Welcome! Here are the website rules, as well as some tips for using this forum.
Need to contact us? Visit https://heatinghelp.com/contact-us/.
Click here to Find a Contractor in your area.

Parallel piping water heaters

Big Will
Big Will Member Posts: 394
I have a building that needs a set of three commercial water heaters replaced. They are currently piped in parallel. Sort of. The center one gets most of the demand because it has the shortest leg. I know how to pipe even numbers in parallel but how do you pipe three in parallel?


  • Ironman
    Ironman Member Posts: 6,856

    Use "reverse return" and you'll have equal flow no matter what the number is.
    Bob Boan
    You can choose to do what you want, but you cannot choose the consequences.
  • EricAune
    EricAune Member Posts: 432
    edited November 2010
    Ironman's got it

    You cannot pipe three water heaters in parallel.  Reverse return is the proper method.

    **I stand corrected, its late, forgot about the pyramid method...sorry. (see Mark's post)
    "If you don't like change, your going to like irrelevance even less"
  • reverse return

    can you explain what that term means please?--nbc
  • Jack
    Jack Member Posts: 1,045
    Looking at the three assign

    1to the unit which get cold water first, 2to second and 3 to last. With a reverse return you say "first on last off" your goal is to have equal pressure drop across each tank. So if you fed from the left, your hw on that first tank would have the hw continue across to the rt picking up each tank as it goes and.

    These kind of things are hard to describe, but what I do is draw the three tanks and assign the 1,2,3 to the cws with 1 being fed first. Assign hw numbers 1,2,3. In each box your total should be 4. From there you can see how your piping should lay-out.
  • icesailor
    icesailor Member Posts: 7,265
    Parallel Heaters:

    I remember and do it like this. Like accounting.

    FIFO is wrong. First In, First Out.

    FILO is right. First In, Last Out.

    FIFO is easy and for lazies

    FICO is harder, not for lazies.

    If both cold and hot supplies to the water heater come in from one side, FIFO is wrong, and easy. FILO is right and harder.

    If the cold comes in on the right and the hot is on the left of the heaters, FILO is easy and FIFO is harder.

    Hot   (Tank-3)     (Tank-2)     (Tank-1)

    ******(Tank_3)     (Tank-2)     (Tank-1)     Cold

    I also "Try" to keep the drop lengths equal to each other with restrictions being equal. Also, be sure that you don't use any street fittings to go into tees. You can get a hard to explain problem in burbulance when the water is flowing horizontally, then turns vertically, then horizontally again and doesn't have time to sort itself out.

    Use full port ball valves on hot and cold of WH's in case you need to tweak the balance Though I never have had to. And if you take one out of service, you will still have equal flow through the water heaters. And it is easier to pipe multiple boilers and water heaters doing it this way.

    Good Luck.

    I hope I got it right.
  • icesailor
    icesailor Member Posts: 7,265
    Oops, FILO:


    FILO is harder, not for lazies.

    FICO is wrong, I don't know what it is.

  • Mark Eatherton
    Mark Eatherton Member Posts: 5,839
    The three methods of paralleling...

    There are actually three ways of paralleling three or more ANY things (storage tanks, softeners, boilers etc).

    For those who say that you can't parallel 3 tanks using anything but parallel REVERSE return, explain that to the commercial storage tank manufacturers, like A.O. Smith and State Industries, because they still show it in their I&O manuals... Not recommended in my book, but still shown none the less.

    See page 18 of


    for what I refer to as Pyramidic Parallel. Get the picture? Lots of soldered joint, and potential leaks, and if the centers are not dead center, you can expect imbalance in flows. It also creates problems as it pertains to simple tank replacement. If piped as shown, you will have to shut the whole shebang down in order to replace one battery.

    Parallel reverse return, if piped properly, allows for the replacement of a unit, with the other units staying on line. THe key is keeping all piping, unions and isolation valves high enough to allow for isolation and removal, which is a lot simpler than trying to do it on a pyramid... All three heaters see the exact same flow, assuming that the pressure drop of the three units is identical.

    The easiest way to think of parallel reverse return piping is to think of a 3 rung (or however many units you want) ladder. The left hand side of the ladder represents the incoming cold water supply, and the right hand side of the ladder represents the hot water outlet. The cold and the hot flow in the same direction as it is flowing through the batteries. The lower right hand side of the ladder is cut off. Now, take the ladder and turn it sideways (horizontal). The rungs represent the tanks, or batteries. After you have gone through all the tanks, then begins the REVERSE return. Here is a picture.


    The third, most common attempt, and misguided effort, is to pipe them in a parallel DIRECT return method, which means that the cold and hot are flowing in opposite directions (left and right hand sides of the ladder). In this case, my ex brother in law principal kicks in, because water is wet, lazy and stupid, and will follow the path of least resistance, which means tanks 1 and 2 will see the most flow, and three will see little flow, hence little contribution. In order to achieve balance, you must balance the flow through all three tanks with balancing valves. Half hearted at best.

    Here is a sample drawing of direct return piping.


    Pipe them parallel reverse return and you and your customer will not regret it...


    It's not so much a case of "You got what you paid for", as it is a matter of "You DIDN'T get what you DIDN'T pay for, and you're NOT going to get what you thought you were in the way of comfort". Borrowed from Heatboy.
  • Big Will
    Big Will Member Posts: 394
    Sounds easy

    I should have thought of that. But I didn't so thanks all. I was trying in my head to make a pyramid design work and could not come up with a plan. Over thinking it is my excuse.  Thanks
  • croydoncorgi
    croydoncorgi Member Posts: 83
    I think I (strongly) disagree!

    The O/P referred to 'water heaters' - so the reverse return arrangement to even-out the pipe-runs and ensure that all get equal load is fair enough - but only if the system design says they ALL need to run, all the time, to meet the demand for hot water.  (I'm assuming we're all talking about the same thing - direct water-heaters.)  This sort of load is most unlikely in practice: it will usually vary a lot, so one machine can easily satisfy demand (say) 65% of the time, two up to 90% and only need the third when one is broken or there's two whole football teams showering together!  To make this work in practice, you need a bit more complexity in the controls, to hold off machines from running until they're actually needed and to make sure that only hot water is delivered.  For Big Hitter systems in UK, the design could include a big buffer tank (maybe 600 litres) of hot water with however many heater units with recirculating pumps connected to it, plus electronics to turn the pumps on/off according to demanded hot water flow and/or buffer tank temperature.  Eg, Rinnai heaters can be fitted this way and only fire up when they sense water flow.  They can be fed with recirculated already-hot water and heat it to a given output setpoint.

    A proposed solution showed a cascade of condensing boilers.  From my own experience, I can't see how this would work properly (or at all!).  The ESSENCE of success for condensing machines is to minimise the Return temperature and to ensure that the Flow / Return delta-T is near the max for the boiler (usually around 20 C degrees) without exceeding it.  This will give maximum efficiency, because the heat exchanger needs to be as far as possible BELOW 56 C degrees for condensing to occur (Dew Point).  The system pipework is (it appears) the One-Pipe design, so to make the last radiator (fan-coil unit, ..?) on the loop work properly, Flow temperature would need to be mantained all the way round the loop (by correct pump selection, flow-adjusters, etc.)  This inevitably means that the Flow / Return delta-T will be small and the boiler(s) won't condense very often.  Other issues with a boiler cascade are sequencing, fault situations and load-control.   It makes no sense to run ALL the boilers ALL the time, even if they are independently cycling on and off on their internal thermostats.  It would be usual to over-equip so that (say) 3 boilers of a 4-cascade could meet 100% of the design max demand. It also makes no sense to leave a boiler 'in the line' with its shunt-pump running if it's not capable of firing.  Also, the boiler firing sequence needs to be rotated, to ensure equal use.  All this needs a proper electronic sequencer that responds to varying loads, rotates the lead/lags and deals properly with faults.  When you've got that, you plumb all the boilers into a common low-loss header ('hydraulic separator').

    And, unfortunately, none of this works optimally with a One-Pipe system.  You need Two-Pipe, where all the heat users (radiators, etc) are plumbed in parallel across a Flow and Return pair, not a single pipe.

    Sorry if this all sounds a bit dictatorial.  Or maybe I'm saying things that everyone knows perfectly well....  It's just that what was actually said in this particular thread make no operational sense to me.
  • icesailor
    icesailor Member Posts: 7,265
    I think I (strongly) disagree with you:

    First off, these are domestic potable water heaters. But it doesn't matter because I just so happen to have an account with three Weil/McLain WGO-6's tied together as a proper reverse return, indoor/outdoor reset and sequencing burners. One, two or three burners may run on any given day depending on temperature, one or more will run. They, the burners take weekly turns on who is in charge. In 10 years, since I have been paying attention, I have never seen a difference in temperature between the boilers. regardless if they are running or not.

    As far as potable water heaters are concerned, it seems to me in observation and in practice that what keeps them balanced is the temperature of the water. That the thermal weight of the water makes the lightest and hotest rise out of the hottest heater and they will all balance out. It is done by gravity. Delta T's whatever. All I know is that when I fugured out how you could suck off three Roth tanks with a two pipe system, the supply goes into all three tanks and the return only goes in one tank, Vio-lah!! Or how ever one would write it. When the tank with the return starts to get higher than the other two, it makes it suck harder off the other two, or it is easier to suck off the one with the return. It will always balance out because if one gets higher, the suction will be easier on the fuller tank and not until they all become even, will the suction change. You can put a tigerloop on as suction and no return and they will all draw down together.

    I have done two water heaters as a balanced system and the tanks were always hot at the same point on both tanks. I once re-piped a mistake someone made where they used a 60 Gal. electric and a 40 gal. under stairs. That's why the shorter one. It was in series. The problem was that one had to run out first before the second one could come on. So, with no other alternative, I piped them as "balanced with equal pressure" on both heaters, the hot water flowed out of both and the smaller tank would start as soon as the smaller tank cooled.

    I think that Mark Etherton was describing what happens when you do three in series. It doesn't work. You need to do it as a reverse return.

    You may not see that it works or not if you install it. I do often because I get called when something doesnt work and I have to figure it out.

    Sadly, this has become incoherent because I have had numerous distractions while writing this. Sorry.
  • croydoncorgi
    croydoncorgi Member Posts: 83
    Still confused

    If you've got a sequencer ( or the heaters include some intelligence and talk to each other) then there's no problem - they WILL load-balance, sequence, fault-manage, etc.

    But that's not what the original posting said.

    And the illustration using heating boilers on a one-pipe cascade STILL makes no sense to me without the inclusion of a sequencer (and is mostly irrelevant to a water-heating applicatoin anyway).  One-pipe mostly won't work with high efficiency boilers where you need to maximise delta-T to around 20C degrees - or, if it can be made to work - how?.
  • Mark Eatherton
    Mark Eatherton Member Posts: 5,839
    Blimey mate...

    What's a bloke to do?

    I think that what we have here is an international failure to communicate.

    Remember, our cylinders over here think they are tanks. Not Panzer's mind you, just plain old steel glass lined tanks with individual simple poppet on/off controls, a combustion chamber on the bottom and a hole/heat exchanger that runs straight through the middle of the stored water, with no damper to control the flow of heat up the flue pipe, with no ability to be sequenced, other than their set point settings.

    Wahtsa one pipe cascade? Sounds like a venturi system full of dish soap to me?? Is that our primary/secondary piping network that you're referring to?

    If I were to employ our state of the art technology, say 3 Knight boilers, tied to one large reverse indirect DHW heating system, I would use the 2 wire communication bus and they WOULD talk to each other. One would be the boss, and the others would be lower level employees. And listening to them modulate, and trade places as lead, lag etc would sound like a symphony.

    Generally speaking, the original poster sounded to me as if he was trying to fill a large fixed volume tub, like a 1 man, 5 woman hot tub...

    Or, he had numerous drive thru car washes disguised as showers, and when they are all on, you WILL run out of hot water, and at that point you WILL need the fire capacity of all three heaters on line. But I am guessing :-)

    Putting heaters in series guarantees that the heaters number 2 and 3 will rarely fire, which depending upon your point of view, can be good (you'll always have a like new water heater laying around), but heater number 1 will cary most of the load. Problem is, heaters 2 and 3 are losing heat up their flue pipes when they are NOT on, costing a lot in the form of wasted energy. Putting heaters in series also increases the pressure drop by 3 X. Which can be problematic in low pressure considerations.

    If you didn't know, our gas fired tanks are a joke, and energy hogs at that. For the 23 hours a day that they are not actually being used, they are losing heat up the flue pipe. They will fire up quite often without being used just to recover their own standby losses.

    Energy hogs I'm telling you. Fortunately, I haven't installed on in about 15 years.

    What's your cost of energy per therm (100,000 btu's) on that side of the pond? I think natural gas in our area is around $0.50/therm....

    It's not so much a case of "You got what you paid for", as it is a matter of "You DIDN'T get what you DIDN'T pay for, and you're NOT going to get what you thought you were in the way of comfort". Borrowed from Heatboy.
  • croydoncorgi
    croydoncorgi Member Posts: 83
    Communication restored!

    If the 'water heaters' are the big tank-like things (with hole up the middle) I now understand completely.  Inefficiency and heat up the flue, ALL the time - who cares!  (I'll check out gas prices and convert to $ per therm (we buy in kilowatts, megajoules or cubic metres!)

    The confusion about one-pipe systems was due to the diagram you yourself included (http://www.esmagazine.com/ES/2004/06/Files/Images/105630.gif)  Still can't see how the arrangement shown might work.....the Return temp would be too high to make the condensing boilers happy.
  • Mark Eatherton
    Mark Eatherton Member Posts: 5,839
    I see said the blind man....

    as he walked into the wall :-)

    I was trying to show parallel reverse return through the boilers. Those boilers are piped into a secondary distribution system, which is most probably a 2 pipe system (Supply and return, not FLOW, or FEED and return ;-))

    I had an employee once who used to call them flow and return, and would mark his pipes with either an F or an R. One day, we were working together, and I noticed that someone had smudged out the right hand side of the R, so it looked like an F. I asked him to kindly begin referring to the piping as Supply and Return, because the chances of someone mistaking an S for an R were a lot less...

    Let's see, there are 3.413 btu's per watt, so 100,000 divided by 3.413 = 29,300 watts. Divide that by 1000 = 29 KW.

    So, $,50 divided by 29 = $2.00 per KW for our gas. I'm guessing your cost per KW is going to be higher.

    Seems funny that we use the BRITISH thermal unit as a means of energy measure, and your side of the pond uses something metric :-)

    As for the return temps, that is dictated by the system being served, but if it is all a low temp system, then all three boilers would be seeing the same low entering temperature, hence keeping all 3 in condensing mode. How else could it be piped? What other means of piping would be better for providing the lowest water temps to the appliances?

    Curious minds want to know.

    It's not so much a case of "You got what you paid for", as it is a matter of "You DIDN'T get what you DIDN'T pay for, and you're NOT going to get what you thought you were in the way of comfort". Borrowed from Heatboy.
This discussion has been closed.