Low head loss hydronic floor heating
 

We have an excellent thread on DIY hydronic floor heating here. Admittedly I haven't read all of it, but there is tons of good information in there.

One topic I haven't seen come up in the discussion much is minimizing head loss in a hydronic system. I'm curious why low head loss hydronic heating isn't a larger topic. I realize that hydronics requires much less energy to move its heat around vs a conventional gas furnace, but it seems like it can be further optimized. I see many images of people heating a room with a single loop of 1/2" pex as illustrated below:

http://ecorenovator.org/forum/attach...1&d=1349800423


However, if one were to route the plumbing in many parallel lines, the head loss would be significantly reduced.

http://ecorenovator.org/forum/attach...1&d=1349800586



In a properly designed system, the two hydronic floors would both recieve the same flow rate, same thermal transfer and heating abililty. However, the second floor would require a substancially smaller pump to achieve the same flow rate. This means less electricity consumed and more efficient operation.

I think that the only downside here would be all the tees and connections you'd need to make within the floor.


Lets take a look at an example to see what kind of gains could be made. In this example, we'll look at putting hydronic heating into a 10'x10' room.

With a single run of 1/2" pex with 6" spacing, this would require about 170 feet of pex. At a flow rate of 1 gpm (total guess, sorry, I haven't done any calculations for hydronic floor flow rates yet, it might be too high or too low), we'll have a pressure head of 6.6 inches of water.

With a parallel setup of 1/2" pex with 6" spacing, this would also require 170 feet of pex. However, now we have a 1/2" headers on each end of the room and 18 parallel run lines of 1/2" pex. We would like to keep that 1 gpm flow rate. This means we have 1 gpm through the headers, and .055 gpm through each parallel line. The pressure head of this system turns out to be .39 inches of water for the 9 or so feet of header tubing. Pressure head for the parallel lines is a bit more tricky because the flow rate is so low (1gpm / 18 lines = .055 gpm). I haven't been able to find any head loss charts that go anywhere near this low. However, I took the 1 gpm rating and divided it by 18 to get a rough idea (actual head loss will be lower) which gives us .23 for the parallel lines. You then add these together for a whopping .62 inches of water.

.62 inches of water vs 6.6 inches of water to do the exact same thing to the exact same room. Add in a bunch of tees and crimp/cinch clamps.

Besides having trouble finding a pump small enough to give us the required flow rate at this rediculously low pressure head, does anyone else see any major problems? Ideas and thoughts are quite welcome.

All those tee fittings will increase the required head significantly through friction losses. I don't have a Crane or Cameron book that includes PEX but I would use 1.38 or so for the K value of each tee fitting. You also run the risk of short-circuiting your heat transfer if for some reason there is a flow obstruction or something to absorb more heat on one side of the room than the other (such as a rug covering the middle).

Welcome to the site mephistopheles!

Excellent point. The tees will definitely cause a restriction. I had not thought of that. I would imagine we'd still be well under the 6.62 inches of water of the initial setup. I'll try to find a crane catalog and see what all the tees would introduce. However, we could fairly easily fix this by enlarging the header pipes if it were an issue. You would just have to use reducing tees.

Flow obstruction IMO is a non issue. If you have something clogging your pipe in the standard version you simply have no heat/flow at all. At least with the parallel setup you would have some heat. But, in either case you would still want to flush the system and try to get it out.

Another alternative would be to reduce the number of parallel lines. Instead of 18 parallel lines perhaps you reduce that down to 9 or 6 and just make some small loops of those to make up for it.

A quick google search came up with this chart:

from this PDF

http://ecorenovator.org/forum/attach...1&d=1349813000


Lets use the above parallel example of 18 parallel lines and using 1/2" pex for everything. It is going to have to flow through the tee as a run for 7 fittings, and as a branch for 2 fittings. That gives us an effective additional length of:

16 * 2.2 = 35.2ft
2 * 10.4 = 20.8ft

Or a total of 56 extra feet.

56 extra feet at 1 gpm gives us an additional pressure head of 2.3 inches of water. That would bring the total pressure head up to just about 2.9 inches of water even. Still quite a bit less than the 6.6 of the standard setup.

The way I would calculate that would be 34 tee fittings and 2 elbows to get 18 parallel pipes between two headers of equal size. That's 34*2.2+34*10.4+9.4*2. That's 447.2 additional feet. If we assume laminar flow (since we're at speculating a low flow rate) we could half that and assume that layout will cost us about the same as pumping 223 feet of straight pipe. Fittings add up in a hurry!

I was definitely wrong with my initial calculations! I have since revised them. I've come up with an additional 2.3 inches of water for a total of about 2.9 inches of water.

I must say I do not agree with your method of calculation. You're counting the fittings as if they're all in series, and you're even counting them twice. A single molecule of water passing through the system is not going to flow through every single fitting. It should only flow through 18 fittings, 16 straight and 2 elbow. That is why you see a benefit from running parallel pumbing lines.

All connector restrictions in your diagram will be elevated by the many new paths the water has to flow threw, there is also more pipe on the floor in this design so it would increase the head, the water weight would be more of a concern then the flow restriction as the flow rate and resistance will be lowered threw each pipe. I think.

this is the same system as ice rinks. there is usually a shut-off valve at each return location, usually every year or so at start-up most returns are shut off to keep all pipes pressurized and a main header drain is opened or the air eliminator in the system will remove air which becomes lodged in one of the pipes. Usually most hydronic system will have some air migrate into the pipes through the non-heating season. At the initial start-up the (sounds like a river in my pipes) air will be pushed into the expansion tank. The more parallel runs, the more chance of cold spots.

If you had an indirect return I would agree with your calculations. With a direct return flow, however, you will have short-circuiting of the system and all the flow will go through the shortest branch; to keep that from happening you would need to assume the system will operate as a series flow system.

http://ecorenovator.org/forum/attach...1&d=1349875140

You will develop cold spots overtime in parallel flow systems--of that there is no (cheap/easy) cure!

Why are assuming that the branches aren't all the same length? I would think that it wouldn't be very difficult at all to design the system to have equal length parallel lines.

Why do you say that you'll develop cold spots over time?

I could believe the freshly heated hot water would head down the first sections available,as the flow is so slow it could or would find the path of least resistance.
I think the coiled line is superior now, after some thought.
I want too make this system as well, as i have leaned,here, it is the best method for heating.

Yes, it will flow down the path of least resistance. The path of least resistance is a balanced flow between the parallel lines. Each line has the same pressure loss across it if analzed individually, so you should have equal flow amongst them all.

I have a friend with a system like this in his warehouse space. 14 parallel loops starting and returning to manifolds on a common utillity wall. All loops start and end with valves. I can only guess that these valves allow some flow ballancing/adjustment. The manifolds are 2" copper feeding 3/4 pex. I have no idea if it adds up to a low head overall.

The fluid will always choose the path of least resistance, thus short circuiting the sections that are further away. The only way to "balance" this would be the addition of throttling valves in each "loop". In practice and real world, more than just a few loops are practically impossible to balance.

Yes, it will choose the path of least resistance. Once it starts going down one tube lengh it will start seeing increased resistance as is normal of any liquid flowing in any tube, thus it'll start creating a pressure head and that pressure head will push liquid down another leg, and so on and so forth.

What this does is mimic exactly what a solar panel does. They use the exact same pricipal except in this case I'm shedding heat instead of collecting heat. They do not have issues with flow balancing, they don't have issues with high pressure head or they would be made differently.

The sketches show a parallel system with supply on one side and return from the opposite side. The flow path from supply through any one branch to return is exactly equal to the flow path through any other branch. The system will self balance, even if the flow velocity through the branches (I'm too lazy to do the calculation) is laminar.

I expect possible trouble from two sources.

The water velocity may be too low to push trapped air out of the branches. Then you will have unbalanced flow.

The pressure drop will be extremely low in the branches. If there is any elevation differences, temperature / gravity effects need to be considered.

Thanks JRMichler. I hadn't thought about air bubbles. That is a good point. You would just have to make sure you can achieve the necessary velocity to get rid of air in the lines. This could hopefully be done by shutting down other loops in a home run type setup, but it would have to be calculated.

I re-ran my calculations for the same floor, but using 6 parallel lines (similar to the 5 parallel lines shown in the image below) since the TEE fitting pressure head was so high in the previous example.

Looking back, the tubing pressure head added up to a measly .62 inches of water, and the pressure head from the TEEs was 2.3 inches of water (but, I believe it was calculated incorrectly). Anyway, it makes sense to reduce the number of tees. This will reduce the pressure head the TEEs create, but it will also increase the pressure head from the tubing. So, this is an exercise in finding the sweet spot between too many tees, and not enough parallel plumbing runs.

I'll run through the calculation step by step.

This is the planned plumbing layout (roughly). Each parallel run is the same length and therefore will have equal flow.

http://ecorenovator.org/forum/attach...1&d=1350407799


Alright, lets assume we have the 1gpm flow rate coming into the flooring. In reality, on the inlet side, as the water travels down the header pipe more and more of the flow is being diverted down the parallel lines, so this isn't 100% true. On the outlet side, it is the exact opposite, you won't get an actual full 1 gpm until the last parallel line has teed into the header. So, we're going to use a little more than half the tubing length as a guess. Each header pipe is roughly 9 feet long, so we'll just say 10 feet for our calculations to make it easier.

Using this spec sheet: http://huduser.org/portal/publicatio...sign_guide.pdf

We can see that at a flow rate of 1gpm through 10 feet of 1/2" PEX tubing is going to create a pressure head of .17 psi. To convert psi to inches of water you multiply by 2.3, so we have .39 inches of water for the header tubing.

At this point, our flow diverges into 6 parallel lines that are about 28.5 ft long. Each line will see a flow rate of .167 gpm. Sadly, our spec sheet doesn't go quite this low, so we'll have to use the .2 gpm pressure head. The pressure head for 28.5 feet of 1/2" pex at .2 gpm is .065 inches of water.

Now, we still have to add the pressure loss from the TEEs that need to be put into the system. Last time this is where a sizable amount of the pressure head came from. This time we have reduced the number of TEEs from 36 down to 12. The water flowing through the system will flow straight through a TEE 4 time, and flow through the TEE as an elbow 2 times. To get the pressure head these TEEs create, we must use the chart below:

http://ecorenovator.org/forum/attach...1&d=1349813000


So, we have 4 'run tees' for a total of 4 X 2.2ft = 8.8 ft of additional tubing pressure head. The flow rate these fittings will be seeing is exactly the same as the header pipes will see because as the water flows down the header pipes more and more is being routed down each parallel branch. So, we can calculate it at a little more than half the 1gpm flow rate. We'll use .6 gpm for these. According to our spec sheet above, that gives us .14 inches of water pressure head.

We also have 2 'branch tees' for 2 X 10.4 = 10.8 ft of additional tubing. Now, these branch tees are only going to see the amount of flow going down each parallel line, so the flow rate is .167 gpm. So, we have 21.6 ft at .167 gpm which gives us a pressure head of .049 inches of water.

So, now we know all our pressure heads and can add them together.
header tubing: .39
parallel tubing: .065
run tees: .14
branch tees: .049
Total: .644 inches of water

In the previous calculations, I didn't break apart the tee fitting calculations, and that made for some pretty huge differences in calculation. I believe I have it correct now. This is 1/10th the power it takes to pump the same flow rate of water through a single loop of 170 feet of 1/2" pex tubing. You should get the same exact heat transfer.

The only disadvantages I see are:
1) Tee fittings are a possible point of leakage. You could possibly put them below the floor so you could fix them if need be.
2) Tee fittings add additional cost to the project.
3) You can't configure the parallel tubing runs as well as you can with a single line. Normally, the inlet and therefore hottest tubing is put along the outside wall which helps keep the room slightly more comfortable.
4) Purging air from the system may prove more difficult.

The advantages are fairly obvious:
1) Drastically reduce pumping power required which means lower energy consumption and a much smaller pump required, both of which save money.


When I get some time I'll probably run a cost analysis on this setup and compare the two.

It sure looks like false economy to me. By having to move everything to below the floor(to be able to fix leaks), it won't be as efficient, heat wise. Drastically reducing pumping power will make it harder to purge. The flow loss through a tubing bend is less than using tees. Don't forget too that if your using water, you'll always have air being introduced into the system either through make up water or just from the water it self. Calcs are fine but sometimes real world works differently for some reason.

Thankfully its not false economy. I've been talking an engineer I work with (who has done hydronic heating installs) and pressure head increases with the square of flow velocity. So, reducing flow velocity reduces pressure head very quickly.

The reason I am posting this is because I don't have real world experience with hydronic floor heating. On paper this sounds great. Talking to the few people I know who have done hydronic floor heating they say its totally possible but don't like it for the reasons I've listed (mainly the connections in the floor). I'm asking if anyone has any experience or ideas that says this is a bad idea because so and so. I have yet to really see any reason saying that this is a horrible idea because of whatever reason.

I don't think having connections in the floor is a huge issue especially if you pressure test the system before putting the flooring in. I'm on a well system so I could just hook it up to my well and crank the pressure up. Or, maybe you just pay a bit more attention to the connections you make and make sure your crimp tool is properly calibrated. I'll have to do some reading on reasons pex connections fail to help reduce the chance of failure.

With your 1/10 the pump power calculation does it allow the use of a smaller pump then the 135w one ?

Since your looking for possible reasons why you shouldn't do it.

If you are worried about the T fittings failing enough that you would consider exposing them for a projected repair maybe you shouldn't do it.
I think durability should be of the up-most concern, how much is the gain are you expecting, if you cut up and T fit the pipes ? If its not substantial then for reliability reasons i would go with a one piece line or as near to as possible.

The 135W pump is for my solar hot water panel water loop. The solar loop will put heat into a water tank in the basement (which isn't made yet). This loop I'm talking about in this thread would be a seperate loop that will pull heat out of that tank to heat the house.

I wouldn't say I'm worried about anything at this point. Tons of plumbers make PEX connections inside walls every day and don't worry about them leaking. Mains water supply runs at about double to triple the pressure versus a closed loop hydronic setup, so that also minimizes risk.

I'm confident at this point as long as its pressure tested before the flooring goes over it, that it should last many many years. I probably wouldn't be as confident if I were pouring concrete over it, but I am not using concrete. I could further reduce risk by again reducing the amount of TEEs in the system. I would like to run one some more calculations with just two or three parallel runs to see what the differences are.

Anyway, I still have a lot of calculations left to do. This shows the pressure head for a single 100 square foot room. My house is roughly 1600 square feet. I'll have to design the system both ways and see what the difference is, or find some way to estimate it accurately.

I've been researching the idea that purging the parallel system would be troublesome because you need enough flow velocity in each parallel branch to move air pockets. This appears to not really be that large of an issue. The typical flow velocity needed to move air is a recommended 2 ft per second. This is the velocity required for inclined piping with water flowing downard trying to push the air down. For horizontal lines that are in a floor, the required flow velocity is actually much less.

According to this paper, the flow velocity for 3/4" HDPE pipe (it is a paper on ground source heat pumps so that is as small of pipe as they tested), the required flow velocity for horizontal piping is only .9 feet per second. I've copied the chart from the paper below, and you can see as you reduce the pipe diameter, the required flow velocity is reduced. For 1/2" piping I'd imagine I only need around .6-.7 feet per second.

http://ecorenovator.org/forum/attach...1&d=1350482474

Another interesting aspect here is that 50-60 degrees is the actual worst angle for air purging according to the paper.

Anyway, to achieve .7 feet per second in 1/2" pex tubing, you need a flow rate of .4 gpm. Under normal conditions we're flowing .167 gpm. So, we need to boost the flow rate by about 2.5X to purge the air. This doesn't seem like it should be all that hard to do IMO. If you have multiple zones on a home run type system, you should be able to boost any one loop by 2.5X the flow rate.

I've still been thinking about this. Another downside to having multiple parallel loops in one room is a reduction in the easy of dealing with obstructions and/or oddly shaped rooms. Careful thought must be put into the parallel loop length if you have anything the tubing must be routed around.

I had a concern which you already answered....reduced flow rate in the parallel scheme. So, my next wonderment is....is the heat shed from the parallel loop the same as one continuous loop? IOW, is the return temp lower? I ask since I have no idea at the moment.

I believe it is. The flow rate is still the same, and the surface area of the pipe is the same. Flowing X amount of water over Y amount of surface area should net the same heat transfer.

Couldn't you just run five or six parallel loops in the floor, run them all to the garage or basement or wherever your manifold is, and join them all there before connecting them to the manifold? That would give you similar flow characteristics, and all your couplings would be easily accessible and fixable.

Also, as far as pump size is concerned, I don't know how small a pump you want to use, but I'm running this one:

0015-MSF2-IFC - Taco 0015-MSF2-IFC - 00R 3-Speed Cast Iron Circulator - Integral Flow Check, 1/20 HP

It's completely silent and runs at speed 1 (78W). I've got over 1000 ft of tubing with the longest loop being 250ft, and the pump has no problem achieving 1gpm flow, which is plenty for my purposes. The pump shouldn't need to run more than 15 minutes out of an hour, good insulation provided, which would put it roundabout 460Wh a day.

Yeah, that would work too. Its a bit more tubing to run, but safer from leak problems.

This might present a problem if you're using a zone control valve though. You'd have to tie them together before the valve somehow. Not a big deal really.

I haven't even come close to choosing a pump yet. However, I'm really thinking that I'd like to go with an ECM pump due to increased efficiency and variable speed. This will probably end up being a Grundfos Alpha or Wilo ECORFC.

I have another thread about ECM pumps here:
http://ecorenovator.org/forum/applia...cm-motors.html

Great thread. The system I installed under the basement floor uses 3 parallel circuit with a single pump, I think it is a Taco 9, uses 135 watts.

See graphic of the system below. Works well, I used about 450' of Pex tubing and about 70 ' of copper. Only used 45 degree fittings to limit pressure drop.

Great to hear I'm not the only one thinking along these lines. How is the floor working out for you so far?

I used to in stall hydronic floor systems with my father years ago.... this was the preferred layout due to the way it averages out heat distribution.

Additionally we would decrease the spacing between parallel runs near exterior walls or other heat sinks to improve comfort

My system works well, I was concerned with the air entrapment issue using a parallel arrangment. By installing a common manifold, I was able valve off one or more circuits to purge the air, as well as balance flows for even temperatures. See pics below of the Manifold I used.

There are actually 3 circuits, not 2 as shown in the graphic.

I'm bringing this thread back. I'm pretty close to starting to work on the office hydronic floor. I am going to go with a few parallel lines in it. I don't know quite how many yet. Somewhere between 3 and 6 is most likely. I'm also still trying to decide between 1/2" and 3/8" pex for it.

Remind us how many ft2 the floor will be? When we design these for boilers we are using trying to get a 20F dT and this usually meant we needed about 250' of 1/2' tube per loop. Head loss is quite reasonable and I recently did an 8 loop floor with a Alpha pump and the display goes back and forth from about 3gpm to a power consumption, when it settled somewhat down, of 12-15w. It is hard to get much better as this was the whole house.

This specific room is about 14x12, so 168 sqft. The whole house is about 1600. I'm also planning on using 3/8" tubing for the floor versus 1/2" due to space constraints.

The flow through each of the parallel circuits (each of the legs) will only be the same, if the diameter of the supply conduit is slightly larger than any of the parallel legs.

Here is why.

Imagine a flow circuit with the parallel "legs" (all the same diameter) being vertical and a supply manifold "header" is at the top of each and a return manifold is on the bottom. The parallel circuit is fed water from the upper left.

If the manifold and each of the legs are the same diameter, then the parallel legs closest to the inlet get the most water flow (heat). The further you are from that inlet, then the lesser the flow. This is because there is a pressure drop along that manifold.

The reason for this is that there is a finite resistance in the manifold "header" tubing. But if you put it in slightly larger in diameter, then the pressure drop along the length is negligible compared to any one parallel resistance.

In physiology, this is why supply arteries (aorta for example) are always larger than the downstream arteries (femoral, iliac, renal, etc). The downstream arteries are all in parallel. Such a system keep the pressure at each downstream artery virtually the same and thereby flow through any one organ system is simply regulated by the resistance of the arterioles (smallest arteries) in that organ. This is an example of a constant pressure system with flow being regulated simply by resistance and being directly proportional to changes in pipe diameter.

If the supply manifold (aka "header") is the same size as the parallel circuits, increasing pressure in the system somewhat negates the pressure drop along the manifold pipe. But in general, radiant floors are low pressure systems.

Flow always goes into the parallel circuit with the lowest resistance (electricity or water). In the above parallel system, there will be flow in each of the parallel legs, but the flow will be highest in the legs that are closest to the inlet supply.

By increasing the manifold diameter slightly, you decrease the resistance in the manifold markedly. Only a 11% increase in diameter will decrease resistance in half (double the flow). This is because flow is proportional to radius to the forth power.

Having a larger manifold diameter also allows you to have parallel circuits of differing lengths have similar (not identical) flows.

You can model this as either a constant flow inlet - or constant pressure inlet system, but the results will be the same.


Steve

Just read this discussion.
What I didn't read in this discussion, are these two things:
1. The heat transfer from water to the pipe is depending on water velocity.
As soon as waterflow is so low, that the flow isn't turbulent anymore but it becomes a laminar flow, the heat transfer from water to the inner pipe wall is reduced considerably.
So by dividing the flow over 18 parallel pipes (as in the first example), the water velocity is only 1/18 th of the original value, so there's a good chance velocity is too low to keep a turbulent flow.
So heat transfer might be much lower than expected.
2. It seems nobody was taking Bernouilli's law into account.
Let's assume the first lay out with the 18 parallel pipes.
Assume the input is at the bottom right of the drawing and the output is top left.
Since a radiant floor heating is in one horizontal surface, there's no difference in height between the pipes.
Then according to Bernouilli's law: 1/2*rho*(velocity)^2 + Pstatic = constant.
This leads to the fact that (contrary to what many people think), the flow is not evenly distributed over the 18 pipes. The highest flow will be in the most left pipe en the lowest flow in the most right pipe.

My manifolds are always piped supply opposite to the return or "reverse return". Flow is partially defined by pump speed which means that the dT can be designed, to some extent. While turbulence does help in heat transfer, 50% or more will still occur with laminar flow. When the heat loss of a particular building gets close to passivHaus standards, it is very difficult to pump fast enough to have a high reynolds number and meet the design criteria for the system.

If we are using the heat pump as a heat source and you want to have a 10C dT over the loop (typical with a boiler), some the capacity of the buffer tank will be quickly overcome. The HP may have some trouble keeping up. The dT over the loop will probably have to be lowered which means shorter loop lengths.