Hot water is less dense than cold water, making it more buoyant. In this sense it will float on the cooler water. It will displace cooler water which then sinks—because it is denser. This can become an endless loop driven by the hot water in your water heater. All that is required for this loop to function is for the water heater to be below the fixtures that you want hot water at, and for there to be a bit of an “assist” to increase the loop’s performance. This assist consists of having the return pipe from the remote location being smaller than the supply pipe. So if you run a ¾ inch pipe to the remote location, you would then run a ½ inch pipe back to the water heater. The supply side comes off the top hot side of the tank and returns to the bottom of the tank to a “T” installed at the drain outlet of the tank.
Years ago I found some information about how the last 10 feet or so had to be uninsulated in order to make the loop function. You can forget that information now. The loop restarts itself perfectly (immediately) even when all lines are heavily insulated. I am pretty sure it has to do with pressure differentials created by the bigger and smaller pipes. There are probably ways to tweak the system further but for practical purposes we do not have to get all scientific about it. I will leave the nuances to the geeks out there.
This “thermosiphon system” is ideal for when you have a water heater in the basement and you have a remote sink that you want “immediate” hot water at. Sinks and other fixtures that are far away from the water heater can waste a lot of water while one waits for the hot water to get there. Of course to have hot water there immediately it is going to cost you a bit. There are no free lunches.
I have blogged in the past about how simple this can be to do, but this post will clarify some of the information in that post. It will also show how the system can be set up to provide hot water while at the same time maintaining the water in the loop at a high enough temperature to discourage bacteria growth.
Since the need for hot water is minimal while we are sleeping, wouldn’t it be nice to install a timer on the system so that we could stop the circulation periodically? As long as the water is not allowed to stagnate the bacteria typically are not a problem. Most systems with an actual pump have timers that can be set so that the pump only cycles when you want it to. The same can be done with a passive, or thermosiphon system. This can be done by installing an electronic valve controlled by a timer—of course then it is not quite as passive as it was. These low voltage valves can be operated for pennies a year and are very effective. The valve is normally closed and requires power to open. By reducing the amount of “open time” we can reduce the number of pennies even further.
The timer & transformer, the electronic valve, the sensorHeat trap nipple used as a “check valve” at the return to the tank
One of the difficult problems of a thermosiphon system is when there are fixtures in the home that are not included on the loop. This can happen with fixtures that are at the same level as the water heater (bathrooms and laundries at the basement level etc). When you run hot water at those fixtures you will not only stop the operation of the thermosiphon loop that serves the remote location, but you can even start the flow to reverse. When it reverses you can get very hot water off the bottom of the tank going to the remote location followed by very cold water as the cold water is fed into the bottom of the tank. This problem can be controlled by installing a sensor that can tell when hot water is being drawn by the other fixtures. The sensor is a “open-on-rise” switch that closes as the temperature falls again. When the sensor closes (because the water temperature starts to go down after use) the valve opens and allows the thermosiphon to function again.
I should also note that there should not be any fixtures on the loop either before or after the furthest point. Anything on the loop should come off the end of the loop otherwise those pressure differences I referred to become a problem again.
The general idea of super-insulating the pipes—the reality of unintended consequences.
Another key component to the successful operation of the thermosiphon loop is to super-insulate the pipes that run to the remote location so that the pipes hold the heat for the considerable amount of time that other points of use might keep the valve closed—or when the timer is programmed to let the valve close (such as during the night). (More on my learning curve with super-insulating the pipes later.) This would be hugely beneficial for systems with pumps as well. It also means that the loop does not have to be operational all the time during the day either–perhaps running ½ the time–again further reducing those pennies. The hotter these pipes can be maintained, the less the returning water will cause the water heater to fire to bring it up to temperature.
Any recirculating water system is going to make your water heater run more frequently, but because there is a bigger supply of water at a higher temperature the water heater will not likely run as long—even if it runs a little more frequently. How efficient the system will be depends on water costs, energy costs, the type of water heater and how well insulated you can get everything. There are advantages with gas fired water heaters due to quicker recovery rates but that can easily be offset with an electric water heater that is much easier to heavily insulate. It is not really possible to insulate a gas water heater to the levels that an electric water heater can. I am not talking about adding thin and meaningless fiberglass batt insulation around the tank, I am talking about adding significant 3 to 6 inches of high-R foam around the heater. Several years ago I added 2″ to mine–the next heater will get more.
Rigid foam insulation fitted around the tank
Regarding the argument that some manufacturer’s void the warranty if you add insulation, this may be a risk that is worth taking. Who cares if adding insulation shortens the life of the water heater if the insulation has saved you hundreds of dollars in energy costs. I for one, would like to be enlightened as to why adding insulation would shorten the life of the heater. I am assuming that all access points, data plate, warning labels, drain, TPRV etc remain accessible. Not insulating the tanks is probably more about these later issues than it is about the life of the tank.
Most water heaters, even ones meeting modern energy requirements, actually have insignificant amounts of insulation around them—typically no more than R-8 to R-14.
This is where it starts to get even more complicated. We want to keep our water at the taps controlled to below 120 degrees Fahrenheit to prevent scalding injuries. The problem with that temperature is that it is ideal for the growth of bacteria—including Legionella Bacteria. It is typically recommended that tank temperatures be maintained at 140 degrees Fahrenheit to control bacteria growth in the heater. If your tank is at 120 degrees Fahrenheit for the week you are gone on vacation you have actually created a potential incubator. Some authorities consider bacterial infections caused by water heaters to be hugely under reported.
The solution is to install a thermostatic mixing valve.
I think it is a good idea to install one at the water heater so that all delivered water after that point is at a safe temperature. The valve dilutes the hot water to whatever you adjust the control valve–anywhere between 112 and 120 degrees Fahrenheit will generally be satisfactory. Some people insist on having water hotter than that to the dishwasher and in that case you might want to tap into the hot water prior to the mixing valve for that appliance. However, most modern dishwashers boost the water temperature at the dishwasher and therefore this may not be as necessary as one might think.
Temperature gauge and Thermostatic Mixing Valve
In my own house I have a mixing valve at the water heater that covers both bathrooms (upper and lower) and the laundry (lower). There is another mixing valve at the end of the long recirculation loop for the kitchen.
In a way you can think of the recirculation loop as merely an extension of the water heater–both are very hot water to control bacteria growth, while the two mixing valves keep the water safe to prevent scalding.
But what about the costs of storing all that water at such high temperatures?
Keep in mind what I said before about there being no free lunches. Additional costs associated with energy consumption to maintain the heaters at a higher temperature can easily be offset by super-insulating the storage tanks, perhaps even reducing the size of the tank needed or at least having more available hot water to dilute for use at fixtures. Is it more costly to keep an 80 gallon tank at 112-120 degrees Fahrenheit or a 50 gallon tank at 135-140 degrees Fahrenheit? Again, I will let the geeks look into that question. But the bottom line is that the tank is safe from bacteria growth and appropriate temperature at fixtures is achieved by mixing valves.
Expecting to save a lot of energy and having safe hot water in our homes may at best be a compromise.
A note on super-insulating the hot water pipes of the recirculation loop.
In my case I decided to run both the supply and return loops close together, with each line conventionally wrapped with foam pipe insulation–about R-4. The two pipes were run inside a 7” metal duct that I then filled with spray foam—like “Great Stuff.” I assembled the pipe in 3 foot sections and sprayed the foam inside the pipe as I assembled it. Each pipe length had four ¼” holes drilled along its length to spray the foam into the pipe through. I could look in the open end of the pipe to keep track of how it was filling up. My run of pipe is 44 feet. THIS TAKES A LOT OF CANS OF FOAM! And now we are at the “learning curve.” Basically it amounts to not knowing that spray foam in a can requires air and humidity to cure–without those ingredients the spray foam all RETURNS TO LIQUID!
The pipe opened up to reveal a mess
Whodathunkit! So now I had 44 feet of duct with a liquid mess in the bottom and I had to start over. Using all those ¼” holes as a pilot hole I drilled 2-1/8” holes with a hole saw over the entire length. Through these bigger holes I could then layer the spray foam in over the entire length adding a little each day. This gave the foam a chance to cure with air around it. It makes me wonder how often this has happened to other people when they think they are successfully insulating inside a cavity, when really it is just turning back into a liquid again.
If I were to do it all over again I would build a bigger box around the pipes and insulate it with cellulose fiber for a fraction of the cost.
Charles Buell, Real Estate Inspections in Seattle
If you enjoyed this post, and would like to get notices of new posts to my blog, please subscribe via email in the little box to the right. I promise NO spamming of your email
Recently subscribed to the blog – enjoy the articles, thanks for posting. Am considering a ductless heat pump (mini-split) for an all electric home I plan to put on the market. Advertising included in utility bills from Seattle City Light encourages consideration of these systems. Do you see these systems often in the course of your travels & do you have an insight as to the pros/cons. Also, any thoughts on the flat panel radiant wall mounted heaters that are sometimes used in conjunction w/the ductless to warm remote rooms (saw used in one open house (was a green construction remodeled home in Wallingford).
Jim, I do see a fair number of them and I think they are great for small spaces. When the house is large I think there are zoning issues that can be difficult to address with some rooms getting too cold or allowed to get too cold. Houses should be maintained a relatively even temperature to avoid condensation/mold issues. I like true radiant panels quite a bit—kind of like the old radiant electric ceiling heat. They heat objects and not the air so that you feel more comfortable quicker and lower air temperatures.
You didn’t mention which heat trap nipple you used at the bottom of the tank, a cold side or a hot side? Or does it even matter much as long as you have either flowing in the right direction? Thanks.
Charles, I recently installed T.S. in my father in laws house. His house is a house in a house with air pocket in between walls and ceilings. I insulated entire runs the best that I could with pipe insulation. The problem that I have been told I have is the temperature inside the house has been elevated about 6 degrees when the air conditioner is turned off for extended lengths of time. Normal temperature is in mid 70’s except when temperatures are above or below greatly. Should I try to sandwich bigger pipe insulation on top of existing to try to rectify this problem. Thanks, Dave
Dave, I am not sure what you mean by T.S. I have never heard of any appreciable rise in temp due to a recirculation loop that is insulated normally. I personally would look to other causes of this rise in temperature. What is the temperature of the water being circulated?
Charles, T.S. was my shortened version for thermosiphon system. I don’t have the actual temperature because I don’t have any thing to measure it with. Before installing the system he had the water heater set on the highest hot setting on water heater, the next day after I installed it I changed to one of the warm settings but without stopping out there I won’t be able see it or try to get actual temperature. I still think that a good part of this problem is from having this to tight with his double wall system and air chamber in between. I will let you know when I find out the temperatures . Dave
Hi Charles, Thanks very much for this excellent write up on pumpless hot water recirculation systems. I recently realized that we have (I think) a similar system in our home when I went to change out our 22 year old hot water heater last week. I noticed for the first time in our 5 years being in the home that there was a 1/2″ copper pipe leading to a tee fitting at the hot water heater drain. I thought “what the heck could this be?” and immediately turned off the ball valve up near the ceiling on this line. I noticed the next morning that it took forever for the hot water to warm up but did not connect the incidents until later in the day when talking to my wife on the phone about a similar incident when she was giving the kids a bath. I looked up diagrams online for what someone would use a circuit attached to the drain with a tee for and I discovered the heat recirculation system. I immediately searched near all the fixtures for an extra circuit with a pump attached and have not found a pump (yet). So, I assumed it is a pumpless sytem although here is why I am commenting on your blog: after I swapped the water heater, it took me a few days to figure out how to attach the recirculation tee fitting to my new replacement one since it came with a plastic drain valve with incompatible plumbing to the tee from my old one. Finally, after $45 in crazy sharkbite and garden hose attachment fittings from Ace Hardware, I realized I could just unscrew the plastic fitting and pop my old tee on there (duh). Anyways, I ended up having to add 2 90º fittings and jog the 1/2″ pipe about 1 foot to the right to meet up with the tee. So, I got it all hooked up and it still takes FOREVER for the hot water to warm up in the far away faucets. I thought maybe air in the lines was stopping recirculation but I think I have cleared all that out and we are still having trouble. I studied your blog a bit further and realized that we do not have a check valve in our circuit, atleast not at the hot water heater as far as I can tell. It is just a 1/2″ pipe leading into the top of a tee which connects to the hot water tank via a 3/4″ copper pipe with male threads on both ends. Is it possible that my hot water heater has a check valve that only allows the hot water to flow out instead of in as I need it to for the recirculation to work? Or, is there a pump hidden some where in my circuit that failed when I had the valve shut off for a week? Or, are the 2 90º elbows I added creating some restriction? Or, or, or? I really would appreciate your insight. Sincerely, Tim from Michigan
It is possible you have some air in the line between the water heater and the distant point of use—either in the supply or return side. Also the heater may have come with heat trap nipples that are interfering—but I would not think that is a likely cause. You may have to call a plumber that understands these thermosiphon type recirc loops (and keep calling until you find one that does). If you configured the setup the way it was with the previous heater it should work. Did you take out the old plastic drain valve. I did on mine. If you post pictures of your setup I might see something that is a problem.
Hello Charles, Glad to find your insights. I have been plumbing for over 30 years and from time to time ponder this TS system. I have only put in maybe 3 in that time with no pump. I see on jobs several with pumps and no insulation which of course makes the heater run quit a bit more. I just put one in this week on a new house and it is not working right. Sluggish and reverse flow from bottom of heater when faucet is used. I now realize that I have 2 problems — only 1/2 inch on supply and also I didnt take out the rubber flaps in the supply nipples on top of the heater– I will correct that. I also wonder if there is a difference in where the loop is connected on the far end—I connected with a tee in the kitchen sink cabnet instead of under the floor. This is a one story with 40 foot run— no other fixtures on loup. The return line is sloped down –dropping about one foot in the 40′ back to the heater. Thanks for any input. Merry Christmas! Charlie from Greenville Michigan.
Yes, I think any heat traps would have to be removed. I had ball type in mine that I had to remove. I think you will need some sort of back flow device on the return side but likely not if you have 3/4″ supply all the way to the farthest point. I found that the system would not restart itself with the type of check valve I installed—but more sensitive ones may exist. My T off the loop is below the floor—but I should not think that would matter as long as it is 3/4″ all the way to the T (plumbing in my area is not sloped so the system can be drained like it would be in your area likely). My entire loop is 3/4″ with only the last 101 to 12 feet being 1/2″ to create a pressure difference. My run is close to what yours is. and has no appreciable slope most of the way. So you said there are no other fixtures on the line—does that mean all fixtures in the home come off of this line?
Some is a bit confusing though. The text doesn’t seem to match the image, and the image seems to be incorrectly drawn? The way you describe the system in the text seems right, but the image seems to show incorrect placement of some items.
The temp sensor and timer valve in the image seem to be connected with a line, indicating their operation is coupled, when it is not. The timer valve is also installed on the main hot water supply to the loop fixtures instead of the return line, which means when it closes, you completely lose all hot water to them, instead of just stopping the loop operation. Correct? This would stop all water flow to the loop fixtures instead of just loop flow.
Although the text is basically correct, the paragraph that mentions when fixtures not part of the loop are operated, those remote fixtures will first ‘get very hot water off the bottom of the tank’, followed by cold water. This might throw readers off, as the water in the bottom of the tank should not be ‘very hot’, the top should. I suggest saying “Without a check valve at the bottom of the tank, hot water to those fixtures will also be supplied out of the bottom of the tank as well, which will turn cooler as cold water refills the bottom of the tank.”
On this same subject, in your comments, you indicate that with the temperature sensing valve for the non-loop fixtures, no check valve is needed. What is the location of the valve that is controlled by this sensor? Where the check valve is? This would not have the same operation as the check valve would. If the in-loop fixtures are operated with the temp valve in the open position, water will flow to those fixtures from both the top and bottom of the tank, and the bottom supply line (normally the loop return line) will soon be cold. I don’t think this system can operate without that check valve.
Also, loop operation is stopped in the original design when a non-loop fixture is operated due to the check valve. As such, if one of these non-loop fixtures is used for an extended period, such as a shower, the water in the in-loop supply line will get cold. Suggestion: Make the non-loop fixtures use the loop return line as their hot water supply. Although technically not ‘in’ the loop, they would increase loop flow when in use rather than stopping it. This should have minimal impact on the starting temperature of those non-loop fixtures as water in the loop return line should be pretty hot by design. (Might require another check valve if the mixing valves will allow reverse-hot water flow)
Hi John, I am not sure where the confusion is, but I will attempt to answer your questions as best I can.
The drawing does indeed describe my system pretty accurately. I have read back through the text and can find nothing inconsistent between the two—but that is always possible I suppose.
While the line from the temp sensor looks like it is going to the timer it is actually going to the automatic valve that happens to have the timer associated with it. Both are in the correct position to do what they need to do. Basically the only reason for the the valve is to prevent reverse flow when fixtures not on the loop are operated—otherwise the valve is always open and circulating. The timer is merely a way to close the valve so that it does not circulate all night long. Typically the timer closes the valve around 10 pm and opens again at 5 am or so. The thing about a thermosiphon loop is that the entire tank is the same temp (or very close). Of course as soon as you start to use hot water cold water enters the bottom of the tank but is quickly heated by both the heating element and the return of hot water to the bottom of the tank.
I assure you with this approach a check valve is not necessary and finding one that does not stop or slow the weak flow of the thermosiphon is difficult. The system does indeed work and has for several years now without the slightest of hiccups.
If one of the non-loop fixtures is used for a long period of time the thermosiphon loop does not cool off because the valve would be closed and stay closed until until water has stopped being used at the non-loop location. Typically the turn around time is about 10 minutes or so for the valve to open itself again.
With many years of fiddling with this system, I can tell you, check valves need to be avoided as much as possible.
In three years, I have never experienced not having hot water anywhere when I needed it and always immediately present at the remote location.
You do not need a check valve in your system because your temp valve is closing when needed, preventing reverse loop flow. This would normally be accomplished with the check valve, but your setup does the same thing.
However, according to your diagram, in the position where you have that valve, when it closes you not only shut off the loop, you shut off the remote fixtures from the hot water supply, forcing them to reverse-pull off the bottom of the hot water tank through the return line. Although not ideal, it would still work. As you said, it hasn’t caused any issues.
As for check valves, it’s unlikely anyone will find exactly what’s needed here, but a modified check valve will definitely work.
A check valve using a floating plastic ball is what’s needed, provided it is installed correctly, which is vertically. The ball floats upward, completely opening the path. When high reverse flow is encountered, the ball is forced and held downward, closing off the the flow. More than likely, a check valve of this type has the ball on a spring, but the spring can be removed. The low pressure thermosiphon will never be able to open a spring loaded check valve, or even a flap valve. The floating ball valve is pretty much the only option for a passive system. A check valve can also be made by using a suitable plastic ball, but the plastic would need to be food grade and able to handle 140F. I’ve actually made a few doing this to mix hot and cold water as makeshift inexpensive mixing valves, and also to supply a hot tub using two tankless heaters, while isolating them from each other, etc.
This is the same type of ball valving system used on some older auto transmissions. A floating rubber or plastic ball works in this manner. The reverse is also employed. Gravity holds a small ball-bearing in a hole until flow comes from underneath it, displacing it.
I agree the floating valve idea should work, but the only one I could find were heat-trap-nipples. The problem with those is they are actually not designed to “completely” shut the water off, so enough water leaked by to reverse the flow anyway. Not sure why they are designed this way but the ones I tried are. I was hopeful they would be the perfect solution. 🙂
For anyone installing this system. After doing some engineering research, it appears that PTFE (Teflon) or PVDF (Super Teflon) would provide the most durable, chemically inert, and heat-resistant check valve ball for making or modifying your own check valve. Most suppliers seem to only have solid PTFE, which will NOT float, and so can’t be used. Suppliers seem to have hollow PVDF, which does float. “GlobalBallSales” on eBay stocks ~.750″ (20mm actually) hollow PVDF balls. I bought some today…..paid $5 each shipped. Happy building!
Purchased watts recirculating system,first operated with timer on 24/7,then changed to 12 on 12 off,tried next to 15 minute on 45 minute off with hot water operating well.finnally turned pump off only relying on sensor valve,which is installed furthest from hot water tank,To my amazement all faucets still provided hot water within 10 to 15 seconds.Called watts to ask why this is possible and they had no answer. They only thing I can think of is sensor valve furthest away needs no additional pressure other the my well pump which cycles between 40 and 60 psi.Your comments would be greatly appreciated.
Well then, it is simply thermo-siphoning just like the installation my article is about. You do not really need a pump. It will likely reverse flow if you have any fixtures at the same level as the water heater though.
Lots of plumber just do not know about these thermo-siphon systems and many many installations it is not possible to get the water heater lower than all the fixtures. I am quite sure if you run your sink next to your water heater for a few minutes you would likely see the flow of hot water reduce in temperature to the upper level—-to the point of becoming cool as it draws water from the bottom of the tank which is cooler because that is where the cold incoming water goes into the tank. This is especially true if it is an electric water heater. With a pump you do not have to think about all this stuff.
In a comment up above, on 17 January, 2017, you wrote “With many years of fiddling with this system, I can tell you, check valves need to be avoided as much as possible.” Yet your illustration shows a check valve right before the hot water return line enters the bottom of the water heater. Can you further explain this possible discrepancy?
Also, as per your article, the hot water supply should be 3/4″ and the hot water return should be 1/2″ in order to create a pressure differential. By this you mean the fact that as water circulation speed increases, water pressure decreases, correct? And thus the water travels from a place of high pressure (3/4″ pipe) to a place of low pressure (1/2″ pipe) which leads back into the bottom of the water heater.
If (that is a big if) the above is correct, does that explain why it is OK to insulate the hot water return? On first thought, I would have thought that by not insulating the return hot water line and letting its water dissipate its heat, that would lead to the water becoming less buoyant/more dense, and thus sinking back to the basement water heater. The benefit is a stronger return current. The cost is increased heating needing to be done by the water heater.
The problem with check valves in a thermo-siphon system is they do not open easy enough because the force is much less than when a pump is installed. As the pressure diffentials, by reducing the pipe size you can fully insulate the pipe as desired—otherwise you would have to leave the last 10 feet or so un-insulated in order to create a temperature induced pressure differential I want the pipe fully insulated, hence the approach I have taken. And it works. 🙂
I understand what you are saying about the problem with check valves in a thermo-siphon. However, I see that in your diagram, you have a check valve just before the hot water return pipe terminates at the bottom of the water heater. This is what I have the question about.
Thanks Travis, The drawing has been through several incarnations and I just had not taken the time to get rid of the check valve. I have to warn you though, without it, you will have to have an electronic valve like the drawing shows. It is not possible, in simple terms, to demonstrate all the nuances of possible differences if other approaches are taken.
Just sharing that in our house the heat traps were causing an issue with a gravity hot water recirc system. Once the heat traps were removed, it worked. The problem is that the heat traps caused enough resistance in the line through the water heater that the path of least resistance for a hot water tap when opened became the back flow of water through the recirc loop. So cold water was back flowing through the recirc loop mixing with some hot water from the water heater’s normal path and no tap ever got hot. It was a bit of a mess and I had to really convince the plumber to take the traps out and now it works just as before.
That is what I have found as well, and of course they are not functional on a water heater with a recirc loop anyway—except perhaps when there is a timer involved. If that is a concern, installation of a loop type heat trap as opposed to a ball type heat trap would likely still work as long as the return pipe is reduced in size the last 10 feet at the water heater.
Hi Charles, I have an electric water heater on the same level as all of my fixtures with a circulation pump and line installed. Recently I have had a pin hole leak in my copper pipe that created quite the mess which I have since repaired. Are there any health concerns with removing the pump and capping off the two lines at the heater? Or just not turning on the pump? I don’t want to tear into the wall and cap the line at the circulation line at the furthest fixture?
In my opinion no water lines, especially on the hot side, should be left with standing water in them. I recommend consulting with a licensed plumber to determine how best to abandon your recirculation loop.
Hi Charles, I have a 30 year old gravity recirc system, water heater in basement, all fixtures on one floor/ranch house, which was fitted to a new water heater last year. Everything was working well until recently: I have instant hot water at all remote sinks and showers, but when someone is showering, the kitchen sink water goes to luke-warm and never reaches even close to maximum temp. The kitchen sink is the closest fixture to the water heater; located almost directly over it. I have noticed that normally the kitchen sink hot water is not instantly hot like in the bathrooms; it takes 20 seconds or so to come up to full temp when no other hot water is running. But the luke warm thing is relatively new. I don’t believe there are any mixing valves or check valves installed in the system. I appreciate any thoughts you might have.
Does the new heater have the re-circulation loop going to the bottom of the tank? If all fixtures are not on the same loop, it is common for locations off the loop to stop or even reverse.
Hi Charles, Nice site! Can a thermosiphon system operate at all *through* a thermostatic mixing valve? I already have a thermosiphon system operating through a dedicated return from the most remote fixture (a second floor sink), but there are other fixtures between the hot water heater (in the basement) and that remote sink, and it is impractical to get to those lines. I would like to add a mixing valve at the heater, but this would require mixed water to run through the recirculation loop, which I don’t think is possible without a pump. Can you confirm? Thanks!
Really? In your diagram, the entire recirculation loop is completed *before* any hot water goes through a mixing valve. All the water in in the loop looks to be up to 140F, in bright red. In my case, the mixing valve would be immediately upon exit of the heater, so the entire loop would be post mixing, with temps up to only 120F. While 120F is more than enough for a thermosiphon effect, do the mixing valves allow actual mixing if there is no flow (demand) from a fixture?
I’m attempting to create a thermosyhon system. I understand that removing the heat trap nipple on the hot side of hot water heater is necessary. But would it still work if I leave the heat trap nipple on cold water supply? I can’t figure out why it would matter.
I have a bathroom in the bonus room above the garage that I am having trouble with water lines freezing. I installed a recirculating pump under the sink, and that solved that problem. However, I still have the cold line to the shower freezing up. Should I install a second (return) line from the shower back to the sink? I don’t remember exactly how it all got plumbed, but I think the hot and cold lines came to the sink and then I teed of to feed the toilet and shower. I know that it would help to wrap the lines as well, but have avoided having to open the wall and dig through the blown-in insulation. I suspect part of the problem is that the water lines are laying on the garage ceiling instead of in the insulation.
Do you think that a thermosiphon loop could work for a 1 car garage floor PEX system? Behind our garage is the laundry room that has a crawl space. The crawl space is below grade and the walls are well insulated/foamed. We heat the crawl space to about 80 degrees F, and the heat radiates thru the floor heating the laundry room. The garage floor PEX is a single loop with the manifold mounted in this crawl space. After reading your re-circulating without a pump article I’m intrigued. And, since I haven’t installed the heat source and circulating pump yet, to heat the garage, I’m thinking that maybe I simply mount a heat exchange/radiator in the crawl space and see if it will thermosiphon. The output of the crawl space heater can be increased to meat the added demand. Any thoughts and ideas are welcomed.
Hello, Couple thoughts that come to mine, is if you want more pull why not just drop the return to 1/2 and leave it cold last 10-15 feet? Also the bacteria in copper lines will not live if the temps drop down on a line due to cut off or other. Copper is a natural antiseptic that will not allow bacteria to grow. You can buy copper mixers and keep water in them with the lid on for months and still drink it. The hot water tank is not copper so keeping it at 140 is not a bad idea.
I have a heat exchange radiator that I placed in my Hot air return. It’s about mid height of the electric boiler which is only a few feet away. Would it work to keep hot water in radiator without a pimp. I have a circ pump but find it’s too fast and keeps boiler element on too long.
I just had a thermoshiphon system installed with a copper repipe on my house. Single story ranch with pipes in the attic. All 3/4 hot lines and the 1/2″ return line insulated. The heater is in the garage and about 6′ lower than the pipes in the attic. Bathrooms are 35′ from the heater and every morning the water is hot within 10 seconds of flow. The kitchen is 20′ off the header in the other direction and takes about 20 seconds to see hot water, as it is not in the loop, but hot water is now closer to where it tees off. All I needed was hot water in the bathrooms right away, where it used to take 3 min. to run through my old galvanized piping.
Thanks for generating the dialogue through this site. I have been running my own designed thermosyphon system (based on physics theory) in a 2 story split for about 13 years. I started by retro-fitting the entire hot water plumbing system from a mixed series/parallel flow to a continuous series flow, starting at the closest fixture and looping through the lower level to each outlet in progression – ending at the 2nd story showers (all 3/4″ copper). At the end of the line I “T”d in with a 1/2” copper return line, connecting again with a “T”, to the cold inlet line at the top of the tank (since this line communicates with the bottom of the hot water tank via a dip tube). Upon purging the air from the return line I had almost immediate flow and instant hot water at all locations. The next problem is (of course) back-flow in the return line which causes cold water mixing with the hot at the shower head and eventually would contaminate the entire hot line if an up-stream tap were turned on for any length of time. Aware that this would happen I installed a copper flapper valve from the local hardware store. The force of the convection flow is so minimal that it could not push the valve open, so I had to slant the valve so it was partially open. The system worked at about 80% efficiency but the clinking of the copper valve was annoying with water running. I found that I could insulate the entire run (including return) and did not lose the circulation. Checking gigajoule gas consumption for one year prior and one year following showed virtually energy neutral. I experimented with various flow actuated electronic valves and the system was getting so complicated I decided to scrap all electronic valve ideas and revert back to a simple valve that would be hyper-sensitive to changes in flow direction in the return line. After many years of experimentation and searching I finally designed a valve that works perfectly and the system is now 100% reliable. About a year and a half ago I retrofitted a single story which already had a semi-series flow, starting with ¾” pex and reducing to ½” about half way. I connected a ½” pex return line at the farthest faucet and “T”d into the cold inlet line at the top of the tank as above with my newly designed valve inserted into the return line just before the water tank. After purging air from the return line – voila instant hot everywhere. The home owner had the hot water tank replaced about 6 months ago and lost the instant hot water. Ironically, they complained that it now took longer than before to achieve hot water. This is what happens when you become accustomed to the luxury of instant hot at the tap – once you have it anything else is unacceptable. The problem turned out to be simply that the new tank had “heat traps” installed in both the inlet and outlet fittings. Upon removal of these interferences instant hot was restored. PS” I do not buy the theory that flow from the 3/4″ line to the 1/2″ line is creating any pressure differential at the flow rates involved. This would have to invoke Bernoulli’s principle which is not significant until higher flow rates are encountered. I am convinced that the thermal energy in the hot water tank is powering most of the thermosyphon effect. This opens the thought that this system should work on a level playing field with no sloping return required. Open for discussion and experimentation…… PPS: How do we impose this on the entire nation – enormous water saving!!
Doug this is a timely comment. I too have struggled with the backflow issue and like you found the brass flapper type does not work. I figured out a solution that gets most of the way there by simply using a cold side floating ball type heat trap nipple. These do not completely cut off all the water so the flow still reverses some. My 100% solution was to take the ball out of the nipple and place it between two valves. The top valve has a conical seat that seals completely and the lower valve does not. No idea if the ball ever goes that far or not—and the lower valve may not be necessary at all. As to Mr Bernouli, I have not idea but I am pretty sure whenever you have height differences, temperature differences and size differences (volume), you will also have pressure differences. It works regardless. As to getting everyone on board with this approach is getting more and more difficult as modern energy codes outlaw the use of Thermosiphon systems—no clue why. Perhaps they are assuming no controls—timers etc. I am attaching a picture of where the ball is in my system and the bypass to the right is merely a loop I can test the system without the backflow.
I recently undertook this endeavor and let me just say THANK YOU!!! I bought a new build and it was 50′ of 3/4 pex to the upstairs shower. Took over a minute to get warm water and just forget about using a sink. I was going to buy a recirc pump or run a dedicated 1/2 line up to the 2 showers but I worried about lack of flow and pressure. Then i stumbled upon this thread and changed my entire plan. Now i have instant water at all of my sinks that use to take minutes before. I guestimate I was wasting 5 gallons a day minimal. I insulated all the 3/4 and the return pipe with r4 insulation. I teed into the 3/4 before it heads upstairs and ran the 1/2 pex back to the heater with a 1′ drop per 10 ft. I used a 1/2″ brass swing valve at a the base of the tank at a slight angle. I hear it close when faucets are ran and insulating it really helped.
I’m now finishing up my solar hookup to the bottom element only. Have it set at 125, top at 115. Matched the panels to the ohms of a 120v 2000w element with the help of David Poz (https://www.davidpoz.com/) On my way to free, instant hot water! Thanks
I appreciate your response to “hot water circulation”. I have employed the thermosyphon principle to design and implement a passive convection flow hot water system in my 2 story home since 2007. The initial design utilized a brass swing valve as you have used, but the system did not perform to my satisfaction. It was reliable only about 85% of the time and it did produce the clinking noise you are experiencing with the hot water running. I have experimented with changing the slope angle multiple times with varying degrees of success. Because the diminutive force of convection flow is not strong enough to open the brass valve wide in static mode and the pressure differential did not shut the valve off completely with hot water flowing in the supply line; the efficiency was not satisfactory enough to promote the design either ‘patent wise’ or simply to publish on the internet.
Since that time, I have designed a return line valve that makes the system perform 100% of the time. The efficiency of the loop has also increased. About 2 years ago I installed a similar system in a bungalow with minimal changes in the supply loop, and it produced results beyond my expectations. Note that I do not plumb the return line into the bottom of the tank; I “T” into the inlet line at the top of the tank. You may also note that the ‘heat traps’ installed into the newer hot water tanks will virtually stop the convection flow and they must be removed.
I have also learned that the main driving force behind the passive convection flow is the thermal energy in the hot water tank creating the thermosyphon action’
With proper design and insulation, passive circulatory hot water systems can be every bit as energy efficient as a well insulated water tank with heat traps, and will increase the hot water capacity and promote significant water savings – not to mention the comfort and convenience of having “hot water waiting” when you open the faucet. When I first installed the passive circulatory system in my home, I kept records of the gigajoule gas consumption over a 2 year period = one before and one after installation. The energy consumption was virtually the same for both periods. It is not an exact measure as I did not isolate the gas consumption of the hot water alone, nor did I measure the water consumption; but it gave me a pretty good indication that we were not using any more or any less natural gas energy for that period in the home.
Hope this helps those out there who attempt to employ this innovative design in their own homes. It would be beneficial for the environment if an organization were to come aboard and promote this design nationwide. Think of the water savings alone and the savings in the sewage treatment that could be achieved.
The main reason to go to the bottom of the tank is the last 10-20 feet need to be reduced ins pipe size and if it goes to the top of the tank, if you had reduced it, it goes back to 3/4″ with the dip tube. Some people say that last 10 feet should not be insulated but I have not found that to be true at all, as mine is heavily insulated.
Your arguments are all correct, there are many factors controlling the flow of fluid in a pipe. The difference in density due to temperature change has a significant effect on the flow rate if there is a corresponding change in elevation. Hence the cooling effect will definitely have an influence if the return line is a lower temperature than the supply line, again provided there is a change in elevation so gravitational forces can come into play. Your experience, and mine, on insulating the return line to reduce the thermal energy loss just goes to prove there are other forces in play that offset the loss of this density / height change factor (although we have some effect). As far as the last 10 feet being un-insulated, unless it is accompanied by a change in elevation, I don’t think it would have any effect.
As for the reduction in diameter of the return line creating the flow; Bernoulli’s principle outlining the pressure drop in a fluid flowing in a pipe with a reduced cross section area requires that there be a flow in the pipe to create the effect. A familiar example of this principle, utilized in the flow of air, as applied to the venturi of a carburetor or the airfoil of an airplane, requires laminar flow to be significant to create a useable force. In addition, when first starting the hot water flow, pipe diameter has no effect because there is no “Bernoulli Effect” when the fluid is static in the pipe. So some other force must be coming into play because the flow starts spontaneously and almost immediately. Once fluid flow starts there will be some pressure change due to pipe diameter change but at the flow rates we are talking about I don’t believe it is a significant force in the equation. All it takes to stop the convection flow in the system is a heat trap left in the hot water tank, so the forces creating that flow must be fairly small.
That leaves the question: What is driving the circulation in our thermosyphon system? I don’t believe there is a simple answer to that question, but I believe the thermal energy in the hot water storage tank plays an important part – particularly when the tank is being fired. I don’t profess to be an expert in this field, I only can verify that the system works. My knowledge related to the circulation of hot water started with some understanding of the physics involved and some inherent logic, but also empirically with the application of the theory and then observation of the results.
I do know that, besides the convenience factor, there is a huge savings in water and a corresponding saving in sewage treatment. I did a ‘rough’ calculation on the water savings in my 2 story home based on 2 person occupancy, and I came up with about 4 cubic feet of hot water wasted waiting for hot at each station on a given day (that’s based on 15 station cycles – hand washings, showers, etc.). That’s 30 US gallons (113 litres) per day – 11,000 gallons/year (41K litres). A community of 50 thousand single family dwellings could save a potential of over half a billion gallons of water in a year conservatively.
Please take the foregoing conclusions with a “grain of salt” Charles, because it is all based on theory and has not been proven (at least not by me)! I stand to be corrected on any assumptions made in the foregoing arguments.
Charles, this is really interesting, the comments too. Can you look over my post in this plumbing forum where I describe the theory a bit? Do you see gaps or errors I should correct?
Jamie, I could not get into the plumbing site to leave a comment so here you go. Certainly uninsulated flows faster and heavily insulated over its entire length will still work fine and self start fine even though there is very little temperature difference. Temp difference is only one thing that drives the thermosiphon and by decreasing the pipe size for the last 10 feet or so just before it gets back to the tank creates a pressure difference that is enough to keep it going.
Hello and thank you for the useful information provided here. My thermosiphon (T.S) system is not working as good as it did before replacing one of the water heaters recently. In some places I have to wait long time for hot water. Due to size of the house 2 water heaters are installed together, call it no. 1 and no. 2. cold water goes into no.1 which feeds no 2 supplying the house with hot water. Return line of T.S. goes to the bottom of no 2. no. 2 was replaced couple of weeks ago and then the problems with T.S. system started. I guess this system is not common around here because I didn’t find a plumber familiar with it so instead wasting time troubleshooting I thought maybe I can add a pump. So here is the question: can I add a pump to the existing T.S system? If so where?
Thank you for your reply offering possible clue. I have opened all the faucets a ran water but did not detect ant air flow. What is the best way find out if there is air in the system and how do I bleed out the air? Thanks again for sharing your wealth of knowledge.
Here is the solution for having to run the hot water for a long time to get hot water to a distant faucet: We have used this for three years and it works beautifully! Google: Laing EcoCirc under sink hot water pump recirculating system.
About 10 years ago I got information about the thermosyphing system I ran a 3/4 line with a one-way check valve and a manual shutoff from the furthest point of hot water to the drain in my hot water heater. No other add ons. It worked fine. I recently replaced my hot water tank. This system wouldn’t work. Replaced the 3/4 line with 1/2. Removed the heat valves in the tank lines. Did everything I could to remove air from the system. still not working. First, why? Second, can I put a recirculating pump in the return line from the faucette to the drain of the tank?
It will certainly work with a recirculation pump. Can you send me pictures of the setup at the heater? Full views. Something obviously must be different. It should still work but without being there I may not have an answer for you.
I had my hot water tank replaced and now my thermosiphon doesn’t work. It takes a long time for hot water to reach the furthest fixtures. Do I need to “prime” the thermosiphon, somehow?
In mine, the distant faucet is the highest point in the system so it auto vents opening the faucet. If you have high spots, you would have to install a way to bleed the air out at those spots. Over time the oxygen will be absorbed into the water.
Hi Charles, thanks for your dedication to this post, I’m glad to see that it is still active! I have a question for you… I am considering installing a Towel Warmer in my upcoming bathroom renovation (floor above the tank and 10′ to the left, lol) and I thinking about having it continuously circulate via a thermosiphon.
I was thinking about a 3/4″ line off the top of the tank rising up to the top section of the towel rack, the exit on the opposite lower corner draining via a 1/2″ line back to the bottom of the tank. I would insulate the 3/4″ supply, but leave the 1/2″ return uninsulated to aid flow, knowing there will be small energy loss.
I don’t believe that I would need any controls on this setup because it is a closed loop, other than an isolation valve for maintenance…. Does this seem to be a sensible idea?
It is an interesting idea and it likely would work. Depending on where the heater is and what else is going on, it could reverse flow I suppose. How about just running it through the towel warmer before the hot water goes anywhere else—use of hot water anywhere in the house would then heat the towel bar. Just a thought. Love to know how your idea works if you try it.
In my house there’s a three-quarter inch copper supply line for the upstairs hot water in the bathrooms. At the end of that run I teed off a ½ inch PEX line which runs about 60 feet back to the water heater two floors down in the basement. In the basement, the PEX connects to a shut off valve followed by a spigot followed by a one-way valve and then the whole thing ties into the cold water inlet about 18 inches above the water heater (see picture).
With this configuration I can connect a hose to the spigot and drain the air out of the system which is critical for the convection to work properly. When I first put the return line in I didn’t do a thorough job purging the air and the hot water convection didn’t start working for several weeks. Apparently it took that long for the air to work its way out of the system. Next time I drained the water in the house I did a better job purging the air and the convection loop started up quickly.
I can also close the valve and open the spigot to test that the one-way valve is working properly. The one-way valve prevents cold water from flowing back up the convection line when someone opens up the hot water upstairs. The one-way is a ball type with no spring and is available from Amazon (search on “pex check valve”).
Everyone in the house is thrilled to have warm water coming out immediately the faucets and showers upstairs. In fact the water is downright hot within a few seconds which is more than we really need. What I was after is for the water to not be cold so I’ve experimented with partially closing the valve in the basement to reduce the convection flow.
When I started I was skeptical that convection would be enough to circulate the water but figured I could add a pump if necessary. It appears that no pump is needed and I am very pleased with how the whole system works.
Thanks to everyone who posted here for helping me figure this out.
oh where do i begin, new house, new rheem hybrid heat pump, copper trunk line in conditioned attic with pex supplys to faucets, return is pex in the foundation going to a grundfos auto adapt recirc pump that doesnt sense anything so i keep it at 100% and switch it on for 3 mins and off for 10 mins. meanwhile the rheem works great in the summer in heat pump mode, cools the garage , but in winter it’s too cold in the garage so i switch it into energy saver mode, big mistake. it turns on the element, satisfies the thermostat, then the recirc pump kicks on, temp leaving heater is 125 and returns at 110. when the pump is off there may be siphoning going on, but i can say in the 10 minutes that the pump is off, the water temp in the hot line drops another 10 degrees, then when the recirc pump kicks back on it’s shooing even colder water into the tank, and the water heater turns on again. if i don’t run it at these intervals, it will take 2min 30 seconds to get hot water at the furthest away faucet. where do i start trying to resolve these issues or is the the nature of the beast? single story 3000 sq ft residence. i’ll even pay someone that knows what they’re talking about for their time if we can do a video call, i’ll show you everything you want to know about this new constructed house, thanks to anyone in advance.
The pump will likely need to run continuously during the hours you are likely to use it. As to the cold garage, that is just one of the many things you were not told when you were sold on the idea of one. There are kits that can be added to take the cold exhaust to the exterior, but then you have to think about replacement air to the garage. An ideal solution is going to be difficult.
thanks for the quick reply! the heater sits in a small closet inside the garage, there’s a 1″ gap under the door for inlet air, and the exhaust is running to a small piece of flex duct into a 12×12 vent into the garage. when it’s 105 outside, it’s 83 in the garage which is good but not superb, garage is 900 sq ft, insulated garage doors, etc. it looks like this grunfos pump is sucking from the return and dumping it into the 2nd return inlet on the heater, is that normal? i tried the auto adapt setting and temperature setting with the sensor and after 2 weeks it doesnt “learn” turns on when it wants and gives me 2 minutes of warm water at best. i have the pump on a timer, 3 mins on, 10 mins off, that keeps the lines primed but boy does this foundation suck the heat out in the winter. i can’t seem to find the right setting on the rheem hybrid, energy saver mode used 12kWh yesterday! heat pump uses 3-5 but makes the garage an ice box during the winter. there is no way to vent the cold air to the outside at this point, my main problem is the recirc loop, i’m a strong believer now that the return loop should be encased with the main trunk, hevily insulated and kept out of the foundation, leave it the attic. i wish the main trunk could dip down from the ceiling and get as close as possible to the faucets, this would really make the water hot after just a few seconds. i’m not sure of any devices that can properly sense the in/out temp and turn on the pump in a given time range. those wifi outlets run out of programming options. i actually had 2 manual timers on the pump. one to turn it on from 6a-midnight, and the other like i said before, on for 3 mins, off for 10. there has to be thermosiphoning going on for the grundfos to detect water movement thinking someone is turning on the tap. I just wish there was a better way to supply a home with a single water heater.
I received your pictures, but unintentionally delete! Could you resend please? I’m having a problem understanding how to handle the issue for fixtures on the same level as the water heater. All my distance relative fixtures are on a single 3/4″ cooper feeder line. The one basement bathroom draws directly from this line. If the issue is this will cause the thermosiphon to reverse, how and at what location do I resolve this. Can we discuss? JUST SENT 3 PICTURES TO CHARLES@BUELLINSPECTIONS.COM
You will need to install the auto-shut-off valve in the line that comes from the remote locations when the same level fixtures are used. Of course if you could figure out some sort of backflow valve that actually worked under the low flow conditions of the thermosiphon, you might not need to auto valve—I gave up and went to the assurance of the electronic valve.
[…] ways of getting hot water faster are to have a circulation pump added or thermosiphon loop, but I don’t have any firsthand experience installing one of those systems. Talk to your […]
Recently subscribed to the blog – enjoy the articles, thanks for posting. Am considering a ductless heat pump (mini-split) for an all electric home I plan to put on the market. Advertising included in utility bills from Seattle City Light encourages consideration of these systems. Do you see these systems often in the course of your travels & do you have an insight as to the pros/cons. Also, any thoughts on the flat panel radiant wall mounted heaters that are sometimes used in conjunction w/the ductless to warm remote rooms (saw used in one open house (was a green construction remodeled home in Wallingford).
Jim, I do see a fair number of them and I think they are great for small spaces. When the house is large I think there are zoning issues that can be difficult to address with some rooms getting too cold or allowed to get too cold. Houses should be maintained a relatively even temperature to avoid condensation/mold issues. I like true radiant panels quite a bit—kind of like the old radiant electric ceiling heat. They heat objects and not the air so that you feel more comfortable quicker and lower air temperatures.
You didn’t mention which heat trap nipple you used at the bottom of the tank, a cold side or a hot side? Or does it even matter much as long as you have either flowing in the right direction? Thanks.
With the automatic valve and temp sensor, you do not need the check valve at the bottom—I should have made that clearer.
Charles, I recently installed T.S. in my father in laws house. His house is a house in a house with air pocket in between walls and ceilings. I insulated entire runs the best that I could with pipe insulation. The problem that I have been told I have is the temperature inside the house has been elevated about 6 degrees when the air conditioner is turned off for extended lengths of time. Normal temperature is in mid 70’s except when temperatures are above or below greatly. Should I try to sandwich bigger pipe insulation on top of existing to try to rectify this problem. Thanks, Dave
Dave, I am not sure what you mean by T.S. I have never heard of any appreciable rise in temp due to a recirculation loop that is insulated normally. I personally would look to other causes of this rise in temperature. What is the temperature of the water being circulated?
Charles, T.S. was my shortened version for thermosiphon system. I don’t have the actual temperature because I don’t have any thing to measure it with. Before installing the system he had the water heater set on the highest hot setting on water heater, the next day after I installed it I changed to one of the warm settings but without stopping out there I won’t be able see it or try to get actual temperature. I still think that a good part of this problem is from having this to tight with his double wall system and air chamber in between. I will let you know when I find out the temperatures . Dave
Hi Charles,
Thanks very much for this excellent write up on pumpless hot water recirculation systems. I recently realized that we have (I think) a similar system in our home when I went to change out our 22 year old hot water heater last week. I noticed for the first time in our 5 years being in the home that there was a 1/2″ copper pipe leading to a tee fitting at the hot water heater drain. I thought “what the heck could this be?” and immediately turned off the ball valve up near the ceiling on this line. I noticed the next morning that it took forever for the hot water to warm up but did not connect the incidents until later in the day when talking to my wife on the phone about a similar incident when she was giving the kids a bath. I looked up diagrams online for what someone would use a circuit attached to the drain with a tee for and I discovered the heat recirculation system. I immediately searched near all the fixtures for an extra circuit with a pump attached and have not found a pump (yet). So, I assumed it is a pumpless sytem although here is why I am commenting on your blog: after I swapped the water heater, it took me a few days to figure out how to attach the recirculation tee fitting to my new replacement one since it came with a plastic drain valve with incompatible plumbing to the tee from my old one. Finally, after $45 in crazy sharkbite and garden hose attachment fittings from Ace Hardware, I realized I could just unscrew the plastic fitting and pop my old tee on there (duh). Anyways, I ended up having to add 2 90º fittings and jog the 1/2″ pipe about 1 foot to the right to meet up with the tee. So, I got it all hooked up and it still takes FOREVER for the hot water to warm up in the far away faucets. I thought maybe air in the lines was stopping recirculation but I think I have cleared all that out and we are still having trouble. I studied your blog a bit further and realized that we do not have a check valve in our circuit, atleast not at the hot water heater as far as I can tell. It is just a 1/2″ pipe leading into the top of a tee which connects to the hot water tank via a 3/4″ copper pipe with male threads on both ends. Is it possible that my hot water heater has a check valve that only allows the hot water to flow out instead of in as I need it to for the recirculation to work? Or, is there a pump hidden some where in my circuit that failed when I had the valve shut off for a week? Or, are the 2 90º elbows I added creating some restriction? Or, or, or? I really would appreciate your insight.
Sincerely,
Tim from Michigan
It is possible you have some air in the line between the water heater and the distant point of use—either in the supply or return side. Also the heater may have come with heat trap nipples that are interfering—but I would not think that is a likely cause. You may have to call a plumber that understands these thermosiphon type recirc loops (and keep calling until you find one that does). If you configured the setup the way it was with the previous heater it should work. Did you take out the old plastic drain valve. I did on mine. If you post pictures of your setup I might see something that is a problem.
Hello Charles,
Glad to find your insights. I have been plumbing for over 30 years and from time to time ponder this TS system. I have only put in maybe 3 in that time with no pump. I see on jobs several with pumps and no insulation which of course makes the heater run quit a bit more. I just put one in this week on a new house and it is not working right. Sluggish and reverse flow from bottom of heater when faucet is used. I now realize that I have 2 problems — only 1/2 inch on supply and also I didnt take out the rubber flaps in the supply nipples on top of the heater– I will correct that. I also wonder if there is a difference in where the loop is connected on the far end—I connected with a tee in the kitchen sink cabnet instead of under the floor. This is a one story with 40 foot run— no other fixtures on loup. The return line is sloped down –dropping about one foot in the 40′ back to the heater.
Thanks for any input.
Merry Christmas!
Charlie from Greenville Michigan.
Yes, I think any heat traps would have to be removed. I had ball type in mine that I had to remove. I think you will need some sort of back flow device on the return side but likely not if you have 3/4″ supply all the way to the farthest point. I found that the system would not restart itself with the type of check valve I installed—but more sensitive ones may exist. My T off the loop is below the floor—but I should not think that would matter as long as it is 3/4″ all the way to the T (plumbing in my area is not sloped so the system can be drained like it would be in your area likely). My entire loop is 3/4″ with only the last 101 to 12 feet being 1/2″ to create a pressure difference. My run is close to what yours is. and has no appreciable slope most of the way. So you said there are no other fixtures on the line—does that mean all fixtures in the home come off of this line?
Hello! Nice page!
Some is a bit confusing though. The text doesn’t seem to match the image, and the image seems to be incorrectly drawn? The way you describe the system in the text seems right, but the image seems to show incorrect placement of some items.
The temp sensor and timer valve in the image seem to be connected with a line, indicating their operation is coupled, when it is not. The timer valve is also installed on the main hot water supply to the loop fixtures instead of the return line, which means when it closes, you completely lose all hot water to them, instead of just stopping the loop operation. Correct? This would stop all water flow to the loop fixtures instead of just loop flow.
Although the text is basically correct, the paragraph that mentions when fixtures not part of the loop are operated, those remote fixtures will first ‘get very hot water off the bottom of the tank’, followed by cold water. This might throw readers off, as the water in the bottom of the tank should not be ‘very hot’, the top should. I suggest saying “Without a check valve at the bottom of the tank, hot water to those fixtures will also be supplied out of the bottom of the tank as well, which will turn cooler as cold water refills the bottom of the tank.”
On this same subject, in your comments, you indicate that with the temperature sensing valve for the non-loop fixtures, no check valve is needed. What is the location of the valve that is controlled by this sensor? Where the check valve is? This would not have the same operation as the check valve would. If the in-loop fixtures are operated with the temp valve in the open position, water will flow to those fixtures from both the top and bottom of the tank, and the bottom supply line (normally the loop return line) will soon be cold. I don’t think this system can operate without that check valve.
Also, loop operation is stopped in the original design when a non-loop fixture is operated due to the check valve. As such, if one of these non-loop fixtures is used for an extended period, such as a shower, the water in the in-loop supply line will get cold. Suggestion: Make the non-loop fixtures use the loop return line as their hot water supply. Although technically not ‘in’ the loop, they would increase loop flow when in use rather than stopping it. This should have minimal impact on the starting temperature of those non-loop fixtures as water in the loop return line should be pretty hot by design. (Might require another check valve if the mixing valves will allow reverse-hot water flow)
I modified your image.
What are your thoughts?
Thanks!
Hi John, I am not sure where the confusion is, but I will attempt to answer your questions as best I can.
The drawing does indeed describe my system pretty accurately. I have read back through the text and can find nothing inconsistent between the two—but that is always possible I suppose.
While the line from the temp sensor looks like it is going to the timer it is actually going to the automatic valve that happens to have the timer associated with it. Both are in the correct position to do what they need to do. Basically the only reason for the the valve is to prevent reverse flow when fixtures not on the loop are operated—otherwise the valve is always open and circulating. The timer is merely a way to close the valve so that it does not circulate all night long. Typically the timer closes the valve around 10 pm and opens again at 5 am or so. The thing about a thermosiphon loop is that the entire tank is the same temp (or very close). Of course as soon as you start to use hot water cold water enters the bottom of the tank but is quickly heated by both the heating element and the return of hot water to the bottom of the tank.
I assure you with this approach a check valve is not necessary and finding one that does not stop or slow the weak flow of the thermosiphon is difficult. The system does indeed work and has for several years now without the slightest of hiccups.
If one of the non-loop fixtures is used for a long period of time the thermosiphon loop does not cool off because the valve would be closed and stay closed until until water has stopped being used at the non-loop location. Typically the turn around time is about 10 minutes or so for the valve to open itself again.
With many years of fiddling with this system, I can tell you, check valves need to be avoided as much as possible.
In three years, I have never experienced not having hot water anywhere when I needed it and always immediately present at the remote location.
Hello.
Okay, I understand what is happening now.
You do not need a check valve in your system because your temp valve is closing when needed, preventing reverse loop flow. This would normally be accomplished with the check valve, but your setup does the same thing.
However, according to your diagram, in the position where you have that valve, when it closes you not only shut off the loop, you shut off the remote fixtures from the hot water supply, forcing them to reverse-pull off the bottom of the hot water tank through the return line. Although not ideal, it would still work. As you said, it hasn’t caused any issues.
As for check valves, it’s unlikely anyone will find exactly what’s needed here, but a modified check valve will definitely work.
A check valve using a floating plastic ball is what’s needed, provided it is installed correctly, which is vertically. The ball floats upward, completely opening the path. When high reverse flow is encountered, the ball is forced and held downward, closing off the the flow. More than likely, a check valve of this type has the ball on a spring, but the spring can be removed. The low pressure thermosiphon will never be able to open a spring loaded check valve, or even a flap valve. The floating ball valve is pretty much the only option for a passive system. A check valve can also be made by using a suitable plastic ball, but the plastic would need to be food grade and able to handle 140F. I’ve actually made a few doing this to mix hot and cold water as makeshift inexpensive mixing valves, and also to supply a hot tub using two tankless heaters, while isolating them from each other, etc.
This is the same type of ball valving system used on some older auto transmissions. A floating rubber or plastic ball works in this manner. The reverse is also employed. Gravity holds a small ball-bearing in a hole until flow comes from underneath it, displacing it.
Thanks for your explanation!
I agree the floating valve idea should work, but the only one I could find were heat-trap-nipples. The problem with those is they are actually not designed to “completely” shut the water off, so enough water leaked by to reverse the flow anyway. Not sure why they are designed this way but the ones I tried are. I was hopeful they would be the perfect solution. 🙂
For anyone installing this system. After doing some engineering research, it appears that PTFE (Teflon) or PVDF (Super Teflon) would provide the most durable, chemically inert, and heat-resistant check valve ball for making or modifying your own check valve. Most suppliers seem to only have solid PTFE, which will NOT float, and so can’t be used. Suppliers seem to have hollow PVDF, which does float.
“GlobalBallSales” on eBay stocks ~.750″ (20mm actually) hollow PVDF balls. I bought some today…..paid $5 each shipped.
Happy building!
Purchased watts recirculating system,first operated with timer on 24/7,then changed to 12 on 12 off,tried next to 15 minute on 45 minute off with hot water operating well.finnally turned pump off only relying on sensor valve,which is installed furthest from hot water tank,To my amazement all faucets still provided hot water within 10 to 15 seconds.Called watts to ask why this is possible and they had no answer. They only thing I can think of is sensor valve furthest away needs no additional pressure other the my well pump which cycles between 40 and 60 psi.Your comments would be greatly appreciated.
Roger Karstofsky
Is your water heater below the level of the sink where you are getting continuous hot water?
Yes all sinks are almost a straight shoot up from water heater!
Well then, it is simply thermo-siphoning just like the installation my article is about. You do not really need a pump. It will likely reverse flow if you have any fixtures at the same level as the water heater though.
If that is the case then why is the pump being sold at all?Also sink in lower utility room next to water heater works just fine as well.
Lots of plumber just do not know about these thermo-siphon systems and many many installations it is not possible to get the water heater lower than all the fixtures. I am quite sure if you run your sink next to your water heater for a few minutes you would likely see the flow of hot water reduce in temperature to the upper level—-to the point of becoming cool as it draws water from the bottom of the tank which is cooler because that is where the cold incoming water goes into the tank. This is especially true if it is an electric water heater. With a pump you do not have to think about all this stuff.
Thanks for your help on this matter I will share this with others who are unaware of this technology.
Thanks again Roger Karstofsky
Hi.
In a comment up above, on 17 January, 2017, you wrote “With many years of fiddling with this system, I can tell you, check valves need to be avoided as much as possible.” Yet your illustration shows a check valve right before the hot water return line enters the bottom of the water heater. Can you further explain this possible discrepancy?
Also, as per your article, the hot water supply should be 3/4″ and the hot water return should be 1/2″ in order to create a pressure differential. By this you mean the fact that as water circulation speed increases, water pressure decreases, correct? And thus the water travels from a place of high pressure (3/4″ pipe) to a place of low pressure (1/2″ pipe) which leads back into the bottom of the water heater.
If (that is a big if) the above is correct, does that explain why it is OK to insulate the hot water return? On first thought, I would have thought that by not insulating the return hot water line and letting its water dissipate its heat, that would lead to the water becoming less buoyant/more dense, and thus sinking back to the basement water heater. The benefit is a stronger return current. The cost is increased heating needing to be done by the water heater.
Thoughts?
Thanks for the very interesting article.
The problem with check valves in a thermo-siphon system is they do not open easy enough because the force is much less than when a pump is installed. As the pressure diffentials, by reducing the pipe size you can fully insulate the pipe as desired—otherwise you would have to leave the last 10 feet or so un-insulated in order to create a temperature induced pressure differential I want the pipe fully insulated, hence the approach I have taken. And it works. 🙂
Hi.
Thanks for your reply.
I understand what you are saying about the problem with check valves in a thermo-siphon. However, I see that in your diagram, you have a check valve just before the hot water return pipe terminates at the bottom of the water heater. This is what I have the question about.
Thanks.
Thanks Travis,
The drawing has been through several incarnations and I just had not taken the time to get rid of the check valve. I have to warn you though, without it, you will have to have an electronic valve like the drawing shows. It is not possible, in simple terms, to demonstrate all the nuances of possible differences if other approaches are taken.
Just sharing that in our house the heat traps were causing an issue with a gravity hot water recirc system. Once the heat traps were removed, it worked. The problem is that the heat traps caused enough resistance in the line through the water heater that the path of least resistance for a hot water tap when opened became the back flow of water through the recirc loop. So cold water was back flowing through the recirc loop mixing with some hot water from the water heater’s normal path and no tap ever got hot. It was a bit of a mess and I had to really convince the plumber to take the traps out and now it works just as before.
That is what I have found as well, and of course they are not functional on a water heater with a recirc loop anyway—except perhaps when there is a timer involved. If that is a concern, installation of a loop type heat trap as opposed to a ball type heat trap would likely still work as long as the return pipe is reduced in size the last 10 feet at the water heater.
Hi Charles, I have an electric water heater on the same level as all of my fixtures with a circulation pump and line installed. Recently I have had a pin hole leak in my copper pipe that created quite the mess which I have since repaired. Are there any health concerns with removing the pump and capping off the two lines at the heater? Or just not turning on the pump? I don’t want to tear into the wall and cap the line at the circulation line at the furthest fixture?
In my opinion no water lines, especially on the hot side, should be left with standing water in them. I recommend consulting with a licensed plumber to determine how best to abandon your recirculation loop.
Hi Charles, I have a 30 year old gravity recirc system, water heater in basement, all fixtures on one floor/ranch house, which was fitted to a new water heater last year. Everything was working well until recently: I have instant hot water at all remote sinks and showers, but when someone is showering, the kitchen sink water goes to luke-warm and never reaches even close to maximum temp. The kitchen sink is the closest fixture to the water heater; located almost directly over it. I have noticed that normally the kitchen sink hot water is not instantly hot like in the bathrooms; it takes 20 seconds or so to come up to full temp when no other hot water is running. But the luke warm thing is relatively new. I don’t believe there are any mixing valves or check valves installed in the system. I appreciate any thoughts you might have.
Does the new heater have the re-circulation loop going to the bottom of the tank? If all fixtures are not on the same loop, it is common for locations off the loop to stop or even reverse.
Hi Charles,
Nice site! Can a thermosiphon system operate at all *through* a thermostatic mixing valve? I already have a thermosiphon system operating through a dedicated return from the most remote fixture (a second floor sink), but there are other fixtures between the hot water heater (in the basement) and that remote sink, and it is impractical to get to those lines. I would like to add a mixing valve at the heater, but this would require mixed water to run through the recirculation loop, which I don’t think is possible without a pump. Can you confirm?
Thanks!
it works it through mine
Really? In your diagram, the entire recirculation loop is completed *before* any hot water goes through a mixing valve. All the water in in the loop looks to be up to 140F, in bright red. In my case, the mixing valve would be immediately upon exit of the heater, so the entire loop would be post mixing, with temps up to only 120F. While 120F is more than enough for a thermosiphon effect, do the mixing valves allow actual mixing if there is no flow (demand) from a fixture?
I’m attempting to create a thermosyhon system. I understand that removing the heat trap nipple on the hot side of hot water heater is necessary. But would it still work if I leave the heat trap nipple on cold water supply? I can’t figure out why it would matter.
Thanks in advance
In my experience it will still work if you leave them in on both sides—the balls in them rattle and drive you crazy though because of the low flow.
I have a bathroom in the bonus room above the garage that I am having trouble with water lines freezing. I installed a recirculating pump under the sink, and that solved that problem. However, I still have the cold line to the shower freezing up. Should I install a second (return) line from the shower back to the sink? I don’t remember exactly how it all got plumbed, but I think the hot and cold lines came to the sink and then I teed of to feed the toilet and shower. I know that it would help to wrap the lines as well, but have avoided having to open the wall and dig through the blown-in insulation. I suspect part of the problem is that the water lines are laying on the garage ceiling instead of in the insulation.
Your best option is going to be to get those pipes inside the building insulation envelope. Anything else I am afraid is going to be a bandaid.
Do you think that a thermosiphon loop could work for a 1 car garage floor PEX system?
Behind our garage is the laundry room that has a crawl space.
The crawl space is below grade and the walls are well insulated/foamed.
We heat the crawl space to about 80 degrees F, and the heat radiates thru the floor heating the laundry room.
The garage floor PEX is a single loop with the manifold mounted in this crawl space.
After reading your re-circulating without a pump article I’m intrigued.
And, since I haven’t installed the heat source and circulating pump yet, to heat the garage, I’m thinking that maybe I simply mount a heat exchange/radiator in the crawl space and see if it will thermosiphon.
The output of the crawl space heater can be increased to meat the added demand.
Any thoughts and ideas are welcomed.
I would have to know more about your installation—but it might.
Hello,
Couple thoughts that come to mine, is if you want more pull why not just drop the return to 1/2 and leave it cold last 10-15 feet? Also the bacteria in copper lines will not live if the temps drop down on a line due to cut off or other. Copper is a natural antiseptic that will not allow bacteria to grow. You can buy copper mixers and keep water in them with the lid on for months and still drink it. The hot water tank is not copper so keeping it at 140 is not a bad idea.
Thanks, JOn, and you are correct on the copper. My return is 1/2 for the last 10 feet.
I have a heat exchange radiator that I placed in my Hot air return. It’s about mid height of the electric boiler which is only a few feet away. Would it work to keep hot water in radiator without a pimp. I have a circ pump but find it’s too fast and keeps boiler element on too long.
I don’t have an answer for you on that. You should be aware of newer information that energy code not prohibits the use of thermosiphon systems.
I just had a thermoshiphon system installed with a copper repipe on my house. Single story ranch with pipes in the attic. All 3/4 hot lines and the 1/2″ return line insulated.
The heater is in the garage and about 6′ lower than the pipes in the attic. Bathrooms are 35′ from the heater and every morning the water is hot within 10 seconds of flow. The kitchen is 20′ off the header in the other direction and takes about 20 seconds to see hot water, as it is not in the loop, but hot water is now closer to where it tees off. All I needed was hot water in the bathrooms right away, where it used to take 3 min. to run through my old galvanized piping.
Day 2 and it’s working well.
Thanks for generating the dialogue through this site.
I have been running my own designed thermosyphon system (based on physics theory) in a 2 story split for about 13 years. I started by retro-fitting the entire hot water plumbing system from a mixed series/parallel flow to a continuous series flow, starting at the closest fixture and looping through the lower level to each outlet in progression – ending at the 2nd story showers (all 3/4″ copper). At the end of the line I “T”d in with a 1/2” copper return line, connecting again with a “T”, to the cold inlet line at the top of the tank (since this line communicates with the bottom of the hot water tank via a dip tube). Upon purging the air from the return line I had almost immediate flow and instant hot water at all locations.
The next problem is (of course) back-flow in the return line which causes cold water mixing with the hot at the shower head and eventually would contaminate the entire hot line if an up-stream tap were turned on for any length of time. Aware that this would happen I installed a copper flapper valve from the local hardware store. The force of the convection flow is so minimal that it could not push the valve open, so I had to slant the valve so it was partially open. The system worked at about 80% efficiency but the clinking of the copper valve was annoying with water running.
I found that I could insulate the entire run (including return) and did not lose the circulation. Checking gigajoule gas consumption for one year prior and one year following showed virtually energy neutral.
I experimented with various flow actuated electronic valves and the system was getting so complicated I decided to scrap all electronic valve ideas and revert back to a simple valve that would be hyper-sensitive to changes in flow direction in the return line. After many years of experimentation and searching I finally designed a valve that works perfectly and the system is now 100% reliable.
About a year and a half ago I retrofitted a single story which already had a semi-series flow, starting with ¾” pex and reducing to ½” about half way. I connected a ½” pex return line at the farthest faucet and “T”d into the cold inlet line at the top of the tank as above with my newly designed valve inserted into the return line just before the water tank. After purging air from the return line – voila instant hot everywhere.
The home owner had the hot water tank replaced about 6 months ago and lost the instant hot water. Ironically, they complained that it now took longer than before to achieve hot water. This is what happens when you become accustomed to the luxury of instant hot at the tap – once you have it anything else is unacceptable. The problem turned out to be simply that the new tank had “heat traps” installed in both the inlet and outlet fittings. Upon removal of these interferences instant hot was restored.
PS” I do not buy the theory that flow from the 3/4″ line to the 1/2″ line is creating any pressure differential at the flow rates involved. This would have to invoke Bernoulli’s principle which is not significant until higher flow rates are encountered. I am convinced that the thermal energy in the hot water tank is powering most of the thermosyphon effect. This opens the thought that this system should work on a level playing field with no sloping return required. Open for discussion and experimentation……
PPS: How do we impose this on the entire nation – enormous water saving!!
Doug this is a timely comment.
I too have struggled with the backflow issue and like you found the brass flapper type does not work. I figured out a solution that gets most of the way there by simply using a cold side floating ball type heat trap nipple. These do not completely cut off all the water so the flow still reverses some. My 100% solution was to take the ball out of the nipple and place it between two valves. The top valve has a conical seat that seals completely and the lower valve does not. No idea if the ball ever goes that far or not—and the lower valve may not be necessary at all. As to Mr Bernouli, I have not idea but I am pretty sure whenever you have height differences, temperature differences and size differences (volume), you will also have pressure differences. It works regardless. As to getting everyone on board with this approach is getting more and more difficult as modern energy codes outlaw the use of Thermosiphon systems—no clue why. Perhaps they are assuming no controls—timers etc. I am attaching a picture of where the ball is in my system and the bypass to the right is merely a loop I can test the system without the backflow.
I recently undertook this endeavor and let me just say THANK YOU!!! I bought a new build and it was 50′ of 3/4 pex to the upstairs shower. Took over a minute to get warm water and just forget about using a sink. I was going to buy a recirc pump or run a dedicated 1/2 line up to the 2 showers but I worried about lack of flow and pressure. Then i stumbled upon this thread and changed my entire plan. Now i have instant water at all of my sinks that use to take minutes before. I guestimate I was wasting 5 gallons a day minimal. I insulated all the 3/4 and the return pipe with r4 insulation. I teed into the 3/4 before it heads upstairs and ran the 1/2 pex back to the heater with a 1′ drop per 10 ft. I used a 1/2″ brass swing valve at a the base of the tank at a slight angle. I hear it close when faucets are ran and insulating it really helped.
I’m now finishing up my solar hookup to the bottom element only. Have it set at 125, top at 115. Matched the panels to the ohms of a 120v 2000w element with the help of David Poz (https://www.davidpoz.com/) On my way to free, instant hot water! Thanks
Hello Shawn: (and Charles Buell)
I appreciate your response to “hot water circulation”. I have employed the thermosyphon principle to design and implement a passive convection flow hot water system in my 2 story home since 2007. The initial design utilized a brass swing valve as you have used, but the system did not perform to my satisfaction. It was reliable only about 85% of the time and it did produce the clinking noise you are experiencing with the hot water running. I have experimented with changing the slope angle multiple times with varying degrees of success. Because the diminutive force of convection flow is not strong enough to open the brass valve wide in static mode and the pressure differential did not shut the valve off completely with hot water flowing in the supply line; the efficiency was not satisfactory enough to promote the design either ‘patent wise’ or simply to publish on the internet.
Since that time, I have designed a return line valve that makes the system perform 100% of the time. The efficiency of the loop has also increased. About 2 years ago I installed a similar system in a bungalow with minimal changes in the supply loop, and it produced results beyond my expectations. Note that I do not plumb the return line into the bottom of the tank; I “T” into the inlet line at the top of the tank. You may also note that the ‘heat traps’ installed into the newer hot water tanks will virtually stop the convection flow and they must be removed.
I have also learned that the main driving force behind the passive convection flow is the thermal energy in the hot water tank creating the thermosyphon action’
With proper design and insulation, passive circulatory hot water systems can be every bit as energy efficient as a well insulated water tank with heat traps, and will increase the hot water capacity and promote significant water savings – not to mention the comfort and convenience of having “hot water waiting” when you open the faucet. When I first installed the passive circulatory system in my home, I kept records of the gigajoule gas consumption over a 2 year period = one before and one after installation. The energy consumption was virtually the same for both periods. It is not an exact measure as I did not isolate the gas consumption of the hot water alone, nor did I measure the water consumption; but it gave me a pretty good indication that we were not using any more or any less natural gas energy for that period in the home.
Hope this helps those out there who attempt to employ this innovative design in their own homes. It would be beneficial for the environment if an organization were to come aboard and promote this design nationwide. Think of the water savings alone and the savings in the sewage treatment that could be achieved.
Sincerely, Doug Fyfe (dougfyfe@shaw.ca)
The main reason to go to the bottom of the tank is the last 10-20 feet need to be reduced ins pipe size and if it goes to the top of the tank, if you had reduced it, it goes back to 3/4″ with the dip tube. Some people say that last 10 feet should not be insulated but I have not found that to be true at all, as mine is heavily insulated.
Hello Charles:
Your arguments are all correct, there are many factors controlling the flow of fluid in a pipe. The difference in density due to temperature change has a significant effect on the flow rate if there is a corresponding change in elevation. Hence the cooling effect will definitely have an influence if the return line is a lower temperature than the supply line, again provided there is a change in elevation so gravitational forces can come into play. Your experience, and mine, on insulating the return line to reduce the thermal energy loss just goes to prove there are other forces in play that offset the loss of this density / height change factor (although we have some effect). As far as the last 10 feet being un-insulated, unless it is accompanied by a change in elevation, I don’t think it would have any effect.
As for the reduction in diameter of the return line creating the flow; Bernoulli’s principle outlining the pressure drop in a fluid flowing in a pipe with a reduced cross section area requires that there be a flow in the pipe to create the effect. A familiar example of this principle, utilized in the flow of air, as applied to the venturi of a carburetor or the airfoil of an airplane, requires laminar flow to be significant to create a useable force. In addition, when first starting the hot water flow, pipe diameter has no effect because there is no “Bernoulli Effect” when the fluid is static in the pipe. So some other force must be coming into play because the flow starts spontaneously and almost immediately. Once fluid flow starts there will be some pressure change due to pipe diameter change but at the flow rates we are talking about I don’t believe it is a significant force in the equation. All it takes to stop the convection flow in the system is a heat trap left in the hot water tank, so the forces creating that flow must be fairly small.
That leaves the question: What is driving the circulation in our thermosyphon system? I don’t believe there is a simple answer to that question, but I believe the thermal energy in the hot water storage tank plays an important part – particularly when the tank is being fired. I don’t profess to be an expert in this field, I only can verify that the system works. My knowledge related to the circulation of hot water started with some understanding of the physics involved and some inherent logic, but also empirically with the application of the theory and then observation of the results.
I do know that, besides the convenience factor, there is a huge savings in water and a corresponding saving in sewage treatment. I did a ‘rough’ calculation on the water savings in my 2 story home based on 2 person occupancy, and I came up with about 4 cubic feet of hot water wasted waiting for hot at each station on a given day (that’s based on 15 station cycles – hand washings, showers, etc.). That’s 30 US gallons (113 litres) per day – 11,000 gallons/year (41K litres). A community of 50 thousand single family dwellings could save a potential of over half a billion gallons of water in a year conservatively.
Please take the foregoing conclusions with a “grain of salt” Charles, because it is all based on theory and has not been proven (at least not by me)! I stand to be corrected on any assumptions made in the foregoing arguments.
Charles, this is really interesting, the comments too. Can you look over my post in this plumbing forum where I describe the theory a bit? Do you see gaps or errors I should correct?
bit.ly/recirculating
How come I couldn’t paste my link in here?
Jamie, I could not get into the plumbing site to leave a comment so here you go. Certainly uninsulated flows faster and heavily insulated over its entire length will still work fine and self start fine even though there is very little temperature difference. Temp difference is only one thing that drives the thermosiphon and by decreasing the pipe size for the last 10 feet or so just before it gets back to the tank creates a pressure difference that is enough to keep it going.
Hello and thank you for the useful information provided here.
My thermosiphon (T.S) system is not working as good as it did before replacing one of the water heaters recently. In some places I have to wait long time for hot water. Due to size of the house 2 water heaters are installed together, call it no. 1 and no. 2. cold water goes into no.1 which feeds no 2 supplying the house with hot water. Return line of T.S. goes to the bottom of no 2. no. 2 was replaced couple of weeks ago and then the problems with T.S. system started. I guess this system is not common around here because I didn’t find a plumber familiar with it so instead wasting time troubleshooting I thought maybe I can add a pump. So here is the question: can I add a pump to the existing T.S system? If so where?
You can put the pump most anywhere on the line where you can get power to run it. If it worked before, you may just have air in the line somewhere.
Thank you for your reply offering possible clue. I have opened all the faucets a ran water but did not detect ant air flow. What is the best way find out if there is air in the system and how do I bleed out the air?
Thanks again for sharing your wealth of knowledge.
And when you say I can put the pump anywhere in the system, should it be on the supply line or return line?
I do not think it really matters.
Here is the solution for having to run the hot water for a long time to get hot water to a distant faucet: We have used this for three years and it works beautifully!
Google: Laing EcoCirc under sink hot water pump recirculating system.
About 10 years ago I got information about the thermosyphing system I ran a 3/4 line with a one-way check valve and a manual shutoff from the furthest point of hot water to the drain in my hot water heater. No other add
ons. It worked fine.
I recently replaced my hot water tank. This system wouldn’t work. Replaced the 3/4 line with 1/2. Removed the heat valves in the tank lines.
Did everything I could to remove air from the system. still not working.
First, why? Second, can I put a recirculating pump in the return line from the faucette to the drain of the tank?
It will certainly work with a recirculation pump. Can you send me pictures of the setup at the heater? Full views. Something obviously must be different. It should still work but without being there I may not have an answer for you.
I had my hot water tank replaced and now my thermosiphon doesn’t work. It takes a long time for hot water to reach the furthest fixtures.
Do I need to “prime” the thermosiphon, somehow?
You might have air in the line somewhere—send me pictures of your set up and I will try to help you.
Thanks, Charles. Here is a photo of my setup. Any idea of how to bleed off air (or “prime”?) the system?
In mine, the distant faucet is the highest point in the system so it auto vents opening the faucet. If you have high spots, you would have to install a way to bleed the air out at those spots. Over time the oxygen will be absorbed into the water.
Hi Charles, thanks for your dedication to this post, I’m glad to see that it is still active!
I have a question for you… I am considering installing a Towel Warmer in my upcoming bathroom renovation (floor above the tank and 10′ to the left, lol) and I thinking about having it continuously circulate via a thermosiphon.
I was thinking about a 3/4″ line off the top of the tank rising up to the top section of the towel rack, the exit on the opposite lower corner draining via a 1/2″ line back to the bottom of the tank. I would insulate the 3/4″ supply, but leave the 1/2″ return uninsulated to aid flow, knowing there will be small energy loss.
I don’t believe that I would need any controls on this setup because it is a closed loop, other than an isolation valve for maintenance…. Does this seem to be a sensible idea?
Thanks!
It is an interesting idea and it likely would work. Depending on where the heater is and what else is going on, it could reverse flow I suppose. How about just running it through the towel warmer before the hot water goes anywhere else—use of hot water anywhere in the house would then heat the towel bar. Just a thought. Love to know how your idea works if you try it.
Thanks, Charles. Here is a photo of my setup. Any idea of how to bleed off air (or “prime”?) the system?
In my house there’s a three-quarter inch copper supply line for the upstairs hot water in the bathrooms. At the end of that run I teed off a ½ inch PEX line which runs about 60 feet back to the water heater two floors down in the basement. In the basement, the PEX connects to a shut off valve followed by a spigot followed by a one-way valve and then the whole thing ties into the cold water inlet about 18 inches above the water heater (see picture).
With this configuration I can connect a hose to the spigot and drain the air out of the system which is critical for the convection to work properly. When I first put the return line in I didn’t do a thorough job purging the air and the hot water convection didn’t start working for several weeks. Apparently it took that long for the air to work its way out of the system. Next time I drained the water in the house I did a better job purging the air and the convection loop started up quickly.
I can also close the valve and open the spigot to test that the one-way valve is working properly. The one-way valve prevents cold water from flowing back up the convection line when someone opens up the hot water upstairs. The one-way is a ball type with no spring and is available from Amazon (search on “pex check valve”).
Everyone in the house is thrilled to have warm water coming out immediately the faucets and showers upstairs. In fact the water is downright hot within a few seconds which is more than we really need. What I was after is for the water to not be cold so I’ve experimented with partially closing the valve in the basement to reduce the convection flow.
When I started I was skeptical that convection would be enough to circulate the water but figured I could add a pump if necessary. It appears that no pump is needed and I am very pleased with how the whole system works.
Thanks to everyone who posted here for helping me figure this out.
oh where do i begin, new house, new rheem hybrid heat pump, copper trunk line in conditioned attic with pex supplys to faucets, return is pex in the foundation going to a grundfos auto adapt recirc pump that doesnt sense anything so i keep it at 100% and switch it on for 3 mins and off for 10 mins. meanwhile the rheem works great in the summer in heat pump mode, cools the garage , but in winter it’s too cold in the garage so i switch it into energy saver mode, big mistake. it turns on the element, satisfies the thermostat, then the recirc pump kicks on, temp leaving heater is 125 and returns at 110. when the pump is off there may be siphoning going on, but i can say in the 10 minutes that the pump is off, the water temp in the hot line drops another 10 degrees, then when the recirc pump kicks back on it’s shooing even colder water into the tank, and the water heater turns on again. if i don’t run it at these intervals, it will take 2min 30 seconds to get hot water at the furthest away faucet. where do i start trying to resolve these issues or is the the nature of the beast? single story 3000 sq ft residence. i’ll even pay someone that knows what they’re talking about for their time if we can do a video call, i’ll show you everything you want to know about this new constructed house, thanks to anyone in advance.
The pump will likely need to run continuously during the hours you are likely to use it. As to the cold garage, that is just one of the many things you were not told when you were sold on the idea of one. There are kits that can be added to take the cold exhaust to the exterior, but then you have to think about replacement air to the garage. An ideal solution is going to be difficult.
Now that I see the picture, is that not already being ducted to the exterior? With the door closed, how is it getting make up air?
Please send the timer and sensor information we just talked about to this email. Thank you for sharing your knowledge and experience on this subject!
I see no email address. Just email me at charles@buellinspections.com
thanks for the quick reply! the heater sits in a small closet inside the garage, there’s a 1″ gap under the door for inlet air, and the exhaust is running to a small piece of flex duct into a 12×12 vent into the garage. when it’s 105 outside, it’s 83 in the garage which is good but not superb, garage is 900 sq ft, insulated garage doors, etc. it looks like this grunfos pump is sucking from the return and dumping it into the 2nd return inlet on the heater, is that normal? i tried the auto adapt setting and temperature setting with the sensor and after 2 weeks it doesnt “learn” turns on when it wants and gives me 2 minutes of warm water at best. i have the pump on a timer, 3 mins on, 10 mins off, that keeps the lines primed but boy does this foundation suck the heat out in the winter. i can’t seem to find the right setting on the rheem hybrid, energy saver mode used 12kWh yesterday! heat pump uses 3-5 but makes the garage an ice box during the winter. there is no way to vent the cold air to the outside at this point, my main problem is the recirc loop, i’m a strong believer now that the return loop should be encased with the main trunk, hevily insulated and kept out of the foundation, leave it the attic. i wish the main trunk could dip down from the ceiling and get as close as possible to the faucets, this would really make the water hot after just a few seconds. i’m not sure of any devices that can properly sense the in/out temp and turn on the pump in a given time range. those wifi outlets run out of programming options. i actually had 2 manual timers on the pump. one to turn it on from 6a-midnight, and the other like i said before, on for 3 mins, off for 10. there has to be thermosiphoning going on for the grundfos to detect water movement thinking someone is turning on the tap. I just wish there was a better way to supply a home with a single water heater.
I received your pictures, but unintentionally delete! Could you resend please?
I’m having a problem understanding how to handle the issue for fixtures on the same level as the water heater. All my distance relative fixtures are on a single 3/4″ cooper feeder line. The one basement bathroom draws directly from this line. If the issue is this will cause the thermosiphon to reverse, how and at what location do I resolve this. Can we discuss?
JUST SENT 3 PICTURES TO CHARLES@BUELLINSPECTIONS.COM
You will need to install the auto-shut-off valve in the line that comes from the remote locations when the same level fixtures are used. Of course if you could figure out some sort of backflow valve that actually worked under the low flow conditions of the thermosiphon, you might not need to auto valve—I gave up and went to the assurance of the electronic valve.