we also want to avoid heat loss in normal times.

Nick

 

[daestrom]

Disclaimer: not trying to sell tubes.......just a discussion. I hate
these conflicts of interest too but what can you do?


Finally a frost plug may offer the last stage of freeze protection
(with ball valves at the tank on the send and return lines to stop the
water shooting off the roof - if worst comes to worst that is!)

I put this in a paper at http://www.swingsys.com/030227_DHWpaper.pdf
if you want to review and comment. Untested as of yet. Armchair theory
at best + nick's analysis..... no freezing where I live The system
seems to look cheap and robust at a first glance though.....

In your paper, the drawing shows the return from the heat-pipes going to the
domestic hot-water heater outlet (i.e. the penetration that connects to the
'top' of the water in the tank). While in the discussion under 'Basic
Circulation and Control', third paragraph, you explain that solar heat
returns, "...to the bottom of the tank via the internal tank fill pipe." I
think you want to say it returns to the top of the tank via the hot-water
outlet pipe. This would perhaps make the recovery time better as the top of
the tank would be heated first. And it would allow the pump suction to
remain cooler for as long as possible. Or maybe not, as I read it, you are
depending on the return going into the bottom of the tank to minimize losses
at night by only circulating water from the bottom of the tank. Well,
either way, the drawing and words don't match up so you might want to change
one of them.

Your ideas for over temperature immunity are novel. Many folks come up with
schemes to prevent boiling, whereas you allow boiling, just limit the amount
of liquid available and accomodate this. If we use Nick's dimensions for
the header of 6"x6"x8', we have 2 ft^3 of water at an initial temperature of
say 140 degF. That's about 123 lbm of water. If the 'mains' pressure is 50
psig, then the steam formed from boiling will be 123 lbm of vapor at 50
psig. That much vapor has a volume of 820 ft^3. Allowing for the intitial
space of the header, the expansion tank will have to accomodate the
remaining 818 ft^3. Of course, many homes fed from municipal water supplies
here in the US don't have any sort of check-valve between the street supply
and the hot-water heater so in those situations, you *could* let the steam
pressure simply force water back out to the 'mains'. For those with pumped
wells, I don't think this option would work as they have checkvalves. Such
a large expansion volume could be a problem, remember it must be kept
'empty' when not boiling to have room for this expansion. And I have no
idea what boiler/plumbing code has to say about this

daestrom



daestrom

 

[Nick Pine]

If your gibberish above is trying to tell us it happens at -4,686 degF, then
it is simply wrong and does not fit the well established experimental data.

Perhaps you are trying to say "I do not understand your calculation," or
"I have no short-term memory." See my posting of yesterday (at 4:53 EDT
on 7/11, to which you responded for an explanation of the gibberish.

Nick

 

[News]

Disclaimer: not trying to sell tubes.......just a discussion. I hate
these conflicts of interest too but what can you do?

Nick, thanks for presenting those calculations - they were a bit over
my head...no pun intended. A few notes: the evacuated tubes are almost
always plumbed for ~3/4" tube. (Metric 22mm to be exact). This of
course would aid the thermosyphon effect. In any case the 1/2" numbers
are still very attractive.

"In short it takes heat from the stored water in the tank below and
emits
this to the night sky "every" night. Not very efficient at all."

Evacuated tube heat pipes are very efective thermal diodes. The losses
are quite small - especially when compared to a glazed flat plate.

"A pipe stat senses the water in the panel is
on 0C - the pumps runs"

sure but added cost and complexity - the idea here is ultimate
simplicity and elegance leading to low cost, added reliability and
ultimate consumer acceptance.

"however it does not extract heat from the tank, so a
motorised valve will have to be in place."

motorized valve?! forget it!

"Moving water is less likely to
freeze. Another stat senses if the water is about to freeze, then a
the
pumps cuts out and the panel water empties into an open vented tank in
the
attic. Dependent on local climate of course."

Holy smokes!! This is beginning to look like a patchwork quilt not
the right direction here....

"If freeze protection is the goal, keep in mind the density of water
stops
increasing with dropping temperatures when you get to about 39 degF.
From
there on down to 32, the density actually decreases (i.e. 34 degree
water
'floats' above 39 degree water). If your water nears 39 degF, your
thermosyphon will stop, allowing the water in upper section to
freeze."

good point - i don't think this is well understood. This is why the
system would be engineered to keep the temp above 45 deg - and as nick
points out - this may not be so difficult.

Finally a frost plug may offer the last stage of freeze protection
(with ball valves at the tank on the send and return lines to stop the
water shooting off the roof - if worst comes to worst that is!)

I put this in a paper at http://www.swingsys.com/030227_DHWpaper.pdf
if you want to review and comment. Untested as of yet. Armchair theory
at best + nick's analysis..... no freezing where I live The system
seems to look cheap and robust at a first glance though.....

Looks simple enough. Let us know when you have tested it and the efficiency
too. Instead of using the bottom of the tank you could use a separate solar
tank feed the main tank, then certainly no mixing of the water that may be
cooling away.

This is similar to a Uk system where the collector is made of non plastic,
so the water freeze without any problems. The pump is run from a PV cell,
so is only running when there is adequate sun. It at times radiates heat
back to the sky. However this is deemed OK because of the simplicity. The
"overall" efficiency is reasonably high.

 

[daestrom]

Thinking about the Aug 2003 HP article, it might be a lot more.
And the tank may slowly destratify by diffusion over time,
especially with some thermosyphoning flow.

Well that will all depends on the exact connections. Andy was talking about
connecting the return line to the inlet so the internal dip-tube would carry
the water down to the bottom of the tank without mixing with the top water.
With a small amount of usage letting in cold (supply temperature) water at
the bottom and taking warm water off the top, I can see the bottom
temperature for the riser dropping very soon to the point where
thermo-siphoning would stop. Have to warn the homeowner to *not* use *any*
hotwater when the power goes out.

The header pipe has about 1.6 ft^2 of surface. With R4 insulation inside
vs outside the header box, RC = 1000 hours, so it might cool from 60 to
40 in 340 hours, when it's -10 F outdoors.


That seems very unlikely, given the number of things that have to go wrong
simultaneously. How does the tank get to 60 F?

You tell me, that was your number.

An average homeowner would
notice 60 F showers, so he needs to be on vacation without having drained
the pipes first, AND it has to be very cold outdoors for a long time, AND
the power fails or the backup water heating system fails or the circ pump
fails or its cloudy for several days or the house becomes unheated.

But if they were to actually *use* any hotwater when the power goes out, the
supply water will cool the bottom of the tank well below 60 before the
homeowner's shower was affected at all. A cloudy day or two and the lower
part of the tank would be quite cool just from supply water. But the
homeowner would get hot water from the upper, electric element. Short
showers or just sink usage would be at upper thermostat temperature. Yet
average tank temperature would be quite a bit lower. And the bottom of the
tank could be close to 40 due to cold supply temperature. Around here (NY)
our city water temp. is about 38 in the wintertime for a couple of months
straight.

So a cloudy day, take a short shower, then lose power to the circ water pump
that night. Doesn't sound very implausible at all. Typical weather,
typical usage, one failure. Unless you just don't shower on cloudy days
(another warning to the homeowner??

Another thought... Exactly what kind of flow can be induced through this
circ water pump when its off anyway? The thermo-siphon calcs assume its
similar to a length of pipe. If it were a centrifugal pump, that would be
reasonable since the casing isn't much of an impediment when its off. But
if it were a screw or rotory type, then it might be a serious impediment to
natural circulation.

With inside insulation, the header needs (40-(-10))0.4 = 20 Btu/h to avoid
freezing at -10 F. That might come from 32' of bare 3/4" indoor pipe with
6.3 ft^2 of surface and 9.4 Btu/h-F of conductance to T (F) house air, with
(T-40)9.4 = 20, so T = 42 F.

Frankly, I doubt that insulation inside the header vs. outside would make
that much difference (UA of 0.4 vs 4). But I suppose you have the numbers
on this. Such a small size, one can hardly use flat heat transfer equations
anymore, may have to shift to radial calculations since the inside surface
area is so much different than the outside area.

And what about conduction of the metal in the heat pipes? I understand the
fluid/vapor inside the heat pipes won't be circulating, but you would still
have a number of metal pipes sticking out the bottom of the header box. Do
they have some sort of 'thermal break' in the metal construction?

And if the pump were running at night when it failed, the riser and return
temperatures would be nearly identical. Assuming the force flow is ten
times the expected natural circulation flow, the temperature difference
would only be 1/10 of the NC delta T or 0.27 degF. How long would it take
to build up natural circulation (i.e. a delta T)? When there is no flow,
the water in the riser and return both cool/warm the same amount so it won't
develop a difference in temperature very fast. Natural circulation systems
typically have the two legs of piping offset vertically to start the flow.
Putting the inlet and outlet on the same elevation may leave the system
stalled for some time. If the return line elevation includes the tank's
internal dip tube (which is heated warmer than the riser outside tank over
the same elevation), then it's conceivable the water in the dip-tube will
actually try to rise and seat the check-valve. This would effectively stop
NC flow from starting at all.

Given typical mid-west or new england weather, IMHO I think the system would
freeze up the first circ-water pump stops out for 8 hours or more at night.
I believe freeze protection is more complicated than this because this
design is flawed (as far as freeze protection is concerned).

daestrom

 

[daestrom]

the idea is that the energy picked up by the internal piping 'losses'
will offset the external header losses. The BTUs required to feed the
thermosyphon cycle are picked up from the internal envelope of the
building. This also presumeably where Nick gets his 68F from. As long
as the building is warm the pipes in the building won't freeze...type
of thing.

So the riser piping is bare and the return is insulated? Otherwise the
riser and return would both pick up the same amount of heat. In fact with
return cooler than riser, it would pick up more heat and reduce the delta T
for natural circulation??

When the water is warmer than the house, this will radiate heat to the house
and only absorb heat from the house when the hot-water is cooling off?

daestrom

 

[Nick Pine]

The circulation can happen in either direction, and thermally equal legs
seem to work more reliably than unequal ones, according to Steve Baer's
experience, with no significant metastability problems or delays. There's
always enough "noise" to start a flow going, electrically-speaking.

A stragetic amount of the riser would need to be exposed to the house
heat source and allow an appropriate BTU input...

We might use a cold water tempering tank (eg a well pressure tank or
4"x10' of PVC pipe) or a heat tape with a thermostat.

...This probably only needs to be conservativly 100BTU or so...

Btu/h. Sigh.

Nick

 

[daestrom]

way he calculated the collector area. The insulated return - albiet
so slightly cooler - would be "more insulated". The effect could
possibly be acheived by simply reducing the insulation thickness for a
portion of the riser. If at all. The system may well thermosyphon
through each tube alone - or somehow set up its own circulation due to
the rotation of the earth ;-). I know in my place the 3/4" vertical
hot water pipe outlet is always hot hot hot on the second floor above
due to thermosyphoning within the pipe itself.

While internal circulation within the riser and return may well happen, that
wouldn't create any flow through the header box. To get circulation through
the header you need to somehow keep the riser warmer than the return. The
storage tank will do this for a while unless some cold supply water is
admitted to replace warm water drawn off by the homeowner. Stripping
insulation from the riser and leaving insulation on the return could also
work if enough of the vertical run is in the warm house (and the house stays
warm ;-). But warming the water halfway up the riser is not as effective
because it has less elevation to 'work' through.

A more accurate calculation for the density of water in this range is
((-8.5010e-5*T+4.6829e-3)*T+62.384) where T is in degrees F and the results
are in lbm/ft^3. This shows that with warmer water (say 80 - 100 degF) the
change per degree is more so you can get more driving head for a given delta
T. (this is what I used in these calculation).

If the return goes down the dip tube inside the storage tank, say a distance
of 5 feet, and the top of the tank is appreciably warmer than the riser (say
100 degrees while the riser is 80 for an average dip-tube temp of 90), then
this little section of pipe will create a dp of about 0.539035 lbf/ft^2,
but in the wrong direction. If the rest of the elevation above the tank is
a total of 16 feet, then the return leg will have to be about 76 degF, *just
to balance the dip tube*. Given Nick's calculations, it might take another
3 degrees to develop the desired flow. So the riser might have to be kept
about 7 degrees warmer than the return to keep up the flow. (or put another
way, the return would have to be 7 degrees cooler). And that has to be over
most of its length!!!

This would be a case for putting the return into the bottom of the tank
through a fitting at the bottom of the tank. This way the return will *not*
be warmed that last five feet and you won't lose that driving head. With
the dip tube, even though it ends at the bottom of the tank, it is warmed
that last five feet or so and that hurts the overall natural circulation.
'Course, this all depends on how much you're going to need that driving head


The same calculation of the dip tube issue using 70 at the top of the tank
and 50 at the bottom shows a disaster. The dp caused by an average dip tube
temperature of 60 and riser of 50 is 0.23657 lbf/ft^2. Sound better right??
But, even with the return cooled down to 40.44 and the riser at 50, no
natural circulation would occur (16 feet of elevation with 50 and 40.44 degF
gives a delta P of only 0.2237 lbf/ft^2). Thermosyphoning would try and
reverse, but your check-valve would stop all flow.

If the tank stratifies a lot more than 20 degrees, lets see, maybe 120 on
the top and 60 on the bottom?? The dp caused by a 5 ft dip tube with
average 90 degrees and riser of 60 degrees is 1.4581 lbf/ft^2. And again,
with the riser at 60, no natural circulation can occur. With the top of the
tank at 120, the most stratification you can have and just balance the dp
would be about 54 degrees from top to bottom (120 on top and 65 on bottom).

Like I said before, warn the homeowner *not* to use any water when the power
is out. Cold supply will cool the bottom of the tank quickly and stall the
thermosyphon.

As you may guess, in a former career, I had a lot of experience with natural
circulation systems Sorry for rambling on, but when I get a problem
laid out in Excel, I can't help but tweak all the variables looking for what
makes a difference

Turns out the dip tube is one of those things you wouldn't have thought of,
but has an impact. Might be better off returning into a bottom fitting.
This would add another 5 feet or so of elevation as well.

How many 'heat-pipes' are you thinking about brazing onto the header? As
far as conduction of heat out passed the header's insulation are they like a
hollow metal pipe? What diameter & wall thickness? Might be interesting to
see how much heat might be conducted away from the header at night through
the heat pipe's metal.

daestrom

 

[Nick Pine]

Seems unlikely, without a return path through another pipe.

Then again...

While internal circulation within the riser and return may well happen, that
wouldn't create any flow through the header box.

I could imagine a little flow. If thermosyphoning can occur within
a single pipe, why not within the header pipe as well? But that seems
like a matter of hope, vs a more pro-active and predictable design.

To get circulation through the header you need to somehow keep the riser
warmer than the return.

Or vice-versa, assuming no check valve (as in the Aug 2003 HP system.)

The storage tank will do this for a while unless some cold supply water is
admitted to replace warm water drawn off by the homeowner

Not exactly something to count on either.

Stripping insulation from the riser and leaving insulation on the return
could also work if enough of the vertical run is in the warm house (and
the house stays warm ;-)

This might work better with both pipes bare.

But warming the water halfway up the riser is not as effective because
it has less elevation to 'work' through.
Agreed...

Like I said before, warn the homeowner *not* to use any water when the power
is out. Cold supply will cool the bottom of the tank quickly and stall the
thermosyphon.

I agree that it doesn't seem like a good idea to depend on such warnings,
but a cold tank might be OK, if the supply and return pipes are warmer...

Another freezing scenario: it's cold outdoors, and the kids have exhausted
the solar hot water, leaving 38 F water in the tank bottom, and then the
power fails. If the house heating fails at the same time, we drain the pipes
and go somewhere else, eg Florida. But suppose the house is still warm, and
we are all sitting around the woodstove with candles... We might avoid
collector freezing, if the pipes inside the house have enough conductance
to warm house air, to a) supply the header heat loss to the outdoors and b)
warm the water that flows out of the tank to a thermosyphoning temperature.

Roughly speaking, suppose the pipes have a conductance of G Btu/h-F to
70 F house air and the pipe water is about 50 F and the header has 1 Btu/h-F
of conductance to -10 F outdoor air. Then the house air has to supply
(50-(-10))1 = 60 Btu/h to the header. Suppose the flow in the pipe loop
Q = 1000dT lb/h, which supplies 1000dT^2 Btu/h of heat to the header.
Then dT = sqrt(60/1000) = 0.245 F, which makes Q = 245 lb/h, which means
we have to heat 245 lb/h from 38 to 50 F. (70-50)G = 60+(50-38)245 = 3000
makes G = 150 Btu/h-F, eg 40' of fin tube pipe Not too practical,
altho this might work with slightly cooler pipe water. Then again, we'd
also like it to work with a slightly cooler house.

Most of that heat (98%) is needed to warm the cold water that comes out
of the tank. Without that requirement, G = 3 Btu/h-F, eg 10' of bare pipe
and 0.5 gpm of flow. Is there a way to add some sort of bypass pipe between
the supply and return pipes to complete the loop at the bottom and avoid
having to heat the cold tank water? It might tend to short out the pumped
solar loop in normal times. Perhaps this could work with a normally-open
check valve that closes when the pump runs but allows thermosyphoning
through the bypass pipe with the pump off.

Nick

 

[daestrom]

I could imagine a little flow. If thermosyphoning can occur within
a single pipe, why not within the header pipe as well?

Because from what I can tell of his design, the header is horizontal, not
vertical.

Or vice-versa, assuming no check valve (as in the Aug 2003 HP system.)

But his paper shows a check valve in the return line as it taps into the
incoming supply right before the dip-tube.

Not exactly something to count on either.

My point exactly.

This might work better with both pipes bare.

Both pipes bare would cause them both to pick up heat from the warmer house.
Since the return line is expected to be colder, it would possibly pick up
*more* heat. And warming the return line is against natural circulation.

I agree that it doesn't seem like a good idea to depend on such warnings,
but a cold tank might be OK, if the supply and return pipes are warmer...

Another freezing scenario: it's cold outdoors, and the kids have exhausted
the solar hot water, leaving 38 F water in the tank bottom, and then the
power fails. If the house heating fails at the same time, we drain the pipes
and go somewhere else, eg Florida. But suppose the house is still warm, and
we are all sitting around the woodstove with candles... We might avoid
collector freezing, if the pipes inside the house have enough conductance
to warm house air, to a) supply the header heat loss to the outdoors and b)
warm the water that flows out of the tank to a thermosyphoning

temperature.

Provided you only warm the riser and not both, yes. But as we get into
these situations, I think we're straying from one of his goals of keeping
things simple and low maintenance. Have to give the homeowner a list of
warnings and instructions of how to handle variety of circumstances. You or
I wouldn't mind, but John Q. Public might.

Roughly speaking, suppose the pipes have a conductance of G Btu/h-F to
70 F house air and the pipe water is about 50 F and the header has 1 Btu/h-F
of conductance to -10 F outdoor air. Then the house air has to supply
(50-(-10))1 = 60 Btu/h to the header. Suppose the flow in the pipe loop
Q = 1000dT lb/h, which supplies 1000dT^2 Btu/h of heat to the header.
Then dT = sqrt(60/1000) = 0.245 F, which makes Q = 245 lb/h, which means
we have to heat 245 lb/h from 38 to 50 F. (70-50)G = 60+(50-38)245 = 3000
makes G = 150 Btu/h-F, eg 40' of fin tube pipe Not too practical,
altho this might work with slightly cooler pipe water. Then again, we'd
also like it to work with a slightly cooler house.

Most of that heat (98%) is needed to warm the cold water that comes out
of the tank. Without that requirement, G = 3 Btu/h-F, eg 10' of bare pipe
and 0.5 gpm of flow. Is there a way to add some sort of bypass pipe between
the supply and return pipes to complete the loop at the bottom and avoid
having to heat the cold tank water?

Well, he did mention using a dip tube to promote stratification in the tank.
I suggested actually having the return enter the tank at the bottom (to
avoid dip-tube warming when the tank is stratified). One penetration on
each side at the base of the tank might do it (one goes-in and one
goes-out). Then we only have to warm the water in the pipe and a small
layer at the base of the tank. Interestingly, with these numbers, the
return water will actually be warmer than the tank for a while.

And keep in mind, the Q=1000dTlb/h assumed all 16 feet of vertical run was
at the indicated temperatures. If the return is kept insulated, that's not
too much of a problem there, but if the riser is actually being heated along
its length, not all its length will be warm and the resulting driving
head/flow will be less.

And we don't have to heat the riser all the way from 38 to 50 all at once.
If the flow is very low (due to similar densities of riser & return), then
we just have to heat the riser faster than the return (remove insulation
from only riser). As the temperature goes above 40.44, it will slowly
buildup a dP and initiate flow. And said flow will draw in more 38 degree
water from tank/return. This denser water will tend to reduce/stall flow.
There will be a dynamic equilibrium where flow only increases as fast as the
riser heating can support warming the incoming water. As the incoming water
warms, the flow will increase.

The only problem with all this is, "How long will it take to get enough flow
established to prevent freezing?" The more bare pipe (or fins); the lower
in the overall elevation where the major heating takes place; and the colder
the return leg is kept, the faster flow will establish. But of course, as
the house cools (if it cools at all), the less flow.

It might tend to short out the pumped
solar loop in normal times.

Well this is a problem with using the dip-tube return as well. It requires
the water entering the tank from the dip tube to somehow circulate with the
upper regions. Of course, being warmer, it would tend to 'turn-over' the
tank volume and that may be enough. And with forced circulation one might
have a sort of header to help distribute the flow that wouldn't be as
effective with the lower flows of natural circulation. Sort of a compromise
between *not* stratifiying the tank with forced flow, but letting it
stratify under emergency, natural circulation conditions.

Perhaps this could work with a normally-open
check valve that closes when the pump runs but allows thermosyphoning
through the bypass pipe with the pump off.

Now that's an interesting idea, maybe a lift-check installed upside down so
it hangs open when the pump is off.

daestrom

 

[Nick Pine]

The bypass pipe might connect the collector supply and return lines just
above the tank in the basement, with your hanging check valve that only
allows flow from return to supply when the pump is running but allows slow
flow in either direction when the pump is off. To collect more house heat,
the supply and return lines might both be bare inside the house.

Steve Baer says thermosyphoning systems with thermally-equal legs are more
reliable than those with one biased leg, since they can easily flow in either
direction. A designer who decides to help water flow from A to B may design
a stuck system, if it wants to flow from B to A.

Nick